Diachrony and Dialects: Grammatical Change in the Dialects of Italy (Oxford Studies in Diachronic and Historical Linguistics) [1 ed.] 9780198701781, 0198701780

Citation preview

Diachrony and Dialects

O X F O R D S T U D I E S I N D I A C H R O N I C A ND H I S T O R I C A L L I N G U I S T I CS general editors Adam Ledgeway and Ian Roberts, University of Cambridge a d v i s o ry e d i t o r s Cynthia Allen, Australian National University; Ricardo Bermúdez-Otero, University of Manchester; Theresa Biberauer, University of Cambridge; Charlotte Galves, University of Campinas; Geoff Horrocks, University of Cambridge; Paul Kiparsky, Stanford University; Anthony Kroch, University of Pennsylvania; David Lightfoot, Georgetown University; Giuseppe Longobardi, University of York; David Willis, University of Cambridge r e c e n t ly p u b l i s h e d i n t h e s e r i e s 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, Chris 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

Diachrony and Dialects Grammatical Change in the Dialects of Italy

Edited by PAO LA BENIN CÀ ADAM LEDGEWAY N I G E L VI N C E N T

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 Paola Benincà, Adam Ledgeway, and Nigel Vincent 2014 # the chapters their several authors 2014 First Edition published in 2014 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: 2013949645 ISBN 978–0–19–870178–1 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.

‘promuovere l’esplorazione scientifica dei dialetti italiani ancora superstiti, sia col raccoglierne materiali genuini e nuovi, sia col dar mano ad illustrarli’ (Graziadio Isaia Ascoli, ‘Proemio’ all’Archivio Glottologico Italiano, 1873) ‘in molti dialettofoni persiste ancora un sentimento di vergogna o imbarazzo, nonostante i tentativi da parte dei linguisti negli ultimi decenni di ribaltare questa valutazione’ (Mair Parry, Parluma ’D Còiri 2005)

For Mair, friend, colleague, and dialectologist per eccellenza, in recognition of her efforts to put dialects in their rightful place within the Italian and the general linguistic tradition

Contents Series preface Preface: For Mair Parry Anna Laura Lepschy and Giulio Lepschy List of abbreviations Notes on contributors 1 Similarity and diversity in the evolution of Italo-Romance morphosyntax Nigel Vincent 1.1 1.2 1.3 1.4 1.5 1.6

Introduction Dialects Diachrony Sources of data History and theory: analysis and synthesis in the verbal system The chapters

xi xii xiv xx 1 1 2 3 5 7 19

Part I. Verbal Structures 2 The development of the southern subjunctive: morphological loss and syntactic gain Adam Ledgeway and Alessandra Lombardi 2.1 Introduction 2.2 Morphological marking of the indicative/subjunctive opposition in the south 2.3 Syntactic marking of the indicative/subjunctive opposition in the south 2.4 Summary and conclusions 3 Perfective auxiliation in Italo-Romance: the complementarity of historical and modern cross-dialectal evidence Michele Loporcaro 3.1 Introduction 3.2 The basics: perfective auxiliation and the unaccusative hypothesis 3.3 Diachrony: intersections between modern dialect comparison and historical data 3.4 Bringing the historical and modern cross-dialectal evidence together 3.5 Conclusion

25 25 28 34 44 48 48 49 56 63 68

viii

Contents

4 Passive and impersonal reflexives in the Italian dialects: synchronic and diachronic aspects Michela Cennamo 4.1 4.2 4.3 4.4 4.5 4.6

Introduction Passive and impersonal reflexives in Standard Italian Passive and impersonal reflexives in the Italian dialects Some diachronic data Converging constraints on impersonal si/se Some unsolved issues: the status of the reflexive morphemes si and ci 4.7 Conclusions 5 On the personal infinitive in Sicilian Delia Bentley 5.1 5.2 5.3 5.4

Introduction The personal infinitive in early Sicilian The personal infinitive in modern Sicilian Conclusion

6 Glimpsing the future: some rare anomalies in the history of the Italo-Romance and Gallo-Romance future and conditional stem, and what they suggest about paradigm structure Martin Maiden and John Charles Smith 6.1 6.2 6.3 6.4

The facts Semantic or phonological causation? An accidental association? The mechanism of the change

7 Person endings in the old Italian verb system Laura Vanelli 7.1 Introduction 7.2 Verb endings in old Italian 7.3 Concluding remarks

71 71 72 76 83 91 92 95 96 96 97 109 114

116 116 125 126 128 131 131 133 148

Part II. (Pro)nominal Structures 8 The evolution of Italo-Romance clitic clusters: prosodic restructuring and morphological opacity Diego Pescarini 8.1 Introduction 8.2 The emergence of the mirror order 8.3 Separability

155 155 156 161

Contents 8.4 8.5 8.6 8.7 8.8

Allomorphy Suppletion Root incorporation Italian Conclusions

9 Subject clitics and macroparameters Ian Roberts 9.1 9.2 9.3 9.4 9.5 9.6 9.7

Introduction Micro- and macroparametric variation Macroparameters and markedness Parametric hierarchies Northern Italian subject clitics and the null-argument hierarchy The difference between subject and objects Conclusion

10 Sicilian 1st and 2nd person oblique tonic pronouns: a historical and comparative examination Rosanna Sornicola 10.1 Conservation and innovation in Romance pronominal systems 10.2 1sg and 2sg oblique tonic pronouns in early and modern Sicilian: Romance equivalents and etymological hypotheses 10.3 Textual evidence from early literary Sicilian 10.4 Textual data, areal data, and diachronic developments 10.5 Conclusions 11 Patterns of variation and diachronic change in Piedmontese object clitic syntax Christina Tortora 11.1 11.2 11.3 11.4

Introduction Piedmontese object clitic syntax Relatable facts in Fassano and Spanish Diachronic change in Piedmontese object clitic syntax revisited: the Functional Hierarchy Hypothesis 11.5 Conclusions 12 Gender assignment and pluralization in Italian and the Veneto John B. Trumper 12.1 Recent work on the theme 12.2 Pluralization, gender assignment, and shift in the Veneto (Neo-Venetian) 12.3 Conclusion

ix 163 167 171 173 175 177 177 181 189 191 195 199 200 202 202 205 210 215 216 218 218 219 225 232 239 241 241 249 256

x

Contents

13 Kind-defining relative clauses in the diachrony of Italian Paola Benincà and Guglielmo Cinque 13.1 13.2 13.3 13.4 13.5 13.6 13.7 13.8

Introduction Kind-defining relatives Properties of post-copular kind-defining relatives Deriving the properties of kind-defining relatives Post- and pre-copular position (canonical and inverse predication) Raising and matching derivations of kind-defining relatives Post-copular relatives and agreement Conclusions and speculations

14 Synchronic and diachronic clues on the internal structure of ‘where’ in Italo-Romance Nicola Munaro and Cecilia Poletto 14.1 14.2 14.3 14.4 14.5 14.6

Introduction Theoretical background The formatives of ‘where’ in northern Italian dialects Matching the structure of locative PPs Ubiquitous ‘where’: relative, interrogative, and prepositional forms Concluding remarks

References Index

257 257 262 265 270 273 274 275 277 279 279 281 283 289 292 299 301 333

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 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

Preface For Mair Parry The following introduction by Laura and Giulio Lepschy is opened by Laura, and followed up by Giulio. My first encounter with Mair was at the end of the 1960s, when Christina Roaf asked me to replace her for a term at Somerville while she took a sabbatical. Mair had already got a first degree in French with accessory Italian (Honours Class I) at Aberystwyth and was now embarking on a second degree in Oxford, specializing in Italian, where her results were to be equally outstanding. I was very glad to be returning to Somerville, which was my old college, and to be holding the kind of one-to-one tutorial which I had enjoyed with Christina Roaf. I went up every Thursday and had five students taking different topics. As I had majored in French, I was pleased that Mair and I had the same background, and also that she shared my interests in the overlap between linguistic and literary problems, which was one of the reasons why I had always enjoyed the questione della lingua, and I hoped that our tastes would also coincide here. After the Oxford tutorials we kept in touch and several years later, when Mair was teaching in Aberystwyth, I was invited there, and at the same time to Bangor, as External Examiner in Italian. I was very happy to see Mair in her Welsh context and to get to know her family, her husband, David, and her two small children, Rhys and Catrin. Staying with them was extremely enjoyable and linguistically fascinating, with Welsh being normally used within the family. I was also very intrigued by Mair’s interest in Piedmontese, a language which I had encountered as a child in Turin but never really learnt to speak. My experience both in Aberystwyth and in Bangor, where I was looked after by Elizabeth Schächter, was very instructive, as I came to understand how Welsh could be respected as a first language, even when non-Welsh speakers were examiners. On one occasion for instance, a student had translated Italian into Welsh and this had to be further translated into English for me. The friendship with Mair and her family was established; we saw each other regularly over the years and shared friends in the UK and in Italy. When Mair moved to Bristol University I felt there was another special link, as that was where my academic career had started. I had been Junior Fellow in Italian before the Italian Department existed, but when Mair joined the Department it was a thriving unit, which I enjoyed going to visit on the occasion of conferences etc., as I had spent a

Preface

xiii

happy two years there and was pleased to see how Italian had developed and how Mair had added to its distinction. So, in Oxford, Aberystwyth, and Bristol we had plenty of experiences to share. And now over to Giulio. It is a pleasure and an honour to take part in this celebration of Mair. I met her when I was teaching at Reading, in late 1968, and Laura introduced her to me as one of her best students. She took her place in that extraordinary group of British scholars who cultivated, in a way I found most congenial, an interest both in modern linguistic theory and in more traditional historical and philological studies, applied, in particular, to Italian and the Italian dialects. It was satisfying to follow Mair’s intellectual and academic progress, from her 1985 Ph.D to her research on the clitics and on negation, to her participation in the SAVI project with its theoretically innovative and stimulating implications. Her work had a formidable impact (to use a word which for a person of my age is still intellectually respectable), in terms both of hard scholarship (as in the 1997 Maiden and Parry Dialects of Italy volume) and of enthusiastic organization of the memorable Bristol Italian Dialectology meetings. I should like to end on a personal note, which I have mentioned on other occasions (including in my preface of Mair’s description of the dialect of Cairo Montenotte, Parluma ’D Còiri). In 1992 I gave a paper at a meeting in honour of Glanville Price, organized at Gregynog by Mair and other colleagues. On that occasion I was unwise enough to confess to a sense of unease at the thought of a great European novel written not in Italian by a native speaker of Milanese such as Manzoni, but directly in Milanese. I should have put my comment differently, and I remember that Mair asked me in a cordial (but sharp) tone whether I would have been equally uneasy if this great European novel had been written in Welsh. On the surface, it was easy to answer that my problem is that I do not know Welsh, while Mair does. But on a deeper level the question is more complicated, and I am still brooding over it. For which (as for many other things) I am grateful to Mair. Laura and Giulio Lepschy

List of abbreviations The Opera del vocabolario italiano (OVI) database contains details of most texts cited in this collection. It is available at: http://artfl-project.uchicago.edu/content/ovi-bibliography. * ? = # √ 1 2 3 I/II/III/IV A abl Abr. acc/acc Adv agr Agr(S/O)(P) AIS ALF Alt.2 an. AO ARTESIA ASIt Asp aux/Aux BA Bar. BCC BG BO Bus. C CA Cal.

(i) ungrammatical form; (ii) unattested form dubious form or usage cliticized to internal morphosyntactic boundary root (i) 1st person; (ii) subject relation (i) 2nd person; (ii) object relation (i) 3rd person; (ii) indirect object relation first/second/third/fourth conjugations subject of a transitive clause ablative Abruzzese accusative adverb(ial) agreement features (subject/object) agreement (phrase) Karl Jaberg and Jacob Jud (1928–40), Sprach- und Sachatlas Italiens und der Südschweiz Jules Gilliéron and Edmond Edmont (1902–10), Atlas linguistique de la France one subvariety of Altamurano (central Pugliese dialect of Altamura, upper southern Italy) animate province of Aosta/autonomous region of Valle d’Aosta Archivio testuale del siciliano antico, ed. Mario Pagano, Opera del vocabolario italiano Atlante Sintattico d’Italia (http://asit.maldura.unipd.it/) Aspect (head position) auxiliary province of Bari Barese, dialect of central Apulia Borer–Chomsky Conjecture province of Bergamo province of Bolzano Bussoleno (western Piedmontese dialect, province of Turin) complementizer (position) province of Cagliari Calabrian

List of abbreviations Cat. CB CF CH ChioseTes Cho CI cl CL CM Cmb. Cmp. Cms. CN coll. comp cond conjug. Cos. CP Cpd. CR CS Ctz. dat/dat Dec. def. DEI DialSGreg dim dir. DisjP DO D(P) E EN Eng. EPP ExistP f F FE FHH

xv

Catalan province of Campobasso clause-final province of Chieti Giovanni Boccaccio, Teseida delle nozze d’Emilia: Chiose chômeur clause-initial clitic province of Caltanissetta clause-medial Colombian (Spanish) Campidanese Comasco (western Lombard dialect, northern Italy) province of Cuneo colloquial complementizer conditional conjugation Cosentino (northern Calabrian dialect of Cosenza, extreme south of Italy) Complementizer phrase Campidanese (dialects of southern Sardinian) province of Cremona province of Cosenza Catanzarese (central Calabrian dialect of Catanzaro, extreme south of Italy) dative Giovanni Boccaccio, Decameron definite Carlo Battisti and Giovanni Alessio (1951), Dizionario etimologico italiano (5 vols) Domenico Cavalca (?), Dialogo de Sam Gregorio composito in vorgà diminutive direct disjunction phrase direct object determiner (phrase) auxiliary ‘be’ (Latin esse) province of Enna English Extended Projection Principle existential phrase feminine feature (i) province of Ferrara; Feature Economy Functional Hierarchy Hypothesis

xvi

List of abbreviations

Fin(P) Flo. Foc(P) FP Fr. FR Frl. fut G GDFC GE Gen. ger Ger. Gk. H HAS HR IFoc(P) IG IM imp impf imprs ind indef. indir. inf Inf. Infl IP IS It. LAS Lat. LF Lgd. LibrAmm LibrCred LibroDare Lig. Lmb. loc Log.

finiteness (phrase) Florentine focus (phrase) functional projection French province of Frosinone Friulian (dialects spoken in northeast Italy) future gender (features) Glossaire du parler français au Canada province of Genoa Genoese gerund German Greek auxiliary ‘have’ (Latin habere) higher adverb space Croatia informational focus (phrase) Input Generalization province of Imola imperative imperfect impersonal indicative indefinite indirect infinitive Dante Alighieri, Inferno, Divina Commedia Inflection position (head of IP) inflection phrase province of Isernia Italian lower adverb space Latin Logical Form Lengadocien (Occitan dialects of Languedoc, southern France) Libro d’amministrazione dell’eredità di Baldovino Iacopi Riccomanni Libricciolo di crediti di Bene Bencivenni Libro del dare e dell’avere di Noffo e Vese figli di Dego Genovesi Ligurian Lombard locative Logudorese (dialects of northwest Sardinia)

List of abbreviations m Mac. MI mid. Mil. Min. Fel. mod. mod. It. mrk MS n N Nap. neg NIDs NO nom NU Num Nuo. O OBJ Occ. ocl/OCL opt OR OT OVI p P PA Pad. Par. pass PassGen Pav. PC PD PE PF Pie. pl PLD P(P)

masculine Maceratese (dialect of Marche, central Italy) province of Milan middle Milanese Minucius Felix modern modern Italian marker province of Massa and Carrara neuter (i) noun; (ii) nominal head; (iii) number (features) Neapolitan negator northern Italian dialects province of Novara nominative province of Nuoro number Nuorese (dialect of northeast Sardinia) (i) old; (ii) object object Occitan object clitic optative province of Oristano province of Olbia-Tempio Opera del vocabolario italiano person (i) predicate; (ii) parameter province of Palermo Paduan (Venetan dialect of Padua, northeast Italy) Dante Alighieri, Paradiso, Divina Commedia passive La Passione Pavese (southwestern Lombard dialect) province of Piacenza province of Padua primary (sub-)ending phonological form Piedmontese plural primary linguistic data preposition(al phrase)

xvii

xviii

List of abbreviations

PR pret prf pro prog prs PrW pst Pt. ptp/PtP PV PZ RæR. RC refl./refl reg. retr. REW RI Ro. RO S Sal. sbjv scl/SCL SE sg Sic. SMT SO Sp. spec. Spec Srd. SS st. SUBJ superl SV Swi. t TA Tardif. Tesoro volg.

province of Parma preterite perfect null pronominal argument progressive present prosodic word past Portuguese participle province of Pavia province of Potenza Ræto-Romance province of Reggio Calabria reflexive regional retroherent Wilhelm Meyer-Lübke (1935), Romanisches etymologisches Wörterbuch province of Rieti Romanian province of Rovigo subject of an intransitive clause Salentino (dialect of southern Apulia) subjunctive subject clitic secondary (sub-)ending singular Sicilian Strong Minimalist Thesis province of Sondrio Spanish specific specifier position Sardinian province of Sassari standard subject superlative province of Savona Switzerland trace (of moved element) province of Taranto Tardif, Jules (1866). Monuments historiques. Cartons des rois. Paris: J. CRYCR. Giamboni, Bono, Il tesoro di Brunetto Latini volgarizzato da Bono Giamboni

List of abbreviations ThLL

xix

Thesaurus linguae latinae, editus auctoritate et consilio academiarum quinque germanicarum: Berolinensis, Gottingensis, Lipsiensis, Monacensis, Vindobonensis, Lipsiae, in aedibus Teubneri, 1900–. TLIO Tesoro della lingua italiana delle origini, a cura dell’Opera del vocabolario italiano, sotto la direzione di Pietro Beltrami (www.ovi.cnr.it) TN province of Trento TO province of Torin top topic (marker) Top(P) topic (phrase) Tor. Torinese T(P) tense (phrase) Trad. Frising. Traditiones Frisingenses trans. transitive Trist. Ricc. Tristano Riccardiano Tsc. Tuscan TV thematic vowel UD province of Udine UG Universal Grammar unacc. unaccusative unerg. unergative v(P) light verb (phrase) V (i) verb; (ii) vowel Ven. Venetan VI province of Vicenza Vn Dante Alighieri, Vita nuova Vnz. Venetian VP (lexical) verb phrase vr province of Verona V2 verb second syntax Zag.2 one subvariety of Zagarolese (Laziale dialect, province of Rome)

Notes on contributors Paola Beninca` is Professor Emerita of Linguistics at the University of Padua, and before that was Professor of Linguistics at the University of Milan, and researcher at the Centre for Italian Dialectology of the National Research Council in Padua. Her research interests include synchronic and diachronic Romance syntax and morphology, and the history of linguistics. Recent works include chapters of the Grammatica dell’italiano antico (ed. G. Salvi and L. Renzi, Il Mulino, 2010) and the collection of papers Mapping the Left Periphery (The Cartography of Syntactic Structures, vol. 5), which she co-edited with Nicola Munaro (OUP, 2010). She has coordinated the online database ASIt (‘Atlas of Italian Dialect Syntax’). Delia Bentley is Senior Lecturer in Italian Language and Linguistics at the University of Manchester. She is the author of Split Intransitivity in Italian (Mouton de Gruyter, 2006). She has also published articles on modality, subjecthood, and auxiliary distribution in Sicilian, Sardinian, and Italian. From 2010 to 2013 she directed an AHRC-funded research project investigating existential constructions across the whole range of Italo-Romance dialects, results of which are reported in Existential Constructions in Cross-Linguistics Perspective, a special Issue of the Italian Journal of Linguistics, 2013, co-edited with Francesco Ciconte and Silvio Cruschina. Michela Cennamo is Professor of Linguistics at the University of Naples ‘Federico II’. She is the author of numerous articles on the grammar and history of argument structure, auxiliary selection, passives, impersonals, and reflexives in Italian and in the Italian dialects. In her research she has paid particular attention to the Latin evidence for the early stages of these constructions and their subsequent history in Italo-Romance. Her articles have appeared in books and journals such as Archivio glottologico italiano, Transactions of the Philological Society, and Zeitschrift für romanische Philologie. Guglielmo Cinque is Professor of General Linguistics at the University of Venice, with interests in syntactic theory, typological linguistics, and Romance linguistics. He has been involved in teaching and administration at the University of Venice and has taught at various universities abroad (Paris VIII, Vienna, Geneva, Harvard, UCLA, ENS Paris, and Brussels). He is the co-editor of The Oxford Handbook of Comparative Syntax (OUP, 2005, with Richard Kayne), and author of Adverbs and Functional Heads (OUP, 1999). He has also edited several collections of papers on the ‘cartographic’ approach to syntax, of which he is one of the principal exponents.

Notes on contributors

xxi

Adam Ledgeway is Professor of Italian and Romance Linguistics at the University of Cambridge, and a Fellow of Downing College, Cambridge. His research interests include Italian dialectology, the comparative history and morphosyntax of the Romance languages, Latin, syntactic theory, and linguistic change. His recent publications include Grammatica diacronica del napoletano (Niemeyer 2009); Syntactic Variation: The Dialects of Italy (CUP, 2010, co-edited with Roberta D’Alessandro and Ian Roberts); The Cambridge History of the Romance Languages, vol. 1: Structures; vol. 2: Contexts (CUP, 2011, 2013, co-edited with Martin Maiden and J. C. Smith); and From Latin to Romance: Morphosyntactic Typology and Change (OUP, 2012). Anna Laura Lepschy is Emeritus Professor of Italian at University College London, an Honorary Fellow of Somerville College, Oxford, and holds visiting positions at Cambridge and Toronto. She is the author of numerous articles on Italian literature, language, and art. She has edited several collections, including Multilingualism in Italy Past and Present (with Arturo Tosi, Legenda, 2002), Languages of Italy: Histories and Dictionaries (Longo, 2007), and Into and out of Italy: Lingua e cultura della migrazione italiana (with Adam Ledgeway, Guerra, 2010). In 2010 she was awarded the British Academy’s Serena Medal for her lifelong contribution to Italian studies. Giulio Lepschy is Professor Emeritus at the University of Reading, Honorary Professor at University College London, and Visiting Professor at the University of Toronto. He is a Fellow of the British Academy and of the Accademia della Crusca. His main research interests are in Italian linguistics and dialectology, and the history of linguistics, themes on which he has published more than a dozen collections of essays, including most recently Parole, parole, parole e altri saggi di linguistica (Il Mulino, 2007) and Tradurre e intraducibilità: quindici seminari sulla traduzione (Aragno, 2009). Alessandra Lombardi is an Affiliated Lecturer in the Department of Italian at the University of Cambridge. She graduated from the University of Calabria and obtained her Ph.D from the University of Manchester with a thesis entitled ‘The grammar of complementation in the dialects of Calabria’. From 2000 to 2005 she worked as a researcher on the AHRC-funded project Sintassi degli antichi volgari d’Italia (SAVI). She has published articles on the syntax of old Italian and on aspects of the morphosyntactic structure of southern Italian dialects and Sardinian. Michele Loporcaro is Professor of Romance Linguistics at the University of Zürich and a Fellow of the Academia Europaea. His research interests include the phonology, morphology, and syntax of Italo-Romance varieties, historical Romance linguistics, and linguistic historiography. In addition to numerous articles he is the author of Grammatica storica del dialetto di Altamura (Giardini, 1988); L’origine del raddoppiamento fonosintattico (Francke, 1997); Sintassi comparata dell’accordo

xxii

Notes on contributors

participiale romanzo (Rosenberg & Sellier, 1998); Cattive notizie: la retorica senza lumi dei mass media italiani (Feltrinelli, 2005; 3rd edn, 2010); Profilo linguistico dei dialetti italiani (Laterza, 2009; 2nd edn, 2013). Martin Maiden is Professor of the Romance Languages at the University of Oxford, the Director of the Research Centre for Romance Linguistics, University of Oxford, and a Fellow of Trinity College, Oxford. He is also a Fellow of the British Academy. His main research interests lie in the historical and comparative linguistics of the Romance languages, especially Romanian and Italo-Romance linguistics, and in morphological theory, on all of which he has published extensively. He co-edited The Dialects of Italy (Routledge, 1997, with Mair Parry) and The Cambridge History of Romance Languages (CUP, 2011, 2013, with Adam Ledgeway and J. C. Smith). Nicola Munaro is Assistant Professor of Linguistics and Dialectology at the University of Venice, and a collaborator on the online ‘Atlas of Italian dialect syntax’ (ASIt). His research interests range from Italian dialectology to comparative syntax, from both a synchronic and a diachronic perspective. His articles mostly address issues involving the left periphery of the clause. He contributed several chapters to the Grammatica dell’italiano antico (ed. Salvi and Renzi, Il Mulino, 2010). He has published the monograph Sintagmi interrogativi nei dialetti italiani settentrionali (1999) and co-edited Mapping the Left Periphery (The Cartography of Syntactic Structures, vol. 5; OUP 2010, with Paola Benincà). Diego Pescarini is Principal Investigator and Co-ordinator for the project ‘A grammatical survey of Italian dialects’ at the University of Padua and before that was Marie Curie Fellow at the University of Bristol. His research focuses on (micro-) variation within morphosyntactic systems of Italo-Romance, in particular in relation to clitic pronouns, on which he has published articles in Atti del sodalizio glottologico milanese, Linguistic Inquiry, and Vox Romanica. He edited La negazione: variazione dialettale ed evoluzione diacronica (Unipress, 2008, with Federica Cognola), Studi sui dialetti della Calabria (Unipress, 2009), and Studi sui dialetti delle Marche (Unipress, 2013, with Jacopo Garzonio). Cecilia Poletto is Professor of Romance Linguistics at the University of Frankfurtam-Main. Her research topics include diachronic and synchronic micro-variation of Italo-Romance dialects, and the modelling of this data within the ‘cartographic’ framework. She is a member of the research group which set up the online syntactic atlas of the dialects of Italy (ASIt). Her publications include La Sintassi del soggetto nei dialetti italiani settentrionali (Unipress, 1993), The Higher Functional Field: Evidence from the North Italian Dialects (OUP, 2000), and Word Order in Old Italian (OUP, 2014).

Notes on contributors

xxiii

Ian Roberts is Professor of Linguistics at the University of Cambridge and Professorial Fellow at Downing College, Cambridge, having previously held chairs at the University of Wales, Bangor and the University of Stuttgart. His books include Verbs and Diachronic Syntax (Kluwer, 1993), Syntactic Change (CUP, 2003, with Anna Roussou), and Diachronic Syntax (OUP, 2007). He has written many articles on topics in synchronic and diachronic syntax, as well as on the syntax of the Germanic, Romance, and Celtic languages. In 2007 he was elected a Fellow of the British Academy. John Charles Smith is a Fellow and Tutor in French Linguistics at St Catherine’s College, Oxford, following lectureships at the Universities of Surrey, Bath, and Manchester, and visiting appointments in Paris, Limoges, Berlin, Melbourne, and Philadelphia. His main field of interest is historical morphosyntax, and he has published on agreement, case, refunctionalization, deixis and pronouns. He is Secretary of the International Society for Historical Linguistics and co-editor of The Cambridge History of the Romance Languages. In 2007 he was created Chevalier dans l’Ordre des Palmes Académiques for services to the French language and French culture. Rosanna Sornicola is Professor of General Linguistics at the University of Naples ‘Federico II’. She is the author of more than 100 articles on the internal and external history of Italian and the dialects of southern Italy and Sicily, on pronouns, word order, and information structure, and on the history of general and Romance linguistics. Her books include Sul parlato (Il Mulino, 1981) and Stability, Variation and Change of Word-Order Patterns over Time (Benjamins, 2000, co-edited with Erich Poppe and Ariel Shisha-Halevy). She is a founding editor of Bollettino linguistico campano. Christina Tortora is Professor of Linguistics at the City University of New York, having previously taught at the University of Michigan and been a Visiting Fellow at the University of Padua. She has published articles both on the syntax of modern Italian and dialects and on regional variation in modern American English. She edited The Syntax of Italian Dialects (OUP, 2003) and is currently working on a comparative grammar of the Piedmontese dialect of Borgomanero to be published by OUP in 2014. John B. Trumper is Professor of Linguistics at the University of Calabria, having previously taught at the universities of Padua and Pavia. He has conducted research into and published on a wide range of topics, including name studies (plants, places, and people), grammatical gender, forensic phonetics, and Albanian and its contacts with the dialects of southern Italy. His books include Toponomastica calabrese (Gangemi, 2000, with Marta Maddalon and Antonio Mendicino), and

xxiv

Notes on contributors

Vicende storico-linguistiche del tirreno cosentino (Pellegrini, 2006). He is the editor of the Vocabolario calabro (vol. 1: A–E, Laterza, 2001). Laura Vanelli is Professor of Linguistics at the University of Padua. Her research encompasses the linguistics and grammar of Italian and Italian dialectology (particularly northern dialects, together with Raeto-Romance varieties, also in comparison with other Romance languages). She has published numerous articles on the synchrony and diachrony of Italo-Romance syntax, morphology, and phonology. Her books include I dialetti italiani settentrionali nel panorama romanzo: studi di sintassi e morfologia (Bulzoni, 1998), Nuova grammatica italiana (Il Mulino, 2004, with Giampaolo Salvi), Linguistica friulana (Unipress, 2005, with Paola Benincà) and Grammatiche dell’italiano e linguistica moderna (Unipress, 2010). Nigel Vincent is Professor Emeritus of General and Romance Linguistics at the University of Manchester, following retirement from the Mont Follick Chair in Comparative Philology, and a Fellow of the British Academy and of the Academia Europaea. He has held visiting appointments at the Universities of Copenhagen, Pavia, and Rome, and an Erskine Fellowship at the University of Canterbury (NZ). His publications include The Romance Languages (with Martin Harris, 1988) and articles on morphosyntactic change, with special reference to Latin, Italian, and the dialects of Italy. He co-directed with Mair Parry and Robert Hastings the AHRCfunded project Sintassi degli antichi volgari d’Italia (SAVI) (2000–2005).

1 Similarity and diversity in the evolution of Italo-Romance morphosyntax NI GE L VI NCE N T

1.1 Introduction As the words themselves imply, ‘dialects’ and ‘diachrony’ have something in common. They share the Greek prefix dia- ‘across’ and with it the fact that they involve comparing and establishing links between linguistic material of different types. In the case of dialectology, such comparisons work across the regional and social dimensions of languages, while for diachrony the key parameter is time and the goal is to understand how and within what limits language may change as time passes. But the two are more closely linked than that, since the source of the dialectal diversity lies precisely in the patterns of change that have caused the local varieties to diverge from the original unity of their ancestor. Such diversity in the fields of phonology, morphology, and vocabulary has long been the staple of research in general and Romance historical linguistics, but much less attention has been devoted to the same issues within the domain of grammar. The dialects of Italy constitute a particularly fertile terrain for the investigation of these links, since not only is there a considerable range of surviving dialectal variety but for many of the dialects there is a long and rich textual tradition, which permits the reconstruction of a wide range of constructional types and of their historical trajectories. Add to this the richness of our textual knowledge of the source language, Latin, and we have a context for research which it would be hard to better. The purpose of the present volume, therefore, is to assemble a group of the field’s leading specialists to address these issues. In so doing, we aim both to take stock of the current state of the field and of the fruits of recent research and to set out new results and new questions to help move forward the frontiers of our knowledge and understanding. This first chapter sets the scene, lays out some of

2

Nigel Vincent

the crucial issues and concepts and, in so doing, introduces the more specialized fare that awaits the reader in the remainder of the volume.

1.2 Dialects In the Italian context the term ‘dialect’ has taken on a rather special sense. As elsewhere, its principal reference is geographical, although the units so labelled correspond to organizational entities varying widely in size and ranging from small villages and municipalities—for example Cairo Montenotte, the site of the principal work of the dedicatee of this volume (Parry 2005), has a population of some 14,000— through to provinces (Neapolitan, Paduan, Sienese) and regions (Piedmontese, Sicilian, Sardinian). Moreover, some dialects that have become well known in the general linguistic literature correspond to no modern administrative unit but to a loosely defined geographical area, as for example is the case with Salentino (Calabrese 1993). At the level of the region we also find recognition of the internal diversity of dialects within these areas in expressions like ‘Lombard dialects’ or in sub-classifications such as western vs eastern Lombard. There is too the notion of a transitional dialect. Cairese is once more a case in point, showing properties both of Ligurian and of Piedmontese, thereby reflecting its geographical position close to the border and on a main communication route between these provinces. Besides such differences in the spatial range of dialects there are differences of register and type of usage, corresponding to what, extending and refining the use of the dia- prefix, are sometimes termed the diamesic, diastratic, and diaphasic dimensions of language (Berruto 1993). Where a local dialect like Cairese may display very little in the way of variation according to speaker or social context, nor evidence a rich written tradition, the dialects of centres like Milan, Naples, or Venice exhibit a wide range of sociolinguistic variation and have long literary and non-literary written histories. Any one of these, and indeed others, might have become the language of Italy—and thus now be known by the label ‘Italian’—if the political history of the peninsula and its associated islands had panned out differently. All of which amounts to saying that the dialects that are the focus of attention in this volume are not dialects of Italian, pace the title of a classic work such as Rohlfs (1966–9), but simply dialects, or even more simply languages, spoken and written in Italy. What unifies them is their shared ancestry as descendants of Latin and hence their membership of the family of Romance languages, and this supersedes other classifications. Thus, it is traditional within the study of Romance to recognize subgroupings like Ibero-Romance and Italo-Romance that in turn roughly correspond to geographical entities such as the Iberian and Italian peninsulas. And, using the criteria that determine such sub-groupings, we can recognize a dialect as a member of one group even when it has, so to speak, strayed away from home, as with the Catalan dialect still spoken in the Sardinian town of Alghero or the Gallo-Italic dialects that

The evolution of Italo-Romance morphosyntax

3

took root in a number of villages in southern Italy and Sicily at the time of the Norman invasions. But even putting such exceptional cases on one side, there are still at best blurred edges to these classifications. It is common, for example, to treat Sardinian as a separate development from Latin rather than as a member of the ItaloRomance group, but Sardinian dialects nevertheless share many properties with the dialects of the mainland. Or again, Friulian is standardly considered part of the RætoRomance sub-branch but shares features with other dialects in the northeast of Italy. Both in this chapter, therefore, and in the book as a whole, we will understand the term ‘Italo-Romance’ in a broadly inclusive sense and, rather than be concerned with the nuances of dialect classification (for which see e.g. Maiden and Parry 1997), we will focus attention on the lessons that data from dialects—whether large or small, richly documented or not—can tell us about the processes of morphosyntactic change that have led to the modern state of affairs, and about the ways such processes can be modelled within, and thereby illuminate, linguistic theory.

1.3 Diachrony The historical perspective provides a natural explanation for the fact that the dialects of Italo-Romance (sensu largo) exhibit a considerable degree of overlap in morphosyntactic structure, the consequence either of shared inheritance from Latin or of shared innovation with respect to Latin. Verbal systems, from which most of the exemplification in this chapter is drawn, display very clearly this balance between retention and innovation (for more general discussion and exemplification of this distinction see the contributions to Maiden, Smith, and Ledgeway 2011). Thus, all dialects have finite verb forms which are the direct descendants of the Latin inflected present and imperfect and, in their early stages at least, also of what in the Latin grammatical tradition is called the perfect. They all too have non-finite forms such as the infinitive and the past participle which can for the most part be traced back to Latin, with due allowance being made for the effects of sound change and analogy. At the same time they also exhibit a variety of innovative structures largely as the result of the grammaticalization of a range of lexical verbs in construction with one or more of the inherited non-finite forms. In some cases—perhaps most evidently in the futures and conditionals—the grammaticalization process has advanced sufficiently for the forms in question to have undergone univerbation and to have been reintegrated into the synthetic part of the system (see Vincent 2013 and references cited there). Other constructions have reached different stages along the historical route from transparent and compositional verbal syntax to opaque, non-compositional periphrasis. Indeed, this contrast between compositionality and non-compositionality is one of the themes we will develop in this chapter. Even when the modern dialectal structures are different, often strikingly so, from Latin, it is remarkable that in the overwhelming majority of cases the ingredients out

4

Nigel Vincent

of which those structures are composed are of Latin origin. In other words, within the verbal system there is relatively little in the way of borrowing, and most of what borrowing there has been is late and between the dialect and Italian, once it had become the literary and educated standard, rather than between dialects. Instances of this kind include the progressive built out of the auxiliary sc-tè, the cognate of stare ‘stand, be’, plus the gerundio ‘gerund’ cited by Parry (2005: 199–200) beside the same dialect’s autochthonous construction with essere ‘be’ plus locative plus che ‘that’; the route by which in Altamurano dovere ‘must’ has developed a future meaning discussed in Loporcaro (1988: 288); and the multiple grammars for clitic pronouns in Piedmontese described by Tortora in her chapter in this volume. There has also been borrowing (in both directions) in those cases where an Italo-Romance dialect and a language from another family coexist within the same community; on such ‘outward’ borrowings from Italo-Romance, see now the studies collected in Breu (2011). Phenomena of this kind, however, remain peripheral both geographically and systemically speaking. For the most part, then, the uniformity of the source material makes the dialects into a set of natural diachronic experiments in which the same core menu of ingredients has come together in different combinations or structures in different places, and thus afforded a series of minimal diachronic pairs or sets which in turn can shed valuable light on the mechanisms of morphosyntactic change, and on the various theoretical systems which have been used to model these changes. Thus, tenere ‘hold’ in construction with a past participle is in a number of southern dialects on the road to a perfective or resultative meaning, while in some Piedmontese and Lombard dialects the same combination has an iterative or continuative meaning (see }1.5.2.3 for further discussion of this pattern). Similarly, many dialects have developed a construction in which an infinitive can occur with an overt nominative noun phrase as subject, as discussed by Bentley in her chapter, but only a subset of those dialects has matched this change with the spread of overt agreement morphology from the finite to the non-finite system. Or again, all ItaloRomance dialects, like the Romance languages in general, have innovated systems of unstressed or clitic pronouns that have come to occupy preverbal or postverbal positions, but there is an enormous amount of regional variety in the order and position of these clitics, as can be seen here in the chapters by Pescarini, Roberts, and Tortora. Investigators therefore find themselves faced with three logically distinct tasks, though not all of these are necessarily undertaken in one and the same study: (a) the accurate description of the empirically attested patterns; (b) the analysis of such patterns within the terms of a general linguistic theory; (c) the explanation of the reasons for why these patterns have developed.

The evolution of Italo-Romance morphosyntax

5

Although, when set out like this, we seem to have three independent stages to the scientific endeavour, the boundaries between them can sometimes be blurred or conflated, depending on the methodological and ideological stance of the individual researcher. Thus, for some (a) and (b) converge since the descriptive terminology— often but not always drawn from traditional grammar—and the theory are taken to be one and the same. For others a descriptive category like ‘subject’ may be a useful pre-theoretical way to get a handle on the patterns of the data, but the theoretical analysis will then subsume this notion within a system of configurations and constraints. And for many, to frame the empirical data within an overarching theory is already in some sense to explain it: what is sometimes called ‘internal’ explanation. For others, the concept ‘explanation’ is reserved for external factors derived from the way language is organized in the mind or embedded in its socio-communicative context. These are very big and complex ontological and epistemological issues which we will not go into further here, but the reader should bear in mind that the chapters collected here do not offer a uniform take on the way these parameters of scientific research interact with each other. In that sense our chapters are as illustrative of the diversity of theoretical and metatheoretical approaches that can be deployed within modern syntactic research as they are of the variety of empirical material available to researchers.

1.4 Sources of data Both in common parlance and in much traditional dialectological research, dialects are taken to operate at a lower register level than standard languages, but as we have noted the Italian situation is complicated by the fact that what are called dialects may sometimes display as wide a range of diamesic and diastratic variation as many standard languages. The evidence for this moreover stretches over a long timespan, since many of the Italo-Romance dialects have documentary histories going back to the twelfth and thirteenth centuries. In this section we briefly review the nature of the depth and variety of the Italian textual tradition (see Vincent, Parry, and Hastings 2004 for more detailed treatment). The earliest texts commonly agreed to be in an Italian vernacular are the so-called Placiti cassinesi (see Ledgeway 2011), datable to the years 960–63. But before that, a word about the Latin background. Two points are relevant here. First, it needs to be emphasized that Latin itself is not a unified language, but rather one that has its own complex internal, and textually attested, history. Indeed, the period of attestation of Latin from the 7th century bc down to the middle of the first millennum ad is longer than the timespan which we conventionally label Italian (for details see Clackson and Horrocks 2007). And interestingly, some of the first examples of patterns that emerge in Romance go back to the earlier texts within that period only to disappear from sight in the Classical period. This diachronic disconnect is something that has been well noted

6

Nigel Vincent

in the Italian historical linguistic tradition, as witness the following remark by Maria Luisa Porzio Gernia (1982: 26): Il problema della scomparsa della declinazione nelle lingue romanze merita di essere ripreso immergendo quel complesso di mutamenti assai vario, che lega il latino al romanzo, in un movimento di deriva che ha le sue radici nel latino arcaico. In questo senso le origini romanze si riallacciano all’antica storia di Roma. [The problem of the disappearance of declension in the Romance languages deserves to be revisited, with the complex and varied set of changes which link Latin to Romance being seen in the context of an overall pattern of drift which has its roots in archaic Latin. In this sense the origins of Romance reconnect with the early history of Rome.]

For those who adopt this view, the reason for this discontinuity is to be found in the codification of what we now call Classical Latin as a language of literary and cultural prestige to some degree divorced from the currents of popular usage that provide the ancestors of many Romance patterns (Rosén 1999). Other scholars—perhaps most notably and most thoroughly Adams (2013: 862–4)—are less persuaded. Just as we need to be cautious in our use of Classical Latin, so too we must avoid the reification of something called Vulgar Latin as an independent language existing beside or after Classical Latin. Rather, what we have is a rich network of potential evidence, and a clear indication of the outcome in the form of the many dialects that are attested in the modern Romance, and in our particular case Italo-Romance, speaking world. Within the Italo-Romance area, the textual tradition is very varied. Not surprisingly, it is the largest administrative centres like Milan, Florence, Naples, Venice and so forth that yield the richest collections. But medieval Italy was characterized by a constantly shifting network of city-states so that, even outside what are now the largest centres, there are frequently many texts on which the diachronic dialectologist can draw. At the other extreme, for very small locations, we can expect that little or nothing will survive from much before the 19th century and sometimes even more recently. This is the case, for example, with Cairo Montenotte. In instances, there may be a relatively rich body of texts but they may be rather stylized and removed from what we presume the everyday language must have been like. This is true for the first attested phase of Sardinian, where the majority of early texts are administrative and legal, and often quite repetitive and formulaic. Even so, we may find precious nuggets. For instance, while it is common in early texts in many parts of Italy to find examples where the finite complementizer che or ca ‘that’ is repeated, only in these early Sardinian texts do we find the repetition of the non-finite complementizer de (Vincent 2006). However, in many areas we are more fortunate in that not only is there a long textual tradition, but it is also a very varied one. Thus in Venice, from the 13th century onwards we have an ample assortment of creative writing, administrative documents, transcripts of court proceedings, and numerous private letters (see e.g. Stussi 1965). Similarly, in compiling his monumental historical grammar of

The evolution of Italo-Romance morphosyntax

7

Neapolitan, Ledgeway (2009a) was able to draw on extensive texts going back into the early years of the 14th century. Nor would the chapters in this volume by Bentley and by Sornicola have been possible without the evidence derivable from the long documentation of Sicilian. For some phenomena, indeed, the early texts may provide the only evidence that a construction has ever existed. Thus, the so-called inflected infinitive is only attested on the mainland in texts from the Naples area before the 18th century.

1.5 History and theory: analysis and synthesis in the verbal system As a way of introducing the themes and issues of the volume, in this section I will consider in some more detail a longstanding question in the history of Romance— and indeed general—linguistics: how are we to understand the interaction of analytic and synthetic forms within the evolving Romance verb system? I will not repeat here the arguments adduced in Vincent (1997a) on the same question, but will move on from the conclusions there: namely (a) that the terms ‘analytic’ and ‘synthetic’ only make sense if applied at the level of individual constructions and not to languages as a whole; (b) that they define a scale rather than a binary divide; and (c) that changes identified as moving from analysis to synthesis are, both logically and empirically, different in kind from those going from synthesis to analysis.1 It is however worth noting that there is current work elsewhere which continues to treat the issue at the level of the language, and indeed Italian has been invoked in this connection. Thus, in a recent statistical study of the patterns of analyticity and syntheticity in the history of English, Szmrecsanyi (2012: 659) writes: ‘we knew before that Russian is a particularly synthetic language, and that, say, Italian is less synthetic and more analytic than Russian’ and ‘even eight centuries ago English was less synthetic than, say, Italian or its contemporary cousin, German’. The innovative aspect of Szmrecsanyi’s study lies in his use of corpus-based statistical methods, and to achieve this, items have to be pre-assigned to categories, namely analytic word tokens (closed-class function words such as complementizers, auxiliaries, and pronouns), synthetic word tokens (items which carry grammatical markers or clitics), and simultaneously analytic and synthetic word tokens (inflected auxiliaries). However, our concern here is with the gradual nature of the boundaries between these classes, as borne out for example in the distinction between main verbs, light verbs, and auxiliaries (see }1.5.1 below), and the nature of the links between such items and those like infinitives and participles with which they co-occur in these new constructions. Arguably, resolution of these issues should precede efforts to quantify, as indeed Schwegler (1990: 193–5) had already concluded. Our aim here therefore is to point the way, in a necessarily 1 Further arguments in support of these conclusions are to be found already in Schwegler (1990) and in the more recent discussion in Ledgeway (2012a: ch. 2).

8

Nigel Vincent

preliminary fashion, towards the cross-dialectal differences attested within the analytic constructions that are emerging or have emerged. All of the examples that we will discuss in detail are ones in which either an analytic construction has come to replace a synthetic one, as in the case of the (imperfective) passive, or the analytic construction sits side by side with a synthetic one, creating possibilities for contrast, for example between the passato remoto and the passato prossimo, or between the synthetic forms and the newly grammaticalized content to be seen, say, in the progressives. We should not however assume that the traffic has to be all one way. Two other options are, first, that the synthetic formation may extend its domain, and second, that a synthetic pattern may, as it were, reassert itself and displace an analytic one. An example of the first is the way a number of Italian dialects have developed a system of inflectional endings on non-finite as well as finite forms, as with ONap. essendono ‘be.ger.3pl’ or poteremo ‘be able.inf.1pl’. As Loporcaro (1986) shows, a key feature of this change is the development of an agglutinative type of morphology within the non-finite system. It is not coincidental that such a morphological pattern is not the kind inherited from Latin, but is precisely a more transparent and readily segmentable pattern of morphological formation, nearer the analytic end of the spectrum even while remaining within the boundary of the word. In this respect the patterns which evolve here are not that far removed from the extension of endings from one etymologically justified place within the finite system to another position where there is no etymological support, as in the complex interactions of the third person plural endings in -ro and -no analysed in Vanelli’s chapter. At the same time, as Bentley demonstrates in her chapter, the syntactic pattern of a non-finite form with an overt nominative subject may emerge in dialects such as Sicilian where there is no agreement marking on non-finite forms, a fact which suggests that the genesis of the infinitive + subject constructions is the prior change, and that some dialects subsequently come to reflect this in their morphology while others do not. Clear cases of a synthetic formation repelling the analytic one are not easy to come by within Italo-Romance, but as evidence that this type of change can also happen, and therefore should not be a priori ruled out, we may cite the recent developments in Rioplatense Spanish documented in Fløgstad (2012). Here, in the usage of younger speakers, as seen in the following examples drawn from her paper, the same verb form fui ‘be/go.1sg.pret’ is used both in the general experiential sense (1) and to express a specific event (2): (1) Fui a Gualeguaychú (Rioplatense Sp.) go.pret.1sg to Gualeguaychú ‘I have been to Gualeguaychú (at some point in my life)’

The evolution of Italo-Romance morphosyntax

9

de visita a Gualeguaychú hace unos 3 años (Rioplatense Sp.) (2) Yo fui I go.pret.1sg of visit to Gualeguaychú ago some 3 years ‘I went for a visit to Gualeguaychú about three years ago’ However, for older speakers, as indeed more generally in European and Latin American Spanish, the analytic form he ido ‘have.prs.1sg go.pst.ptp’ would be required in (1). In short, where once there was a contrast, now there no longer is, and the form that has won out is the synthetic one. We may contrast this with the converse development in northern Italian dialects such as Cairese in which it is the analytic form which has replaced the synthetic passato remoto (Parry 2005: }69). In terms of the division of labour sketched in }1.3, it is best to see labels like ‘analytic’ and ‘synthetic’ as part of our pre-theoretical vocabulary in terms of which to classify and organize the data, and thus prepare the way for a more refined and precise series of research questions, namely: (a) Typology: Are all the constructions we have called analytic of the same kind, or do we need a more delicate sub-classification of the types and their diachronic relations to each other? (b) Compositionality: How do the different parts of an analytic construction fit together? How does the separation of the formal constituents of the construction relate to their semantic transparency? And how do these relations change over time? (c) Paradigmaticity: What is the nature of the interface between the forms called synthetic and those called analytic? In particular, how do they relate to the concept of the paradigm? The following sections briefly address each of these in turn. 1.5.1 Typology As an example of the issues that arise in this connection, let us start with causative constructions. It is common in the typological literature to distinguish between synthetic and analytic causatives (see e.g. Shibatani and Pardeshi 2002). Synthetic formations of this kind need not detain us, since there are none of these attested within the empirical domain of interest in this volume. The nearest we come are some of the Latin verbs in -eō such as moneo¯ ‘warn’, in origin a causative on the stem ‘think’ (cf. mens, mentis ‘mind’). However, already in Latin the Indo-European origins of this pattern had been obscured, and we find instead most commonly a biclausal construction with various main verbs such as efficio¯ ‘cause, bring about’ and iubeo¯ ‘order’ followed by a full finite CP complement introduced by ut ‘that’. Among these is the verb facio¯ ‘do’ as in (3) (which is taken from the compilation in Chamberlain 1986):

10 (3)

Nigel Vincent adicies super esicia, facies pour.fut.2sg over quenelles make.fut.2sg ferveat (Lat., Apicius, De re coquinaria II.i.5) boil.prs.sbjv.3sg ‘pour over the quenelles, bring to the boil’

ut comp

More reduced in structure and yet still biclausal is the pattern found in English where make takes a small clause complement, as in he made [the children dance]. Arguably this is the structure we should attribute to examples such as (4) where facit ‘he makes, portrays as’ governs two coordinated clauses, one headed by the infinitive laudare and the other by the participle conloquentem:2 (4) Polyphemum Homerus . . . cum ariete etiam conloquentem Polyphemus.acc Homer.nom with ram.abl even with.talk.prs.prt.acc facit eiusque laudare make.prs.3sg his=and praise.inf fortunas (Lat., Cicero, Tusculanae disputationes 5, 115) fate.acc.pl ‘Homer even has Polyphemus talking to a ram and praising its fate’ Out of beginnings such as this there emerges the familiar pan-Romance (except Romanian) monoclausal causative construction in which the infinitive and the ‘light’ verb fare/faire/hacer etc. combine to create a complex predicate. It is this last which is of particular interest here, for three reasons: first because it is attested across the whole of Italo-Romance and throughout the period for which we have texts. The examples in (5) are a small sample, intended for illustrative purposes only; they are by no means systematic or comprehensive. (5)

a. i an fò intrè i’ ómi (Cairese, Parry 2005: 224) scl have.3pl make.pst.ptp enter.inf the men ‘they let the men come in’ b. Maria at fattu tuncare Mary has make.pst.ptp close.inf ‘Mary made someone close the door’

sa the

janna (Srd., Jones 1993: 277) door

c. fe’ he.made

abrosciare circa otto case burn.inf about eight houses (15th-c. Nap., Ferraiolo Cronica, 116r.26; Ledgeway 2009a: 899) ‘he caused about eight houses to burn down’

2 This example is from Robustelli (2000), who provides a rich documentation of this construction in Latin, old Italian, and the modern language.

The evolution of Italo-Romance morphosyntax d. lo the

facea portare li soy dei made carry.inf the his gods (15th-c. Sal., Libro di Sidrac 2 r.20; Sgrilli 1983) ‘the king Botus had his gods carried . . . ’

e. me me=

re king

11

Botus . . . Botus

fè he.made

arsaltare a i suoi compagni attack.inf to the his companions (Pad., Ruzante, La Moscheta, 625, c.1530; D’Onghia 2003) ‘he had/made his friends attack me’

f. la the

cerca el fa fare grande search he makes make.inf big (Mil., Bonvesin La vita di Sant’Alessio, 18r.22, c.1300; Wilhelm 2006) ‘he caused a great search to be made’

Second, complex predicate constructions display systematic differences from temporal and aspectual periphrases and thus demonstrate the need for a theoretical classification more sophisticated than simply analytic vs synthetic. In particular, the causative verb contributes part of the argument structure of the clause, leading to a consequent demotion of what otherwise would have been the subject of the infinitival verb. Such argument sharing is not typically part of the profile of an auxiliary (for more discussion of these differences see Rosen 1997; Abeillé and Godard 2010). Third, and most striking in the context of the present discussion, is the way the historical profile of the descendants of fare ‘make’ (and a few other verbs such as lasciare ‘let’, vedere ‘see’) remains constant over many centuries. There are some adjustments over time to the rules governing position of clitics, as well documented by Robustelli (2000), and in this respect there is also dialectal variation (D’Onghia 2003), but there is no evidence of further shifts of meaning or of the causative and the main verb fusing (univerbation). These facts are fully in accord with the hypothesis advanced by Butt and Lahiri (2002) that light verbs show systematically different diachronic profiles from auxiliaries and do not follow the clines of grammaticalization which are characteristic of the latter. This in turn raises the question of how their mode of combination is to be represented and how it differs from the way an auxiliary combines with its associated non-finite form, a question to which we now turn. 1.5.2 Compositionality By ‘compositionality’ we mean the principle whereby in general in linguistic expressions we expect the interpretation of a whole phrase or sentence to be built up out of the interpretation of its sub-parts, be they words or phrases. Exactly how this is to be achieved in all circumstances is a complex issue and one we cannot go into here; see Vincent (2013; forthcoming) for further discussion and references. For present purposes, however, the definition provided by Szabó (2012: 64) will suffice: ‘The meaning

12

Nigel Vincent

of a complex expression is a function of the meanings of its constituents and the way they are combined.’ What this means when we come to look at analytic constructions, whether complex predicates or those involving auxiliation, is that we need to investigate the role of both parts. This may seem to verge on the statement of the obvious, but in fact a remarkable amount of the general literature on the phenomenon of grammaticalization has concentrated on only one half of the story, namely the semantic shifts undergone by auxiliaries. To address the issue, we consider briefly the developments of perfects and of progressives, before looking at the behaviour of two different constructions with tenere, one in the south and the other in the north. 1.5.2.1 Perfects A significant proportion of the contributors to this volume have, at some point in their careers, explored the many descriptive and theoretical issues thrown up by the evolution of Latin habere ‘have’ and esse ‘be’ as complementary auxiliaries within the system of Italo-Romance perfect periphrases, and we will not enter further into these questions here (see the chapter by Loporcaro for further discussion and references). Rather, we need simply to underscore the continuity of the past participle within this construction. In origin a deverbal adjective, it had already come in Latin to have a perfective and passive value, and hence was a natural candidate for a role in the newly emergent periphrastic structure (for detailed discussion and exemplification in different varieties and periods of Latin, see now Adams 2013: 615–51). The two parts of the periphrasis thus compose naturally, with the auxiliary expressing the deictic or absolute tense and the participle the relative tense (Vincent 2011). One immediate consequence of the genesis of the new periphrasis, in Italian traditionally called the passato prossimo, is to set up a contrast with the inherited synthetic perfect form, or the passato remoto. The traditional labels, however, are not the most appropriate; the distinction is rather between a simple past and a present perfect expressing current relevance (Harris 1982). And yet hardly has the new periphrasis emerged than it begins to assume the function of referring to the past tout court, a recurrent type of change that has affected a number of languages and which has been dubbed ‘aoristic drift’ (Schaden 2012). And here Italy divides, with this drift having affected principally northern dialects, and in consequence the northern variety of the standard language, while the role of the synthetic form is increasingly dominant the further south one goes. From our perspective, the crucial consequence of this development is the loss of compositionality: a present auxiliary and a perfect participle cannot be composed to yield the value of past. Not for nothing did Spencer (2001) label this construction a ‘syntactic idiom’. Within the same timespan and in the same general geographical area as the passato remoto was falling out of use (and see Buono 2011 for some intriguing diamesic and diastratic dimensions of this process), a new pattern was coming into existence, usually known by the name of the equivalent formations in French as the passé surcomposé. The following representative examples are from the dialect of Cereda in

The evolution of Italo-Romance morphosyntax

13

the province of Vicenza and are reported in Poletto (2008: 502; see also Pisano 2010 for a similar development in modern Sardinian): (6)

a. co go bio when I.have have.pst.ptp ‘when I had eaten’

magnà (Cereda) eat.pst.ptp

b. go bio visto el papa (Cereda) I.have have.pst.ptp see.pst.ptp the Pope ‘I have had the experience of seeing the Pope’ The value of the auxiliary complex in (6a) is [past], equivalent to avevo in standard Italian, so is non-compositional in the same sense as discussed above under the label ‘aoristic drift’, yet at the same time the way it combines with the past participle magnà ‘eaten’ to yield a past perfect or pluperfect reading is eminently compositional, a phenomenon we have discussed elsewhere under the heading of ‘Dahl’s paradox’ (Vincent 2013: 131–2). In (6b), on the other hand, we have a reading that appears more compositional: the speaker has had the experience, as expressed through the perfect construction, and that experience was in the past. It is tempting to see in the emergence of the surcomposé a compensation for the loss of the simple past and the shift of the analytic perfect to assume the latter function. However, two reasons not to follow that line of argument are first that the aspectual values of the surcomposé are subtly different from those of the simple perfect periphrasis (Schaden 2007; Poletto 2008) and second that, in at least one area, Cosenza (northern Calabria), the simple past has also been lost but no surcomposé has emerged. 1.5.2.2 Progressives Progressives constitute a further good example of the interplay of innovation and retention. Latin of course did not have any progressive constructions of this kind, and an intriguing suggestion by Haverling (2010) is that part at least of the motivation for the Romance developments lies in a shift of meaning of the Latin imperfect, with consequent effects on the interpretation of the Romance imperfect. Four etymologically distinct progressive patterns can be distinguished: (i) the copula esse ‘be’ combined with a gerund or participle, e.g. OFr. se ne fust cumbatant (Chanson de Roland 1770) ‘if he was not fighting’ and Srd. so travallande ‘I am working’ (Jones 1993: 83); (ii) stare ‘stand’ plus the gerund attested in Spanish and across the whole of ItaloRomance, although, as we have noted, in some dialects it may be a calque from the standard language rather than a genuinely autochthonous development (Parry 2005: 199–200); (iii) various kinds of locative expression, particularly in the north those built from dietro a ‘behind’ + infinitive (cf. Poletto 2008); (iv) andare ‘go’/venire ‘come’ + gerund (Squartini 1998), patterns which emerge in high literary registers and are found already in medieval texts such as the Milanese of Bonvesin but not in the modern colloquial dialects.

14

Nigel Vincent

Once again we put on one side the contribution of the various auxiliaries, all of which fall into cross-linguistically well-attested categories, but underscore instead the fact that the accompanying non-finite form, in this instance the gerundio or present participle, with its open-ended relative tense or aspectual value, is entirely appropriate to the expression of incomplete or ongoing activities. 1.5.2.3 The two constructions with tenere It is instructive now to compare two different constructions in which the verb tenere, originally meaning ‘hold, keep’, comes to assume the auxiliary function. The first is exemplified in a number of southern dialects, as in the Neapolitan example (7a) and the Avellino example (7b), cited by Rohlfs (1969: 127) and discussed by Ledgeway (2009a: 626): (7)

a. saccio ca tien@ lu fuoc’ I.know that you.keep the fire ‘I know that you have the fire lit’

allumato (Nap.) light.pst.ptp

b. teneva na funtana frabbecata (Avellino) he.kept a fountain construct.pst.ptp ‘he had constructed a fountain/he had a fountain constructed’ Here we may argue that tenere is on its way to becoming a temporal/aspectual auxiliary but, pace Rohlfs, it is not there yet and is closer to the types of habere ‘have’ construction in Latin discussed above. A very similar situation for the Pugliese dialect of Altamura is described by Loporcaro (1988: 299–301). However, even if this construction has not yet been fully grammaticalized, the role of the past participle is consonant with its intrinsically perfective meaning and is consistent with its becoming grammaticalized in the future, as has happened in modern Portuguese. Compare now the construction built out of the same grammatical and lexical ingredients which is found in a number of Piedmontese and Lombard dialects as in (8) (examples from Ricca 1998): (8)

a. E chiel me teniva aggrezà (Pie., G. G. Alione 1521) and that.msg me= kept annoy.pst.ptp ‘And that man kept annoying me’ b. ten spendasciaa (Mil., G. P. Lomazzo 1589) keep.prs.3sg waste.money.pst.ptp ‘he keeps on wasting money’ c. l’ altra ten butò l’ acqua fuora del burchielo the other keeps throw.pst.ptp the water out of.the boat (Pad., Ruzante 1532) ‘the other kept on bailing out the boat’

The evolution of Italo-Romance morphosyntax

15

Rohlfs (1969: }733) also mentions this construction and assigns to it a durative value (‘significato durativo’). He draws his examples from the grammar of the dialect of Castellinaldo (province of Cuneo) by Toppino (1926), who strikingly translates this construction with the andare ‘go’ periphrasis: thus, u ten dić ‘egli va dicendo’, cosa tenstu fait ‘cosa vai facendo?’, i tnia dit tra mi ‘io andavo dicendo’, i ténu braiá ‘van gridando’. Rohlfs also cites an example from Bandello (2,11), il marito teneva detto che allumasse il fuoco, which he glosses as ‘persisteva nel dire’. In similar vein Salvioni (1919: 537) discusses the Milanese tegní battuu, which he reports as meaning both ‘andare insistendo’ and ‘continuare a insistere’. He also notes that the construction, while particularly frequent in earlier writings such as those of the Milanese poets Maggi (1630–99) and Balestrieri (1714–80), is no longer found in contemporary Milanese. This second construction is much more challenging for compositional approaches, since the past participle is not usually consistent with a progressive or continuative meaning. In this respect, one can contrast, as Ricca (1998) does, the English construction keep (on) doing, where the auxiliary is from the same semantic class but the non-finite form is a gerund or a participle. Ricca also suggests a possible way out of this dilemma when he observes that translations with the andare ‘go’ construction do not capture the real meaning here, which is iterative rather than progressive or durative. The iterative interpretation, with its implication of an event that has happened at least once before the time of the narration, is perhaps therefore less paradoxically expressed by means of a past participle. Whatever should prove the best account, this example demonstrates very well the central theme of this volume: the comparison and contrast between similar ingredients and their different regional developments, and the lessons that may be drawn therefrom for our general theory of linguistic change. We can now turn back to the question we posed at the end of }1.5.1: how does the mode of semantic composition of the two elements of a complex predicate differ from what takes place in constructions with a temporal or aspectual auxiliary? As we have just seen, in the latter case, for auxiliaries the most natural solution is to see them as either expressing the relevant tense/aspect feature or equivalently, in a cartographic approach, as occupying a tense/aspect head, and thus as having over time separated themselves from their source verb. Such separation is formally marked for avere ‘have’ in those varieties which contrast auxiliary with a main verb of possession averci, into which a locative clitic (ci) has been obligatorily incorporated, but it is an assumption that it would be necessary in any case to make on semantic grounds. With essere ‘be’ the issue may seem less clear, but even here arguments have been adduced to show that the perfective auxiliary needs to be distinguished from the copular or predicational use (Abeillé and Godard 2000). For complex predicates, on the other hand, Butt and Lahiri (2002) propose a single underspecified lexical entry that unites the main verb and the light verb. The unity of the two items at the lexical level accounts for the fact that light verb constructions

16

Nigel Vincent

persist without change of meaning over long periods of time, and also for the fact that light verbs do not develop idiosyncratic morphology which distinguishes them and serves to separate them from their lexical sources in the way that auxiliaries often do. 1.5.3 Paradigmaticity The following comment by Loporcaro (1988: 275) is revealing: I costrutti perifrastici del primo tipo—vale a dire i tempi composti dell’indicativo e del congiuntivo, e le perifrasi passive—avrebbero potuto esser descritti nel capitolo sulla morfologia, in quanto in sincronia fanno parte del sistema verbale non diversamente dai tempi non composti. [The periphrastic constructions of the first kind—that is to say the compound tenses in the indicative and the subjunctive, and the passive periphrases—could have been described in the chapter on morphology insofar as, from the synchronic point of view, they play a role in the verbal system which is no way different from that of the non-compound tenses.]

A statement of this kind presupposes two things: first, a circumscribed set of verb forms that constitute the ‘sistema verbale’, in effect the traditional paradigm, and second, that membership of this paradigm is not restricted to synthetic verb-forms but may encompass some but not all the analytic ones. When it comes to choosing between those which qualify and those which do not, Loporcaro distinguishes those where the auxiliary expresses the grammatical features of voice, tense, mood, and person from those which contribute modal or aspectual meanings. On this account, perfect and passive are in but progressive is out; mood is in but modality is out. Causatives are not dealt with, but it seems reasonable to predict that they too would be out. That analytic forms can be part of the core paradigm is much the less controversial of these assumptions. There is clear evidence already in Latin if we consider the contrasts and the sub-paradigm they form, shown in Table 1.1. TABLE 1.1. Latin verbal paradigm (laud- ‘praise’) imperfective perfective

active

passive

laudat laudauit

laudatur laudatus/-a est

As discussed in Börjars, Vincent, and Chapman (1997), while the imperfective verb forms have recourse to contrasting sets of inflections to express the difference between active and passive, for the perfective series the synthetic active forms are matched by a set of passive periphrases. It is a short step from this to conclude that the verbal paradigm of the modern languages should be filled out with a perfect periphrasis. Since the modern passives are also periphrastic, this will create the possibility of a compound periphrasis (Table 1.2).

The evolution of Italo-Romance morphosyntax

17

TABLE 1.2. Italian verbal paradigm (lod- ‘praise’) imperfective perfective

active

passive

loda ha lodato

è lodato/-a è stato/-a lodato/-a

Further ramifications are possible. Thus, as we have seen, some dialects have a surcomposé which expresses past perfect, and this generates another form of compound periphrasis as in (6) above. Moreover, as Poletto (2008: 501) notes, the additional aspectual dimension that the surcomposé offers can be added to all tenses, moods, and voices; hence when it is combined with the passive we will get a case of triple auxiliation or a double compound periphrasis: (9)

co so bio when I.am have.pst.ptp ‘when I had been called’

sta be. pst.ptp

ciamà (Cereda) call. pst.ptp

With developments such as this, the periphrastic part of the paradigm comes close to being the larger part of the system. When we come to consider what is inside and what is outside the verbal paradigm, progressives occupy an intriguing middle ground. Loporcaro’s criterion for inclusion is that the auxiliary should do no more than function as a piece of morphology (‘funziona da puro elemento morfologico’); if it adds meaning (‘una sfumatura semantica’) in its own right, then it falls ouside the core system. While it is not difficult to grant him this conclusion in the case of modals, let alone causatives or perception constructions, it is much harder to see that the aspectual value added in a progressive is any more or less a semantic feature than is pastness or futurity. However, there is perhaps a case for excluding them on formal grounds, since progressives do not freely combine with other periphrases. The attempt to join progressive and perfect in (10a) or progressive and passive in (10b) fails: (10)

a. *Giorgio è stato dormendo (It.) Giorgio is be.pst.ptp sleep.ger ‘Giorgio has been sleeping’ b. *Giorgio è stato essendo criticato (It.) Giorgio is be.pst.ptp criticize.pst.ptp ‘Giorgio has been being criticized’

We may contrast here the acceptability of the corresponding English translations as evidence that in a different verbal system progressive and perfect or passive may combine. We do not, however, have to go as far as English to see this. Consider the

18

Nigel Vincent

Sardinian example in (11), which is taken from Jones (1993: 141); for further examples and discussion see Remberger (2006: 272–81). (11)

So istatu travallande (Srd.) I.am be.pst.ptp work.ger ‘I have been working’

This in turn would mean that the boundary of the verbal paradigm is to be drawn differently in different dialects, though the exact nature of this grammatical terrain remains to be explored. Other contexts in which we find differences in combinability of constructions between dialects concerns the use of venire ‘come’ as a passive auxiliary. In the standard language, synthetic forms of both essere ‘be’ and venire ‘come’ are available: (12)

a. Giorgio era/veniva criticato dai suoi colleghi (It.) George be.impf/come.impf criticize.pst.ptp by.the his colleagues ‘George was (being) criticized by his colleagues’ b. Giorgio fu/venne criticato George be.pst/come.pst criticize.pst.ptp ‘George was criticized by his colleagues’

dai by.the

suoi his

colleghi (It.) colleagues

When we come to express a perfect passive, as in (13), essere ‘be’ is still fine but now venire ‘come’ fails: (13) Giorgio è stato/*venuto George be.prs.3sg be.pst.ptp/come.pst.ptp ‘George has been/*has come criticized’

criticato (It.) criticize.pst.ptp

We find a similar restriction with andare ‘go’ in the modal passive in (14): (14)

a. La questione va the.fsg issue.fsg go.prs.3sg ‘The issue should be discussed’

discussa (It.) discuss.pst.ptp.fsg

b. La questione andava discussa (It.) the.fsg issue.fsg go.impf.3sg discuss.pst.ptp.fsg ‘The issue should have been discussed’ c. *La the.fsg

questione issue.fsg

è be.3prs.3sg

andata go.pst.ptp.fsg

discussa (It.) discuss.pst.ptp.fsg

It is hard to see a phonological or morphological explanation for these ungrammaticalities; both venire ‘come’ and andare ‘go’ have past participles, venuto and andato, which are available in other contexts. The explanation is also unlikely to be semantic given the successful reference to the past in (12) and (14b). An alternative view, consistent with the differences we have noted above in respect of the

The evolution of Italo-Romance morphosyntax

19

progressive, is that new periphrases do not enter a language all at once but gradually expand their domain of applicability over time, and can work their way through to different parts of the paradigm in different places at different rates. Thus, the patterns which are not available in the standard language are attested in the Engadine (Ræto-Romance) dialects of Scuol (15a,b) and Casaccia (15c) (Manzini and Savoia 2005: ii.621–2): (15)

a. la jakka ez the.fsg coat.fsg be.prs.3sg dalla pløvdʒa (Scuol) by.the.fsg rain.fsg ‘The coat got soaked by the rain’

ɲyda come.pst.ptp.fsg

b. e sOŋ ɲy I be.prs.1sg come.pst.ptp ‘I was called by him’

klOˈma call.pst.ptp

c. l ys l e the door clt be.prs.3sg ‘The door has been opened’

ˈɲi come.pst.ptp

daˈd by

blettʃa soak.pst.ptp.fsg

el (Scuol) him

dzaˈra (Casaccia) open.pst.ptp

A relevant factor here is expressive necessity. In these dialects, as elsewhere in Ræto-Romance, the ‘come’ verb is the only passive auxiliary. If the examples in (15) were not acceptable, the feature complex [perfect, passive] would remain unexpressed and inexpressible. At this point, however, it is time to bring our own reflections on these themes of inter-dialectal variation and contrast to a close and point the reader in the direction of the individual studies collected here.

1.6 The chapters Some of the chapters in this volume have already figured at various points in the preceding discussion. This section, however, introduces all of them more systematically with a view to showing how they illustrate and amplify the themes we have been exploring. Inevitably in a volume of these dimensions the studies do not provide a comprehensive comparative (morpho)syntax of the dialects of Italy. Indeed, in our current state of knowledge, that would be difficult to achieve. Rather, what these investigations do is to set out some of the dimensions and diversity and some of the techniques that can be used to explore them. They are grouped into two broad categories: verbal and (pro-)nominal. Under the first heading we start with a group of chapters that look in different ways at the interaction between morphology and syntax. Thus, in their contribution ‘The development of the southern subjunctive: morphological loss and syntactic gain’, Adam Ledgeway and

20

Nigel Vincent

Alessandra Lombardi examine how the traditional claim that the subjunctive has been lost in many southern dialects needs to be balanced against the presence of a range of subtle syntactic effects which are the counterpart of subjunctives elsewhere. Syntax and morphology are thus seen as alternative and complementary ways of giving expression to grammatical content. In similar vein, the chapter by Michele Lorporcaro, ‘Perfective auxiliation in Italo-Romance: the complementarity of historical and modern cross-dialectal evidence’, explores some of the finer ramifications of the interaction between the perfect auxiliaries in an analytic construction which in one sense can be said to replace the Latin synthetic perfect but which in so doing develops a range of special properties, the full extent of which can only be appreciated if we have access to diachronic and dialectal evidence. These two empirical dimensions are also to the fore in the investigation by Michela Cennamo of ‘Passive and impersonal reflexives in the Italian dialects: synchronic and diachronic aspects’, which builds on her own extensive earlier work on these constructions. Delia Bentley, by contrast, in ‘On the personal infinitive in Sicilian’, examines a structure which is a Romance development, with variants both inside and outside the ItaloRomance domain, and in which the morphology has not, so to speak, caught up with the syntax in the way that it has in other dialects such as Neapolitan and Logudorese. The verbal section of the volume concludes with two chapters that move away from morphosyntax into the domain of inflectional morphology. The title of the contribution by Martin Maiden and John Charles Smith says all that needs to be said by way of presentation: ‘Glimpsing the future: some rare anomalies in the history of the Italo-Romance and Gallo-Romance future and conditional stem, and what they suggest about paradigm structure’. The technique here is a classic one, well known to historical linguists for more than a century—the use of the irregular and anomalous forms as a key to unlocking the historical patterns of regularity. This work is of a piece with other studies that the authors and their colleagues and co-workers have carried out into the morphomic or ‘pure’ morphological structures of Romance verbal inflection. In ‘Person endings in the old Italian verb system’, Laura Vanelli focuses her attention on the desinences that attach to the stem and the semiotically driven patterns of morphological realignment that underpin the differences between the earlier and later stages of the modern language. We come next to a cluster of chapters on the pronoun systems of Italo-Romance. Diego Pescarini shows in ‘The evolution of Italo-Romance clitic clusters: prosodic restructuring and morphological opacity’ how the techniques of analysis and segmentation are relevant not only in relation to verb morphology stricto sensu but also apply to the clusters of unstressed pronouns, another typical development where the etymological material is Latin but the organization of the system is Romance. How that organization plays out in respect of syntax is the concern of Ian Roberts in ‘Subject clitics and macroparameters’. Following on from his earlier work in this area, he shows how the patterns of variation in these systems of the dialects of northern

The evolution of Italo-Romance morphosyntax

21

Italy both illuminate and are illuminated by the organization of Universal Grammar. These cross-dialectal studies are followed by two contributions which narrow the focus to particular dialects, Rosanna Sornicola’s ‘Sicilian 1st and 2nd person oblique tonic pronouns: a historical and comparative examination’ and Christina Tortora’s ‘Patterns of variation and diachronic change in Piedmontese object clitic syntax’. They are valuably complementary in that, while the latter looks at atonic pronouns in a northern dialect (and thereby adds more detail and a new perspective to the picture traced by Roberts and Pescarini), Sornicola looks at the hitherto rather less studied question of how stressed object forms develop in the early stages of a southern dialect, and the nature of the changes from the Latin forms that this involves. In a category of its own is John B. Trumper’s investigation of ‘Gender assignment and pluralization in Italian and the Veneto’. Although this phenomenon has generated a specialist literature when it comes to standard Italian—not least because rules for gender assignment are of great interest to language learners!—this study is rare if not unique in the way it pursues the questions into the dialects. An added dimension here comes from the effects of language contact between the dialect and the standard language. Our final two chapters bring us back to syntax. Paola Benincà and Guglielmo Cinque tackle the very little-studied topic of ‘Kind-defining relative clauses in the diachrony of Italian’. The most prominent feature of their account is the absence of change in the more than seven centuries that this structure has been attested. Such diachronic stability is reminiscent of the behaviour of the causative construction discussed above, and reminds us that a complete theory of diachrony needs to account for stasis as well as change. The chapter ‘Synchronic and diachronic clues on the internal structure of “where” in Italo-Romance’ by Nicola Munaro and Cecilia Poletto is striking for the way it brings together techniques from traditional etymology and modern theoretical syntax. Their conclusion opens up a perspective not only for specialists in Italo-Romance but also for those whose concern is diachronic morphosyntax more generally. It is very much to both these audiences that we hope the contributions gathered in this volume will appeal. In so doing, they follow the trail already blazed to such excellent effect by our colleague and friend, Mair Parry.

Part I Verbal Structures

2 The development of the southern subjunctive Morphological loss and syntactic gain* A D A M L ED G E W A Y A N D A L E S S A N D R A L O M B A R D I

2.1 Introduction It has traditionally been observed (Rohlfs 1968: 301–2; 1969: 61–2) that in an area extending from Sicily as far north as central Italy, including most of Lazio, southern Umbria, and a large part of the Marche, the (present) subjunctive has been replaced by the indicative (on the fate of the imperfect subjunctive, see }2.2.1 below). Consequently, where most Romance varieties morphologically mark an indicative/subjunctive contrast on the verb in cases like (1) and (2), southern Italian dialects (3) and (4) neutralize this modal distinction through the indiscriminate use of an indicative form in both cases:1 (1)

(2)

a. Vediamo we.see

che that

ballano they.dance.ind

b. Speriamo we.hope

che that

ballino they.dance.sbjv

bene (It.) well

a. Te you=

disi I.say

que that

vendrà (Lgd.) he.will.come

b. Te you=

disi I.say

que that

vengas (Lgd.) you.come.sbjv

bene (It.) well

* This study is dedicated to Mair, who has been a generous and inspirational colleague and a dear friend to both of us for many years. 1 In what follows, free translations of the examples are only provided where the sense is not immediately inferable from the glosses.

26 (3)

Adam Ledgeway and Alessandra Lombardi a. m me= b. e it.is

(4)

au they.have mmeʎʎo better

a. Crèdisi you.believe b. Vogliu I.want

ke that

ca that

chi that

ittu said

ke that

ve you.come.ind

ttu you

lu him=

vve (Borgorose-Spedino (RI), Lazio) you.come.ind

truvamu? (Maratea (PZ), Basilicata) we.find.ind

scúmbisi you.finish.ind

This is the standard view in the literature (cf. Ledgeway 2003: }2.4), witness Cordin’s (1997: 89) observation that ‘[i]n many contexts where Tuscan (and standard Italian) would select a subjunctive, dialects of central and southern Italy employ an indicative instead’. Now, while it is true that the (present) subjunctive has been lost in the dialects of southern Italy, the question naturally arises whether the modal distinction traditionally encoded by the indicative/subjunctive morphological opposition has been entirely jettisoned in these same dialects, or whether the relevant grammars have come to mark the distinction by some other means. Indeed it has not gone unnoticed in the literature that the indicative/subjunctive morphological distinction may have significant repercussions on syntax (for an overview, see Quer 2009a). For instance, it has been observed in some Occitan dialects (5a) and Italian (5b) that, whereas in indicative contexts the 2nd person singular subject pronoun may be null or overt in accordance with the usual pragmatic considerations, in subjunctive contexts only the overt pronoun (< Lat. tu) can be licensed (Benincà 1993: 258–63; Cardinaletti 1997: 51–2; 2004: 126–8; Oliviéri 2009). (5)

a. kaɽeɽˈia it.would.require

ke that

b. Vogliono they.want

*(tu) you=

che that

*(ty) you=

u sabese (Menton, Alpes-Maritimes) it=knew.sbjv

faccia do.sbjv

tutto (It.) everything

Another example concerns the binding of anaphors such as the Italian 3rd person possessive proprio ‘one’s own’ which, in accordance with Principle A of the Binding Theory, has to be locally bound within its immediate minimal domain. This is not possible in (6a) since the anaphor is in the embedded indicative clause and hence too distant from its intended antecedent Gianni, the matrix clause subject. Ostensibly, these same considerations ought to apply with equal force to the minimally different (6b); yet in this case the matrix predicate now selects for a subjunctive clause, a choice which exceptionally extends the minimal binding domain of the anaphor to include the matrix clause (Giorgi 1983: 335–41; Manzini and Wexler 1987).

The development of the southern subjunctive (6)

27

a. *[Giannii suppone [che tu sei innamorato della propriai moglie]] (It.) Gianni supposes that you are.ind in.love of.the own wife b. [Giannii suppone [che tu sia innamorato della propriai moglie]] (It.) Gianni supposes that you be.sbjv in.love of.the own wife

Similar explanations apply to the following Catalan contrast where the subjunctive (7b), but not the indicative (7a), complement proves transparent for the purposes of the long-distant dependency between the matrix negator no ‘not’ and the embedded n-word ningú ‘nobody’ (Quer 2009b: 1786). (7) a. *No han confessat que han vist absolutament ningú (Cat.) no they.have confessed that they.have seen absolutely nobody b. No not

vull I.want

que that

saludis you.greet.sbjv

absolutament absolutely

ningú (Cat.) nobody

Our final example involves contrasts like those illustrated in (8a,b), where C(omplementizer)-drop is licensed in irrealis contexts such as subjunctive complements, but crucially not in indicative complements (Wanner 1981; Giorgi and Pianesi 1997; 2004; Poletto 2001; Scorretti 1994; Giorgi 2009: 1840–45; 2010: 43–7; Ledgeway in press a). (8)

a. Temo (que) pueda durar mucho/Digo *(que) puede durar mucho (Sp.) I.fear that may.sbjv last.inf much I.say that may.ind last.inf much b. Sembra (che) voglia piovere / Dico *(che) vuole piovere (It.) it.seems that it.wants.sbjv rain.inf I.say that it.wants.ind rain.inf

Although data like these highlight how in particular cases there may be a strong correlation between morphological marking for indicative/subjunctive and specific structural phenomena, it remains to be understood whether the relevant modal distinction can be independently licensed and marked through syntax alone.2 Below we shall explore how in the dialects of southern Italy the indicative/subjunctive distinction, although no longer robustly marked (if at all) morphologically, continues to be subtly marked in several areas of the syntax which have hitherto gone unnoticed. In essence, we shall highlight how a loss at the level of morphology has, in turn, been offset by a gain in syntax.

2 In what follows we informally refer to this modal distinction in terms of a realis/irrealis opposition, although recognizing that this is nothing more than a crude yet convenient label for a series of modal values traditionally encompassed by the indicative/subjunctive contrast (for discussion, see Quer 2009b: 1779–80).

28

Adam Ledgeway and Alessandra Lombardi

2.2 Morphological marking of the indicative/subjunctive opposition in the south 2.2.1 The subjunctive In (1) and (2) above we noted how the indicative/subjunctive opposition represents a core distinction of the Romance verb system. Admittedly even among the standard languages there are cases of loss or attrition of the traditional morphological distinction. Such is the case of French (9a), where phonomorphological attrition has given rise to a systematic neutralization of the present indicative/subjunctive opposition in the singular and 3rd person plural of verbs of the first conjugation. Also relevant here is the observation that since the 19th century the imperfect subjunctive has all but been lost from most registers of the language and is now systematically replaced with the present subjunctive (9b; cf. Jeppesen Kragh 2010).3 Similarly, in Romanian, with the exception of a fi ‘to be’, the subjunctive is only distinctly marked, and then not in all verbs, in the 3rd person (10a,b). (9)

a. Je I

vois see

/ /

veux want

qu’ that

elle she

danse (Fr.) dances.ind/sbjv

b. Il fallait qu’ il partît / fût parti ) il parte / soit parti (Fr.) it required that he left.sbjv / was.sbjv left he leave.sbjv be.sbjv left (10) a. Văd I.see b. Văd I.see

că / that /

vreau I.want

să that

plătesc / pay.ind.3pl /

că that

plăteşti (Ro.) you.pay.ind/sbjv să that

vreau I.want

plătească (Ro.) pay.sbjv.3pl

Quite different though is the situation found in the dialects of southern Italy, where the subjunctive would appear to have been systematically replaced by the indicative; witness the representative modern dialect examples in (11): (11)

a. Me me=

pare it.seems

’a the

ca that

b. Crit@ I.think

ca that

s@ self=

c. Prima before

ca that

parlava he.spoke

d. Vuliti you.want

chi that

freva fever

n’ from=

cci there=

è is.ind

av’ has.ind

scesa (Nap.) descended

a to

aju I.have.ind

chiusu closed

vaju I.go.ind

iu? (Sic.) I

scì? (Bar.) go.inf a the

porta (Cos.) door

3 On the partial replacement of the imperfect subjunctive with the imperfect indicative in centralnorthern varieties of peninsular Spanish, see Fernández-Ordóñez (2010: 95–104).

The development of the southern subjunctive

29

The situation in the modern dialects, however, is much more complex than this overly simplified yet widespread view (cf. Rohlfs 1968: }559; 1969: }681; Cordin 1997: 89–90). First, until the latter half of the 19th century the subjunctive still constituted a robust and productive morphological category of the verb (Loporcaro 1999a: 70–73; Ledgeway 2009a: 501–3), as the following representative pre-19th-century Neapolitan examples illustrate: (12)

iuve helps.sbjv

a. Se vuole chesta acqua te if you.want this water you= sanare (Nap., BagniR 329; 13th c.)4 heal.inf b. determinaro they.resolved

puro also

che that

fosse it.was.sbjv

c. conveniva che lo desse it.required that it= he.give.sbjv omo (Nap., De Rosa 12r.6; 15th c.) man

a to

a to

abattuto (Nap., LDT 258.9; 14th c.) pulled.down ccualeche some

d. dicendono al deto eletto che venesse (Nap., CDT 5–6; 16th c.) saying.3pl to.the foresaid elected.one that he.came.sbjv e. aspettammo che se facciano we.wait that selves =make.sbjv pacionielle (Nap., Basile 80; 17th c.) kids f. è it.is

nnecessario necessary

che that

lo it=

granne big

li the

ddica (Nap., Oliva 70.54; 18th c.) I.say.sbjv

By the modern period, however, the present subjunctive had fallen entirely out of use, except for some lexical residues (cf. managgia! ‘damn it!’ < male ‘evil’ + aggia ‘have.3sg.sbjv’) and the 3rd person of non-first-conjugation verbs in some dialects of central Salento (Mancarella 1998: 99–117; Loporcaro 1997: 345–6; Bertocci and Damonte 2007), although even here the indicative is progressively replacing these limited uses of the subjunctive among younger speakers (Damonte 2006). 4 This chapter uses the following abbreviations: BagniR = M. Pelaez, ‘Un nuovo testo dei Bagni di Pozzuoli in volgare napoletano’, Studj romanzi 19 (1928): 47-134. Basile = Michele Rak (ed.), Giambattista Basile, Lo cunto de li cunti (Milan: Garzanti, 1986). CDT = P. Bianchi et al., ‘Cronaca dei tumulti del 1585’, in I’ te vurria parlà: storia della lingua a Napoli e in Campania (Naples: Tullio Pironti, 1993), 246–53. De Rosa = V. Formentin (ed.), Loise De Rosa, Ricordi (Rome: Salerno, 1998). LDT = N. De Blasi (ed.), Libro de la destructione de Troya: volgarizzamento napoletano trecentesco da Guido delle Colonne (Rome: Bonacci, 1986). Oliva = F. Oliva, ‘Aminta’, in C. C. Perrone (ed.), Francesco Oliva: opere napoletane (Roma: Bulzoni, 1977), 1–97. RDF = D. McArthur (ed.), ‘ “Il Romanzo di Francia”: une version du “Libro di Fioravante”, édité d’après le manuscrit unique conservé à la Bibliothèque nationale’ (doctoral thesis, University of Paris).

30

Adam Ledgeway and Alessandra Lombardi

Second, in the dialects of the extreme south, which make only limited use of the infinitive (Calabrese 1993; Ledgeway 1998; Lombardi 1997; 1998), the demise of the subjunctive affects not only the present but also the imperfect, which are both systematically substituted by the present indicative: (13) Ti sciali yourself= you.enjoy

/ /

scialavi you.enjoyed

ma that

’i them=

guardi (Ctz.) you.watch.ind

Finally, in the dialects of the upper south, by contrast, the imperfect subjunctive has not been entirely eradicated, but continues to be employed, alongside its frequent conditional uses, with its original subjunctive value. Rohlfs (1968: 301), for example, tells us that in this area ‘the present subjunctive has largely been lost, and is replaced by the indicative or by the imperfect subjunctive’. Although there is some truth in Rohlfs’s observations, they are still far off the mark. For instance, it is not the case that where we would expect a present subjunctive in Italian, we indiscriminately find either a present indicative or imperfect subjunctive in the dialects. Similarly, although these dialects do have an imperfect subjunctive verb form, it too is frequently replaced by the imperfect indicative. Rather, as argued in Ledgeway (2003: }2.4; 2009a: 503–11; 2009b: 10–11; cf. Leone 1995: 38–41), the imperfect subjunctive survives as a highly marked form licensed in just two contexts. The first of these concerns root and embedded jussive clauses (14a–d), although even here the corresponding indicative forms are not unknown (14e,f ). (14)

a. Dijje tell=him

che that

b. Nisciuno nobody

se self=

c. Dunacci give=him

l’ordine the.order

d. Dicci tell=him

chi that

cce there=

jesse! (Abr.) he.went.sbjv

muvesse! (Nap.) moved.sbjv.3sg

si self=

ca that

trasissa! (Cos.) he.entered.sbjv

mittissi he.put.sbjv

dintra! (Sic.) inside

e. Cumannajeno a uno ca turnava ’a miez’ â terra (Nap.) they.ordered to one that he.returned.ind to middle to.the Earth f. Diciaccìllu tell=to.him=it

ca that

trasa he.enter.ind

i the

panni! (Cos.) washing

The other bastion of subjunctive usage regards volitional contexts, where the indicative is robustly attested whenever the main clause predicate is in the present (15a), alternates freely with the subjunctive when the matrix predicate is in the past (15b), but is excluded whenever the matrix predicate itself occurs in an irrealis form such as the conditional (15c) in accordance with a rule of modal attraction between matrix and embedded predicates (Ledgeway 2009a: 507; cf. also Giorgi 2009: 1855–6).

The development of the southern subjunctive (15) a. Vuogliu I.want

ca that

ti spusi (Cos.) yourself= you.marry.ind

b. Vulìa I.wanted

ca that

ti spusave/spusasse (Cos.) yourself= you.married.ind/sbjv

c. Vulesse/Vulerra I.should.like

ca that

31

ti spusasse/*spusave (Cos.) yourself= you.married.sbjv/ind

In summary, the subjunctive has disappeared entirely from the dialects of the extreme south, while in those of the upper south the original imperfect paradigm survives as a generalized irrealis modal form in two highly restricted contexts. As a consequence, the morphological opposition between the indicative and subjunctive has been largely or entirely neutralized in the south today. 2.2.2 Dual complementizer systems The dialects of southern Italy are traditionally reported to display a Greek-style dual complementizer system.5 Accordingly, these dialects distinguish between two complementizers: ca on the one hand, and a second complementizer variously realized as che, chi, cu, or (m)u/(m)i/(m)a on the other. While the former heads clauses selected by declarative and epistemic predicates (henceforth the ‘indicative’ complementizer), the latter is employed after predicates such as volitionals that typically characterize the state or events of their complements as unrealized at the time of speaking (henceforth the ‘irrealis’ complementizer). For example, consider the following minimal pairs where the declarative vs irrealis reading of the complement clause is marked by the differing shape of the C(omplementizer) head and not in the morphological realization of the verbal dependent which occurs in the indicative in all cases (cf. also (3)):6 (16)

a. m a domm@ ditt@ ka me= has one said that ‘I was told that you are coming’ b. e it.is

mmeʎʎ@ better

ke that

ve (Colledimacine (CH), Abruzzo) come.ind.2sg

vie come.ind.2sg

tu you

5 Calabrese (1993); Lombardi (1997; 1998); Ledgeway (1998; 2004; 2005; 2006; 2007; 2012b); Damonte (2002; 2005; 2006; 2010); Roberts and Roussou (2003: 88–97); Manzini and Savoia (2005, i.455–501, 650–76); Vecchio (2010). 6 As argued extensively in Ledgeway (2004; 2005), we take the opposition in the complementizer system to involve a morphological distinction, according to which from an underlying underspecified form [k-] the differing indicative and irrealis surface forms are derived by checking specific modal features in different C-head positions. In the case of southern Calabrian and Messinese, the relevant opposition presumably involves a morpholexical distinction (e.g. Cal. ca vs mi) with a differential spell-out of different feature bundles at PF.

32

Adam Ledgeway and Alessandra Lombardi

(17) a. m Onn@ me= they.have b. vujj@ I.want (18)

k@ that

a. Ti dicu you= I.tell b. Ti rissi you= I.told

d@tt@ said

ka that

ve (Guglionesi (CB), Molise) he.comes.ind

vi come.ind.2sg

ca that

staju stand.ind.1sg

’u that

vai go.ind.2sg

jendu (Melicucco (RC), Calabria) going rá there

Syntactically, the two complementizers lexicalize different positions within the C-domain (Ledgeway 2004; 2005), as witnessed by their respective positions in relation to topicalized and focalized constituents in the following Salentino examples. In particular, the indicative complementizer lexicalizes the highest complementizer position Force, from where it precedes topicalized and focalized elements (19a), whereas the irrealis complementizer follows these same elements inasmuch as it lexicalizes the lowest complementizer position Fin (19b). (19)

a. Aggiu I.have

tittu [Force said

ca [TopP/FocP ca

la Lia the Lia

/ /

crai] [TP tomorrow

ene]] (Sal.) comes

b. Oyyu [TopP/FocP lu libbru / crai] [Fin cu [TP lu kkatta lu Mariu]] (Sal.) I.want the book / tomorrow cu it= buys the Mariu Unsurprisingly, in view of the demise of the subjunctive in the south, some have maintained that the morphological distinction in the complementizer system might have contributed to the weakening of the original modal distinction on the verb (Tekavčić 1980: ii.446; Vincent 1997b: 176). However, as argued in Ledgeway (2009a: }12.2.23; 2009b; 2012b), in most dialects of the upper south and Sicily (though not in dialects of the extreme south, including in the Sicilian province of Messina) the relevant distinction in the complementizer system has also been lost following the widespread generalization, from as early as the 19th century in some areas (Ledgeway 2009a: 501–2), of either the original indicative (20)–(22) or irrealis complementizer (23)–(25) to both clause types; witness the following minimal pairs from a variety of southern dialects (cf. also (3a,b) above):7 (20)

a. m@ sO me= they.are b. e it.is

mmeʎʎ@ better

dditt@ said

ka that

vi@ (Vastogirardi (IS), Molise) you.come.ind

ka that

iss@ he

ve comes.ind

7 For detailed discussion of the demise of the dual complementizer systems in the dialects of the upper south, see Ledgeway (2012b).

The development of the southern subjunctive (21)

a. m Onn@ me= they.have b. vOʎʎ@ I.want

(22)

ven@ (Ruvo di Puglia (BA), Puglia) he.comes.ind

ven@ he.comes.ind ka that

ven@ (Belmonte Mezzagno (PA), Sicily) he.comes.ind

ka that

tu you

ven come.ind.2sg

a. m enn@ me= they.have

ritt@ said

k@ that

vve (Monteroduni (IS), Molise) he.comes.ind

k@ that

vvje you.come.ind

a. m anu me= they.have b. e it.is

(25)

ka that

mmeɟɟu better

b. vOʎʎ@ I.want (24)

ka that

a. m@ ˈrissIru me= they.told b. e it.is

(23)

ðitt@ said

mmi@ɟɟu better

a. m an@ me= they.have b. vOɟɟ@ I.want

k@ that

33

ðittu said

ki that

vˈvi@nIsI (Laino Borgo (CS), Calabria) you.come.ind

ki that

ttu you

ˈvi@nIsI come.ind.2sg

ditt@ said tu you

k@ that

veŋ@ (Aidone (EN), Sicily) he.comes.ind

veŋ@ come.ind.2sg

2.2.3 Summary of findings In the dialects of the south the original morphological opposition between indicative and subjunctive has been seriously eroded both at the level of the clause and at the level of the sentence through the levelling of formal distinctions in the complementizer and the verb, respectively. More specifically, in dialects of the extreme south there still operates a robust distinction in the complementizer system, whereas the verb has lost all traces of subjunctive marking (cf. 18,19). In dialects of the upper south, by contrast, the complementizer now generally presents an invariant form and the verb, apart from two restricted contexts in conjunction with the erstwhile imperfect subjunctive (cf. 14,15), also presents the generalization of the indicative (cf. 20–25). It would thus appear that, unlike the dialects of the extreme south, those of the upper south have all but lost all marking of the original indicative/subjunctive opposition. These results are summarized in Table 2.1.

34

Adam Ledgeway and Alessandra Lombardi

TABLE 2.1. Morphological marking of indicative/subjunctive Indicative/subjunctive modal distinction

Dialects of upper south Dialects of extreme south

Complementizer

Verb

 þ

(+) 

2.3 Syntactic marking of the indicative/subjunctive opposition in the south From the above discussion, it appears legitimate to conclude that in the dialects of the upper south, at least, the indicative/subjunctive contrast has been entirely eradicated from the relevant grammars. In this section we shall demonstrate that closer scrutiny of these dialects reveals how such a conclusion does not accurately reflect the facts, insofar as other reflexes of this traditional modal contrast surface at the level of syntax in the differential positions occupied by the verb and the complementizer. 2.3.1 Verb movement Following the seminal work of Cinque (1999) building on the earlier Split-Infl Hypothesis of Pollock (1989) and Belletti (1990), both in turn influenced by the pioneering observations of Emonds (1978), Infl/T is now commonly interpreted as a general label for the rich inflectional area of the sentential core made up of a series of distinct functional projections. Consequently, the semantically related adverb and auxiliary contrast in Italian examples like (26) can now be understood as distinct positional reflexes (namely, specifier and head positions) of the same underlying functional projection. (26)

a. Gianni [AspHabitual [Spec Gianni

solitamente] usually

b. Gianni [AspHabitual [Spec Ø] Gianni

soleva was.wont

Ø

[v-VP fumava]] (It.) smoked

[v-VP

fumare]] (It.) smoke.inf

The various adverbs and their associated functional projections making up the inflectional core of the clause immediately above the v-VP complex can be broadly divided into two ‘spaces’, here termed the Higher Adverb Space (HAS) and the Lower Adverb Space (LAS), as represented schematically in (27): (27)

( . . . [C that) [HAS SUBJ Adv . . . [LAS Adv . . . [v-VP SUBJ Vlexical OBJ]]](])

Broadly speaking, the HAS comprises modal functional projections variously spelled out by speaker-oriented adverbs marking such categories as evidential and epistemic modality (e.g. apparentemente ‘apparently’), whereas the LAS principally includes

35

The development of the southern subjunctive

aspectual functional projections lexicalized by adverbial classes such as perfective sempre ‘always’, with temporal projections and associated adverbs (e.g. TPast-allora ‘then’, TAnterior-già ‘already’) variously scattered across both spaces (28). (28)

[HAS Ida apparentemente allora [LAS diceva già Ida apparently then said already sempre [v-VP diceva bugie]]] (It.) always lies ‘Ida was apparently in those days already always telling lies’

Now, in contrast to English, Romance is standardly described as exhibiting overt verb movement, as superficially revealed by English/Italian contrasts like (29a,b) where the Italian verb chiama, but not the English calls, has vacated the VP to move to a position to the left of the lower VP-adverb spesso/often: (29)

a. Michael

[VP often calls] (Eng.)

b. Michele chiama

[VP spesso chiama] (It.)

Armed with our assumptions about a universal fixed hierarchy of adverb positions and corresponding functional projections, we can now establish a much more nuanced interpretation of Romance verb movement than these familiar broadbrush treatments which assume tout court all varieties to pattern uniformly in exhibiting overt movement (cf. Cinque 1999; Ledgeway 2012a: }4.3.2; in press b). Thus, although it is true that the Romance finite lexical verb leaves its base position to vacate the v-VP complex (witness its position to the left of manner adverbs like bien/ bene/bine ‘well’ situated at the right margin of the LAS), it raises to different functional projections within the T-domain across Romance, as illustrated by its differential position with respect to distinct adverb classes in (30).8 (30)

[V-Medial [LAS . . .

[HAS . . . a. Elle

connaît peut-être *connaît

b. Lei

*conosce forse

c. Ea

*ştie

d. Ella *conoce

déjà

conosce già

[VP V . . . *connaît

]]]]

la

recette (Fr.)

?conosce la

ricetta (It.) reţeta

(Ro.)

receta

(Sp.)

poate

?ştie

deja

ştie

tal vez

?conoce

ya

conoce

la

e. Idda *canuscia forze ?canuscia ggià canuscia ’a ricetta (Cos.) she knows perhaps knows already knows the recipe(=the) ‘Perhaps she already knows the recipe’

8 On the different extent of Romance verb movement, see Lois (1989); Pollock (1989); Belletti (1990: 44–5); Kayne (1991); Cinque (1999: 152); Cornilescu (2000: 89–92); Alboiu and Motapanyane (2000: 22–4); Tortora (2002; 2010); Zagona (2002: 162–24, 168–70); Ledgeway and Lombardi (2005: 103–6); Fedele (2010); Schifano (2011); Ledgeway (2012a: }4.3.2; in press b).

36

Adam Ledgeway and Alessandra Lombardi

For example, in French (30a) the finite verb raises to the highest available position within the HAS (Rowlett 2007: 106–7), whereas in Romanian, Spanish,9 and Calabrian (and other southern dialects), the verb typically raises only as far as the LAS (30c–e).10 Italian represents an intermediate case (30b), where the finite lexical verb targets a clause-medial position sandwiched between the HAS and the LAS (Cinque 1999: 31, 110–11, 180 n. 80; Tortora 2002; Ledgeway and Lombardi 2005: 87–8). Thus, taking Cosentino as broadly representative of all southern dialects (see further the data discussed in Ledgeway and Lombardi 2005), we take verb movement in the south to be typically rather low, on a par with Romanian and Spanish.11 In this regard, it is interesting to note that verb movement in the dialects of southern Italy

9 Paola Benincà (p.c.) observes that in the case of Spanish the apparently low position of the verb might be a superficial effect of intrinsically quantificational adverbs such as ya ‘already’ and siempre ‘always’ raising to a left-peripheral non-contrastive focus position (cf. also Garzonio and Poletto’s (2011) recent analysis of Marchigiano dialects, and their comparative remarks on Spanish and Catalan). Some evidence to this effect might be that, whereas in southern Italian dialects and Romanian the low position of the verb and the adverb can be directly proven inter alia through the existence of clitic interpolation structures (namely, cl + Adv + V; cf. also discussion of (34a,b) below), such interpolation structures are absent from Spanish. Of course, while the left-peripheral-focused analysis might be appropriate for Spanish adverbs such as ya and siempre, it is not clear how this same analysis will carry over to non-quantificational lower adverbs. Also problematic are Spanish and Catalan examples like (i.a,b) where, assuming the existence of a single left-peripheral non-contrastive focus position SpecIFocP (Ledgeway 2010: 41–4; Cruschina 2012: 117), here lexicalized by a preverbal quantifier subject, the preverbal adverb in these examples can only be understood to occur in situ with the verb occurring in the LAS (pace Garzonio and Poletto 2011).

(i)

a. [IFocP Nadie nobody

[HAS

[LAS siempre always

llega [v-VP llega tarde]]]] (Sp.) arrives late

[HAS [LAS encara dubta [v-VP dubta qui b. Si [IFocP algú if somebody still doubts who ‘If someone still doesn’t know who to vote for’

votar]]]] (Cat.) vote.inf

Identical considerations carry over to Spanish indirect interrogative structures like (ii), where the fronted wh-item cuánto, an exponent of the lowest specifier position in the C-domain (Rivero 1978; Cruschina 2012: ch. 4) or the highest position in the I-domain (Goodall 1991; Zagona 2002: 247–8), necessarily places the preverbal adverb ya in the LAS (although Paola Benincà points out the ungrammaticality of similar sequences in the speech of other Spanish speakers, perhaps indicating a degree of dialectal variation): (ii)

llevo [v-VP llevo en México]]]]] (Sp.) Preguntaron [ForceP que [IFocP cuánto [HAS [LAS ya they.asked that how.much already I.spend in Mexico ‘They asked how long I’ve already been in Mexico’

Moreover, assuming default raising of the adverb to a left-peripheral non-contrastive focus position for Spanish, but not for Italian, for example, incorrectly presupposes that the unmarked pragmatic interpretation of such adverbs is different in the two languages (focused in the former but not the latter; cf. Cruschina 2012: 64–6, 173, 219). This is clearly an area which requires future research; in what follows, we shall simply assume, following Lois (1989), Cinque (1999), and Ledgeway and Lombardi (2005), that Spanish exhibits low verb movement. 10 Lois (1989); Cinque (1999: 152); Ledgeway and Lombardi (2005: 86–9, 102 n. 12); Monachesi (2005: 178); D’Alessandro (2010: 35–6). According to Cyrino (2010); verb movement in Brazilian Portuguese also appears to target the LAS (in Cyrino’s analysis, the lower T2 position above adverbs like bem ‘well’). 11 An apparent exception is Neapolitan, where the finite lexical verb appears to raise to a clause-medial position as in Italian (Ledgeway 2009a: 777–83).

The development of the southern subjunctive

37

displays an interesting asymmetry between irrealis and non-irrealis clauses: whereas the verb raises to a position in the LAS in both root and indicative clauses (31a), in irrealis clauses the verb exceptionally targets the highest available position within the HAS (31b), a property shared, not by chance, with Romanian (32a,b) which also exhibits reduced subjunctive morphology and a dual complementizer alternation. (31)

a. (Dicianu they.say b. Vunonu they.want

(32)

a. (Spun they.say

ca) that ca that

[HASLello Lello [HASLello Lello

că) [HAS that

[LASsempe always fatica works.ind

[LAS mereu always

fatica [v-VP fatica]]] (Cos.) works.ind [LASsempe *fatica [v-VP fatica]]] always

munceşte [v-VP munceşte]]] (Ro.) he.works.ind

b. Vor să [HAS muncească [LAS mereu *muncească[v-VP muncească]]] they.want that he.works.sbjv always The parallel between southern Italian dialects and Romanian illustrated here is revealing in that it demonstrates how the Cosentino verb, despite lacking any formal subjunctive marking, nonetheless mimics the syntax of the Romanian verb which, at least in the 3rd person, also marks a formal modal distinction on the verb. By the same token, the Cosentino examples in (33) highlight how the formally ambiguous verb form faticassa, which can function both as a conditional (33a) and as a subjunctive (33b), raises to low and high positions in accordance with its respective indicative/irrealis modal readings. This observation underlines how V-movement is ultimately licensed, not by morphological distinctions, but by the differing modal interpretations of the verb alone. (33)

a. (Pienzu ca) [HAS Lello *faticassa [LAS sempe faticassa [v-VP faticassa]]] I.think that Lello always would.work

(Cos.)

b. Vulìa ca [HAS Lello faticassa [LAS sempe*faticassa[v-VP faticassa]]] (Cos.) I.wanted that Lello worked.sbjv always The exceptionally high position of the verb in irrealis clauses is further substantiated by clitic–adverb interpolation structures (Ledgeway and Lombardi 2005). In particular, in root and indicative clauses (34a) the low position of the verb is directly evidenced by the fact that complement clitics, which presumably raise to a functional head in the HAS, can be separated from the verb by an intervening LAS adverb. Crucially, though, in irrealis clauses adverb interpolation is invariably excluded, since the verb must raise to the HAS together with its associated clitic(s) (34b). (34)

a. (Pienzu ca) [HAS un mi [LAS mai chiamassa [v-VP chiamassa]]] (Cos.) I.think that not me= ever he’d.call ‘(I think that) he’d never call me’

38

Adam Ledgeway and Alessandra Lombardi b. Vulessa ca [HAS un mi chiamassa [LAS mai *chiamassa I’d.like that not me=he.call.sbjv ever [v-VP chiamassa]]] (Cos.) ‘I’d like him never to call me’

This novel dimension of Romance verb movement, which is not peculiar to Cosentino but is widely found among the dialects of the South (35),12 including those of the Extreme South which otherwise maintain a modal opposition in the shape of complementizer (36),13 force us to conclude that the indicative/subjunctive distinction is still very much alive in the dialects of southern Italy, albeit marked at the level of syntax through verb movement, rather than morphologically through the traditional indicative/subjunctive contrast. (35)

a. (Penze ca) [HAS ji li [LAS (cande) sembre cande) [v-VP cande ssi he.thinks that I them= I.sing.ind always I.sing.ind these canzune]]] (Arielli (CH), Abruzzo) songs ‘(He thinks that) I always sing these songs’ b. Si ssa canzone ji piace, [HAS ocche le cande [LAS sembre if this song to.him= pleases opt.mrk it= he.sing.ind always (*le cande) [v-VP cande]]] ‘If he likes this song, then may he always sing it’

(36)

a. (Tice ca) [HAS l’Anna [LAS (u sapìa) già u sapìa [v-VP sapìa]]] (Sal.) he.says that the.Anna it= knew.ind already it= knew ‘(He says that) Anna already knew’ b. Speru cu [HAS u sape I.hope that it= he.knows ‘I hope that he already knows’

[LAS già already

(*u sape) [v-VP sape]]] it= he.knows

Indeed, it has recently been shown in Ledgeway (in press a) that in the dialects of northern Salento exceptional V-raising in irrealis clauses under cu-drop systematically

12 For further discussion, see Hart (2006: }2), D’Alessandro and Ledgeway (2010: 2053–6), Ledgeway (2009b: 12–14; 2012b), and Taylor (in preparation). 13 Following ideas in Alexiadou and Anagnostopoulou (1998), we take high V-movement in varieties like Salentino to check the EPP feature on T, thereby explaining the lack of a preverbal subject position in these dialects (viz. [FinP cu [HAS (*SUBJ) {V+T} [LAS . . . V]]]). This assumption is not necessary for southern Calabrian and Messinese dialects if reflexes of modo lexicalize T (Ledgeway 1998; Damonte 2010), rather than Fin, since preverbal subjects can be assumed to lexicalize SpecTP to the immediate left of modo (viz. [FinP Ø [HAS SUBJ modoT V [LAS . . . V]]]). In these varieties irrealis marking in the T-domain is therefore expressed discontinuously through the combination of a reflex of modo, which can be assumed to realize syncretically a number of the highest functional modal projections of the HAS, and the verb, which raises to the highest available position immediately below modo (cf. also Taylor, in preparation). Together through this discontinuous complex, {modo+V} in the T-domain can, in turn, license and render visible a marked irrealis feature on Fin in the C-domain.

The development of the southern subjunctive

39

correlates with consonantal lengthening of the initial consonant of the embedded verb (37b). There it is maintained that this consonantal lengthening, the output of an erstwhile sandhi phenomenon (viz. raddoppiamento ( fono)sintattico) triggered by the once-present irrealis complementizer cu (37a), has been reanalysed synchronically as an innovative phonomorphological marker of irrealis modality, or better subjunctive, on the embedded T head hosting the verb. Significantly, such intrinsic consonantal lengthening of the verb is only found in the case of irrealis clauses, but, crucially, never on verbs in root or indicative complement clauses (37c). There thus follows a strict correlation between V-raising to T and consonantal lengthening, which in these particular varieties can be readily interpreted as a robust diagnostic for high V-movement to the HAS. (37)

a. No sta not prog sentu la the

vvesciu I.see voci voice

l’ora [CP cu [HAS [s : ]entu [LAS ancora *sentu [v-VP the.hour that I.hear.ind again tua]]]] (Maruggio (TA), N.Salento) your

b. No sta vvesciu l’ora [HAS [s : ]entu [LAS ancora *sentu not prog I.see the.hour I.hear.ind again [v-VP sentu la voci tua]]] the voice your ‘I can’t wait to hear your voice again’ c. (Sta ddicu ca) [HAS [LAS ancora [s(* : )entu] [v-VP sentu la voci tua]]] prog I.say that still I.hear the voice your ‘(I claim that) I can still hear your voice’ 2.3.2 Dual complementizer systems In }2.2.2 above we observed how, in contrast to the dialects of the extreme south, the vast majority of the dialects of the upper south have now jettisoned the traditional morphological distinction in the complementizer system in favour of the generalization of one of the two original complementizers, be it ca (20)–(22) or che (23)–(25). It might then seem appropriate to treat these dialects on a par with standard Romance varieties like Italian, where, on the basis of the distribution of topics, the finite complementizer has been argued (Rizzi 1997) to lexicalize the highest head Force (38a), in contrast to non-finite prepositional complementizers which lexicalize the lower Fin position (38b). (38)

a. So I.know

[ForceP che that

[TopP la data] the date

[TP l’ ho sbagliata]] (It.) it= I.have mistaken

b. So [TopP la data] [FinP di [TP averla I.know the date of have.inf=it ‘The date, I know that I got it wrong’

sbagliata]] (It.) mistaken

40

Adam Ledgeway and Alessandra Lombardi

If correct, then the structural representation of the left periphery of the sentential pairs in examples like (20)–(25) will be identical, with the complementizer occupying the highest head in all cases. However, as argued in Ledgeway (2009b; 2012b), the loss of an overt morphological alternation in the complementizer system cannot a priori be assumed to imply a shift from a dual to a single complementizer system. By way of example, consider Cosentino, which belongs to the vast majority of upper southern dialects which have generalized the complementizer ca to both indicative clauses (39a) and irrealis clauses, regardless of whether the verb occurs in the indicative (39b) or the imperfect subjunctive (39c). (39)

a. A he.has b. Iddu he

dittu said

ca ca

sgarrati (Cos.) you.err

vo wants

ca ca

ci for.her=

fazzu I.do

’a the

spisa (Cos.) shopping

c. Diciacìllu ca mi priparassa ’i tell=her.it ca me= she.prepare.sbjv of ‘Tell her to prepare me something to eat’

mangià (Cos.) eat.inf

Now, alongside the canonical variety of Cosentino in (39), we must also recognize an archaic variety with a traditional ca/chi alternation in the complementizer system (Ledgeway 2000: 284 n. 6; 2009b: }3.1), albeit subject to significant diagenerational variation, inasmuch as the irrealis complementizer chi is only found among older speakers, and even among these is now normally replaced by the more frequent ca: (40)

a. Un not

pienzu I.think

ca ca

vi you=

canuscia he.knows

buonu (archaic Cos.) well

b. Vulìa chi (/ca) m’ accumpagnassa a ra casa (archaic Cos.) I.wanted chi ca me= he.accompany.sbjv to the home ‘I wanted him to take me home’ Younger speakers generally only have a passive competence of this use of chi, preferring instead the undifferentiated ca in all contexts. There is, however, one exception to this generalization where all Cosentino speakers, regardless of age, employ the distinctive irrealis complementizer chi, namely in lexicalized optative expressions: (41)

a. Chi tti vija mpisu a nu chiaccu! (Cos.) chi you I.see.sbjv hanged to a rope ‘Would that I could see you hanged from a rope!’ b. Chi bbò bena chi want.sbjv come.inf ‘Let the worst happen!’

’a the

fine end

du of.the

munnu! (Cos.) world

The development of the southern subjunctive

41

In view of the fossilized nature of such expressions (that include otherwise exceptional relics of the now defunct present subjunctive) which invariably require chi in the speech of all speakers, there can be little doubt that the original irrealis complementizer in Cosentino must have been chi. However, even if the use of chi remains entirely optional, albeit subject to diagenerational restrictions, syntactically its distribution is not so free. More specifically, the use of chi is intimately tied to residual indicative/subjunctive distinctions on the verb (cf. discussion of (14)–(16)): whenever the embedded verb occurs in the indicative (42a,b), ca proves obligatory, but when the verb occurs in the subjunctive (42c,d), ca and chi freely alternate, at least syntactically if not diastratically. (42)

a. Vuogliu I.want

ca /*chi ca /chi

mi me=

risbigli you.wake.ind

b. Vulìa I.wanted

ca /*chi ca /chi

mi me=

risbigliavi you.woke.ind

a at

ri the

sette (Cos.) seven

c. Vulìa I.wanted

ca /chi ca /chi

risbigliassa you.woke.sbjv

a at

ri the

sette (Cos.) seven

d. Vulissa/Vulerra I’d.like

mi me= ca /chi ca /chi

mi me=

a at

ri the

risbigliassa you.woke.sbjv

a at

sette (Cos.) seven

ri the

sette (Cos.) seven

Ca thus functions as an all-purpose complementizer compatible with both morphologically indicative and subjunctive forms, whereas chi represents a modally marked complementizer only licensed by a morphologically subjunctive verb form. These facts strongly suggest that the driving force behind the generalization of ca is to be sought in the gradual replacement of the subjunctive with the indicative: following the pattern of declarative complement clauses which are introduced by ca containing a verb in the present indicative, ca would have first spread, albeit still alongside chi, to irrealis complement clauses whose verb was also in the present indicative. Subsequently, the complementizer ca would have generalized to irrealis clauses containing a verb in the past tense, initially still freely alternating between indicative and subjunctive, giving rise to the present-day situation where ca has all but replaced chi.14 Translating these observations in terms of the fine structure of the left periphery sketched in (19a,b) for the Salentino dual complementizer system, it is logical to assume that the modally undifferentiated complementizer ca occupies the highest complementizer position Force (43a), while the modally marked complementizer chi lexicalizes the lowest Fin head (43b), the position standardly assumed to be responsible for licensing marked modal values. 14 We thank Mair Parry (p.c.) for suggesting to us this interpretation of the effects of the loss of the subjunctive on the generalization of the complementizer ca.

42 (43)

Adam Ledgeway and Alessandra Lombardi a. [Force ca

[TopP/FocP . . .

[Fin . . .

[TP Vind/sbjv

b. [Force . . .

[TopP/FocP . . .

[Fin chi

[TP Vsbjv/*ind

. . . ]]]] . . . ]]]]

Tempting though this analysis may seem, there is reason to believe that it is not an accurate reflection of the structure of the Cosentino complementizer system, nor of other southern dialects that have generalized a single complementizer. In particular, if all-purpose ca invariably lexicalized Force, then we should expect it to be able to precede all fronted topicalized and focalized constituents. By way of example, consider the following Cosentino sentences containing hanging topics ((a) examples), scenesetting adverbs ((b) examples), clitic left-dislocations ((c) examples), contrastive foci ((d) examples), and fronted quantifiers ((e) examples): (44)

a. Dicia he.says

ca ca

Mariu Mariu

’u him=

unn’ not

parra speaks

cchiù no.more

nullu (Cos.) nobody

b. Dicia ca doppu ’a guerra papà unn’ era cchiù ’u stessu (Cos.) he.says ca after the war dad not was more the same

(45)

c. Dicia he.says

ca ca

’e mulingiane the aubergines

d. Dicia he.says

ca ca

Cicciu Cicciu

e. Dicia he.says

ca ca

’ngunacosa something

sgarra errs

ca ca

Mariu Mariu

b. *Vuonnu they.want

ca ca

doppu ’a after the

ca ca

(nun not

addi he.must

a. *Vuogliu I.want

c. *Vulissa I’d.like

unn’ not

unn’ not

vo he.wants

cchiù (Cos.) more

Micheluzzu) (Cos.) Micheluzzu fa (Cos.) do.inf

’u parra him= speaks festa party

’u canciellu the gate it=

’e them

cchiù more

nullu (Cos.) nobody

mangiamu (Cos.) we.eat

’u they.fix.sbjv

cunzassaru (Cos.)

d. *Vulìa ca na bella sagna priperassa (nun nu purpettune) (Cos.) I.wanted ca a nice lasagna he.prepared.sbjv not a meatloaf e. *Vulimu we.want

ca ca

’ngunacosa something

ni us=

duna (Cos.) he.gives

Despite expectations, ca may only precede topicalized and focalized constituents when it introduces a declarative clause (44a–e), but not when it introduces an irrealis complement (45a–e), a pattern readily replicated in other dialects with a single complementizer such as the Sicilian dialect of Mussomeli (CL; S. Cruschina, p.c.), illustrated in (46) and (47), and Neapolitan (cf. Ledgeway 2009a: }24.1.3):

The development of the southern subjunctive (46)

43

a. Dissi ca dopu pranzu si viniva a pigliari u café (Mussomeli (CL)) he.said ca after lunch self= came to take.inf the coffee b. Dissi ca cucuzzi un si nn’ a mai mangiatu (Mussomeli (CL)) he.said ca courgettes not self= of.it= he.has ever eaten c. Dissi ca nni tia tuppià (no nni Pina) (Mussomeli (CL)) he.said ca at your.place he.knocked not at Pina’s) d. Dissi he.said

(47)

ca ca

nuddu nobody

vitti he.saw

p’ in

a the

strata (Mussomeli (CL)) street

a. *Vuagliu ca dopu pranzu nni pigliamu u cafè (Mussomeli (CL)) I.want ca after lunch ourselves= we.take the coffee b. *Vulissi I’d.like

ca i cucuzzi si mangiassi (Mussomeli (CL)) ca the courgettes self.them= he.eat.sbjv

c. *Vuliva ca u pitrisinu ci chiantava (no l’ accia) (Mussomeli (CL)) I.wanted ca the parsley there= he.planted not the celery d. *Vuliamo we.want

ca nuddu ci veni a festa (Mussomeli (CL)) ca nobody there= comes to.the party

Significantly, the same behaviour is found with irrealis complements introduced by the overtly irrealis complementizer chi in archaic Cosentino: (48) a. *Vulissa chi ’u canciellu ’u cunzassaru (archaic Cos.) b. *Vulìa chi na bella sagna priperassa (nun nu purpettune) (archaic Cos.) Indeed, the only way to rescue irrealis examples like (45a–e) and (48a,b) is to resort to right dislocation (49a,b) and focus in situ (49c,d). (49)

a. b. c. d.

Vuonnu ca mangiamu doppu ’a festa (Cos.) Vulissa ca (/chi) ’u cunzassaru ’u canciellu (Cos.) Vulìa ca (/chi) priperassa na bella sagna (nun nu purpettune) (Cos.) Vulimu ca ni duna ’ngunacosa (Cos.)

In short, this asymmetry in the distribution of topicalized and focalized constituents in accordance with the differing declarative/irrealis values of ca, coupled with the distributional parallelism between chi and ca in its irrealis functions, provides strong proof that dialects with apparently a single complementizer may still operate a dual complementizer system below the surface. To be sure, we claim that dialects like Cosentino and Mussomelese have not lost the original distinction in the C-system, but continue to display a dual complementizer series. In particular, the distinction is no longer marked by a morphological opposition in the shape of the complementizer (namely, c- ) ca (ind) vs chi (sbjv)), but rather by a syntactic contrast in the differing positions targeted by the invariant complementizer ca within the C-space: in

44

Adam Ledgeway and Alessandra Lombardi

declarative contexts ca lexicalizes Force, whereas in irrealis contexts it lexicalizes Fin, hence being able to precede fronted topics and foci in the former but not the latter case.15 (50)

[ForceP caRealis [TopP Top [FocP Foc [FinP caIrrealis [TP . . . ]]]]]

These facts lead us to conclude that in its irrealis functions ca is an impostor, albeit a chi disguised as ca which, despite its phonomorphological appearance, retains all the syntactic properties of the original chi it has come to replace. In short, the apparent generalization of ca represents a straightforward case of relexicalization of the original complementizer chi which is now predominantly pronounced as ca.

2.4 Summary and conclusions The evidence reviewed in this chapter has highlighted how the dialects of the south have lost morphological marking for the indicative/subjunctive opposition either on the verb (dialects of extreme south) or on both the verb and in the complementizer system (dialects of the upper south). Although no longer transparent at the morphological level, we have demonstrated how this modal opposition is still very much alive in the syntax of these same dialects, where the relevant distinction manifests itself in the different positions lexicalized by the verb and the complementizer in the T- and C-domains, respectively. In particular, we have observed an asymmetrical distribution of the verb in declarative and irrealis contexts, respectively targeting low and high positions in the sentential core which, in turn, directly correlate with the placement of the (morphologically undifferentiated) complementizer merged in Force in the former case and Fin in the latter, as illustrated in (51a,b): (51) a. [Force caRealis [TopP . . . [FocP . . . [Fin Ø b. [Force Ø

[HAS . . .

Adv [LAS Adv VInd

[v-vP V]]]]]]]

[TopP . . . [FocP . . . [Fin caIrrealis [HAS VInd/Sbjv Adv [LAS Adv VInd/Sbjv [v-vP V]]]]]]]

Following Chomsky’s (2007; 2008) ideas about feature transmission and inheritance between phase heads and their complements (cf. also Ouali 2008), we propose that the modal interpretation of the clause is ultimately licensed by the feature specification of the phasal C-head. Assuming a strong correlation between morphological marking and syntactic licensing (cf. Baker’s (1985) Mirror Principle and Bobaljik’s (2002) Rich Agreement Hypothesis), we take the marked modal reading of irrealis clauses in the standard Romance languages to be licensed by the verb alone which, by virtue of its rich subjunctive morphology, can readily make visible and license under long-distance Agree the relevant modal feature on Fin. In the dialects of the south, by contrast, the modally neutral morphology of the verb is such that the marked modal reading of irrealis clauses can no longer be licensed by a long-distance 15 For a possible explanation why the opposite order Topic/Focus + caSbjv is also ruled out, see Ledgeway (2009b: 19–20).

The development of the southern subjunctive

45

Agree relation between the verb and the Fin head, even when the latter displays a morphologically overt irrealis form as happens in the dialects of the extreme south. Rather, in the absence of distinct irrealis morphology, the verb-form in these varieties must raise to the highest available position within the sentential core, and hence to the left of all higher adverbs, from where it can enter into a strictly local agreement configuration with the complementizer in Fin to license the marked irrealis interpretation (cf. 51b). In short, this local agreement configuration represents an overt reflex of the mechanism of feature transmission between C and T, highlighting how a case of morphological loss has given rise to a case of syntactic gain through the exploitation of different verb and complementizer positions to license the relevant modal opposition. Ultimately, these facts can be used to construct a descriptive typology of irrealis modal marking across Romance based on the interaction of feature licensing and transmission across the C- and T-domains, as sketched in (52). (52) C-domain

T-domain

Morphology Syntax

Morphology Syntax

a*

+

–





b*





–

–

c

–

–

+

–

d

–

–

–

+

e

–

+

–

+

f

–

+

+

–

g

+

+

–

+

h

+

+

+

–

i

+

+

(+)

+

Two generalizations immediately emerge from (52). Firstly, if the complementizer system presents a morphological distinction (e.g. modern dialects of extreme south (52g), early dialects of upper south (52h), Romanian (52i)), then it also presents a syntactic (namely, positional) distinction (hence the unattested option (52a)), but not vice versa; witness the modern dialects of the upper south (52e) that typically display a morphologically undifferentiated form of the complementizer capable of lexicalizing distinct positions within the C-domain. Secondly, in all Romance varieties the verb must always be modally marked within the T-domain (hence the unattested option (52b)), with the result that morphological and syntactic marking,

46

Adam Ledgeway and Alessandra Lombardi

manifested respectively in subjunctive morphology (standard Romance: option (52c)) and high V-movement (southern Italian dialects: options (52e,g)), are almost in complementary distribution, the exception here being Romanian (52i), which still exhibits residual subjunctive morphology in the 3rd person (cf. 10b) as well as high V-movement (cf. 32b).16 Presumably, the Romanian pattern represents an intermediate diachronic stage, where subjunctive morphology is now too weak to independently license the relevant irrealis modal feature on Fin which has to be made visible through overt V-movement. What then of the other potential options (viz. d,f,h) predicted in (52)? Beginning with (52d), in which irrealis modality is marked solely on the verb, a potential manifestation of this option is found in Italian indirect interrogatives, where the embedded verb may appear either in the subjunctive (53a) or the indicative (53b). Significantly, though, in the latter case where the indicative verb may be thought of as mimicking its more formal subjunctive counterpart, only a postverbal subject is possible. Although alternative analyses are available (cf. Giorgi and Pianesi 1997: 235–47), we might take this to imply that, on a par with the modern dialects of the south, the subjunctive reading of the morphologically indicative verb form can only be licensed by exceptionally raising to the highest available position within the T-domain. (53)

a. Mi me=

domando I.ask

cosa what

(faccia) do.sbjv

b. Mi domando cosa fa me= I.ask what does ‘I wonder what Mario is doing’

Mario Mario

Mario Mario

(faccia) (It.) do.sbjv

(*fa) (It.) does

As for option (52f ), in which irrealis modality is marked both morphologically on the verb and syntactically through lexicalization of the lowest C-head, a potential candidate comes from widespread Romance C(omplementizer)-drop structures like those examined in (8a,b) which, at least according to one analysis (e.g. Poletto 2001), instantiate residual V2 structures with raising of the modally marked verb directly to the Fin head. Turning finally to (52h), in which irrealis modality is marked in the C-domain both morphologically through a dual complementizer system and syntactically through the concomitant differential lexicalization of the two complementizer positions, but only morphologically in the T-domain through overt subjunctive morphology on the verb, this option would appear to be instantiated by the early dialects of southern Italy (54a,b). Despite the presence of a dual complementizer system, the 16

Also to be included here are some dialects of central Salento where, in addition to high V-movement in irrealis clauses, it was noted in }2.2.1, following Damonte (2006) and Bertocci and Damonte (2007), that distinct subjunctive forms are preserved in the 3rd person of non-first-conjugation verbs.

The development of the southern subjunctive

47

rich subjunctive marking on the verb in these varieties is sufficient to license the relevant irrealis feature on Fin under long-distance Agree without the need for high V-movement, witness the intervening adverb between the irrealis complementizer che and the subjunctive verb facesse in the old Neapolitan example (54b). (54)

a. scrisse a li frati [ForceP ca [TP volleva isorare lo she.wrote to the brothers that she.wanted marry.inf the fillo]](ONap., RDF 76) son b. iscrisse he.wrote trovare find.inf

a la matre [ . . . ] [FinP che [TP icontinente facesse to the mother that at.once she.made.sbjv la croix de Dio]] (ONap., RDF 10–11) the cross of God

3 Perfective auxiliation in Italo-Romance The complementarity of historical and modern cross-dialectal evidence* M I C H E L E L O PO R C A R O

3.1 Introduction In this chapter, I intend to show how the study of modern and medieval dialects may complement each other as sources of evidence, allowing us to draw one coherent picture. The empirical evidence comes from perfective auxiliation, with focus on Italo-Romance dialects, but cast within the broader Romance picture. I shall show on the one hand that we need modern dialects, for which we have access to richer and potentially more complete data sets, in order to better understand and classify medieval varieties. I shall also show, on the other hand, that we may sometimes need medieval varieties in order to complement the descriptive generalizations that inspection of modern dialects allows us to draw. * I thank the editors for inviting me to offer this chapter, as a sign of esteem and friendship, to an outstanding expert in the kind of issues I discuss here. Previous versions of part of this research have been presented at the Universities of Campobasso (SLE Summer School, July 2007), Leiden (May 2011), and Roma Tre (October 2011). Whenever unreferenced, dialect data come from my own field notes. They are reported in a simplified IPA transcription, CC instead of [C:], and with stress marked only on nonparoxytonic words. Fieldwork in Agnone in July 2007, the results of which are reported in }3.4.1, was organized in the framework of the yearly field trip of the Romanisches Seminar of the University of Zürich—in cooperation with the Phonetics Laboratory and the Phonogrammarchiv—and funded by the Philosophische Fakultät of the University of Zürich (Exkursionskredit), which I gratefully acknowledge. I am also indebted to Paolo D’Achille, Rachele Delucchi, Vittorio Formentin, and Lorenza Pescia for comments on a previous draft (usual disclaimers apply), as well as to the Agnonese speakers who helped me with the organization of the fieldwork and by sharing their native intuitions with me: first and foremost Domenico Meo, as well as Paola Cerimele and Nicola Mastronardi. The corresponding audio files are available at the Phonogrammarchiv.

Perfective auxiliation in Italo-Romance

49

The chapter is organized as follows: in }3.2 I shall briefly take stock of previous work on perfective auxiliation, introducing in particular some questions that remain open given the factual knowledge that has been accumulated so far on cross-dialectal variation over space and time in this structural domain. In }3.3 I shall move on to consider two selected examples from medieval Italo-Romance: I will show that old Florentine and old Romanesco allow us to fill in some gaps that were left in Loporcaro’s (2007) classification of auxiliation systems based solely on modern dialects. Especially in relation to old Romanesco, however, the intrinsic limitations of a corpus-based study of the kind one is forced to carry out on medieval varieties become evident. This brings us to }3.4, where the old Romanesco auxiliary selection rule is corroborated by comparative evidence from two modern dialects: the Molisano variety of Agnone and the dialect of Picerno, a Gallo-Italic colony of Lucania. Finally, }3.5 draws some lessons of methodology from this exercise in combining medieval and modern, synchronic and diachronic evidence.

3.2 The basics: perfective auxiliation and the unaccusative hypothesis The cornerstone of the analysis of perfective auxiliation in Romance—as in other Indo-European languages—is David Perlmutter’s unaccusative hypothesis, which assumes a bipartition of intransitive predicates, as illustrated in (1a,b) with just a few initial items from the lists of one-place intransitives in Perlmutter (1978: 162): (1) a. unaccusative: P{2} b. unergative:

P{1}

burn, fall, drop, sink . . . (It. bruciare, cadere, affondare . . . ) work, play, speak, talk . . . (It. lavorare, giocare, parlare . . . )

As shown in the structural representations in (2a,b)—which adopt the formalism of Relational Grammar, the theory in which Perlmutter’s hypothesis was originally cast—the core argument is an underlying direct object (initial 2) in subclass (1a), while it is an underlying (or initial) subject in subclass (1b):1 (2) a.

2

P

1

P

1 P Cho La nave è affondata ‘The boat sank’

b.

1 1

P

P

1

= subject

Cho

2

= direct object

Gianni ha lavorato ‘John worked’

P = predicate Cho = chômeur

1 I provide in (1) some notational aid on the right-hand side, to help the reader decode the structural representations which adopt Davies and Rosen’s (1988) ‘Predicate Union’ formalism. In-depth familiarity with those structural representations is not assumed here: all relevant structural notions are introduced explicitly.

50

Michele Loporcaro

Syntactic unaccusativity correlates with semantic properties of the predicate (telicity) and the core argument (non-agentivity), while the symmetrical correlations hold for unergative predicates, as shown in (3a,b) (cf. Rosen 1984): (3)

a. La nave è affondata in/*per tre ore ‘The boat sank in/*for three hours’

unaccusative (¼ telic; argument patient)

b. Gianni ha lavorato per/*in tre ore ‘John worked for/*in three hours’

unergative (¼ non-telic; argument agent)

In spite of this statistical correlation, unaccusativity cannot be reduced to semantic properties such as telicity, despite what is claimed by many authors in one form or other (e.g. Aranovich 2003; Cennamo 2002; Cennamo and Sorace 2007; Bentley 2006; Dowty 1991; Mithun 1991; Sorace 2000; van Hout 2004; Van Valin 1990); this issue cannot be considered further here owing to limitations of space (cf. Loporcaro 2011a; 2011b). Rather, I shall concentrate in what follows on purely syntactic aspects of perfective auxiliation. The next ingredient needed for this purpose is Rosen’s (1997) definition of auxiliary predicate: (4)

Auxiliary (Rosen 1997: 112) Auxiliaries are a lexically designated closed class of verbs whose defining property is that they inherit a 1.

Based on these assumptions—the structural contrast in (2a,b) and the definition of auxiliary as a predicate that inherits its subject—the generalization in (5) now suggests itself, which accounts for the distribution of the perfective auxiliaries essere ‘be’ and avere ‘have’ in Italian: (5)

Perfective auxiliation in Italian (Perlmutter 1989: 81) The perfective auxiliary is essere iff: the final 1 is a 2. Otherwise the perfective auxiliary is avere.

The same generalization also accounts straightforwardly for the selection of essere in different kinds of reflexive construction:2 (6)

1,2

P

1

P

direct transitive reflexive

1 P Cho Maria si è lavata ‘Mary washed herself’

2 In the structural representations in (6) and following, reflexiveness corresponds to multi-attachment (cf. Rosen [1981] 1988; 1982): as shown by the notation 1,2, one and the same nominal (the final subject) bears two distinct grammatical relations in the same stratum. Due to general properties of Italian (and the Romance languages), multi-attachment is resolved in the first available stratum, which results in the appearance of the reflexive clitic marker si on the verb.

Perfective auxiliation in Italo-Romance

51

In addition to direct transitive reflexives, as exemplified in (6), there are several types that should be kept distinct, such as retroherent unaccusatives (exemplified in (7); cf. Rosen [1981] 1988; 1982) and the two structural subtypes of indirect reflexives in (8a,b): (7)

2 P 2,1 P 1 P P Cho 1 Maria si è divertita ‘Mary had fun’

unaccusative with retroherent advancement

(8)

a.

b.

1,3 P 1,2 P 1 P 1 P Cho Maria si è risposta ‘Mary answered herself’

1,3 P 2 1,2 P Cho 1 P Cho 1 P Cho Cho Maria si è lavata le mani ‘Mary washed her hands’

indirect reflexives

Monadic indirect reflexives (8a) are reflexive constructions which involve only one nominal core argument, as in unaccusatives (2a), whereas in indirect transitive/dyadic reflexives (8b) two distinct nominals bear the subject and the DO relations initially, as in plain transitive constructions. Based on these premises, the line of research initiated by Perlmutter (1978; 1989) (cf. La Fauci 1988; 1989; 1992; 2004; Rosen 1997; Loporcaro 1999b; 2001; 2007; 2011a; 2011b; Formentin 2001; 2002; Vecchio 2006; Paciaroni 2009) has investigated Romance auxiliary selection across time and space, based on fieldwork on modern dialects and on the study of ancient texts from various areas. In my own work on the topic, I have elaborated on the binary contrast in (1)–(3).3 The starting observation is that reflexive constructions sometimes pattern uniformly with unaccusatives (say, for auxiliary selection or past participle agreement in standard Italian), but sometimes part ways with respect to each other, so that one subset goes with unaccusatives, the other with unergatives. If this is the case, an implicational pattern is observed, which allows one to turn that binary (unaccusative vs unergative) contrast into a scale, as shown in (9) with data from Logudorese Sardinian:4

As Perlmutter (1978: 186) stressed, the unergative/unaccusative contrast corresponds to what in linguistic typology is called active vs inactive alignment: this justifies the lumping together (for present purposes) of unergative and transitive constructions. 4 The divide in (9) separates the constructions requiring auxiliary 'essere ‘be’ (henceforth aux E) from those requiring auxiliary 'aere ‘have’ (henceforth aux H). The data stem from the variety of Bonorva (province of Sassari). 3

52 (9)

Michele Loporcaro a. maria Mary

es is

unaccusative

b. maria Mary

z refl

est is

c. maria Mary

z refl

el is

besti:äa dressed.fsg

dir. tr. reflexive

d. maria Mary

z refl

er is

rispOsta answered.fsg

indir. unerg. refl. aux E

e. maria

z

a

ssamuna:äu

Mary f. maria Mary

refl z refl

has washed.msg the hands a bbuffa:äu una bbirreɖɖa antipassive has drunk.msg a small beer

g. maria Mary

a has

palti:äa left.fsg arreneɣa:äa gotten.angry.fsg

mmaniɣa:äu eaten.msg

(za (the

zal man:Os

minestra) soup)

retroherent

indir. trans. refl. aux H

unergative/transitive

The auxiliation contrast documented by Logudorese is one out of a series of possible options. Before commenting on the empirically observed auxiliation systems, however, let me add that the scale is motivated deductively on structural grounds. Though the following description employs the vocabulary of Relational Grammar, it seems to me that any robust theory of syntax should be able to model this set of data, recognizing a scale such as (9a–g) in order to make sense of the variation in Romance perfective auxiliation. First, among the constructions displaying a reflexive clitic on the predicate, retroherent unaccusatives (7) are closest to plain unaccusatives (2a): both have just an argumental DO. In retroherent unaccusatives, then, reflexivization occurs as the product of a purely syntactic process with no semantic import, since the final subject bears no other argumental grammatical relation. Thus, it is to be expected that in case reflexives do not all pattern alike, retroherent unaccusatives will be those most likely to behave like plain unaccusatives. Next come direct transitive reflexives (9c), whose final subject is also an argumental DO, namely that initialized by the transitive predicate. Again, if reflexives part ways for some syntactic properties such as auxiliation (or past participle agreement or the like), in case the relevant rule is sensitive to initial objecthood, plain transitive reflexives will pattern together with unaccusatives. If, however, the rule is sensitive to initial subjecthood, they will pattern together with the following constructions in accordance with the scale (9d–g). The next type of reflexive construction, indirect intransitive reflexives (9d), share with the preceding one (9c) the property of being monoargumental (or monadic), in that just one nominal (the final subject) is initialized by the predicate, not only as a subject but also in another grammatical relation. However, contrary to (9c), this is not the DO

53

Perfective auxiliation in Italo-Romance

relation.5 But still, they are monadic, and this makes a difference with respect to the next class (9e), biargumental reflexives, where in addition to the final subject there is a distinct nominal that bears the DO relation argumentally. Thus, if a split occurs, separating reflexive constructions for some syntactic property, dyadic reflexives can be expected to behave rather like transitives (9g), with which they share the property of having a subject which is distinct from the initial DO.6 The predictions reviewed here translate into an implicational scale, which appears to be borne out by the empirical data from variation in auxiliary selection across Romance. This variation has been modelled, in Loporcaro (2001; 2007; cf. also Ledgeway 2012a: 321), as shown—with modifications—in (10) (where the different contexts listed in (9) are displayed horizontally): (10)

unacc. retr. a. b. c. d. e. f.

Italian Sardinian ? ? Leccese Spanish

E E E E E

INACTIVE reflexive dir. ind. unerg. trans.

indir. trans.

ACTIVE trans./ unerg. H H H H H H

The question marks in (10c,d) are due to the fact that no modern (Italo-)Romance dialect among those surveyed in the cited literature displays those options. Furthermore, it must be added that (10) is simplified as to both the manifestation and the number of the contrasts involved: first, it does not take into account mixed systems, which display either free variation of ‘be’/‘have’ or person-driven distribution of the two auxiliaries; secondly, it displays only binary (like Italian) or unary options (like Spanish), and does not include triple-auxiliation systems. Both mixed and triple systems, as shown in Loporcaro (2007), can be plotted onto (a slightly modified version of) the scale in (10). This becomes possible, however, only on the assumption that the main syntactic organizing principle of perfective auxiliation, for both mixed and nonmixed systems, is the mapping of entire auxiliary paradigms onto the different clause types listed in (9), whereas (for mixed systems) alternation according to person and/or

5 In indirect reflexive constructions, reflexive 1,3 ! 1,2 advancement is assumed to take place (cf. La Fauci 1988: 82–8; 1994: 53–60 for discussion and supporting empirical evidence). Initialization (cf. Dubinsky 1985) consists in the attribution of a grammatical relation (and at the same time a thematic role) by a predicate to its argument(s). 6 In between, antipassives ((9f ); cf. La Fauci 1984b: 224ff. for a definition) are another subtype of dyadic reflexives, which however behave like indirect transitive reflexives for the syntactic properties investigated here and thus will not be discussed separately.

54

Michele Loporcaro

free variation have to be accounted for in the morphology (an idea developed recently by Corbett (2013) in his study of the different kinds of lexeme split).7 As discussed in Loporcaro (2007), this assumption is at odds with accounts of auxiliary selection which—following Kayne (1993)—focus on exactly which auxiliary, whether ‘have’ or ‘be’, appears in which person and try to account for the occurrence of aux E/H in each cell of the paradigm in syntactic terms. This is still a current procedure in the literature on Romance syntax, including recent work such as D’Alessandro and Roberts (2010), Legendre (2010), and Manzini and Savoia (2005; 2007). Under this alternative approach, however, no reduction of the huge amount of variation in auxiliary selection options seems to be possible. To convey an impression of how wildly perfective auxiliation can vary, I reproduce in (11) the inventory provided by Manzini and Savoia (2005, ii.721; 2007: 225–6): (11)

Manzini and Savoia (2005: ii.721; 2007: 225–6): A(vere), E(ssere), / = free variation, – = (syntactically relevant) contrast, ~ = neither entirely / nor entirely – Dialect/Verb person

1sg

2sg

3sg

1pl

2pl

3pl

a. Poggio Imperiale etc.

E

E

E

E

E

A/E

b. Roccasicura

A/E

E

E

E

E

E

c. Capracotta

A

E

E

A/E

A/E

E

d. Gallo Matese

A/E

E

E

E

E

A/E

e. Monteroduni

A/E

E

E

E

E

A

f. Vastogirardi

A/A–E E

E

E

E

A/E

g. Colledimacine etc.

E

E

A–E

E

E

A–E

h. S. Benedetto del T. etc E

E

A

E

E

A

i. Viticuso

A/E

E

A

E

E

A

j. Sassinoro

A/E

E

A/E

A/E

A/E

A/E

k. Secinaro

A/E

E

A–E

A/E

A/E

A/E

l. Agnone

A/E

E

A–E

A

A

A/E

m. Bisceglie

E

E

A

A

A

A

n. Ruvo

A/E

E

A/E

A

A

A

o. Popoli

E

E

A–E

A

A

A

p. Padula

A/E

E

A–E

A

A

A

7

cf. also Bentley and Eythórsson (2001: 71); Bentley (2003: 93), for a similar view.

55

Perfective auxiliation in Italo-Romance

q. Molfetta

A/E

E

A

A

A

A

r. Monteparano

A

A/E

A/E

A

A

A

s. Minervino M.

A/E

A/E

A–E

A/E

A/E

A/E

t. Morcone

A

A

E

A

A

A

u. S. Giorgio d. Sannio

A

A

E~A

A

A

A

v. Frigento

A

A

E–A

A

A

A

w. Gravina

A/E

A

E~A

A

A

A

x. Viguzzolo

A–E/A A–E/A A–E/A A

A

A

y. Aliano

A

A

A

A

A–E

A

This inventory is not exhaustive, and I could mention several more dialects with different distributions. The point is that, though there are admittedly some tendencies (e.g. for aux H to occur more frequently in the 3sg and 3pl), this distribution is basically unconstrained. As argued in Loporcaro (2007: 185), each of the six cells in the paradigm is an independent variable to which one out of three values can be assigned (H, E, or free variation H/E): this generates 36 = 729 logically conceivable options, if all clause types display the same auxiliary choice. If, furthermore, a doubleauxiliation option is brought into the picture (and it must, since such contrasts do occur in mixed systems too), then one gets 7292 = 531,441. Finally, considering tripleauxiliation systems too would generate 387,420,489 different possible configurations. While it is inconceivable for this empirical variability to be directly encoded into structural (syntactic) contrasts in a one-to-one correspondence—the unavoidable consequences of accounts of mixed systems à la Kayne (1993)—in Loporcaro (2007) it has been shown that the scale in (10) allows one to reduce mixed and triple-auxiliation systems to a relatively small number of syntactically different types. All such systems described so far, in fact, just like binary and unary ones, happen to display just the contrasts that are predicted given (non-discontinuous distributions along) the scale in (10). The scheme in (12) summarizes the triple-auxiliation options reported to date: (12)

Triple-auxiliation systems so far described: Loporcaro (1999b; 2007); Paciaroni (2009)

unacc. a. Trentino b. Alt.2, Zag.2, Mac. c. Castrovillarese

E E E

reflexive retr. dir. trans. indir. unerg. E/H

indir. trans. E/H

E/H

trans./ unerg. H H H

56

Michele Loporcaro

The table is somewhat simplified, as the dialects in (12a,b) display variation of aux H/E, for the clause types indicated there, not throughout the paradigm but only in some persons. But the point remains that those systems, under whatever superficial form (i.e. whatever the specific mix of perfective auxiliaries occurring in the six cells of the person/number paradigm), always seem to represent compromises between two of the systems in (10a–f): (12a), the kind of triple system found in northern Italy, is a compromise between the Leccese type (10e)—which occurs also in several Romansh varieties and many Alpine dialects of northern Veneto—and the standard Italian type (10a). On the other hand (12b), the kind of triple system found in the two upper subdivisions of central-southern Italy, namely Area Mediana (exemplified here with Maceratese and Zagarolese) and the upper south (Altamurano), is a compromise between the standard Italian type (10a) and the Sardinian type (10b), the latter occurring in many dialects of the centre-south. Finally (12c), occurring in the northern Calabrian dialect of Castrovillari, is a mix between the Leccese and the Sardinian types.8 In view of these observations, we now return to the two gaps which appear in (10c,d). I will show that joint investigation of medieval texts and present-day dialect variation, including mixed and non-binary auxiliation systems, provides data that enable us to fill in those gaps.

3.3 Diachrony: intersections between modern dialect comparison and historical data The synchronic classification in (10) intertwines with the modelling of language change. For instance, the Spanish type (10f ) is more recent than several others higher up in the scheme, in the sense that older documented stages of Spanish (as well as other Romance varieties which today display the same unary auxiliation pattern, such as Catalan, Portuguese, or Sicilian) still had two auxiliaries.9 So the question is how exactly the Spanish type came into being (in Castilian, but also in many dialects of Calabria and Lucania and in many mixed systems of central-southern Italy too). In other words, one can ask whether the stages the auxiliation system passed through can be modelled by means of a structural scale such as that in (10). (For old Spanish, Loporcaro’s (2011a) reappraisal of the data reported by Aranovich (2003) shows that this is indeed the case.) This is one sense in which the study of synchrony and diachrony of perfective auxiliation may interact.

Such triple systems are sometimes stable, like Castrovillarese (as described by Pace 1993–94: 129–30), but often represent a delicate transitional stage, within a speech community in which they coexist with binary options: this is the reason for the numerical index in Alt2 and Zag2 in (12b). 9 On the occurrence of ser as a perfective auxiliary in old Castilian see Aranovich (2003): see also Huber (1933: 221) on auxiliary seer in old Portuguese, La Fauci (1992: 202, 207) on essiri in old Sicilian, and Pérez Saldanya (1998: 214) on ésser in old Catalan. 8

57

Perfective auxiliation in Italo-Romance

Another, and quite obvious, possible intersection is that medieval data may display options which just happen not to be documented in present-day dialects. This applies to the gaps in (10c,d). While the prediction was that also the options (10c,d) may occur, in the dataset on which Loporcaro (2007) was based—encompassing modern dialects only—there were no examples of any variety which distinguished between retroherent unaccusatives and direct transitive reflexives (10c), nor of varieties cutting off the scale between direct transitive and indirect unergative reflexives (10d). 3.3.1 Old Florentine As shown in Loporcaro (2011a: 77–81), medieval Italo-Romance allows us to fill in both gaps (albeit with some caveats). An example of (10d)—with avere systematically selected with all reflexives, including direct transitive ones ((13c))—is provided by La Fauci’s (2004) analysis of Dante’s Florentine (in the modern standard language, aux E occurs in the successors of (13a–e) vs aux H in (13f)): (13) a. Fuggito è ogni augel che ’l caldo segue (Rime c 27) ‘All birds that follow the heat have fled’

unaccusative

b. io mi sarei brusciato (Inf. xvi 49) ‘I would have burnt myself ’

retroherent

c. la donna che [ . . . ] ci s’hae mostrata (Vn xxxviii 3) ‘the woman that showed herself to us’

direct aux H trans. refl.

aux E

d. Quand’io m’ebbi dintorno alquanto visto (Inf. xxxii 40) indirect unerg. refl. ‘After I looked around me for a while’ e. poscia che tanti/speculi fatti s’ha (Par. xxix 143–4) ‘after he has made (for himself) so many mirrors’ f. Ma i Provenzai che fecer contra lui/non hanno riso (Par. vi 130–31) ‘But Provençals, who slandered him, did not laugh’

indirect trans. refl.

unergative

Other studies, not limited to the corpus provided by Dante’s oeuvre, confirm this picture partially, in that they document persistence of avere with direct transitive reflexives (13c) well into the 14th century, although those studies also report variability.10 10 The thorny philological issue arises here as to what extent Dante’s corpus can be considered to mirror reliably the syntactic competence of a Florentine speaker born in 1265. The problem is that Dante’s oeuvre did not reach us in autographic form (as opposed to e.g. Boccaccio’s). Thus, any text transmitted under his name is liable to have been modified linguistically in the course of later manuscript tradition. Yet, in our specific case (perfective auxiliation), since Dante’s corpus displays a situation which does not correspond to later Florentine (see (16b,c) below), it is highly unlikely that later copyists would have modified data on auxiliation to make them diverge from their own variety, and to make them seem as consistent as they appear in (13). In other words, the most economical hypothesis is that the contrast (13a,b) vs. (13c–f) has to be attributed to Dante’s Florentine.

58

Michele Loporcaro

According to Jezek (2010: 122), avere, the only option occurring in Dante in this context, is used in other Florentine texts in just a minority of occurrences, as exemplified in (14b,c) (from Jezek 2010: 120–22):11 (14) a. Allora si sono accordati intra anbo due li servi (Trist. Ricc. c. 67, p. 127, 1–2); ‘Then the two servants agreed’ b. Questi due cavalieri s’aveano lungamente amato (Novellino 33, 4–5); ‘These two knights had long loved each other’ e quando s’ebbono insieme salutate (Bono Giamboni, Libro lxiii 7); ‘and when they greeted each other’ c. Bito [ . . . ] s’avea messa la più ricca roba di vaio (Novellino 96, 14–15) ‘Bito had put on the richest clothes of vair’ Jezek’s (2010: 122) results, on the other hand, confirm that retroherent unaccusatives (as exemplified in (14a)), invariably select essere: mentre l’it. mod. non consente mai l’ausiliazione con avere nelle costruzioni con V pronominale, che richiedono sempre essere, in it. ant. questo è possibile [ . . . ], anche se meno frequente, in casi in cui il clitico ha effettivamente valore riflessivo (diretto o indiretto) [ . . . ]. L’ausiliare avere non è possibile invece in costruzioni pronominali inaccusative (ad es. *s’hanno vergognati).’ [while modern Italian never allows have-auxiliation in pronominal verb constructions, which always require be, in old Italian this is possible [ . . . ], albeit less frequent, in cases where the clitic is really reflexive (direct or indirect) [ . . . ]. Auxiliary have is not possible, however, in unaccusative pronominal constructions.]

Thus, also in this description, there is a sharp divide cutting across reflexive constructions between (13b) and (13c), except that the contrast is between categorical selection of aux E in unaccusatives (plain and retroherent) vs avere/essere variation in direct transitive reflexives—which, by the way, results in a transitional stage with three auxiliation options to be added to the triple-auxiliation systems inventoried in (12). A similar result is reached by Rindler Schjerve and Kratschmer’s (1990) study, which provides quantifications, unlike the other authors just mentioned: (15)

‘old Italian’ a. unaccusatives b. retroherent c. dir. trans. refl. dir. trans. reciprocal d. indir. unerg. e. indir. trans. refl. f. unerg./trans.

aux E always 96% 76% 60% no never

aux H never

(Rindler Schjerve and Kratschmer 1990) Total occurrences: 474 (of which 381 = (15b))

data 89% always

11 As customary in studies of Romance syntax, reciprocal and reflexive si-constructions are considered together, since the contrast in Romance is merely one of interpretation.

Perfective auxiliation in Italo-Romance

59

Their corpus is mostly Tuscan, but with exceptions, as some texts are from other regions, which surely brings in some noise. This cannot be checked conclusively, however, as they just give overall percentages with some selected examples, rather than all the data illustrating the auxiliation options synthesized in (15a–f). Concretely, the figure of 96% for the occurrence of aux E with retroherent unaccusatives in (15b) is suspicious for old Florentine, since it implies that in the complementary 4% aux H occurred—a result which is at odds with that of all studies which limit their scope to Florentine (cf. (13b), (14a), and the quotation from Jezek 2010: 122 above).12 Be that as it may, their conclusions confirm that aux essere seems to be spreading into the next construction (direct transitive reflexives and reciprocals), as shown by the figures in (15c).13 As an illustration of the further diachronic development of auxiliation in Florentine, consider now the following examples (discussed by Menoni 1982: 68–75) from a late 14th-century text (Franco Sacchetti’s Trecentonovelle, 1392–97, ed. Pernicone 1946): (16) a. la corda del pesce essendosi rotta, il pesce se n’era ito giù per lo Po ccxvi 47–8; ‘the string of the fish having broken, the fish had swum down the Po river’ b. essendosi sciolta da uno arpione, cominciò a fuggire per la via clix 20–21; ‘having freed herself from a harpoon, she started to flee down the street’ il podestà e il capitano essendosi armati, . . . salirono a cavallo clix 120–22; ‘the podesta and the captain having armed themselves, [they] mounted their horses’ c. essendosi molto ben pettinato e assettata la chioma . . . , tolse la tovagliuola . . . ii 9–11; ‘having combed himself and having arranged his hair, he took the towel away’ d. e di questi cotali fu questo Populo d’Ancona, . . . che, avendosi recato nella mente d’acquistare una roba . . . , giammai non restava che veniva a effetto . . . clxii 23–7. ‘and such was this Populo d’Ancona, who, once he had decided to buy a robe, never gave up before succeeding in doing that’ Direct transitive reflexives (16b) now behave like retroherent unaccusatives (16a) in selecting aux E. This text also offers an example of aux E in a dyadic reflexive (16c), where Dante’s usage exclusively admits avere. But note that in this single example, as Menoni (1982: 75) remarks, this indirect reflexive is coordinated with a direct reflexive, which regularly requires essere in Sacchetti (16b). When no coordination is involved—as in (16d)—one still finds avere. Note also that this coordination effect cf. also Brambilla Ageno (1978: 364–73); Menoni (1982). The authors follow Vincent (1982: 96) in surmising that the spread of aux E may have been favoured by the fact that retroherent unaccusatives (15b), selecting aux E from the outset, accounted for the vast majority of the occurrences of reflexives. 12 13

60

Michele Loporcaro

bears witness to the authenticity of essere selection in direct reflexives in Sacchetti; otherwise they could not attract dyadic reflexives under coordination, if they had not switched to essere in the first place.14 What is beyond doubt is that retroherent unaccusatives (16a) categorically display aux E throughout the history of Florentine, while dyadic reflexives (16d) still had prevalently avere at the end of the 14th century. Auxiliary selection in the clause types in between (along the scale in (9) and (10)) seems to have been undergoing change at that time. Consequently, one might speculate that it is not by chance that Pietro Bembo, in his effort to establish a reference variety of literary Florentine (with his Prose della volgar lingua, 1525; cf. Dionisotti 1966), exemplifies auxiliation with reflexives covering only the clear cases, namely aux E with retroherent unaccusatives (17a) vs aux H with dyadic reflexives (17b): (17)

a. La donna s’è doluta, Voi vi sete ramaricati, Coloro si sono ingegnati (iii 36) ‘The woman complained, You regretted, They did their best’ b. Messosi le mani ne’ capelli [Bocc., Dec. ii 8, 22] invece di dire Avendosi le mani ne’ capelli messe (iii 54) ‘[having] Put his hands in his hair instead of saying Having put his hands in his hair’

Note in passing that Avendosi le mani ne’ capelli messe in (17b) is Bembo’s own prose, not a quote from Boccaccio, which means that aux H with dyadic reflexives could still be actively used, at least in writing, in the 16th century.15 On the other hand, no examples of auxiliation with monadic reflexives with two grammatical relations (direct transitive and indirect unergative) are commented on in Bembo’s Prose, so that one might venture that he may have avoided taking a stance with regard to a slippery domain of usage that was being affected by ongoing change.16

14 A proviso has to be mentioned here, because the two extant manuscripts of Trecentonovelle date back to the 16th century (cf. Pernicone 1946: xxv). Thus, while the text reliably provides evidence of of 14thcentury Florentine grammatical features which were to be subsequently lost (e.g. selection of aux avere in (16d)), it may be that innovations (like selection of aux essere in (16c)) crept into the text in the process of copying (I thank Vittorio Formentin for bringing this point to my attention). 15 Whether this option was still grammatical in the spoken language at that time can hardly be inferred from Bembo’s own usage, since he is well known to remain faithful to 14th-century Florentine for ideological reasons. Note, however, that occurrence of aux avere with dyadic reflexives is documented in other contemporary authors, as witnessed by the following passage from a grammatical treatise published by Bembo’s opponent Lodovico Castelvetro in 1563 (ed. Motolese 2004: 195): (i) il ladro non s’haveva egli stesso avinta la corda al collo, ma il giustiziere ‘it was not the thief who had wound the rope around his own neck, but rather the hangman’

Note also that Castelvetro is not focusing on auxiliary selection here. Caution is required, though, as one must be careful not to judge 16th-century grammars by our modern standards, especially in relation to the requirements of descriptive exhaustiveness. Thus, for instance, Fortunio’s Regole grammaticali della volgar lingua, the first grammar of Italian to appear in print (1516; ed. Marazzini and Fornara 1999), does not mention perfective auxiliation at all. 16

Perfective auxiliation in Italo-Romance

61

Modern standard Italian differs from the auxiliation system prescribed by Bembo based on his 14th-century canonical authors. Three centuries later, Manzoni has the modern system,17 and it seems likely that in the spoken language, usage of aux E in dyadic reflexives was established by the early 17th century at least. For instance, in 1643 Benedetto Buommattei (Della lingua toscana, ed. Colombo 2007: 329) mentions, like Bembo (17b), the occurrence of aux H in dyadic reflexives, adducing examples from the same authority, i.e. Boccaccio, but adding a commentary that suggests that in his own usage aux E was the normal option instead: ‘il verbo avere si truova usato non poche volte in significazione d’essere’ [the verb ‘have’ is found not infrequently used instead of ‘be’]. After quoting examples from Boccaccio (Et avevasi recato il fanciullo in braccio ‘And he had scooped up the child’, vii 3, 35; avendosi l’anel di lei messo in bocca ‘having put her ring into his own mouth’, x 9, 106; e non avendoselo bene saputo legare ‘not having being able to tie it well’, vii 8, 13; et egli s’avesse molto messo il cappuccio innanzi agli occhi ‘and he had put his hood well in front of his eyes’, vii 5, 22), Buommattei comments: ‘dove si vede chiaramente che tutte stanno in luogo di [ . . . ] era, essendosi e fosse’ [where one sees clearly that they all (= those forms of have) stand for ‘was’, ‘being’, ‘were’].

While it still remains to carry out a quantitative diachronic study of exactly which stages the modern situation passed through to come into being, one (quite obvious) lesson from the above can already be drawn: since auxiliary essere with unaccusatives (plain and retroherent) and auxiliary avere with unergatives and transitives were stable throughout the history of Florentine, variation in the in-between categories, with shifting usage (aux H > aux E) over time as shown in (14)–(16), must have resulted in transitional stages with three auxiliation options. This shows 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. 3.3.2 Old Romanesco Returning to the scheme in (10), if Dante’s Florentine seems to fill the structural gap (10d), the other gap left over after scrutinizing modern dialects (at least those displaying binary non-mixed systems), viz. (10c), seems to correspond to old Romanesco as analysed by Formentin (2002). The documentation of this ancient variety, however, is considerably less rich than that available for Florentine. Formentin has taken into account all the available textual evidence (see Formentin 2002: 206–9 for 17

Cf. Brambilla Ageno (1978: 368), discussing examples like the following:

(i) a. dopo essersi . . . levate il vestito delle feste (I promessi sposi iii 58) ‘having taking off their feast-day clothes’ b. Affari intralciati . . . non se n’era mai trovati addosso tanti (xxiv 474) ‘Messed-up matters, he never had had so many’

62

Michele Loporcaro

the corpus) from the 13th century through to the early 16th. Yet he comes up with figures, summarized in (18), which are quite scanty, especially concerning the crucial constructions we are interested in: (18)

a. b. c. d. e. f.

unaccusatives retroherent dir. trans. refl. indir. unerg. indir. trans. refl. unergatives/transit.

aux E always 34 28 0 1 never

aux H never 1 1 1 15 always

Total old Romanesco (Formentin 2002) … 35 29 1 16 …

Compound tenses of unaccusative, unergative, and transitive predicates are well documented, and in those clause types auxiliation is the same as in modern standard Italian. As for reflexive constructions which lie between these two extremes, modern Romanesco is like the modern Florentine-based standard language (and has been such since the 17th century, as shown by Formentin 2002: 238), whereas the medieval variety was quite different: first of all, in dyadic reflexives avere was usually selected (as in old Florentine). Formentin (2002: 236–7) was able to collect 16 occurrences of this construction, 15 of which (one example in (19a)) had aux H, as against only one instance of aux E (19b): (19)

a. se l’haco saputo fà a modo lloro le leie questi hominacci riballi (Castelletti 183, ed. Ugolini 1982) ‘men, those bastards, knew how to make laws in a way that suited them’ b. me so cresa che me scannassi (Castelletti 259, ed. Ugolini 1982) ‘I thought he was going to kill me’

Furthermore, this isolated exception occurs in the very last document of nonTuscanized romanesco di prima fase ‘first-phase Romanesco’ (the Romanesco lines by the servant Perna in Castelletti’s comedy Le stravaganze d’amore, 1587). Thus, as Formentin (2002: 237) observes, this may well be a harbinger of impending change (note that Perna still has the more conservative option with avere alongside, (19a), the only one attested in medieval texts.) In retroherent and direct transitive reflexives, aux E is regularly selected, as exemplified in (20a,b), whereas aux H exceptionally occurs in just one case for each one of the two categories (Formentin 2002: 236–7):18 18 Contrary to (19b), these exceptions cannot be explained away with chronological arguments and thus represent a genuine residue (Formentin 2002: 237):

(i)

a. a’ te facta I(n)ga(n)nare alla toa se(n)sualità (S. Francesca Romana 160v1) ‘you’ve let your sensuality deceive you’ b. granne ora me aio penzato sopra la opera la quale intienni (Cronica xxvii 217) ‘I long thought over the deed you are envisioning’

Perfective auxiliation in Italo-Romance (20)

63

a. migliara de perzone se soco inninocchiate denanti da essa (Cronica xviii 558, ‘thousands of people kneed before her’ ed. Porta 1979) b. p(er) lo q(ua)le te si’ avenenata (S. Francesca rom. 160v1, ed. Vignuzzi 1992) ‘for which you have empoisoned yourself ’

In the case of indirect unergative reflexives, though, Formentin’s corpus yields just one relevant example: (21)

secu(n)do ch(e) se ào lassato (S. Francesca Romana 171 recto 16) ‘in compliance with the dispositions left over for oneself ’ (in one’s last will and testament)

This isolated occurrence seems to tell us that the divide between the two auxiliation options falls between direct and indirect reflexives (i.e. between (18c) and (18d)). Plotted onto the scale in (10), this yields the following result: (22)

unacc. a. b. c. d. e. f.

Italian Sardinian old Romanesco old Florentine Leccese Spanish

E E E E E

INACTIVE reflexive ind. indir. retr. dir. trans. unerg. trans.

ACTIVE trans./ unerg. H H H H H H

This was assumed without discussion in Loporcaro (2011a: 80), where the medieval Italo-Romance data that have been dealt with here in some detail were mentioned in passing in order to compare with them the situation observed in old Castilian. However, registering old Romanesco under (22c) does require discussion: given the limitations imposed by the corpus, and given the fact that the figures are low (and hence not statistically reliable), the only way to check whether Formentin’s reconstruction of this aspect of the syntax of old Romanesco can stand further investigation is to look for independent evidence that may confirm that the option (10c)/(22c)—with a divide between direct and indirect reflexives—does occur in other, and better-documented, Romance varieties.

3.4 Bringing the historical and modern cross-dialectal evidence together This is where inspection of modern dialects comes into play, since such supporting evidence is indeed found in at least two modern southern Italo-Romance varieties—that of Agnone (province of Isernia, Molise), which will be discussed in more detail in } 4.1, since the data have not previously been described, and the dialect of Picerno, a

64

Michele Loporcaro

Gallo-Italic enclave in the province of Potenza, Lucania, whose auxiliation rule has been described by Pescia (2011: 236–41; 2012). As it happens, neither of them displays a plain binary auxiliation option of the Italian or Sardinian kind, since both are mixed systems displaying free variation and/or person-driven auxiliary selection. In other words, they would have been unavailable for comparison with old Romanesco, under theoretical premises alternative to those summarized in }3.2. 3.4.1 Perfective auxiliation in Agnonese Agnonese is spoken in the town of Agnone (Upper Molise). There are several studies on this dialect (cf. Ziccardi 1910; Finarelli 2002–3), including a syntactic study specifically covering auxiliary selection, since Agnone is one of the points investigated in Manzini and Savoia’s (2005: ii.706–8) survey of Italo-Romance morphosyntax. This has already been seen in (11.l) above; the results of their field inquiries into the Agnonese perfective auxiliation system can be summarized as follows: (23) Agnone

1sg

2sg

3sg

1pl

2pl

3pl

A/E

E

A–E

A

A

A/E

As shown in (23), they report for this dialect a system with free variation in the 1sg and 3pl person, as well as a contrast (‘selection of essere or avere according to verbal class’: Manzini and Savoia 2007: 226) in the 3sg only. The data I collected myself on Agnonese in 2007 largely coincide with Manzini and Savoia’s. This free variation in the 1sg for all predicate classes is exemplified in (24): (24)

a. mad@meæ̯n@ so ppartiu̯t@/ai̯ partiu̯t@ ’subb@t@ ‘this morning I left early’

unaccusative

1sg

b. mad@meæ̯n@ m aj arr@ttsi@̯t@/m@ so arr@ttsi@̯t@ ’subb@t@ retroherent ‘this morning I got up early’ c. m@ so ss@kuɵ̯t@/m aj ass@kuɵ̯t@ ku ru makkatiu̯r@ ‘I dried myself (my face) with my handkerchief ’

dir. trans. reflexive

d. m a̯i rediu̯t@ mbatʧ@/ m@ so rrediu̯t@ mbatʧ@ ‘I’ve laughed at myself ’

indir. unerg. refl.

e. m aj ’ve:v@ta/m@ so v’ve:v@ta na bbuttiʎʎa de voi̯n@ ‘I drank a bottle of wine’

dyadic refl.

f. ai ̯ fatiji@̯t@/ffatiji@̯t@ prassia ‘I worked a lot’

unergative (= transitive)

In the 2sg all my informants use ʃi, a form of ‘be’, but they still recall that the corresponding form of ‘have’ was used until one or two generations ago, with all predicate types:

Perfective auxiliation in Italo-Romance (25) ʃi ppartiu̯t@/{e partiu̯t@ 'subb@t@ ‘You left early’

65

unaccusative 2sg [same for all remaining predicate classes, parallel to (24b–f)]

This is confirmed by the early 20th-century study by Ziccardi (1910: 435), whose transcribed texts contain occurrences such as the following: (26) tu jé ̣ ’ččís@ p@ híss@ ru v@tiéll@ ̣ kkju ĝĝráss@ (Ziccardi 1910: 435) ‘You slaughtered the fatter calf for him’ Unfortunately, his texts contain no 2sg instances of unaccusatives or reflexives, but it seems likely that the 2sg might have been another cell of the paradigm displaying free variation at that time. For the plural persons, my data diverge slightly from Manzini and Savoia’s: they report free variation in the 3pl, based on work with one informant (cf. Manzini and Savoia 2005: i, p. ix), whereas my informants divide into two groups. Some of them reject ‘be’ for all verbs in all plural persons, whereas other speakers (more innovative) display free variation in all plural persons, at least with some clause types. Let us, however, concentrate on the 3sg, which is where a syntactically relevant contrast persists, as shown in (23). It is indeed often the case that it is the 3sg that maintains a contrast even within systems with free variation in many (or most) persons of the paradigm. The relevant data provided for the 3sg by Manzini and Savoia (2005: ii.706–8) are as follows: (27) a. e mmə'niu tə/-a ‘S/He has come’ b. s e m'muɐssə/m'mɔssa ‘S/He has moved’ c. s e lla'vεtə ‘He has washed himself’

unaccusative

3sg

retroherent direct transitive reflexive

d.

?

indirect unergative reflexive

e.

?

dyadic reflexive

f. l a ca'mεtə/a dur'miu tə ‘He has called him/has slept’

unergative/transitive

Given the data set in (27), one can conclude that: (a) there is a contrast among different clause types selecting different auxiliation patterns in the compound perfect (aux E in (27a–c) vs aux H in (27f )); and (b) the Agnonese auxiliation system might be situated somewhere between the old Romanesco type (22c)—if it indeed exists—and the standard Italian one (22e). As highlighted by the boxes in (27d,e), however, there

66

Michele Loporcaro

are crucial gaps in the data, with the result that these are not sufficient to pin down the auxiliation rule of Agnonese within the framework schematized in (22). Once the different distribution of the ‘have’/‘be’ morphemes across persons has been factored out, one needs to ask whether its structural organizing principle (in terms of the sets of clause types contrasted via auxiliation) is like that of old Romanesco, Sardinian, or standard Italian. Published dialect texts do not help here, as the corpus they provide is too restricted. There are Agnonese texts available in Papanti (1875) and Ziccardi (1910), but they add up to no more than 800 words. More recently, a dissertation by Finarelli (2002–3: 55–181) includes transcribed dialect texts which span slightly over 10,000 words. A search of this corpus for auxiliated verb forms in the compound perfect yields the results synthesized in (28): (28)

a. Papanti (1875: 303–4): 1 occurrence (trans 3pl: m én@ fatt@ ‘they did to me’) b. Ziccardi (1910: 434–5): 7 occurrences (1 trans. 1sg, 1 trans. 2sg, 1 trans. 3sg; 3 unacc. 3sg, 1 antipassive 3sg) c. Finarelli (2002–3: 55–181): 14 occurrences (1 trans. 1sg, 1 trans. 2sg, 1 trans. 3sg; 1 unerg. 3sg; 1 retroherent 3sg, 1 retroherent 3pl; 1 unacc. 2sg, 7 unacc. 3sg)

As is immediately apparent, (28a–c) do not help us to fill in the gaps in (27d,e). A specifically designed inquiry is necessary, which is what I carried out during a 2007 field trip. Again, it transpires that Agnonese speakers divide into at least two subgroups as far as perfective auxiliation is concerned. The competence of the conservative informants, among the Agnonese speakers I have interviewed, is mirrored in (29.i), from which the second group of speakers (29.ii) diverges through the innovation in (29e): (29)

(i) Agnonese 1

(ii) Agnonese 2

a. ess e ppartiu ta/∗a partiu ta ‘S/He has left’

same

unaccusative

b. essa s e rrəttsiəta/∗s a rrəttsiəta/∗s a rrəttsiəta ‘She got up’

same

retroherent

c. essa s e/∗s a ssəkuɵ ta ku ru makkatiu rə ‘She dried herself with her handkerchief ’

same

direct trans. refl.

d. kella 'femməna s e kkutʃəniəta/s a kutʃətniətə essa seu la ‘That woman cooked for herself ’

same

indir. unerg. refl.

s a/s e H/E

indir. trans. refl.

same

transitive/unerg.

e. sɔrma s a missə/∗s e mmessa ru kappiəllə ‘My sister put on her hat’ f. flumoina j a messa/∗j e mmessa la onna ‘Filomena put him the skirt on’ (i.e ‘she’s the boss)

>

Perfective auxiliation in Italo-Romance

67

More conservative speakers have free variation of ‘have’/‘be’ in the 3sg, only with indirect unergative reflexives (29d), whereas ‘have’ is selected categorically not only with transitives/unergatives (29f) but also in dyadic reflexives (29e). More innovative speakers (29.ii), on the other hand, have extended free variation of ‘have’/‘be’ in the 3sg to dyadic reflexives. Given that the remaining clause types (29a–c) categorically select aux E among all speakers, we are faced with (two different subtypes of) a tripleauxiliation system, since the clause types classified in (9) above select three, rather than two, distinct series of auxiliaries (viz. 30a–c), contrasting with each other in the 3sg only: (30)

a. b. c.

1sg E/H E/H E/H

2sg E E E

3sg E E/H H

1pl H H H

2pl H H H

3pl H H H

Agnonese

The rules accounting for the distribution of perfective auxiliaries in the two varieties are as follows (where aux-i/ii/iii stand for the three options in (30a–c) respectively): (31)

Triple auxiliation in Agnonese 1 i. aux-i if the final 1 is: a P-initial 2 in the clause19 ii. aux-ii if the final 1 is: (a) a non P-initial 2, and (b) the first 2 in the clause. iii. aux-iii elsewhere.

(32)

Triple auxiliation in Agnonese 2 i. aux-i if the final 1 is: a P-initial 2 in the clause ii. aux-ii if the final 1 is: a non P-initial 2 in the clause iii. aux-iii elsewhere.

The two triple-auxiliation systems of Agnonese can now be added to the summary of such systems, thus expanding (12) into (33): (33)

Triple-auxiliation systems (revised)

unacc. a. b. c. d. e.

Trentino Alt.2, Zag.2, Mac. Castrovillarese Agnonese 1 Agnonese 2

E E E (30a) (30a)

retr.

reflexive dir. indir. trans. unerg. E/H

indir. trans. E/H

E/H (30b) (30b)

trans./ unerg. H H H (30c) (30c)

19 The notion P-initial x refers to a nominal bearing the x relation in the first stratum of the predicate sector of some predicate, the predicate sector being defined as the set of strata in which that predicate bears the P relation.

68

Michele Loporcaro

We are now in a position to realize that the crucial gaps (27d,e) in Manzini and Savoia’s (2005: ii.706-8) set of data obscured the fact that Agnonese has to be included among triple-auxiliation systems. Moreover, for our present discussion Agnonese, with both its varieties (33d,e), is of special importance, since it confirms that there are Italo-Romance dialects in which indirect unergative reflexives are kept distinct, in terms of auxiliation, from other monadic reflexives, as was the case in old Romanesco according to Formentin’s (2002) results. 3.4.2 Perfective auxiliation in Picernese A further similar case, confirming the structural plausibility of such a situation, has recently been revealed by Pescia’s (2011: 236–41; 2012) study of the syntax of the dialect of Picerno. Perfective auxiliation in Picernese is exemplified in (34): (34)

a. so ppar'tu/∗adə par'tu ‘I have left’ b. mə so ppən'du/∗m adə pən'du ‘I have repented’ c. mə so dda'va/∗m adə ra'va ‘I have washed myself ’

unaccusative

d. mə so ddeʃpɔʃtə/m adə reʃpɔʃtə ra solə ‘I have answered myself all alone’ e. mə so dda'va/m adə ra'va li pijə ‘I have washed my feet’ f. so mma'a/adə mma'a (la mela) ‘I have eaten (the apple)’

indirect unergative refl.

retroherent direct transitive refl. aux E aux E/H

indirect transitive refl. transitive/unergative

Contrary to Agnonese, auxiliation in this dialect is not sensitive to grammatical person. Rather, as in Italian or French, one and the same auxiliary is selected in all persons of the compound perfect of any given verb. However, the system belongs to the ‘mixed’ category, since free variation is observed, in all persons, with indirect unergative reflexives, with dyadic reflexives, as well as with transitives/unergatives (34d–f). In spite of this different empirical manifestation, the contrast emerging from the two auxiliation options (viz. (34a–c) vs (34d–f )) is the same as in old Romanesco.

3.5 Conclusion The foregoing discussion of two modern Italo-Romance dialects yields independent confirmation for the old Romanesco system as reconstructed by Formentin (2002), thus providing a nice example of fruitful interaction between the study of medieval and modern Romance varieties. Agnonese and Picernese confirm that there are dialects which mark the divide between direct and indirect reflexives, in the same way

Perfective auxiliation in Italo-Romance

69

as old Romanesco seems to do, and this corroborates Formentin’s (2002) conclusions, even if the database on which they are founded is limited and, for the crucial context of indirect unergative reflexives, reduces to just one relevant example. Another significant conclusion is that the morphosyntax of Italo-Romance varieties, both medieval and modern, may differ subtly: this limited exercise thus contributes to the discussion of whether morphosyntactic structure is, and has been in the past, less differentiated across Italo-Romance than the phonetics/phonology of the relevant local varieties (cf. Vincent, Parry, and Hastings 2004: 506–8). The Agnonese case in particular lends itself to some further points of methodology. Note first that discovery of triple auxiliation in Agnonese crucially depends on the completeness of the dataset taken into account. Specifically, the contexts outlined in the scale in (9) must all be considered, or one cannot conclusively classify the dialect at issue—as seen in (27) above—and even recognize that triple auxiliation is at work here. Realistically, this can hardly be expected to materialize in a strictly corpus-based study, because some of the crucial contexts have very low textual frequency in both ancient and modern dialects, as the old Romanesco example has shown.20 In other words, in spite of recurrent claims to the contrary (e.g. Gibson and Fedorenko’s (2010) complaint about ‘weak quantitative standards in linguistic research’), elicitation of grammaticality judgements is still inescapable (provided, of course, that one asks the ‘right’ questions). Another interesting point concerns standardization. Like other triple systems, Agnonese 1 ((33d), the more conservative variety) represents a compromise, in this case a compromise between the old Romanesco and the Sardinian types (22b,c). Not surprisingly, the more innovative variety Agnonese 2 (33e) has shifted towards the standard, a fact that has also been reported for other triple systems (cf. Loporcaro 1999b on the dialect of Zagarolo).21 This shows that syntactic standardization may take place in a way that does not correspond to standard Italian surface morphology at all: standard Italian does not display free variation in dyadic reflexives, yet by extending this free variation to them, Agnonese 2 is becoming syntactically more similar to the standard. Again, this is something we can grasp only under a syntactic approach to mixed auxiliation of the kind summarized in }3.2 above, while alternative views of perfective auxiliary selection, of the kind represented most prominently by Kayne (1993), would miss the parallelism between mixed and non-mixed systems. 20 Examples could easily be multiplied. Thus Longhino’s (2002–3) study of old Paduan shows that aux E/H were selected variably with retroherent unaccusatives, whereas only aux H occurs with dyadic reflexives (9e,f ). However, she was unable to find, in her corpus, any occurrences of indirect unergative or direct transitive reflexives. 21 As stated in the conclusion of }3.1 and in fn. 9 above, most triple systems seem to represent the manifestation of a transitional, and hence unstable, stage. Note that in mixed varieties which preserve an auxiliation contrast in just one or few grammatical persons, standardization may be favoured by the fact that, under such conditions, the evidence for the contrast the acquirer is exposed to is scantier than within non-mixed (binary) systems. This may lead to increased instability, eventually yielding to the pressure of the standard language.

70

Michele Loporcaro

A final point of methodology concerns one aspect which has been touched upon only cursorily in (3) above. All that has been said above about diverging auxiliation options in old Florentine, old Romanesco, Agnonese, and Picernese is about syntax, not semantics. The contrasts between the different subclasses of reflexives, which prove crucial for classifying those contrasting options, are orthogonal to the semantics of the predicates involved. An elementary test will suffice to demonstrate this. Take two Agnonese clauses involving retroherent predicates, like those in (35): (35)

a. marojja s e dd@v@rtiu̯t@/*s a d@v@rtiu̯ta (p@ ttre au̯r@/*ke ttre au̯r@) ‘Maria had fun (for/*in three hours)’ b. la kamoi ̯ʃa s e ss@kuɵ̯ta (k@ ttre au̯r@) ‘The shirt dried (in three hours)’

Their Aktionsart semantics is different, as shown by the test with the adverbial in x time. Yet the auxiliation option is the same, be they telic (35b) or non-telic (35a). The same goes for the pair of indirect transitive reflexives in (36): (36)

a. marojja s a pruweæ̯t@/*s e ppruweæ̯ta v@ʃ’tet@ra p@ ttre au̯r@ /*k@ ttre au̯r@ ‘Maria has tried clothes on for/*in three hours’ b. marojja s a mess@/*s e mmessa l@ skarp@ k@ tʧiŋg@ m@niu̯t@ ‘Maria put her shoes on in five minutes’

This is only expected under an auxiliation rule à la Perlmutter (5) (or (31)–(32)), and under a syntactic view of unaccusativity such as that endorsed in the present chapter. On the contrary, competing approaches to Romance auxiliation which claim that this has to be accounted for in semantic terms have problems with the auxiliary choice in reflexives: but this is another story, whose discussion cannot be tackled here.

4 Passive and impersonal reflexives in the Italian dialects Synchronic and diachronic aspects* M I C H EL A C EN N A M O

4.1 Introduction This chapter investigates the morphosyntax of passive and impersonal reflexives in contemporary Italian dialects and in some 11th–15th-century northern (Venetian, Lombard), central (Florentine), southern (Neapolitan), and Sardinian (Logudorese) vernaculars, considering the distribution of morphosyntactic, semantic, and pragmatic characteristics which set several northern varieties apart from central and southern dialects and from Sardinian. It is shown that variational and diachronic data from early Italo-Romance throw light on the diachronic processes leading to the rise of impersonal reflexives in Romance, contributing to a better understanding of some highly debated issues of Italian morphosyntax, such as the nature and function(s) of the reflexive morpheme si ‘self-’, and the function of the clitic ci ‘us; there’ in the impersonal of reflexives. The discussion is organized as follows. Section 4.2 illustrates the use of the reflexive morpheme as a voice marker in Italian and some current theoretical assumptions on its (varying) distribution and functions. Sections 4.3 and 4.4 describe the range of variation in the use of the reflexive morpheme si/se in passive and impersonal patterns in the Italian dialects and the early vernaculars, respectively. Section 4.5 * To Mair, a rigorous and critical mind, a ‘native speaker’ of early northern Italian vernaculars, and a very generous and inspiring scholar and friend. I wish to thank Paola Benincà, Adam Ledgeway, and Nigel Vincent for commenting on an earlier draft of this chapter. I am also grateful to Diego Pescarini and Davide Ricca for discussions on Paduan and Torinese, respectively, and to Ignazio Putzu and Maurizio Virdis for their help with old Logudorese. The usual disclaimers apply.

72

Michela Cennamo

builds on the insights gained from the analysis of the synchronic and diachronic variation, and addresses the theoretical debate concerning the status of the reflexive morpheme in impersonal patterns and the existence of either two types of si (a passive and an impersonal one) or of only one si with different uses.

4.2 Passive and impersonal reflexives in Standard Italian The reflexive pattern in Italian is widely used as a passive and impersonal strategy, marking different types and degrees of argument defocusing (either suppressed or entirely lacking at argument structure), with the reflexive morpheme si signalling a logically implied indefinite, human participant, either A, S, or O—following a wellestablished terminology referring to the clause-nuclear arguments (see e.g. Dixon 1994; Haspelmath 2011)—and the occurrence of an event, depending on the verb and the type of construction (}}4.2.1 and 4.2.2), with variable interpretation (}4.2.3) and function (}4.2.4). 4.2.1 Passive(/impersonal) and impersonal si With passive/impersonal si formed from bivalent verbs, A is suppressed, causing O to occur as subject, and the construction receives various interpretations in accordance with the context, as in (1). It can be either O-oriented, as in its passive function, with focus on the O argument (‘the results can be seen’), or A-oriented, as in its impersonal reading, with focus on an indefinite, non-referential participant (‘one, someone sees/they see the results’) (cf. e.g. Cennamo 1995; Bentley 2006; D’Alessandro 2007): (1)

I risultati si vedono /Si vedono the results self= see / self= see ‘The results are apparent/One sees the results’

i the

risultati (It.) results

Although both the S si V/si V S orders are possible when the subject is [
an.] (with S preverbal if [þgiven] [þdef.], and postverbal if [þnew] [def.]), the order S si V generally favours the passive interpretation (2a), whereas the si V S order most readily licenses the impersonal reading (2b) and tends to occur with [þan.], [þnew] [def.] S (3a; see Cennamo 1995 for a corpus-based investigation): (2)

a. Molte battaglie si sono combattute inutilmente (It., Lepschy 1986: 143) several battles self= are fought.fpl unnecessarily ‘Several battles have been fought unnecessarily’ b. Si sono combattute molte battaglie inutilmente (It.) self=are fought.fpl several battles unnecessarily ‘One/They/We fought several battles unnecessarily’

Passive and impersonal reflexives in the Italian dialects (3)

73

a. Quando il sistema è in pericolo si scelgono persone collaudate / le when the system is in danger self= choose people expert the persone migliori (It.) people best ‘When the system is in danger one chooses people of well-attested competence/people of well-attested competence are chosen’

The S si V order, however, is impossible if S is [þan.] [—def.], as illustrated by the contrast between (3b and 3c; cf. Cennamo 1995: 100; 2010; 2011a; 2011b): (3)

b. Quando il sistema è in pericolo *persone collaudate si scelgono / ?le when the system is in danger people expert self= choose the persone collaudate si scelgono (It.) people expert self= choose ‘When the system is in danger one chooses people of well-attested competence/people of well-attested competence are chosen’ c. Le persone collaudate si scelgono sempre quando il sistema è in the people expert self= choose always when the system is in pericolo (It.) danger ‘Competent people are always chosen when the system is in danger’

If the subject is pronominalized, as in (4), the pattern becomes impersonal: the finite verb occurs in the non-agreeing, 3sg form and in compound tenses the past participle agrees in number and gender with the pronoun li ‘them’ in (4b): (4)

a. Li si them.m= self= ‘One sees them’

vede (< ex. (1) above, It.) sees

b. Li Si è visti (It.) them.m= self= is seen.mpl ‘One has seen them’ In compound tenses, past participle agreement differentiates passive si from impersonal si: (5)

a. Si è pagati poco dall’ azienda per quel tipo di lavoro (It.) self= is paid.mpl little by.the firm for that type of work ‘One is not paid much by the firm for that type of work’ b. Si è pagato molto in quel self= is paid.msg much in that ‘They/We paid a lot in that restaurant’

ristorante (It.) restaurant

74

Michela Cennamo

In the passive pattern (5a) the past participle is in the masculine plural, agreeing with an underlying unexpressed O—reflecting the more general rule whereby in the impersonal pattern the nominal part of the predicate is in the plural (Salvi 1988: 98; Cennamo 1993a: 36; 2010)—and the agent optionally surfaces as a prepositional phrase introduced by the preposition da ‘by’. In the impersonal (active) construction (5b), by contrast, the past participle occurs in the non-agreeing msg form. This pattern of agreement is shared with bivalent verbs with an unexpressed/ omissible object (e.g. pagare, ‘pay’, bere ‘drink’, leggere ‘read’), bivalent verbs with an ‘oblique’ O (e.g. discutere di ‘talk about’, pensare a ‘think about’), and unergative verbs (e.g. camminare ‘walk’, passeggiare ‘stroll’). On the other hand, unaccusatives, equative patterns, and impersonal passives show split agreement in compound tenses, with the finite verb displaying 3sg agreement and the nominal part of the predicate mpl (or more rarely fpl) agreement: (6) Si è contenti/-e/ insegnanti/ ascoltati/ partiti all’ alba (It.) self= is happy.mpl/fpl/ teachers/ listened.to.mpl/ left.mpl at.the dawn ‘One is/They/We are happy/teachers/ listened to/left at dawn’ Si can also mark the taking place of an event with no implied participant, in conjunction with the verb fare ‘do’, e.g. far(si) tardi ‘become late’, with the past participle occurring in the default msg in compound tenses (see e.g. Cennamo 1995): (7) Si fa tardi / si self= makes late / self= ‘It is getting late/it got late’

è is

fatto made.msg

tardi (It.) late

4.2.2 Impersonal of reflexive patterns A characteristic feature of reflexive constructions is the use of the 1pl clitic ci in the sequence ci si for the corresponding impersonal form of reflexive/middle and inherent reflexives, also referred to as ‘lexicalized’ uses of the reflexive (Salvi and Vanelli 2004: 206): (8) Ci si lava / Ci si è lavati/-e (direct reflexive; It.) ci= self= washes / ci= self= is washed.mpl/fpl ‘One washes/has washed oneself/We wash/have washed ourselves’ The nature and function of ci in these patterns has been widely debated. Bentley (2006: 167) regards the sequence ci si as a marker of double argument suppression, with the clitic ci an allomorph of reflexive si, due to the rule banning two occurrences of si. The same view is shared by Salvi (2008a; 2010b), who regards the 1pl clitic ci as replacing the 3rd person reflexive si and si as a marker of an unexpressed indefinite participant. Here we argue that ci is an impersonal marker, signalling an unexpressed A/S, and realising the 1pl referent (speakerþhearer) that in the course of time has become part

Passive and impersonal reflexives in the Italian dialects

75

of the referential domain of si in its impersonal reinterpretation (Cennamo 1993b; 2000). On this view, ci may be regarded as an indefinite pronoun with a variable referential status like si. Unlike si, however, it can only have an existential (inclusive/non-inclusive) interpretation, signalling an underlying A or S, depending on the syntactic valency of the verb (cf. }4.3.4). As we shall see (}4.4), our claim is confirmed by synchronic dialectal variation showing ci in impersonal function, replacing si, in several central and southern varieties, as well as by the use of si followed by the finite verb in the 3sg in Tuscan and some Umbrian varieties where it replaces the canonical 1pl form (Cennamo 1997: 82; 2011a). 4.2.3 Impersonal reflexive: interpretation of the morpheme si The unexpressed participant (A, S, O) signalled by si in impersonal constructions has either a generic or an indeterminate/existential reference, according to the type of event described (Salvi 1988: 98; D’Alessandro 2007). Generic si denotes a group of people to whom a property applies. It can comprise the speaker (1pl inclusive interpretation), as in (9): (9) In questa casa si beve molto in this house self= drinks much ‘In this house they/we drink a lot of wine’

vino (It.) wine

Indeterminate/existential si refers to one or more people in the universe of discourse, whose existence is implied, but whose identity is unknown/non-relevant. It may refer to the speaker (inclusive interpretation): (10)

Ieri si è bevuto molto a casa yesterday self= is drunk.msg much at home ‘Yesterday they/we drank a lot at Mario’s house’

di of

Mario (It.) Mario

The generic or indeterminate/existential interpretation of si reflects, in turn, the temporal reference of the clause and the aspectual nature of the predicate, namely the boundedness of the event (D’Alessandro 2007: 150). Specific time reference and perfective aspect seem to trigger the existential inclusive interpretation of si with all verbs (transitives, unergatives, and unaccusatives; cf. Salvi 1988: 98; Cinque 1988a; D’Alessandro 2007: 135ff. ): (11)

a. Ieri si è lavorato fino a yesterday self= is worked.msg till at ‘Yesterday we (*one, they . . . ) worked till late’ b. Ieri si è partiti alle yesterday self= is left.mpl at.the ‘Yesterday we (*they . . . ) left at 5 p.m.’

tardi (It.) late

cinque (It.) five

76

Michela Cennamo

The inclusive interpretation of impersonal si can be ‘suspended’ in case of specific time reference, if the predicate is not temporally ‘bounded’, as in hypothetical clauses, where the event extends without reaching a final point (Cinque 1988a: 150; D’Alessandro 2007: 154f.). (12)

Ieri, se si fosse arrivati tardi all’ appuntamento, si sarebbe yesterday if self= were arrived.mpl late at.the appointment self= would.be persa una riunione importante (It.) missed.fsg a meeting important ‘Yesterday, if they had arrived late at the appointment’ (also ‘if we had arrived . . . ’), they/we’d have missed an important meeting’

4.2.4 Some controversial issues It has been widely debated in the literature, since the late 1970s (Castelfranchi and Parisi 1976; Napoli 1976), whether a unifiying analysis of the various uses of si is possible, together with the related question concerning the existence of only one si with different functions or of two types of si belonging to different domains of the grammar: reflexive si, whereby the reflexive is an accusative/dative object clitic, and passive and impersonal si, where si is treated as a subject (Burzio 1986) or as an argument (Cinque 1988a; Cennamo 1993a; Bentley 2006: 173; D’Alessandro 2007; Manzini and Savoia 2005: ii.13). We argue for the latter view, although with a different grouping of patterns marked by si, singling out reflexive and passive si versus impersonal si (see also Cennamo 1993a; 2000). As for the impersonal of reflexives, in traditional descriptions (Rohlfs 1968; Lepschy and Lepschy 1988) taken up in some theoretical analyses (e.g. Burzio 1986: 55, 81, n. 47), ci is regarded as a case of dissimilation, the morphological variant of impersonal si in the reflexive/middle/inherent si patterns, whereas in the theoretical literature ci is viewed instead as the reflexive form, followed by impersonal si (Cinque 1995: 193–8; Salvi 1988; Bentley 2006: 173–4). The following sections show that synchronic and diachronic dialectal data appear to support the hypothesis of the existence of two types of si, a reflexive ~ passive si and an impersonal si, resulting from two different diachronic paths (Cennamo 1993a; 2000). At the same time, these data throw new light on the nature and function of the clitic ci in the corresponding impersonal form of reflexives.

4.3 Passive and impersonal reflexives in the Italian dialects In the Italian dialects the 3sg/pl reflexive morpheme se/si and its variants are well attested in passive and impersonal function alongside the 1pl reflexive morpheme ci (and its variants) in some central and southern varieties (for an overview, see Cennamo 1997; 1998; Manzini and Savoia 2005: ii.135–7). Six main parameters of variation can be identified in the dialects in the use of these morphemes as voice markers: (i) the morphological and/or syntactic distinction

Passive and impersonal reflexives in the Italian dialects

77

between passive and impersonal si (}4.3.1); (ii) their grammatical domains (}4.3.2); (iii) the nature of the subject (}4.3.3); (iv) tense/aspect restrictions (}4.3.4); (v) the cooccurrence of the impersonal reflexive with a pronominal object (}4.3.5); and (vi) the interpretation of the reflexive morpheme (}4.3.6; see also Cennamo 1997: 161). 4.3.1 Passive vs impersonal reflexives Most northern dialects distinguish morphologically and/or syntactically anticausative (characterized by A suppression and O subjectivization)/passive si from impersonal si (Cennamo 1997; Manzini and Savoia 2005: ii, ch. 4). The former occurs in the sequence S si V, with a preverbal S, most typically [–an.], [þdef.], and conveying [þgiven] information: (13) La porta la se verze / I libri de aventura i se vende the door scl= self= open.3 the books of adventure scl= self= sell.3 ben (Vnz.) well ‘The door opens/Adventure books sell well’ In these varieties the sequence si V S is banned if the ([def.]) subject conveys [þnew] information (14a). The order si V S, in fact, is only possible if the subject is clause-external, conveying [þgiven] information (14b): (14) a. *La scl=

se verze la porta/ *I se vende i libri (Vnz.) self= open.3 the door scl= self= sell.3 the books

b. La se verze, la porta / I se vende, i libri (Vnz.) scl= self= open.3 the door scl= self= sell.3 the books ‘The door opens/The books sell’ The verb instead occurs in the 3sg with an optional non-agreeing ‘impersonal’ subject clitic (in the dialects which have a subject clitic also for the 3rd person), identical in several varieties with the 3msg subject clitic, as in Florentine (15a; for a general overview, see Cennamo 1997: 153; Manzini and Savoia 2005: ii.19–74). In dialects where 3rd singular and plural are not morphologically distinct, the impersonal pattern becomes overt in compound tenses, where the past participle occurs in the unmarked msg form, as in Paduan (15b) where, depending on the context, the pattern can also have an anticausative interpretation (Cennamo and Sorace 2007: 88): (15) a. (E) s’ apre la porta (Flo.) scl.imprs= self= opens the door ‘One opens the door/We open the door’ b. Se ga verto la porta (Pad.) self= have.3 opened.msg the door ‘One has opened/they/We have opened the door’

78

Michela Cennamo

In some dialects (Torinese, Cairese, Genoese) (Battye 1990; Parry 1998a; 2005), the patterns with and without verb agreement with the postverbal nominal (si V S/si V N) alternate in passive and impersonal function. The pattern with agreement is regarded, however, as less dialectal, reflecting Italian influence, as in (16a) from Torinese (Cennamo 1997: 154): (16)

a. a s katu le scl= self= buy the ‘The cakes are bought’

turte (Tor.) cakes

b. a s kata le scl= self= buys the ‘One buys the cakes’

turte (Tor.) cakes

Interestingly, in some northern varieties (e.g. Cairese and generally in Piedmontese and Ligurian) the morphological/syntactic impersonality of the pattern does not entail its impersonal interpretation (Parry 1998a; 2005): (17)

a. i s bøtu i pjat au so poʃt (Cairo Montenotte (SV), Liguria) scl= self= put the plates at.the their place ‘The plates are put away’ b. u s bøt i pjat au so scl.imprs= self= puts the plates at.the their poʃt (Cairo Montenotte (SV), Liguria) place ‘The plates are put away/One puts the plates away’

In some Piedmontese varieties (e.g. Borgomanero) impersonal si occurs preverbally on a par with subject clitics (18a), whereas anticausative and passive si occur postverbally (18b,c) (Manzini and Savoia 2005: ii.70–71): (18)

a. ki a z droma here scl= self= sleeps ‘Here one sleeps well’ b. i 'rumpusi scl= break=self ‘The plates break’

i the

bei (Borgomanero (NO), Piedmont) well

pjati (Borgomanero (NO), Piedmont) plates

c. lø i vøŋgusi sempri they scl= see=self always ‘They are always seen going . . . ’

pasøndu (Borgomanero (NO), Piedmont) going

Other northern dialects such as Bussoleno (Piedmontese) (19) and Friulian show free variation of pre-/postverbal subjects, although the S si V/si V S order generally conveys a pragmatic distinction, the distribution of given/new information (Cennamo 1997: 157; 1998):

Passive and impersonal reflexives in the Italian dialects (19)

79

I liber d aventura a z vendu / A z vendu i liber d the books of adventure scl= self= sell / scl= self= sell the books of aventura (Bussoleno (TO), Piedmont) adventure ‘Adventure books sell/are sold; One sells/They sell adventure books’

Free alternation of either order, S si V/si V S, in both the anticausative and passive/ impersonal interpretation, instead characterizes central-southern dialects, with the different word order reflecting the distribution of given/new information, as in (20a,b): (20) a. e libbrә ddʒallә sә 'vennenә fatʃilmentә (Nap., S = [+ given]) the books thriller self= sell easily ‘Thrillers sell/are sold easily/One, They sell thrillers easily’ b. sә 'vennenә fatʃilmåntә e libbrә d’dʒallә (Nap., S = [+new]) self= sell easily the books thriller ‘One/They sell(s) thrillers easily’ 4.3.2 Grammatical domains The dialects also differ in the grammatical domains covered by the reflexive in its function as a voice marker (Cennamo 1997: 153–9; Manzini and Savoia 2005: ii, ch. 4). Some northern dialects (e.g. Paduan, Venetian) show maximal exploitation of this strategy, as illustrated in (21), from Paduan, where the reflexive occurs in all grammatical domains. It is used, in fact, in equative structures (21a), in impersonal passive patterns (21b), and in the impersonal of reflexive patterns (21c). It also occurs in the corresponding impersonal form of passive/impersonal patterns with the original subject pronominalized (21d; examples from Cennamo 1997: 155): (21)

a. se ze pu'tei / se ze 'zovani / se ze stai 'zovani] (Pad.) self= be.3 children / self= be.3 young.pl / self= be.3 stayed.mpl young.pl ‘One is a child/One is young/One has been young’ b. se ze par’tii / se ze pa'ga(i) / se ze sta pa’ga(i)/ self= be.3 left.mpl / self= be.3 paid.sg(pl) / self= be.3 stayed.sg paid.sg(pl)/ se se pentise / se se ga pen'tio (Pad.) self= self= repented / self= self= have.3 repented.msg ‘One has left/One is paid/One has been paid/One repented/has repented’ c. se se ga kom'pra na kaza / se ga konda'na i self= self= have.3 bought a house / self= have.3 sentenced.sg the kol’pevoli (Pad.) guilty ‘One has bought oneself a house/One, They (indef.) have sentenced the guilty’

80

Michela Cennamo d. se la ga kom'pra / se li self= it= have.3 bought.sg / self= them= ‘One has bought it/One has sentenced them’

ga have.3

konda'na (Pad.) sentenced.sg

e. se ga ven'duo i libri / se li ga ven'dui (Pad.) self= have.3 sold.msg the books / self= them have.3 sold.mpl ‘One has sold the books/One has sold them’ In contrast, in some southern dialects (e.g. Neapolitan, Cosentino, Sicilian) the reflexive only covers some grammatical domains: it does not occur in equative patterns, the impersonal of reflexives, and impersonal-passive structures (22). Alternative strategies are employed for these grammatical domains, including the indefinite uno/una ‘one’, the collective noun gente ‘people’, the 3pl (indefinite exclusive interpretation), the 1pl (Cennamo 1997: 156): (22)

a. kwann unә å kkrjaturә / unә sә påntә / tʃә pәntimmә (Nap.) when one is young / one self= repents / ourselves= repent.1pl ‘When one is a child/one repents /we repent’ b. 'pavәnә pOkә pә kkillu lavorә / tʃ annә pavatә pOkә (Nap.) they.pay little for that work / us= they.have paid little ‘they do not pay much for that work/they have not paid us much for that work’

Some varieties show instead variable uses, depending on the pattern. For instance, for the impersonal of the reflexive pattern (i.e. ‘lexicalized’ uses of the reflexive) in Milanese, although the type se se V is possible (23), the pattern with only one se is preferred (Nicoli 1983: 168; Manzini and Savoia 2005: ii, ch. 4): (23) se se lava / se se se self= self= washes / self= self= self ‘One washes oneself/If one forgets to pay’

de'mentega forgets

de of

pa'ga (Mil.) pay.inf

4.3.3 Nature of the subject The animacy of the subject too appears to affect the occurrence of the reflexive vis-àvis other strategies. For instance, varieties such as Trentino and Bussoleno do not allow passive/impersonal si with an animate subject in compound tenses (24a), and the corresponding active with an indefinite 3pl subject occurs instead (24b) (Cennamo 1997: 79). The same restriction is at work in some central-southern varieties (Castelpetroso, Agnone (Molise), Naples; Cennamo 1997: 156): (24) a. *se a konda'na i kol'pevoli a tsiŋkwe ani (Trento, Trentino) self= has sentenced the guilty to five years

Passive and impersonal reflexives in the Italian dialects

81

b. i a konda'na i kol'pevoli a tsiŋkwe ani (Trento, Trentino) scl= have sentenced the guilty to five years ‘They (indef.) have sentenced the guilty to five years’ In several varieties from Abruzzo (province of l’Aquila) and Molise (e.g. Castelpetroso; cf. Giammarco 1968: 479–80), northern Puglia (the Gargano peninsula; cf. Melillo 1973: 120–21), Salento (cf. Rohlfs 1956–9: 128), Tuscany (Isle of Giglio; cf. Rohlfs 1968: 186), and Calabria (the province of Reggio Calabria; cf. Rohlfs 1977: 455), in reflexive, anticausative, passive/impersonal, and impersonal functions we find the 1pl reflexive morpheme ci, nci (see also Cennamo 1997: 158–9; Manzini and Savoia 2005: ii.135–8): (25) tʃә 'vinnәnә rә libbrә / tʃ annә vennutә ri libbrә (Castelpetroso self.1pl= sell the books self.1pl have sold the books (IS), Molise) ‘The books are sold/One sells the books’ 4.3.4 Tense/aspect constraints Tense/aspect restrictions play a role in determining the occurrence of the reflexive as a voice strategy. In several varieties the reflexive in impersonal/passive constructions does not occur in perfective contexts, even with inanimate subjects, the restriction applying either to all verbs or only to unaccusatives, as in Palmanova Friulian (data from Laura Vanelli). The corresponding active pattern is employed instead, with either an indefinite 3pl or a 1pl subject. This constraint might reflect the diachronic spread of si in impersonal/passive and impersonal function, attested initially in simplex tenses and later in compound tenses, and might also be related to the differential behaviour of transitives/unergatives vs unaccusatives in clauses with specific time reference and/or (ir)realis modality, i.e. the boundedness of the event (Cennamo 1997: 156–7): (26) a. se sa’ria par’tii ale siŋkwe ma no ge dʒera posto (Pad.) self= would.be left at.the five but not there= was place ‘We would have left at 5 p.m., but there were no seats’ b. ??se ze / dʒera par’tii ale self= is was left.mpl at.the ‘One has/We have left at 5 p.m.’

siŋ’kwe (Pad.) five

c. si a finut di lavo’ra as sis e la sera o sin las self= has finished of work.inf at.the six and the evening we are gone al bar (Palmanova (UD), Friuli-Venezia-Giulia) to.the bar ‘We finished work at 6 and in the evening we went to the bar’

82

Michela Cennamo

Thus, whereas in Paduan with unaccusatives the impersonal reflexive is only possible in irrealis contexts (26a) and hardly acceptable in realis ones (26b), in Palmanova Friulian with unaccusatives in realis contexts the reflexive is banned; instead we find the 1pl (26c). 4.3.5 Impersonal si/se with a pronominal object The co-occurrence of the impersonal reflexive with a pronominal object appears to have different degrees of acceptability across the dialects. It is possible in some dialects, with the reflexive preceding the object clitic (27a). In other dialects, instead, either we find the reflexive pattern without the pronominal object, as in Neapolitan (27b), or a different strategy is employed, either the 1pl or the 3pl according to the context, as in Esperia (southern Lazio) (27c; Cennamo 1997: 157–8): (27)

a. se li vende / se li self= them= sell.3 / self= them= ‘One sells them/One has sold them’ b. se 'vennәnә / s annә self= sell / self= have ‘They are sold/They have been sold’

ga have.3

ven'dui (Pad.) sold.mpl

vennutә (Nap.) sold

c. li 'vinnәnә lestә (Esperia (FR), Lazio) them= sell quickly ‘They sell them easily’ 4.3.6 Interpretation of impersonal si Another parameter of variation concerns the interpretation of the reflexive morpheme in impersonal patterns. In some varieties (e.g. Neapolitan) the impersonal reflexive has a generic or indeterminate/existential non-inclusive interpretation. In other dialects (e.g. Florentine) the impersonal reflexive always has an inclusive reading. In Florentine and other Tuscan varieties (e.g. Sienese, Viareggino) as well as in some Umbrian (e.g. Città di Castello, Amelia) and Marchigiano (e.g. Arsoli) dialects, the impersonal reflexiveþ3sg active has replaced the traditional 1pl synthetic verbal form (Cennamo 1997: 158; 1998: 82; Manzini and Savoia 2005: ii, ch. 4; also AIS map 1278): (28)

Si va (Flo.) self= goes ‘We go’

Interestingly, if the speaker is excluded from the universe of discourse, Florentine uses 3pl rather than impersonal siþactive (Stefanini 1983: 111).

Passive and impersonal reflexives in the Italian dialects

83

4.3.7 Interim summary Unlike the (anticausative/)passive reflexive (but see Manzini and Savoia 2005: ii, ch. 4), the impersonal reflexive is not equally attested across the dialects. In some dialects impersonal si is not found in perfective contexts and with specific time reference, where the implied participant comprises the speaker (inclusive interpretation); therefore si only has either generic or an indeterminate/existential non-inclusive reference. In other varieties, instead, si has an existential inclusive interpretation, and the sequence siþ3sg has replaced the 1pl ending of verbs. The data thus show the existence of a correlation between the use of the reflexive strategy in impersonal function and its (inclusive/noninclusive) interpretation. It might be argued, therefore, that the spread of the reflexive morpheme in impersonal patterns in the dialects depends on the extent to which in the various varieties the 3rd person reflexive has come to include the 1st and 2nd person participants (i.e. the speech act participants) in its referential domain, and on the extent to which the new referents are grammaticalized.

4.4 Some diachronic data1 The analysis of some early Italian vernaculars of the eleventh to fifteenth centuries from different areas (Venice, Lombardy, Florence, Naples, and

1 The bibliographical details of the textual sources and their abbreviations cited in the examples below are as follows: Bonvesin, Disputatio = Disputatio musce cum formica, in Contini, Gianfranco (ed.) (1937), Cinque Volgari di Bonvesin da la Riva. Modena: Società Tipografica Modenese, 27–40; Bonvesin, Miraculis = De quindecin miraculis que debent apparere ante diem iudicij, in Contini, Gianfranco (ed.) (1937), Cinque Volgari di Bonvesin da la Riva. Modena: Società Tipografica Modenese, 41–4; Capitolare = Princivalli, Alessandra, and Ortalli, Gherardo (eds) (1993). Il Capitolare degli Ufficialli sopra Rialto. Nei luoghi al centro del sistema economico veneziano (secoli XIII-XIV). Milan: La Storia; Capitoli = ‘Capitoli della Compagnia di San Gilio’, in Schiaffini, Alfredo (ed.) (1926), Testi fiorentini del Dugento e dei primi del Trecento. Florence: Sansoni, 34–54; Cedola di Marco Granello = ‘Cedola di Marco Granello’, in Stussi, Alfredo (1965), Testi veneziani del Duecento e dei primi del Trecento. Pisa: Nistri-Lischi Stussi, 36–40; Cedola di Marco Michel = ‘Cedola di Marco Michel’, in Stussi, Alfredo (1965), Testi veneziani del Duecento e dei primi del Trecento. Pisa: Nistri-Lischi Stussi, 111–22; Cedola di Pangrati Barbo = ‘Cedola di Pangrati Barbo (Prima)‘, in Stussi, Alfredo (1965), Testi veneziani del Duecento e dei primi del Trecento. Pisa: Nistri-Lischi Stussi, 63–4; Cronica = Ceruti, Antonia (ed.) (1878), ‘Cronica degli Imperadori’, Archivio glottologico italiano 3: 177–243; Cronica fior. = ‘Cronica Fiorentina’, in Schiaffini, Alfredo (ed.) (1926), Testi fiorentini del Dugento e dei primi del Trecento. Florence: Sansoni, 82–150; CSNT = Merci, Paolo (ed.) (1992), Il Condaghe di San Pietro di Silki. Testo logudorese inedito dei Secoli XI-XIII. Sassari: Dessì; RSPS = Piras, Sara Silvia, and Dessì, Gisa. (eds) (2003), Il Registro di S. Pietro di Sorres. Cagliari: Cuec.; Dante, Inferno = Petrocchi, Giorgio (ed.) (1966), Dante Alighieri, La Commedia secondo l’antica vulgata. Volume II. Inferno. Milan: Mondadori; Grisostomo = Förster, Wendelin (ed.) (1880–83), ‘Antica parafrasi lombarda del “Neminem laedi nisi a se ipso” di S. Giovanni Grisostomo’, Archivio glottologico italiano 7: 1–120; LDT = De Blasi Nicola (ed.) (1986), Libro de la Destructione de Troya. Volgarizzamento napoletano trecentesco da Guido delle Colonne. Rome: Bulzoni; Libro = ‘Libro degli ordinamenti della Compagnia di Santa Maria del Carmine’, in Schiaffini, Alfredo (ed.) (1926), Testi fiorentini del Dugento e dei primi del Trecento. Florence: Sansoni, 55–72; Milione = Bertolucci Pizzorusso, Valeria (ed.) (1975), Il Milione di Marco Polo (versione toscana del Trecento). Milan: Adelphi; Milione veneto = Barbieri, Alvaro, and Andreose, Alvise (eds.) (1999), Marco Polo. Il Milione Veneto. Venice: Marsilio; Panfilo = Tobler, Adolphe (ed.) (1886), ‘Il Panfilo in antico veneziano’,

84

Michela Cennamo

Logudoro)2 shows a highly varied picture as regards the use of the reflexive morpheme in passive and impersonal function, and appears to confirm the existence of a correlation between the spread of this strategy in impersonal function and the referential domain of si/se. In our discussion we do not consider ‘impersonal’ se/si in fixed phrases with avalent/monovalent verbs like old Venetian convenire ‘be advisable’, contenere ‘contain, state’, deser ‘become’, and their equivalents in the other vernaculars, well attested in the early varieties investigated, sometimes preceded (according to the verb) by expletive el ‘it’ and its variants in old Venetian, old Lombard, and old Florentine, alternating with non-reflexive forms (e.g. OVnz. (el) se convien(e) ‘(it) self= is.advisable’ ~ conven ‘is.advisable’). 4.4.1 Old Venetian In old Venetian the reflexive mainly occurs in passive(/impersonal) function, most frequently with inanimate subjects and with the order S se V, the only one attested in some texts (e.g. the Testi veneziani) (29a), with very few examples of animate subjects (29b) (Cennamo 2000: 93–7): (29)

a. voio que le dite chase se I.want that the afore.mentioned houses self= afita tute (OVnz., Cedola di Marco Granello, 39.21) rent.3 all ‘I want all the aforementioned houses to be let’ b. L’ Imperador . . . per nessuna arte the emperor by no type podeva aidare (OVnz., Cronica 221.50a) could help.inf ‘The Emperor could not be helped in any way’

de of

medesina drug

se self=

Archivio glottologico italiano 10: 179–243; Passione = Salvioni, Carlo (ed.) (1886), ‘La Passione e altre antiche scritture lombarde’, Archivio glottologico italiano 9: 1–22. Testi Ven. = Stussi, Alfredo (ed.) (1965), Testi veneziani del Duecento e del Trecento. Pisa: Nistri Lischi; Rettorica = Maggini, Francesco (ed.) (1968), Brunetto Latini, La rettorica. Florence: Le Monnier; Ricordi = ‘Ricordi di compere e cambi di terre in Val di Streda e dintorni’, Castellani, Arrigo (ed.) (1982), La prosa italiana delle origini. l: Testi toscani di carattere pratico. Bologna: Pàtron, 215–54; SRS = Guarnerio, Pier Enea (ed.) (1892–94), ‘Gli Statuti della Repubblica sassarese, testo logudorese del secolo XIV, nuovamente edito d’in sul codice e annotato’, Archivio glottologico italiano 13: 1–124. 2 The corpus investigated consists of non-literary texts from the 11th to 15th centuries, from different areas, namely old Venetian, old Lombard, old Neapolitan, and old Logudorese Sardinian, partly available at http://www.ovi.cnr.it/. The choice follows the approach by Vincent, Parry, and Hastings (2004). For old Neapolitan we have benefited also from data from Adam Ledgeway. This chapter uses the following abbreviations for texts cited.

Passive and impersonal reflexives in the Italian dialects

85

The S–se–V order usually favours a passive interpretation, whilst the se–V–S sequence favours an impersonal reading. The distinction between passive and impersonal reflexive, however, is not yet fully grammaticalized, as shown in (30), where either order is possible, although the subject conveys given information: (30)

a. tanto que questi dr. se paga (OVnz., Cedola di Marco Granello, such that these denarii self= pay.3 39.24–5) ‘so that this money is paid’ b. avanti che se paga sta dota (OVnz., Cedola di Marco Michel, 112.22) before that self= pay.3 this dowry ‘before this dowry is paid//one pays/they pay this dowry’

The reflexive in impersonal function is confined to bivalent verbs with an implicit latent object, sometimes preceded by expletive el ‘it’ (and its variants elo/eli): (31) sì co’ elo so as scl.imprs= ‘as one finds it’

se self=

trova (OVnz., Cedola di Pangrati Barbo, 63.8) find.3

In addition, passive/impersonal and impersonal se only occur in simple tenses and in imperfective verb forms, occasionally with the expletive el and a postverbal subject: (32) quando el se leze Henrico when it= self= read.3 Henry ‘when one reads Henry the First’

Primo (OVnz., Cronica, 226.18) first

In 13th-century texts impersonal se never occurs with monovalent verbs and always refers to an indefinite human participant who does not include the speaker. Thus se has either a generic or an indeterminate/existential reference, but never an inclusive one. There are no examples of overt expression of the agent. Impersonal se with monovalent verbs is attested from the mid- to late 14th century and during the 15th century in imperfective tenses only, with noninclusive (33a) and inclusive indeterminate/existential reference (33b) (cf. also Wehr 1995: 106): (33)

a. la calle che se va in Riolto Novo (OVnz., Capitolare 124, 58.18–19) the street that self= go.3 to Riolto new ‘the path through which one goes/people go to Riolto Novo b. torneremo ala provinzia [ . . . ], perché da altra via non we.will.return to.the province since from other way not se pò andar (OVnz., Milione veneto, 35.10) self= can.3 go.inf ‘we will go back to the province [ . . . ] since we cannot go another way’

86

Michela Cennamo

4.4.2 Old Lombard Old Lombard shows a different distribution of the reflexive pattern according to different texts/areas. Passive/impersonal and impersonal se are well attested in Bonvesin de La Riva (13th-c. Milanese) (34a,b) and Grisostomo (14th-c. Pavese), with few examples in the Passione (15th-c. Comasco), where impersonal se is only found with bivalent verbs: (34)

a. Senza quest [ . . . ] no ’s pò haver salvamento (OMil., without this not self= can.3sg have.inf salvation Bonvesin, Miraculis, 32.105) ‘Without which one cannot find salvation’ b. Azò k’ il temp [ . . . ] ke no ‘s pò lavorar (OMil., Bonvesin, So that the time that not self= can.3sg work.inf Disputatio, 39.247) ‘So that during the time when one cannot work’

In the various varieties both S–se–V and se–V–S orders are attested, although only in simplex tenses. S se V with a [—an.] S is the most frequent pattern (35a), in all the texts investigated. There are instead few examples of this pattern with [þan.] S and the se V S order (35b): (35)

a. questo comandamento se rompe in cinque this commandment self= infringes in five Passione 19.30) ‘and this commandment may be infringed in five ways’

modi (OCms., ways

b. qui no se po fa oltro seno pianze here not self= can.3 do.inf other if.not cry.inf madre (OCms., Passione 13.18) mother ‘here one cannot do anything else but cry with his mother’

con with

la the

As for the referential domain of se, in old Comasco it signals an indefinite participant which may include the speaker, as in (36): (36) (quando nu voloma provar una vigna [ . . . ]) Per lo semegliante se when we want assess.inf a vineyard for the resembling self= contempla e guarda l’ oliva e gl’ altri contemplates and sees the olive.tree and the other arbori fruteveli (OCms., Passione, 5.37) trees fruit.bearing ‘(when we want to assess a vineyard [ . . . ]) similarly we look at the olive trees and the other fruit trees’ Impersonal se also occurs with pleonastic el and bivalent verbs (37) and in fixed phrases with monovalent vebs, in pleonastic function (e.g. se convene ‘self= be.advisable’):

Passive and impersonal reflexives in the Italian dialects (37) la qual (leçe) comandava ch’ el no the which law commanded that it= not carne de porcho (OPav., Grisostomo, 49.20) meat of pork ‘according to which (law) pork was not to be eaten’

se self=

87

mangiasse ate.sbjv.3

In Grisostomo (old Pavese) there are few occurrences of impersonal reflexive with monovalent verbs in simple (imperfective) tenses only: (38)

a. ma per le citae se çeva pur peçorando (OPav., Grisostomo, 40.30) but through the city self= went also worsening ‘but in the town things were getting worse’ b. onde se çoga soto a la roçça (OPav., Grisostomo, 15.22) so.that self= plays under to the rock ‘so that one plays/they play under the rock’

4.4.3 Old Florentine In old Florentine the reflexive pattern is very frequent in passive/impersonal function with both S–si–V and si–V–S orders, with [an.] S (39a,b), but with a preference for the order si–V–S with [þan.] S, and optional expression of the agent through a prepositional phrase (39c) (see also Giacalone Ramat and Sansò 2011: 197–8): (39)

a. lo die che’ capitani si chiameranno (OFlo., Capitoli, 44.30–31) the day that.the capitans self= will.call ‘they day when the captains are called’ b. Quando si rachomanda alcuno infermo (OFlo., Capitoli, 40.33–7) when self= recommends some invalid ‘When one recommends an invalid’ c. venne al porto [ . . . ] lo quale si tenea per came to.the port the which self= kept by lo re Carlo (OFlo., Cronica fior., 146.10–11) the king Charles ‘he arrived at the harbour that had been conquered by King Charles’

Impersonal si is mainly attested with bivalent verbs, displaying in some cases a non-agreeing 3sg verb (40), also occurring with avalent/monovalent verbs in pleonastic function (e.g. convenire ‘be advisable’): (40)

si dea soldi XL a’ poveri (OFLo., Libro, 65.32) self= give soldi forty to.the poor ‘one/they (indef.) should give forty coins to the poor’

Passive/impersonal (41a) and impersonal si (41b) with bivalent verbs in compound tenses are rare and do not occur with animate nominals in 13th-century texts, unlike in the 14th century (41c; see also Wehr 1995: 155–6):

88 (41)

Michela Cennamo a. le grandissime cose non s’ arebbono potute mettere in the very.great things not self= would.have been.able put.inf in compimento (OFlo., Rettorica, 26.21–2) action ‘those very important things could not have been done’ b. un dì che s’ era bandito una corte a day that self= was prepared a court di nozze (OFlo., Novellino, 862.12) of wedding ‘one day when a wedding banquet had been prepared’ c. si sono uccisi degli uomini (OFlo., Dec., Introduzione, 17.29) self= are killed some men ‘one/they (indef.) have killed some men/some men have been killed’

Impersonal si with monovalent verbs instead is not attested in early 13th-century texts; this pattern is found only (with both unaccusatives and unergatives) by the end of the 13th century (42a), also with overt expression of the agent (42b; cf. Wehr 1995: 106; Salvi 2008a; 2010b): (42)

a. Iddio, al cui conoscimento non si God to.the whose knowledge not self= per fede (OFlo., Bono Giamboni, Orosio, 359.23) by faith ‘God, who one comes to know only through faith’

viene comes

se if

non not

b. non vuol che ’n sua città per me si not wants that in his town by me self= vegna (OFlo., Dante, Inf. I, A19.126) came.sbjv.3sg ‘he does not want people to come to his city led by me’ In 14th-century Tuscan texts impersonal si with monovalent verbs also occurs in compound tenses, alternating with l’uomo lit. ‘the man’ (Wehr 1995: 156): (43)

a. quando se’ ito uno die e una notte, si when self=is gone.msg one day and one night self= truova acqua (OFlo., Milione, 74.9–10) finds water ‘when one has travelled for one day and night, one finds water’ b. e quando l’ uomo è ito XX giornate (OFlo., Milione, 173.16) and when the man is gone 20 days ‘and when one has travelled for twenty days’

Passive and impersonal reflexives in the Italian dialects

89

As for its interpretation, impersonal si can have both a generic and indeterminate/ existential (both exclusive and inclusive) reading. It may refer, in fact, to both speaker and hearer/interlocutor, as shown by the alternation of si with the 1st person plural: (44)

Demoli [ . . . ] a Iacopo Lanberti, ke gli si scontò we.gave=them to Iacopo Lamberti that to.him= self= deducted de’ denari ke ci dee dare (OFlo., Ricordi, 224.13–14) of.the denarii that us= must give.inf ‘We gave them (= the money) to Iacopo Lamberti, money that was deducted from the denarii that he owes us’

4.4.4 Old Neapolitan Old Neapolitan shows a high frequency of the reflexive pattern in passive/impersonal function. Both S–se–V and se–V–S orders are possible, with S [ an.] (45a,b). S–se–V has higher frequency. The pattern also occurs with the overt expression of the agent introduced by da/per ‘by’ ((46); Cennamo 2000: 97–8; Ledgeway 2009a: 670–76): (45)

a. la festa in quillo tiemplo se the feast in that temple self= ‘the feast was celebrated in that temple’ b. Che de vui se non that of you self= not ‘May you not be forgotten’

(46)

perda lose

celebrava (ONap., LDT 97.25) celebrated

memoria (ONap., LDT 64.18) memory

azò che da tuti li Troyani (li cuorpi) so.that that by all the Trojans the bodies potessero resguardare (ONap., LDT 230.2) could watch.inf ‘so that their bodies could be seen by all the Trojans’

se self=

The texts show the prevalence of simple tenses and imperfective verb forms, with only few examples of the simple past. In the rare examples of compound tenses, only the S–se–V order (with [an.] S) is attested: (47)

quisto nuostro exercito non se avesse this our army not self= had.sbjv potuto guidare (ONap., LDT 202.14) can.ptp lead.inf ‘this army of ours could not have been led’

ben well

Impersonal se is very frequent with verbs which can be construed as bivalent but with an unexpressed object (48), and well attested with monovalent verbs, especially

90

Michela Cennamo

unaccusatives denoting telic change of location (49), although never in compound tenses with the latter: (48) secundo che se cunta de quilli che [ . . . ] (ONap., LDT 205.4) according that self= tells of those who ‘according to what they (indef.)/people say about those who [ . . . ]’ (49) a quista insula non se nce to this island not self= there= non per mare (ONap., LDT 51.2) not for sea ‘one could only reach this island by sea’

poteva could

gire go.inf

se if

With monovalent verbs, impersonal se only occurs in imperfective tenses. It never occurs with specific time reference, which would trigger the 1pl interpretation of the unexpresssed agent with unaccusatives. This interpretation is excluded from the texts investigated (Cennamo 2000: 99). In terms of the interpretation of se, it refers to an unspecified human participant, which does not include the speaker/narrator. This is shown by the fact that se never alternates with the 1pl, but only with the 3pl, and at times the 2sg (Cennamo 2000: 99) 4.4.5 Old Logudorese Sardinian In 11th–13th-century old Logudorese texts the passive/impersonal reflexive occurs both in the S–se–V and se–V–S order, whilst there are no examples of impersonal reflexives with monovalent verbs. The postverbal position of S in these patterns either reflects its pragmatic status or is due to syntactic factors such as the presence of an adverb (50b) or the heaviness of the subject (50c): (50)

a. sa fura [ . . . ] se iudicait (OLog., CSNT 140.5) the theft self= judged ‘the theft was debated in court’ b. uue se fekit tottu suki est iscrittu where self= did all the.that is written daue susu (OLog., CSPS 383.13) from above ‘where all this was done is written above’ c. nareuos progitteu se condennait custu tramutu ki I.tell.sbjv=you why self= condemned this exchange that feki s’ apatissa Tedora cun Petru de Thori (OLog., CSPS 388.1) made the abbess Teodora with Petre of Thori ‘let me tell you why this exchange was disapproved that I, Abbess Teodora, made with Petru de Thori’

Passive and impersonal reflexives in the Italian dialects

91

The expression of the agent is attested later, in 14th- to 15th-century texts (51a), where passive/impersonal si also occurs with animate subjects (51b), occasionally in perfective tenses (51c), albeit only with bivalent verbs, never with monovalent ones. With the latter, impersonal si occurs mainly with unaccusatives like andare ‘go’ (51d), always with a non-referential, exclusive interpretation: (51)

a. sas quales se deuen dare the which self= must give.inf persone (OLog., SRS XXVIII.8) person ‘which must be given by everybody’

daue by

chalunque whatever

b. qui non si podiat hobligare sensa licentia who not self= can force.inf without permission desu p[re]ladu (OLog., RSPS (18r.) 82.16-17) of.the priest ‘who one cannot force without his priest’s permission’ c. qui pius non si li siat dadu who ever not self= to.him= be.sbjv given termen (OLog., RSPS (64r.), 290.19-20) limit ‘that one has never given him a temporal limit’ d. fina assa uia per issa quale se until to.the path through it which self= ad Osilo (OLog., SRS XXXIII.7) to Osilo ‘as far as the path by which one goes to Osilo’

uaet goes

4.4.6 Interim summary The early vernaculars investigated reveal an uneven attestation of the impersonal reflexive, unlike the passive/impersonal one, which occurs in all varieties, most typically with [—an.] subjects—pre-/postverbal, depending on their given/new status—and more rarely and at subsequent stages with [þan.] ones. With monovalent verbs, in fact, impersonal si/se is not found in early Venetian and Logudorese Sardinian texts, where it is attested from the 14th century, and in old Neapolitan and old Logudorese Sardinian it never has an inclusive interpretation.

4.5 Converging constraints on impersonal si/se The analysis of synchronic and diachronic data shows a different distribution and incidence of the reflexive pattern in impersonal function, but also some converging

92

Michela Cennamo

parameters: (i) animacy; (ii) tense/aspect characteristics; (iii) the interpretation/referential domain of the reflexive morpheme; (iv) the grammaticalization of the distribution of the pragmatic notions given/new. More specifically, in passive and passive/impersonal patterns some varieties do not allow these structures with animate subjects, and in the early vernaculars investigated these constructions occur only/mainly with inanimate subjects; there is also a prevalence of imperfective contexts, both synchronically and diachronically. In some contemporary varieties the reflexive cannot occur in perfective tenses and with specific time reference, i.e. in bounded contexts. Indeed, the boundedness of the event might play a role in the absence of the reflexive with unaccusatives in perfective tenses and with specific time reference in some contemporary dialects. The interpretation/referential domain of the reflexive morpheme might have played a role, instead, in the extension/non-extension of the reflexive morpheme in impersonal function, to mark initially A, subsequently and optionally S. In the dialects where the reflexive can only have either a generic or an indeterminate/existential non-inclusive interpretation (e.g. Neapolitan), it occurs in a narrower range of impersonal patterns (for instance, it does not occur in equative impersonal patterns and in impersonal passives, for which other strategies are employed). In the varieties where the reflexive occurs also with an existential inclusive interpretation, it occurs in all the syntactic domains in which it is attested in standard Italian and in perfective contexts as well (e.g. Paduan). Finally, in old Venetian the grammaticalization of the distribution of the pragmatic notions given/new, characteristic of several northern varieties, is already in evidence.

4.6 Some unsolved issues: the status of the reflexive morphemes si and ci 4.6.1 One or two si’s? Synchronic dialectal data give contrasting evidence as to the existence of just one si—with variable function (reflexive, passive, impersonal), interpretable as a variable whose value is fixed either through coreference (as with the reflexive and passive pattern) or through the binding of a generic operator (as with the impersonal structure) (Manzini and Savoia 2005: ii.70)—or of two si’s, either a reflexive and passive si vs an impersonal one or a reflexive vs a passive and impersonal si, with some varieties allowing only either passive or impersonal si (Manzini and Savoia 2005: ii.57). In point of fact, several northern varieties do not differentiate among reflexive, passive, and impersonal si, which occur in the same position as object clitics (cf. }4.4; Manzini and Savoia 2005: ii.64), whilst in other northern, central, and southern dialects (including northern Sardinian dialects), impersonal si occurs in initial

Passive and impersonal reflexives in the Italian dialects

93

position, preceding all types of object clitic, unlike reflexive and passive si (cf. Manzini and Savoia 2005: ii.19–62). Diachronic data, on the other hand, appear to support the claim concerning the existence of two different homophonous si’s: a reflexive ~ passive/impersonal si and an impersonal si. In point of fact, whereas passive/impersonal si is attested in all the early vernaculars investigated, impersonal si is hardly found in old Logudorese (where it is not attested in the early texts), and in both old Logudorese and old Venetian it is found in later texts than passive/impersonal si. Indeed, synchronic and diachronic variation support the findings in Cennamo (1993a; 1993b; 2000) whereby passive and impersonal reflexives are two different structures, reflecting two different diachronic paths which at some point merge. While the passive reflexive is a late Latin development, related to changes in the encoding of voice in the transition from Latin to Romance, the impersonal function of the reflexive with one-argument verbs is a Romance phenomenon, not equally attested in the Romance languages and in the Italian dialects. More specifically, the rise of impersonal si in Italo-Romance may be related to a stage where the reflexive pronoun acquires a non-anaphoric pronominal value, whereby Latin se becomes equivalent to is ‘he’ (52)—a pattern that is well attested in Italo-Romance, both diachronically and synchronically (Rohlfs 1968: }480, 458, n. 1; Manzini and Savoia 2005: ii.121–4)—also referring to 1st and 2nd person participants in its reflexive/middle use in late texts, as illustrated in (53a,b), where se is equivalent to nos ‘we’ and even me ‘I’ (Cennamo 1991: 17; 1993a: 81): (52) ipsi [ . . . ] sibi they self.dat= ‘they trusted him’

crediderunt (Lat., Trad. Frising. 553; Cennamo 1991: 8) believed

(53) a. inter se singuli dissimiles between themselves each different invenimur (Lat., Min. Fel., 18.15; II AD) we.find.pass ‘we find ourselves different from each other’ b. Ego [ . . . ] Adhalhardus [ . . . ] recogitans I Adhalhardus thinking se (Lat., Tardif: N: R 59; Cennamo 1993b: 58) self= ‘I Adhalhardus, [ . . . ] thinking [ . . . ]’ Indeed, a possible starting point for the acquisition of the impersonal function of the continuators of Latin se might have been the non-anaphoric pronominal function and the gradual widening of the referential domain of si/se in the course of time, whereby se/si comes to include the 1st and 2nd person participants, due to the 1pl interpretation that the reflexive morpheme acquires in the course of time in its

94

Michela Cennamo

impersonal function. This parallels a change that had already occurred in Latin in its reflexive/middle use, whereby Latin se = nos and even me in late texts alongside the non-reflexive, anaphoric pronominal use of se, whereby se = is, as illustrated in (52) and (53). As a result of the above-mentioned factors—(i) its use in non-anaphoric pronominal function; (ii) the gradual widening of its referential function; and (iii) the gradual grammaticalization of its ‘new’ referents, speaker and hearer, as in its 1pl interpretation—se/si becomes an indefinite pronoun, and gradually ousts the homo ‘man’ type in the early Italian vernaculars, which could only occur with a generic interpretation, never an existential one (Egerland 2003; Manzini and Savoia 2005: ii.132). Thus the impersonal use of the reflexive morpheme with one-argument verbs and various types of impersonal pattern appears to be related to a series of changes where the Person Hierarchy is one of the features involved (Cennamo 1993a: 81; 1993b: 60). 4.6.2 Impersonal ci The analysis of the synchronic distribution of the 1pl reflexive clitic ci in impersonal function in some Italian dialects points to the use of ci as an indefinite pronoun in some varieties on a par with si. In some varieties ci is equivalent to si in reflexive, anticausative, passive, and impersonal function. We can hypothesize a stage where se and ci become functionally equivalent, competing in some areas, with ci replacing si as a marker of agent defocusing/argument suppression. Thus in the ci–si–V patterns as in ci si lava – ci= self= washes (‘one washes oneself ’), ci si pente – ci= self= repents (‘one repents’), ci may be regared as an impersonal pronoun and si as the reflexive pronoun, following the traditional description. The alternative view, put forward in Cinque (1988a), Salvi (2008a; 2010b), and Bentley (2006), regards ci as the 1pl pronoun and si as the impersonal pronoun. According to Salvi (2008b; 2010a) the ci si lava pattern originates in Tuscan (noi) si lava lit. ‘(we) self= washes’/ci si lava lit. ‘ci= self= washes’ = laviamo lit. ‘we.wash’/ci laviamo lit. ‘us= we.wash’. It is attested in 19th-century northern writers such as Rovani and Nievo (Salvi 2008a: 17; 2010b) and might have developed as a result of the interference between the native language of these northern writers, where the impersonal reflexive only had a generic/indeterminate, exclusive interpretation, and the higher-prestige Tuscan variety on which the literary language is based (Florentine), where the impersonal reflexive pattern could occur also with an inclusive interpretation. Therefore northern writers/speakers chose the system of rules of the higher-prestige variety that had a wider range of functions than their native varieties. This might have led to the innovations introduced by northern writers/ speakers (one of which is the ci–si–V pattern), which then became consolidated in literary and spoken Italian (Salvi 2010b: 7).

Passive and impersonal reflexives in the Italian dialects

95

Synchronic variational data, however, show the occurrence of ci as an impersonal marker in several Italian dialects from various geographical areas. It is therefore more plausible to think of a stage of alternations/competition between si and ci in impersonal function, namely as voice markers (e.g. in Tuscan, where both impersonal ci and si are attested), with ci prevailing in some areas.

4.7 Conclusions The synchronic and diachronic data investigated show that the spread of the morpheme si (and its variants) to impersonal function might be related to the gradual extension of its referential domain and its degree of grammaticalization. There appears to be a correlation between the syntactic contexts in which the impersonal reflexive occurs and its generic or indeterminate/existential inclusive interpretation. Dialects where the reflexive never acquires an indeterminate/existential inclusive interpretation show a limited use of this strategy in impersonal patterns. More specifically, the data point to the existence of two different si’s, reflecting diffferent diachronic paths converging into an apparently unitary pattern. While the passive reflexive is a late Latin development, related to changes in the encoding of transitivity in the transition from Latin to Romance, the impersonal function of the reflexive with monovalent verbs is a Romance phenomenon, not equally attested in the Romance languages and in the Italian dialects, related to a stage where the reflexive pronoun acquires a non-anaphoric pronominal value, whereby Latin se = is, occurring also to refer to 1st and 2nd person participants (se = nos). As for ci in the ci–si–V patterns in standard Italian, it can be regarded as an indefinite pronoun, having an indeterminate/existential interpretation, either inclusive or non-inclusive, rather than the phonologically conditioned allomorph of si.

5 On the personal infinitive in Sicilian* D E L I A BE N TL E Y

5.1 Introduction A number of Romance languages exhibit constructions where an infinitival predicate would seem to have a subject of its own. In all the relevant languages this can be spelled out by a nominative noun or pronoun (Campidanese Sardinian, Catalan, Occitan, Romanian, Spanish, etc., cf. (1a)), and in a subset of these languages the infinitive bears in addition an overt inflection which agrees with the nominative argument, whether covert or overt (Galician, Nuorese Sardinian, Portuguese, old Neapolitan, etc., cf. (1b,c)). (1)

a. Se fué antes self= went.3sg before ‘S/he left before I arrived’

de of

b. Depois de chegarem, after of arrive.inf.3pl ‘After they arrived, we fled’ c. [ . . . ]

yo I

llegar arrive.inf (Cmb. Sp., Kany 1951: 126)

fugimos fled.1pl

per nuy averemo for we have.inf.1pl ‘ . . . so that we may receive orders’

(Pt., Willis 1971: 338)

ordene orders (ONap., Ledgeway 2000: 113)

These constructions constitute an innovation with respect to Latin, where the dependent infinitive can take neither inflection nor a nominative argument.1 We shall refer to the structure illustrated in (1a) as the personal infinitive, whereas

* This is a small token of my immense gratitude to Mair Parry for her infectious passion for the dialects of Italy. I am also thankful to Paola Benincà, Adam Ledgeway, and Nigel Vincent for comments which have improved the quality of this chapter. 1 The historic infinitive, which has a nominative argument, occurs in main clauses.

On the personal infinitive in Sicilian

97

we shall call the structure in (1b,c) the inflected infinitive. Unlike the inflected infinitive (Pountain 1995; Vincent 1996; 1998), the personal infinitive is solely found in non-obligatory control domains, i.e. syntactic domains where the reference of the nominative argument of the infinitive is not determined by the matrix verb (e.g. Cuneo 1997). As a result of this restriction, in general, the personal infinitive figures in non-subcategorized contexts (Ledgeway 2000: 117–25), in particular adverbial clauses and some subject clauses (e.g. clauses following impersonal ‘be’ plus an adjectival predicate, cf. (2)), but not in object complement clauses (in this respect, Sicilian constitutes an exception, as will be pointed out in }5.2). (2)

L’ è in bèlu afâre, êse lasciü cölu lì! scl is a nice thing be.inf there.up that there ‘It is really quite something for him to be up there’ (Cicagna (GE), Liguria, Cuneo 1997: 107)

The personal infinitive is frequently attested with passive or copular ‘be’ (cf. (2)), as well as in unaccusative predications (cf. (3a)), whereas it is more rarely found in unergative and transitive predications (cf. (3b)) (Cuneo 1997: 105). (3)

a. [ . . . ] prüma de sciurtî u fjûre before of come.out.inf the flower ‘before the flower comes out’

(Cicagna (GE), Liguria)

b. L’ a fatu tantu pe stüdjâ i figi scl has done so.much for study.inf the children ‘He did so much so that his children could study’ (Cicagna (GE), Liguria) Across Romance, a lexical nominative argument follows the personal infinitive (cf. (3a,b)), with very few exceptions: for example, one that will be discussed in }5.2. In the present case study, we restrict our attention to the personal infinitive in Sicilian, reporting the findings of the analysis of a corpus of 14th- and 15th-century vernacular texts drawn from the well-known collection of the Centro di Studi Filologici e Linguistici Siciliani, and comparing these results with evidence from Giuseppe Pitrè’s 19th-century short stories and from the modern dialect. We compare the Sicilian structure to its counterparts in other Romance languages. In addition, we look at the way this structure has evolved during the period in question.

5.2 The personal infinitive in early Sicilian In early Sicilian, non-finite predication figures prominently. In her collection of 14thcentury texts (public and private letters, wills, price lists, etc.), Rinaldi (2005: 486–9) finds that the incidence of gerunds and infinitives is particularly high.

98 (4)

Delia Bentley a. Vui, zo non attendend[u] [ . . . ], non lassati perchipiri li you this not doing not let collect the ditti grana per lu dictu Bundu said money by the said Bundu ‘Failing to do this, you do not let the said Bundu collect the said money’ (OSic., Rinaldi, Testi d’archivio, 2005, no. 27, 69–70, 6–10) b. La virga di Aron, ki apersi lu Mari Rubeu a passari the stick of Aron which opened the Sea Red to pass li figli di Isdrael the children of Israel ‘Aron’s stick, which opened the Red Sea for the children of Israel to pass’ (OSic., Rinaldi, Testi d’archivio, 2005, no. 19, 55, 5–6) c. Vi plaza ki kista literra non essiri disscoperta you please.sbjv that this letter not be disclosed ‘May you agree not to disclose the contents of this letter’ (OSic., Rinaldi, Testi d’archivio, 2005, no. 108, 219, 4)

Unlike other early Italo-Romance vernaculars—notably old Neapolitan, the prime Italo-Romance example of a language with inflected infinitives—as well as inflected gerunds and participles (Savj-Lopez 1900; Loporcaro 1986; Vincent 1996; 1998; Ledgeway 1998; 2000; 2009a: 585–90), early Sicilian only exhibits bare (i.e. uninflected) infinitival and gerundival predicates with a nominative argument, the infinitival type being the focus of our discussion. To begin with, we note that, typically, the early Sicilian personal infinitives are intransitive predications (e.g. copular or passive ones, cf. (5a,b)), although there are also rare attestations of the personal infinitive in transitive predications (cf. (5c)). (5)

a. Eu possu raxunivilmenti pensari kistu chelu essiri plui altu ki non esti I can reasonably think this sky be.inf more high than not is ‘I can reasonably think that this sky is higher than it is’ (OSic., Palumbo, Sposizione del Vangelo, 1954: 16, 24–5) b. Si era necessariu di essiri adimpluti li Scripturi if was necessary of be.inf fulfilled the Scriptures ‘If the Scriptures would be fulfilled’ (OSic., Palumbo, Sposizione del Vangelo, 1954: 185, 18) c. Hai factu vutu, congnoscendu tu non lo potiri compliri? have made vow knowing you not it can.inf keep ‘Have you made a vow, knowing that you would not be able to keep it?’ (OSic., Confessionali, in Branciforti, Regole, 1953: 166, 31)

On the personal infinitive in Sicilian

99

In this respect, the distribution of the early Sicilian personal infinitive is comparable with that of the other Romance languages (see }5.1). Cuneo (1997: 108), for example, notes that the personal infinitives of the Ligurian dialect of Cicagna tend to figure in unaccusative structures, although there appear to be no absolute restrictions on unergative and transitive structures with personal infinitives in this dialect. Since the object of transitive personal infinitives is not within the scope of our study, we will henceforth refer to the nominative argument as the argument of the infinitive. A number of facts indicate that the early Sicilian personal infinitive only occurs in clausal domains, thus turning out to be comparable to the inflected infinitive of early Neapolitan in this respect (Vincent 1996). We provide some relevant evidence here. Cross-linguistically the sharing of arguments, as well as operators of root modality and internal negation (negation of the predicate and/or one of its arguments), occurs across syntactic boundaries which are smaller than the clause (Van Valin and LaPolla 1997: 40–52). Accordingly, if the personal infinitive and the preceding inflected predicate occur in separate clausal domains, they should not be able to share arguments, root modality operators, or internal negation (see Skytte 1983: 292–314 with respect to Italian). Indeed, we have found no examples of the structure normally known as clitic climbing (Rizzi 1978), which suggests that a pronominal clitic hosted by the personal infinitive cannot be shared by the preceding inflected predicate. Observe the example in (6a): here the clitic chi ‘there’ is hosted by the infinitive itself, and not by the inflected predicate vulissi ‘wanted’. Argument sharing is instead the norm in causatives with fari ‘make’ and lassari ‘let’, which never exhibit the personal infinitive (cf. (6b)). (6)

a. Et vulissi homu tu non chi essiri and want.sbjv man you neg there= be.inf ‘And one would like you not to be there’ (OSic., Rinaldi, Testi d’archivio, 2005, no. 86, 184, 21) b. Lu dimoniu mi lu fichi the devil me=it=made ‘The devil made me do it’

fari do.inf (OSic., Luongo, Renovamini, 1989: 57–8, 10)

The fact that the argument of the infinitive exhibits nominative case also indicates that it is not an argument of the inflected predicate. Compare (6a) with (6b): whereas in (6a), the infinitive essiri ‘be’ has a nominative argument (tu ‘you’), in (6b) neither of the arguments of the infinitive is nominative, since they are shared by causative fichi ‘made’, which has a nominative argument of its own, lu dimoniu ‘the devil’. Nominative case marking on the argument of the infinitive is spelled out not only on the form of the pronoun (tu in (6a), rather than ti) but also by the lack of the accusative case marker a on the classes of nominal and pronominal objects which can exhibit this marker in early Sicilian as well as in other Romance languages (e.g. Bossong 1991; Sornicola 1992: 467–9; 1997; Rinaldi 2005: 450; Guardiano 2010).

100

Delia Bentley

This type of accusative case marking is not found on the argument of the personal infinitive. When the personal infinitive is negated, negative non follows the nominative argument (cf. (6a) and (7a)). If, on the other hand, it is the preceding inflected predicate that is negated, the personal infinitive is not within the scope of negation (cf. (7b)). (7)

a. Muriu a la cruchi per nui non muriri died at the cross for we not die.inf ‘He died on the cross so that we would not die’ (OSic., Palumbo, Sposizione del Vangelo, 1954: 280, 6) b. Eu non criiu Cristu essiri santu I not believe Christ be.inf saint ‘I do not believe that Christ is saint’ (OSic., Palumbo, Sposizione del Vangelo, 1954: 326, 1–2)

Like the scope of negation, that of root modals is constrained either to the matrix predicate or to the personal infinitive, but cannot encompass both. This is clear in (8a), where possu ‘I can’ has scope over intendiri ‘understand’, and poti ‘he can’ over fari ‘make’, but the personal infinitive essiri ‘be’ is not within the scope of these modals.2 By contrast, in (8b) the modal vuliri ‘want’ has scope solely over the infinitive. (8)

a. Ben possu eu intendiri, et Deu multo melius poti fari, well can I understand and God much better can do un midesmi pani essiri dui volti pani et una volta a same bread be.inf two times bread and one time factu corpu di Cristu made body of Christ ‘I can indeed understand, and much better than I understand God can make the same bread be twice bread and once the body of Christ’ (OSic., Palumbo, Sposizione del Vangelo, 1954: 128, 7–11) b. Sperandu tu vuliri studiari et no lu fachisti hoping you want.inf study and not it did ‘Hoping that you would be willing to study, but you did not do this’ (OSic., Rinaldi, Testi d’archivio, 2005, no. 86, 184, 27)

The evidence in (6a,b) to (8a,b) thus supports the hypothesis that the personal infinitive occurs in a clausal domain of its own. Since tense is a clausal operator (Van Valin and LaPolla 1997: 47), the possibility that the temporal reference of the personal infinitive can be different from that of Observe that vuliri ‘want’ can be a matrix predicate, rather than an operator or serial verb expressing root modality, in which case it does not modify an infinitive in the matrix clause (cf. (6a)). This fact is irrelevant to our claim that the personal infinitive occurs in a clausal domain of its own. 2

On the personal infinitive in Sicilian

101

the inflected predicate should indicate that the two predicates belong to separate clausal domains. The early Sicilian data indeed suggest that the personal infinitive has independent temporal reference (cf. (9a–c)), even though this can arguably be anaphoric, insofar as it is evaluated in relation to that of the matrix clause. Needless to say, the temporal reference of the infinitive is not encoded in its morphology, and normally has to be understood from the context. (9)

a. li bocti ki sustinni lu nostru signur Iesu Cristu supra li spalli, the blows that received the our lord Jesus Christ on the shoulders per nui essiri liberati da li flagelli eterni di lu infernu for we be.inf freed from the pains eternal of the hell ‘The blows which our Lord Jesus Christ received on his back in order for us to be freed from the eternal pain of hell’ (OSic., Palumbo, Sposizione del Vangelo, 1954: 63, 5–6) b. Hai factu vutu, congnoscendu tu non lo potiri compliri? have made vow knowing you not it can.inf keep ‘Have you made a vow, knowing that you would not be able to keep it?’ (OSic., Confessionali, in Branciforti, Regole, 1953: 166, 31) c. Lu profeta Zacharia previdi Cristu essiri passionatu the prophet Zacharia forsaw Christ be.inf sacrificed ‘The prophet Zacharia foresaw that Christ would be sacrificed’ (OSic., Palumbo, Sposizione del Vangelo, 1954: 161, 24–5)

The evidence in (9a–c) supports Ledgeway’s (1998) claim that, whilst not bearing agreement, the personal infinitive is endowed with tense. Since the personal infinitive has tense, we conclude that it constitutes a clausal domain in syntax. With respect to the types of clause which exhibit the personal infinitive in the early Sicilian texts that we have examined, these can be adverbial purpose or causal clauses with the subordinator per (cf. (10a,b) and (10c), respectively). (10)

a. Deu omnipotenti, ki [ . . . ] si mecti a morti per nui viviri eternamenti God omnipotent that self= puts to death for we live.inf eternally ‘Omnipotent God, who [ . . . ] dies so that we shall live eternally’ (OSic., Palumbo, Sposizione del Vangelo, 1954: 11, 28–30) b. Muriu a la cruchi per nui non muriri died at the cross for we not die.inf ‘He died on the cross so that we would not die’ (OSic., Palumbo, Sposizione del Vangelo, 1954: 280, 6) c. Hai negatu li toi parenti, per ipsi essiri poviri? have disowned the your relatives for they be.inf poor ‘Did you disown your family because they were poor?’ (OSic., Confessionali, in Branciforti, Regole, 1953: 172, 3)

102

Delia Bentley

The personal infinitive also figures in subject clauses and, more conspicuously and unexpectedly (see }5.1), in object clauses (cf. (11a) and (11b–d), respectively, as well as (8a,b) and (9b,c)). (11)

a. Cussì esti per necessitati di materia homo muriri [ . . . ] spata ruginiri thus is for necessity of matter man die.inf sword rust.inf ‘Thus, that men should die . . . and swords should rust is a necessity of matter’ (OSic., Palumbo, Sposizione del Vangelo, 1954: 26, 21–2) b. Canuxivi ipsu fari falsu testimoniu knew he do.inf false testimony ‘I knew that he gave false testimony’ (OSic., Confessionali, in Branciforti, Regole, 1953: 141, 31–2) c. Cussì pensu eu Cristu essiri in chelu localmenti thus think I Christ be.inf in heaven physically ‘Thus, I believe that Christ is physically in heaven’ (OSic., Palumbo, Sposizione del Vangelo, 1954: 122, 8–9) d. Li iudei avirianu dictu lu corpu di Cristu essiri statu fantasticu the Judaeans have.cond said the body of Christ be.inf been imaginary ‘The Judaeans would have said that the body of Christ was imaginary’ (OSic., Palumbo, Sposizione del Vangelo, 1954: 243, 24–5)

Finally, we have come across one occurrence of the personal infinitive in an adjectival complement. (12) Si lu regnu di Iudea sia dignu di tucti essiri minati per la spata if the kingdom of Judaea be.sbjv worthy of everyone be.inf led by the sword ‘If the kingdom of Judaea deserves that everyone should be punished by sword’ (OSic., Palumbo, Sposizione del Vangelo, 1954: 48, 3–4) Significantly, we have not found the personal infinitive in obligatory control domains, i.e. complex predications where the reference of the argument of the embedded predicate is determined by the verb in the matrix clause (‘promise’, ‘decide’, ‘try’, etc.). Example (13) illustrates an obligatory control structure without the personal infinitive. (13) Promictimu diri et nunciari lu dimoniu promise.1pl say.inf and announce.inf the devil ‘We promise to warn people against the devil’ (OSic., Renovamini, Luongo, 1989: 20 (4), 25) In being absent from obligatory control domains, the personal infinitive of early Sicilian behaves in the same way as its counterpart in other Italo-Romance dialects and in Romance in general (see }5.1).

On the personal infinitive in Sicilian

103

The restriction of the personal infinitive to non-obligatory control domains leads one to think that the personal infinitive has the function of a switch-reference strategy, indicating that the reference of the argument of the infinitive is not the expected one in the given context, normally the subject of the main clause. If this were the case, complementation with the personal infinitive would be comparable to complementation with the subjunctive in modern Italian (contrast (14a) with (14b)). (14)

a. Voglio che tu / Maria want that you / Maria ‘I want you / Maria to leave’

parta (It.) leaves.sbjv

b. Voglio partire (It.) want leave.inf ‘I want to leave’ Observe, however, that adverbial clauses with the bare infinitive can also involve switch reference in early Sicilian, although this structure only figures rarely in our corpus. (15)

a. Fichi li homini per viviri iustamenti et essiri beati made the men for live.inf justly and be.inf blessed in corpu et in anima in body and in soul ‘He made man (lit. men) for him to live in justice and be blessed in flesh and soul’ (OSic., Palumbo, Sposizione del Vangelo, 1954: 188, 22–4) b. Lu re Saul et li soi persequitaru David per distrudirilu, the king Saul and the his persecuted David for destroy.inf.him per non essiri re di li iudei for not be.inf king of the Judaeans ‘The king Saul and his men persecuted David to destroy him, so that he would not be the king of the Judaeans’ (OSic., Palumbo, Sposizione del Vangelo, 1954: 233, 9–10)

In (15a), the argument of the bare infinitives viviri ‘live’ and essiri ‘be’ is not coreferential with the subject of the predicate of the main clause fichi ‘he made’, but rather with its object. In (15b), the first bare infinitive distrudiri ‘destroy’ does not involve switch reference, whereas the argument of the second bare infinitive, essiri ‘be’, is coreferential with the object of the main clause, i.e. David. The free reference of the covert argument of subordinated infinitival clauses also characterizes other Romance languages with the personal infinitive (Cicagnese, see Cuneo 1997: 112–16; Occitan, Wheeler 1988: 270; and Sardinian, Blasco Ferrer 1988: 159; Jones 1988: 344; 1992: 296).

104 (16)

Delia Bentley a. G’ ô ditu au Giôśe de vegnî tüti inséme cl have said to.the Joseph of come.inf all together ‘I told Joseph that we (we and he) / they (he and they) should come together’ (Cicagna (GE), Liguria, Cuneo 1997: 116) b. Al tocar la pòrta, lo gos se botèt a jaupar at.the knock.inf the door the dog self= start to bark ‘When somebody knocked at the door, the dog began to bark’ (Occ., Wheeler 1988: 270)

This comparative evidence suggests that the pattern illustrated in (15a,b) is indeed a structural possibility of early Sicilian despite the paucity of its attestations. In our corpus, this possibility is confined to adverbial clauses. A change of reference from the subject of the main clause can also be found in finite adverbial clauses with the subordinators per tal ki, azò ki ‘so that’, (im)però ki ‘because’, etc., and the subjunctive. (17) Et eu elessi a vui, per tal ki zo ki vui fachissivu, and I chose to you.pl for such that that which you.pl do.sbjv cui avi intellectu potissi cunucxiri he.who has intellect can.sbjv know ‘And I chose you, so that he who has an intellect would be able to learn what you do’ (OSic., Palumbo, Sposizione del Vangelo, 1954: 82, 17–18) We thus have evidence which suggests that the personal infinitive signals a change of reference from the subject of the main clause, and evidence which indicates that there are other switch-referencing strategies in early Sicilian. Accordingly, we must seek the rationale which lies behind the alternation of the bare infinitive, the personal infinitive, and finite adverbial clauses. In the light of our findings, we suggest that the covert argument of the bare infinitive finds its reference in the main clause, regardless of the syntactic function of the coreferential argument (cf. (15a,b) and (18a)).3 The nominative argument of the personal infinitive need not be coreferential with an argument of the main clause (cf. (10a,b)). However, if it is, it has a disambiguating function, for example when it is not coreferential with the subject of the main clause (cf. (10c)) or there are two possible antecedents (cf. (18b)). Like the nominative argument of the personal infinitive, the subject of finite adverbial clauses need not be coreferential with an argument of the main clause (cf. (17)). If there is such coreference, there is no ambiguity and hence no need for an overt pronoun (cf. (18c)). Needless to say, an overt pronoun can nonetheless be found in unambiguous structures to mark focus, as is the case with (18d), where ipsu is contrasted with Iesu Cristu. 3 The bare infinitive may also have a generic argument or be impersonal, in which case there is no covert argument. An early Sicilian example of this structure is cited in Sornicola (1992: 473): et narrandolo a le sue compagne per essere certo che . . . ‘and she told her companions in order for it to be certain that . . . ’.

On the personal infinitive in Sicilian (18)

105

a. Si dasti venenu ad alcunu per muriri if gave poison to someone for die.inf ‘If you poisoned anyone to make them die’ (OSic., Confessionali, in Branciforti, Regole, 1953: 143, 6) b. Item lu frati di lu summu pontifichi auchidia therefore the brother of the high pontiff killed lu summu pontifichi, per ipsu aviri lu sacerdociu the high pontiff for he have.inf the priesthood ‘Therefore, the brother of the high pontiff would kill the high pontiff in order to gain the priesthood’ (OSic., Palumbo, Sposizione del Vangelo, 1954: 73, 4–6) c. Lu salvaturi dissi una secunda parola [ . . . ]. Et dissila per tal ki the saviour said a second word and said.it for such that ni insignassi ki . . . us taught that ‘The Saviour said a further word [ . . . ]. And he said it so that he would teach us that . . . ’ (OSic., Palumbo, Sposizione del Vangelo, 1954: 85, 20–22) d. Lu cristianu esti mortali, per tal ki ipsu aia di putiri rendiri the christian is mortal for such that he have.sbjv of can offer a Cristu, ki possa muriri per amuri di Iesu Cristu, comu to Christ that may die for love of Jesus Christ as Iesu Cristu muriu per amur so Jesus Christ died for love his ‘Christians are mortal, so that they can offer to Christ their death for the love of Jesus Christ, as Jesus Christ died for his love of them’ (OSic., Palumbo, Sposizione del Vangelo, 1954: 34, 19–21)

Although early Sicilian does allow non-finite complementation with a nominative noun (cf. 7b, 11a,c,d), the nominative argument of the personal infinitive tends to be a pronoun in adverbial clauses. This has an antecedent in the main clause or is a speech act participant (see the pronoun nui ‘we’ in 9a, 10a,b). We have only found three examples which deviate from this pattern, two with passive infinitives (cf. 19a,b) and one with the existential copula essiri ‘be’ (cf. 19c). (19)

a. Tranbudui cadinu in la fossa [ . . . ] per non esseri both fall in the pit for not be.inf facta la confessioni comu si divi. done the confession as one must ‘They both fall in the pit because confession has not been done properly’ (OSic., Luongo, Renovamini, 1989: 66, 13–14)

106

Delia Bentley b. Parterusi li vestimenti soi, mictendu li xorti, share=self the clothes his putting the draw per essiri adimplutu zo ki esti dictu per lu propheta for be.inf fulfilled that which is said by the prophet ‘They drew for his clothes and shared them, so that what was said by the prophet would come true’ (OSic., Palumbo, Sposizione del Vangelo, 1954: 227, 12–13) c. Innocentissimu homu moira piacentissimamenti per essiri innocent.superl man die.sbjv willingly.superl for be.inf salvacioni et liberacioni di li pperduti genti salvation and freeing of the lost mankind ‘A most innocent man should be willing to die so that there would be salvation for the lost mankind’ (OSic., Palumbo, Sposizione del Vangelo, 1954: 11, 25–6)

Example (19c) is hardly representative, since the post-infinitival nouns spell out the pivot (Milsark 1977) of an existential construction. Existential pivots have been claimed to be predicates semantically, or covertly (La Fauci and Loporcaro 1993; 1997; Moro 1997; Francez 2007; 2009), whereas syntactically they are known to behave differently from the subject of other mono-argumental predications (Bentley 2010). Accordingly, only (19a,b) can be considered to be real exceptions to the tendency for the argument of the personal infinitive to be pronominal in adverbial clauses. In the light of the above findings, we propose that, in early Sicilian, non-finite adverbial subordination tends not to introduce a new referent into the discourse. Due to the lack of a restriction on the reference of the argument of the infinitive, which can be provided by an argument with any syntactic function in the main clause, or can be a speech act participant, the infinitive is accompanied by an overt pronoun, if this reference is not unambiguous. Finite adverbial subordination, on the other hand, can introduce a new referent into discourse, and therefore its subject may not be covert or pronominal. This is the principal rationale for the alternation between finite and non-finite adverbial subordination in our corpus of early Sicilian vernacular texts, a rationale which, as will be pointed out below, appears not to have been maintained in modern Sicilian. With respect to the non-finite subject and object complement clauses of early Sicilian, the evidence (cf. (7b), (8b), and (11a–d)) indicates clearly that the nominative argument of the infinitive need not be coreferential with an argument in the main clause. The robust presence of the personal infinitive in object complement clauses sets early Sicilian apart from modern Romance, another counterexample to the modern Romance tendency being found in Sardinian (Jones 1992; Cuneo 1997: 102; Ledgeway 2000: 117–22). Importantly, the object complement clauses which exhibit the personal infinitive are not within obligatory control domains.

On the personal infinitive in Sicilian (20) a. Non

kèlio a vénnere not want to come.inf ‘I do not want you to come’

107

tue you (Nuo., Sardinia, Jones 1992: 295)

b. Non bollu a ddu scí nisciunus not want to it know.inf nobody ‘I do not want anyone to know’ (Cmp., Sardinia, Blasco Ferrer 1988: 159) In Sardinian, object complement clauses with a personal infinitive are preceded by the subordinator a, whilst this is not the case with early Sicilian non-finite complement clauses.4 The lack of a subordinator in these clauses is reminiscent of the accusative and infinitive structure of Latin (Sornicola 1992: 471–2; Rinaldi 2005: 486).5 The same could be said of the possibility for the argument to precede the infinitive, whether it is a noun or a pronoun, although this argument is nominative, rather than accusative, and thus the Sicilian structure is innovative in this respect (see Skytte’s (1983: 293) observations on Italian). To be sure, the position of the subject in the Latin accusative and infinitive is far from fixed, but what matters in this context is its possible occurrence in pre-infinitival position. The possibility for a nominal nominative argument to precede the infinitive, while distinguishing complement clauses from adverbial clauses in early Sicilian, is hardly recorded in modern Romance. In early Sicilian, the lack of pre-infinitival nominal nominative arguments in adverbial clauses might be related to the tendency for these not to introduce a new referent into discourse (regardless of position, we have only found very few attestations of nominal nominative arguments of personal infinitives in adverbial clauses, cf. (19a–c)). However, the contrast between, on the one hand, the structures which do not allow pre-infinitival nominal arguments at all and, on the other, those which do is also likely to have a diachronic and synchronic syntactic rationale. To explore this rationale, we must first take a step backwards and ascertain the syntactic function of the argument of the personal infinitive. We believe case marking not to be a universal subjecthood diagnostic; rather, we take case to be assigned on the basis of the semantic macroroles of arguments (Van Valin 2005: 107–9).6

4

One exception to the lack of a complementizer is the subject complement clause in (5b). Observe, however, that Cecchetto and Oniga (2002) and Ledgeway (2012a: 240) argue that the accusative and infinitive construction involves a null complementizer, comparable with a null version of the English infinitival complementizer ‘for’. In accordance with this analysis, the accusative and infinitive only lacks an overt complementizer. 6 For Van Valin (2005: 107–9), case marking depends on macrorole status, rather than on grammatical relations (although there are specific languages where nominative case is indeed a subjecthood diagnostic). In accusative alignment, the highest-ranking argument in the semantic representation of the predicate of 5

108

Delia Bentley

In accordance with this hypothesis, the subjecthood status of the argument of the personal infinitive cannot be taken for granted. In the absence of inflectional agreement, the only other criterion which can be used to verify the grammatical relation that this argument has to the infinitival predicate is its position. The position of the subject can, of course, vary in accordance with syntactic or pragmatic factors. As is well known, this is the case with both early and modern Romance. However, we expect the behaviour, and hence the position, of the putative subject of infinitival clauses to be comparable to that of the subject of finite clauses. This expectation is borne out in early Sicilian, where the nominative argument of the personal infinitive can precede its predicate, on a par with the subject of other structures. We thus conclude that this argument is indeed a subject. Following Ledgeway’s (2000: 148–52) hypothesis that the nominative pronominal arguments of the personal infinitive are subject clitics, we propose that head marking (Nichols 1986) is by far the predominant strategy in the encoding of the subject in adverbial infinitival clauses, whereas dependent marking is amply recorded in complement infinitival clauses. As we pointed out, the latter type of structure is reminiscent of Latin insofar as the lack of a subordinator is concerned. It now turns out to be conservative also in the marking of the grammatical relation of the argument to the predicate. By contrast, adverbial infinitival clauses are truly innovative structures, where the grammatical relation subject is marked almost exclusively on the head. We return to the synchrony of complement and adverbial non-finite subordination below, when we deal with modern Sicilian. To conclude the discussion of early Sicilian, we consider the textual distribution of the personal infinitive. The comparative analysis of different types of early vernacular texts (volgarizzamenti aimed at different kinds of readership, constitutions of religious orders, legal documents, and letters) suggests that the innovative structure with the personal infinitive (cf. (10a–c)) characterizes the registers which do not strive towards the Latin and Tuscan models. The same cannot be claimed of the conservative structure found in complement clauses (cf. (11a–d)). In brief, in our admittedly small corpus, the adverbial clauses with the personal infinitive have been found in Sposizione del Vangelo (Palumbo 1954) and Confessionali (Branciforti 1953). The Confessionali are not volgarizzamenti, by contrast with the Regole, which, whilst being part of the same collection, do not exhibit the innovative structure.7 Branciforti’s (1953: ix–xxv) introductory notes on the text of the Regole acknowledge their adherence to Latin, as well as the influence of Tuscan, whereas with respect to the

the clause is assigned nominative case by default: this explains the nominative case marking of the argument of the Romance personal infinitives. Interestingly, in Regole (Branciforti 1953) we have found the personal gerund, which is also attested in Renovamini (Luongo 1989), and in Testi d’archivio (Rinaldi 2005). Whether the personal gerund pertains to higher registers or was felt to be less typically Sicilian is a question which we must leave to future research. 7

On the personal infinitive in Sicilian

109

Confessionali, the editor points to their practical function and to the authenticity of the language, ‘conservata in ottimo stato’ (‘very well preserved’) (1953: xxii). Sposizione del Vangelo (Palumbo 1954) is a volgarizzamento from a Latin text. However, its author has both a powerful command of the Sicilian of his time and a clear educational purpose. Indeed, he states that he has translated the text in order to educate a simple, i.e. uneducated, readership (Palumbo 1954: vii–xiv). We have not found any adverbial clauses with the personal infinitive in Renovamini (Luongo 1989), a volgarizzamento from Tuscan, with the exception of (19a), which, however, differs from the structure attested elsewhere, in that it has a post-infinitival nominative argument. The innovative personal infinitive in adverbial clauses is also absent from Rinaldi’s collection of early texts, which are characterized by the influence of other vernaculars and a preference for pan-Italian forms over exclusively Sicilian ones (Rinaldi 2005: 348–9). Significantly, in our corpus, switch reference with the bare infinitive is only attested in Sposizione (cf. (15a,b)) and Confessionali (cf. (18a)). The textual distribution of the innovative adverbial clauses with a personal infinitive does not have a parallel in the distribution of the conservative type of personal infinitive, i.e. that of complement clauses (cf. (6a), (8a,b), (9b,c), and (11a–d)). This was recorded both in the texts which are representative of the lower registers and in Rinaldi’s (2005) collection, which, as we pointed out above, is influenced by other vernaculars. This is shown in (6a) and (8b), as well as in the following example: (21)

Non si sapi mai genti vinchiri et fugiri not one knows ever people win.inf and flee.inf ‘One never hears of people winning and fleeing’ (OSic., Rinaldi, Testi d’archivio, 2005, no. 84, 178, 15)

In the light of the clear-cut diatextual distribution of the personal infinitive of adverbial clauses, we conclude that this is a dialectal innovation, rather than a learned structure, on a par with the inflected infinitive (Loporcaro 1986). This hypothesis is also supported by the attestation of this form in documents from the 18th century which have been shown to exhibit features of the low registers (Mocciaro 1991: 15, 62), as well as its survival in modern Sicilian (La Fauci 1984a: 122–3; Leone 1995: 59, 71–2; Ledgeway 2000: 115–54), which will be the topic of the next section.

5.3 The personal infinitive in modern Sicilian In modern Sicilian (19th century to the present day), the syntactic environments of the personal infinitive are somewhat reduced by comparison with early Sicilian. In particular, this structure still occurs in adverbial clauses (cf. (22a–f)), but it is not found in bare complement clauses (contrast (23a) with (23b)).

110 (22)

Delia Bentley a. Ni nni iemu pi ttu ristari sulu self= cl went for you remain.inf alone ‘We left in order for you to be left alone’ (mod. Sic., La Fauci 1984a: 122) b. Maistà, prima di io assirvari . . . majesty before of I observe.inf ‘Majesty, before I conduct the examination . . . ’ (mod. Sic., Pitré 1875: 40) c. Ddoppu di aviri studiatu after of have.inf studied ‘After they both tried for an hour’

tuttidui both

un’ a

ura hour

(mod. Sic., Pitré 1878: 175)

d. Chistu t’ havi a serviri pri manciari tu sulu e li toi this to.you have to serve for eat.inf you alone and the yours ‘This is for you alone and your family to eat’ (mod. Sic., Pitré 1878: 273) e. Fici jittari arreri lu bannu pri cuncurriri tutti li made send again the call for participate.inf all the Baruna, Cavaleri e Profissura barons knights and professors ‘He had the invitation sent out again so that all the Barons, Knights and Professors would partecipate’ (mod. Sic., Pitré 1878: 191) f. Lu patri ci avia fattu fari un puzzu pi un the father there had made make a well for not cci ammancari l’ acqua to.them miss.inf the water ‘The father had had a well built for them so that they would have water’ (mod. Sic., Pitré 1878: iii.318, cited in Ledgeway 2000: 146) (23) a. *Pensu

think

iddu / Pippinu he / Joseph

’un not

b. Pensu ca iddu / Pippinu think that he / Joseph ‘I think that he is not there’

ci there ’un not

essiri (mod. Sic.) be.inf c’ there

è (mod. Sic.) is

Complement clauses introduced by a complementizer can exhibit a personal infinitive (cf. (24a,b)), and this structure is also found in the regional Italian of Sicily (cf. (24c)), although it is not attested conspicuously in Pitré’s short stories. Finally, in

On the personal infinitive in Sicilian

111

these stories, the personal infinitive can be subordinated to a noun, in which case it is, again, introduced by a complementizer (cf. (24d)). (24) a. Si

dicinu di tu mangiarmi, mi mangi if say of you eat.inf.me me eat ‘If they say that you should eat me, you will eat me’ (mod. Sic., Pitré 1875: iv.165, cited in Ledgeway 2000: 125)

b. Si ricurdau la giuvina di una zia sua aviricci self= remembered the young.f of a aunt her have.inf.her datu una pezza given a cloth ‘The young woman remembered that an aunt of hers had given her a cloth’ (mod. Sic., Pitré 1878: 192) c. Crede di essere Carlo il colpevole believes of be.inf Carlo the culrpit ‘S/he believes Carlo to be the culprit’ (reg. Sic. It., Leone 1982: 137) d. Lu Re, p’ amuri di sò figghiu gualiri . . . the king for love of his son recover.inf ‘The king, for his desire of his son’s recovery . . . ’ (mod. Sic., Pitré 1875: 40) The conservative type of non-finite complementation of early Sicilian (cf. (11a–d)) has therefore been lost, while the type illustrated in (24a–d) hardly figures in our corpus of short stories. By contrast, the innovative type of adverbial subordination (cf. (9a) and (10a–c)) appears to have gained ground and to have evolved in two ways. On the one hand, it admits both pronominal and lexical nominative arguments (cf. (22a–f )). On the other hand, the reference in the discourse of its nominative argument is not constrained in the same way as in early Sicilian. Indeed, the argument of the personal infinitive need not draw its reference from the previous clause or from a speech act participant: hence the contrast between the modern Sicilian examples in (22d–f ) and the early Sicilian ones in (10a–c). Whereas in early Sicilian purpose clauses it is finite subordination that introduces a new discourse referent (cf. (17)), finite purpose clauses are replaced by causatives (cf. (25a,b)) or by the personal infinitive in Pitré’s short stories.8

8 Finite causal and temporal clauses are instead attested in early Sicilian, although, for the sake of brevity, we cannot discuss this point at length here.

112

Delia Bentley

(25) a. Vulia

Povira-bedda pi falla spusari cu lu Riuzzu wanted Povira-bedda for make=her marry with the prince ‘She wanted Povira-bedda in order to make her marry / for her to marry the Prince’ (mod. Sic., Pitré 1878: 34)

b. Prigava a sò matri pi fàricci jiri a fari tanticchiedda di caccia begged to his mother to let=him go to do little.dim of hunting ‘He begged his mother that she would let him go and do a little bit of hunting’ (mod. Sic., Pitré 1878: 133) In modern Sicilian adverbial purpose clauses, therefore, the alternation of bare and personal infinitives is more readily comparable with the Italian alternation of the infinitive and the subjunctive (cf. (14a,b)). (26) a. La

Riggina [ . . . ] pinsau di giustu, pri quitari a lu Riuzzu . . . the queen thought of right for calm.inf to the prince ‘The Queen rightly thought, to calm down the Prince . . . ’ (mod. Sic., Pitré 1875: 33)

b. Chi cci vurrissi pi io nèsciri di ccà? what cl want.cond for I go.out.inf of here ‘What would be necessary for me to escape from here?’ (mod. Sic., Pitré 1875: 13). c. Mè patri pi nasciri io fici un vutu my father for be.born.inf I made a vow ‘My father made a vow so that I would be born’ (mod. Sic., Pitré 1878: 130) Whilst in (26a) the covert argument of the bare infinitive is coreferential with the subject of the main clause, in (26b) the personal infinitive indicates a change of reference. The same is true of (26c), where the post-infinitival position of the pronominal argument is presumably due to focus. Interestingly, we have not found any instances of free or switch reference of the covert argument of the bare infinitive, although this can of course be determined by the matrix verb, as is the case with obligatory control structures. (27)

La prigau pi dacci ‘na surticedda d’ ajutu her begged for give.inf.him a kind.dim of help ‘He begged her to give him some kind of help’ (mod. Sic., Pitré 1878: 134)

On the personal infinitive in Sicilian

113

Free reference with the bare infinitive must nonetheless still be grammatical in some varieties of modern Sicilian, as Leone (1995: 64) mentions this possibility in passing, alongside that of causative structures. Recall now that adverbial infinitival clauses are innovative in early Sicilian, in that they exhibit almost exclusive head marking of the subject. Interestingly, in the 18thcentury texts analysed by Mocciaro (1991: 62), the personal infinitive only exhibits a pronominal argument. As for modern Sicilian, at first sight it would seem that both head and dependent marking are now an option, since the argument of the personal infinitive need not be a personal pronoun (cf. (22c–f)). However, as noted by Ledgeway (2000: 150), the pre-infinitival nominal arguments of these clauses tend to be kinship terms, whereas there is no such restriction on the post-infinitival ones. We have only found two examples of pre-infinitival nominal arguments, which are indeed kinship terms, in complement clauses (cf. (24b,d)). Ledgeway reports the examples in (28a,b), which are constructed on the basis of attested examples with post-infinitival nominal arguments. (28)

a. Partìu senza (so pa’) virillu so pa’ left without his father see.inf.him his father ‘He left without his father seeing him’ (mod. Sic., Leone 1995: 72, cited in Ledgeway 2000: 150) b. Maistà, pi (so’ figghia) stari bona so’ figghia . . . . majesty for your daughter stay.inf well your daughter ‘Majesty, for your daughter to get well . . . ’ (mod. Sic., Pitré 1878: ii.21, cited in Ledgeway 2000: 150)

In our theoretical framework (Van Valin and LaPolla 1997; Van Valin 2005), the imbalance in the distribution of nominal and pronominal arguments of the personal infinitive can be understood in either of two ways. On the one hand, it could have a rationale in discourse: being a new discourse referent, the nominal argument would tend to occur in post-infinitival position. On the other hand, the personal infinitive of adverbial clauses could still be a predominantly head-marking structure of Sicilian, as we saw was the case with the early texts, although dependent marking would not be ungrammatical (cf. (28a,b)). In accordance with the latter hypothesis, the postinfinitival nominal arguments of the personal infinitive in modern Sicilian would not be subjects, since they do not properly alternate with pre-infinitival ones. Rather, they would be arguments which are marked in terms of their semantic relation to the predicate (see fn. 6 above), but are not assigned a grammatical relation. We noted above that the subjecthood status of the argument of the personal infinitive should be ascertained, rather than being taken for granted. We expect the behaviour of the subject of the personal infinitive to be comparable to that of the subject of other clauses. Since we do not find an imbalance in the distribution of

114

Delia Bentley

nominal and pronominal subjects in other clauses, we could assume that the nominal arguments of the personal infinitive are not subjects. This assumption, however, would not capture the possibility for kinship terms to occur in pre-infinitival position (cf. (28a,b)). We thus opt for the former of the above hypotheses, and we claim that the imbalance in the distribution of pronominal and lexical subjects depends on their status in discourse. Speech act participant pronouns are by definition accessible, i.e. situationally available (Chafe 1987; Van Valin and LaPolla 1997: 200–201), uniquely individuated within the discourse context (Benveniste 1966: 230), as well as topicworthy (Wierzbicka 1981: 64). Third person pronouns, whilst not referring to participants in the discourse context, nonetheless refer to active or accessible arguments. Accordingly, whether they have been mentioned before in discourse is virtually irrelevant to their realization as topics or foci. They can, of course, be focalized post-infinitivally (cf. (26c), (29)), but they need not be (cf. (22a,b)). (29)

Tò patri prummisi ‘na Funtana d’ògghiu e [ . . . ] pi nàsciri tu your father promised a fountain of oil and for be.born.inf you ‘Your father promised a fountain of oil [ . . . ] in order for you to be born’ (mod. Sic., Pitré 1878: 115)

The same cannot be said of arguments that are realized lexically (cf. (22e,f)). Kinship terms constitute a possible exception by virtue of their relatedness to the speech act participants (cf. (28a,b)).9 To be sure, focused information units can take a preverbal position in Sicilian. However, only contrastive focus is allowed in preverbal position in embedded contexts (Cruschina 2012: 74–6), and this structure is quite marginal. We thus conclude that both the pronominal and the lexical nominative arguments of the infinitive of adverbial clauses are subjects in modern Sicilian, the difference between the two being that the pronominal ones can mark this grammatical relation on the infinitival head. Non-finite adverbial subordination has thus developed away from the older pattern in modern Sicilian, in that it can introduce a new referent in discourse. The grammatical relation of the argument is still marked on the head, but it can also be marked on the dependent. Whether an analysis along the same lines would suit non-finite complementation in modern Sicilian (cf. (24a,b,d)) remains a moot point due to the paucity of the available evidence.

5.4 Conclusion In this case study, we have discussed the personal infinitive in early and modern Sicilian. We have pointed out that this has properties which are comparable to those of the inflected infinitive (it occurs in clausal domains) and the personal infinitive in

9

I am grateful to Paola Benincà for drawing my attention to this important point.

On the personal infinitive in Sicilian

115

other Romance languages (it is not found in obligatory control domains). We have shed light on a conservative structure of early Sicilian, where the infinitive occurs in complement subject and object clauses, with marking of the subject on the dependent. On the other hand, an innovative structure was also found in early Sicilian, where the infinitive occurs in adverbial purpose or causal clauses, and the subject is marked on the infinitival head. Whilst the former structure is well attested in the early texts, but subsequently lost, the latter structure has evolved into a major switch-referencing strategy of modern Sicilian. Complement clauses with a personal infinitive are also found in modern Sicilian. They are subordinated to a verb or a noun and introduced by a complementizer—a fact which suggests that they are not a direct development of the conservative structure attested in the early vernacular texts, but rather a further innovation of the Sicilian dialect.

6 Glimpsing the future Some rare anomalies in the history of the Italo-Romance and Gallo-Romance future and conditional stem, and what they suggest about paradigm structure* M A R T I N MA I D E N A ND J O H N C H A R L E S S M I T H

6.1 The facts It is a curious fact about the comparative perspective that it can reveal phenomena which are simultaneously rare and common. The morphological development we examine here is rare in the sense that it tends to occur in only one or two lexemes in those varieties where it is found, and is not found in all that many varieties. Yet it is common in that it is a type of deviation from expected patterns of historical and morphological development which is repeatedly attested over the Italo-Romance and Gallo-Romance domains. Indeed, if there is an aberration from the expected development, then it tends to be of this kind. We shall argue that it offers an important insight into how inflectional paradigms are organized, but it is an insight that would be impossible if we focused on individual varieties in isolation, for it would always appear too marginal to be of interest. Only from a comparative perspective can its significance be properly grasped. Emblematic of the development at issue is the behaviour of the Italian verb sedere ‘sit’. Only in this verb is the relevant peculiarity observed, namely that the future and conditional root contains a diphthong: siederà/siederebbe ‘he will/would sit’. This diphthong is historically expected only in those cells of the paradigm in which

* We dedicate this study to our good friend Mair, with great respect and admiration for all she has contributed to the field.

Glimpsing the future

117

stress falls on the root, where it represents the regular outcome of a long defunct process of diphthongization affecting the vowels /e/ and /ɔ/ in stressed open syllables, and producing the stressed diphthongs /je/ and /wɔ/. Since stress falls on the root only in the singular and 3rd person forms of the present and the 2nd person singular present indicative (and in some infinitives), i.e. where the diphthong should occur, and not in unstressed roots such as that of the future and conditional, as can be seen for example from morire ‘die’, and older forms of the verbs muovere ‘move’ and sonare ‘sound’: INF.

PRS.IND 1SG

2SG

3SG

1PL

2PL

3PL

IMPF.IND

FUT

COND

GER

morire

muoio

muori

muore

moriamo

morite

muoiono

moriva

morirà

morirebbe

morendo

muovere muovo sonare suono

muovi suoni

muove suona

moviamo soniamo

movete sonate

muovono suonano

moveva sonava

moverà sonerà

moverebbe sonerebbe

movendo sonando

(older)

The common alternative development is analogical levelling, usually such that the diphthong is generalized throughout the paradigm (or occasionally eliminated altogether, as we shall see later): inf

prs.ind impf.ind fur cond ger 1sg 2sg 3sg 1pl 2pl 3pl suonare suono suoni suona suoniamo suonate suonano suonava suonerà suonerebbe suonando chiedere chiedo chiedi chiede chiediamo chiedete chiedono chiedeva chiederà chiederebbe chiedendo

Sedere ‘sit’, however, goes its own way and fits neither of these two major classes of development:1 inf

prs.ind 1sg 2sg sedere siedo siedi

3sg siede

1pl 2pl sediamo sedete

impf.ind fut cond ger 3pl siedono sedeva siederà siederebbe sedendo

The exceptional extension of an allomorph historically predicted to occur only in stressed present-tense roots into the future and conditional is also attested in other places and periods.2 What is principally involved is the diphthong, but other allomorphs originally associated with stress may also be implicated (e.g. in Tuscan the /o/ of udire ‘hear’). In 13 and 14-century Tuscan texts (data from OVI) we find the following for muòvere ‘move’, nuocere ‘harm’, trovare ‘find’, possedere ‘possess’, and udire ‘hear’:

See also the figures given by Thornton (2011: 376). The allomorph found in the future/conditional is sometimes also shared by the infinitive, but the common denominator across all the anomalous future/conditionals is that they share an allomorph with the present tense; there are numerous cases (e.g. It. sedere ‘sit’) in which the infinitive does not share the allomorph with the future and conditional. 1 2

118

Martin Maiden and John Charles Smith

i. Muoverà for expected moverà ‘(s)he will move’ Reggimento de’ principi di Egidio Romano (1288) inf prs1sg/2sg/3sg/3pl impf.ind fut muov- 11 14 0 1 muoveranno mov0 0 3 1 Bartolomeo da San Concordio (Ammaestramenti degli antichi latini e toscani, 1308) inf prs1sg/2sg/3sg/3pl fut ger 0 muov- 6 11 2 muoverà mov0 0 0 1 Bibbia volgare (14th–15th c.) inf prs1/2s/3sg/3pl prs1/2pl impf.ind impf.sbjv fut cond ger 0 0 muov- 13 30 0 0 1 8 muoverò muoverà muoverae muoveranno mov- 4

6

4

5

7

35

2

4

ii. The type nuocerà ‘(s)he will harm’ (< nuoce ‘(s)he harms’, etc.) for expected nocerà Bibbia volgare (14th–15th c.) inf prs3sg prs2pl nuoc- 4 2 0

1

noc-

0

1

impf.sbjv 0

fut 4 nuocerà nuoceranno

3

4

iii. Odirà/oderà for expected udirà ‘(s)he will hear’ Bartolomeo da San Concordio (Ammaestramenti degli antichi latini e toscani, 1308) inf prs1sg/2sg/3sg/3pl impf.ind fut ger ptp 0 0 od- 0 18 0 1 odirà ud-

16

0

1

2

3

15

Glimpsing the future

119

Neri Pagliaresi (Leggenda di santo Giosafà, 1400) inf prs1sg/2sg/3sg/3pl prs.2pl impf.ind fut pret impf.sbjv ger ptp od- 0 6 0 0 0 0 0 0 1 odiranno ud- 14 0 2 6 4 19 3 9 17 Carteggio dei Lazzari (1320–22) inf prs1sg/2sg/3sg/3pl od- 0 1 ud-

3

0

fut 0 1

cond 1 oderebbe 0

pret 0

ptp 0

1

1

iv. Truovarà/truoverà for expected trovarà/troverà ‘(s)he will find’ Andrea da Grosseto (Trattati morali di Albertano da Brescia volgarizzati, 1268) inf prs1sg/2sg/3sg/3pl prs.1pl impf.ind fut pret impf.sbjv ptp truov- 0 57 0 (1)3 0 0 0 1 truoverà trov- 8 0 1 0 12 1 1 7 Pistole di Seneca volgarizzate (1325) inf prs1sg/2sg/3sg/3pl prs.1pl impf.ind fut pret impf.sbjv ger ptp truov- 0 69 0 0 0 0 0 0 1 truoverai trov- 34 0 3 9 0 25 7 4 63 Statuto dell’Arte della Mercanzia senese (1343) inf prs1sg/2sg/3sg/3pl fut truov- 0 9 1 truovaranno trov8 0 15

impf.sbjv 0

ptp 0

7

16

v. Possiederà for expected possederà ‘(s)he will possess’ Libro del difenditore della pace e tranquillità volgarizzato (1363) inf prs1sg/2sg/3sg/3pl impf.ind fut cond pret ptp possied- 0 5 0 0 0 0 1 possiederà possed- 7 0 2 2 1 1 2 3 This solitary truovava (p. 68) occurs in a context in which a present tense form would be far more natural (the reference is to that which ‘finds itself ’ written in a particular text whose content is otherwise referred to in the present tense). It may be a lapsus calami, involving repetition of the final two letters of truova.

120

Martin Maiden and John Charles Smith

The relative rarity of the synthetic future and conditional forms in modern southern Italian dialects means that there is little opportunity to observe our phenomenon there, but Ledgeway (2009a: 60) observes the following for old Neapolitan, where metaphonic vowels in the root, historically restricted to the stressed roots of the present, may also turn up in the future and conditional (our translation): [I]n earlier phases of the dialect metaphony could, albeit rarely and somewhat erratically, affect the root of the verb (which bore a secondary stress) in the second person singular future and conditional. For example, in Cavalcanti we find forms such as lievarraje ‘you will raise’, tiennarraje ‘you will hold; you will have’, stiennarraje ‘you will stretch out’, vuotarraje [‘you will empty’], meniestrarraje ‘you will dish out’, mbruogliarraje ‘you will mix in’, vidarraje ‘you will see’, as well as alternations such miettarraje/mettarraje ‘you will put’, siervarraje/servarraje ‘you will serve’.

In central and especially northern Italy we find more abundant parallels. Thus at Cascinagrossa (Alessandria; see Castellani 2002) we find: bere ‘drink’ inf

prs.ind3sg prs.sbjv3sg prs.ind2pl impf.ind3sg fut3sg

’baiv

’baiva

’baiva

bu’iva

bu’i

cond3sg

ger

ptp

baivan’ra baivan’raisa bu’anda bu’y

Consider also the future and conditional of ’kras ‘grow’; ’lez ‘read’; sa’ta ‘sit’; ’ten ‘hold’; ’vagg ‘see’; ’vend ‘sell’ ’mov ‘move’: INF

PRS.IND 3SG

PRS.SBJV 3SG

PRS.IND 2PL

IMPF.IND

FUT3SG

COND3SG

GER

PTP

3SG

kras

krassa

krassa

karsi

karsiva

krasanra

krasanraisa

karsanda

karsy

lez

leza

leza

alzi

alziva

lezanra

lezanraisa

alzanda

alzy

sata

sεta

sεta

satε

satava

sεtanra

sεtanraisa

satanda

sata

ten

tena

tena

tni

tniva

tenanra (tnanra)

tenanraisa (tnanraisa)

tnanda

tny

va vaa

vaa

aui

auiva

vianra (vira)

vianraisa (viraisa)

auanda

vst

vend venda

venda

vindi

vindiva

vendanra

vendanraisa

vindanda

vindy

mov

mova

muvi

muviva

muvanra

muvanda

muvy

mova

movanraisa

The substantial majority of verbs do not show the present tense allomorph in the future/conditional, e.g. an’da ‘go’, au’rai ‘want’, du’vai ‘must’, a’sest ‘be present’, ru’sti ‘roast’: INF

anda aurai duvai asest rusti

PRS.IND3SG va vo deva asesta rustessa

PRS.SBJV3SG

PRS.IND2PL

IMPF.IND3SG

FUT.3SG

COND.3SG

GER

PTP

vaa voja deva asesta rustessa

andε auri duvi asisti rusti

andava auriva duviva asistia rustiva

andra aura duvra asistanra rustinra

andraisa auraisa duvraiva asistanraisa rustinraisa

andanda auranda duvanda asistanda rustanda

anda ansy duvy asisty rusti

121

Glimpsing the future

Similarly, in Genoese (Toso 1997) we find, for example, for ta’ʒej ‘be silent’, ’bejve ‘drink’: inf

prs.ind3sg prs.sbjv3sg prs.ind2pl impf.ind3sg fut.3sg cond.3sg

ger

ptp

taej bejve

tae bejve

taedu bevedu

tayu bevyu

tae bejve

taej bevej

taia bejvia bevia

taejva bevejva

taiejva bejviejva∗ beviejva

∗=‘rare’

The dialect of Realdo (Liguria; Garnier 1898: 79–80) shows for ’vejro ‘see’, mu’ri : ‘die’, fi’ni : ‘end’: inf vejro muri fini

prs.ind3sg vejre mœjre finiʃe

prs.ind2pl vi muri fini

prs.sbjv3sg vee mœjre finiʃe

impf.ind3sg fut.3sg vejra vija murija mœjrera finira finija

cond.3sg vejrija mœjrerija finirija

ger veende murende finende

ptp viʃte morte fini

In the Alpine Lombard dialects of Bormio (Longa 1912) and Livigno (Huber 1958), both in the province of Sondrio, we find for the verbs ’bœr ‘drink’, ’tœr ‘take’, ir ‘go’:4 Bormio inf

prs.ind3sg

prs.sbjv3sg

prs.ind2pl

impf.ind3sg

fut.3sg

cond.3sg

bœr tœr

bœf tœ

’bœvja ’tœgja

bo’e to’le

bo’ea to’lea

bœ’ra tœ’ra

bœ’resi tœ’resi

Livigno inf

prs.ind3sg

prs.sbjv3sg

prs.ind2pl

impf.ind3sg

fut.3sg

cond.3sg

bœr

bœf

bœia

tœr

tœ

tœia

boe tole

boœ toœ

bœrε tœrε

bœrɔ tœrɔ

ir

va

vadja

djœt [IMP]

djœ

varε

varɔ

The characteristic suppletive pattern of the verb ‘go’, which in many varieties has a reflex of the Latin verb uadere ‘go, hasten’ limited just to the root-stressed forms of the present indicative, shows that form in the future as well in, e.g., the Lazio dialect of Subiaco (Lindsström 1907; cf. also Crocioni 1907: 55; Navone 1922: 101; Orlandi 1980: 71, 73):

There are signs of recent penetration into the 1/2pl present of ir ‘go’: cf. Huber (1958: 126–7) and the other sources he quotes there. In Bormio (Longa 1912: 347) it permeates virtually all the non-finite forms of this verb. See also Locatelli (1970: 37); Prader Schucany (1970: 211, 214). 4

122

Martin Maiden and John Charles Smith

inf

prs.ind ’vaʎo ’va

prs.sbjv ’vaʎa ’vaʎi

fut va’raʎo va’raʎi

cond va’ria va’reʃti

’vaʎa

impf.ind ’jea ’jii ’jea

’i

’va ’jamu

pret ’i

va’ra

va’ria

’iʃti ’i

’jamu

’jemmu

va’rimu

va’remmu

’jemmu

’jate

’jate

’jeʃte

va’rite

va’reʃte

’jeʃte

’vau

’vaʎanu

’jeenu

va’rau

va’rienu

’iru

ptp

’itu

For similar patterning in this verb elsewhere in Italo-Romance, see also Markun (1932: 351). In many Italo-Romance varieties there occurs in the present tense (and usually in those cells of the present-tense subparadigm in which stress normally falls on the root) a formative, the ‘augment’, immediately following the root and itself bearing stress. The history of this form is discussed in e.g. Maiden (2004), but what is of immediate interest here is that in some Lombard varieties it also percolates— optionally—into the future and conditional (cf. Facchetti 1901: 11–12.; Locatelli 1970: 36; Pagani 1977: 75; Sanga 1979; Beretta 1980: 143–4; Nicoli 1983: 302; Zahner 1989; Biella et al. 1992: xciii–iv; Alberti 1993: 104, 109). Thus the dialect of Cremona (Magri 1995: 71–2), where the augment has the form -is-, the verb ‘finish’ presents the following forms: inf

finìi

prs.ind prs.sbjv finìsi finìsa

impf.ind fut

cond

finìivi

finirès

finìset finìsa finìsa finìs finìsum finìsum

finìivet finirèet finisarèet finìiva finirà finisarà finìivum finirùm finisarùm

finìi

finìi

finìives

finirìi

finisarìi

finirèses

finisarèses

finèses

finìs

finìsa

finìiva

finirà

finisarà

finirès

finisarès

finès

finiròo finisaròo

impf.sbjv ptp finisarès

finirèset finisarèset finirès finisarès finirèsum finisarèsum

finès finèset finès finèsum

finìit

Similar phenomena in respect of the augment occur in some Friulian varieties (Iliescu 1972: 175–4), and in Corsica (cf. Bottiglioni 1932–44: maps for guarirò ‘I will cure’ and finirò ‘I will finish’). Fusina (1999: 150–53) gives the following representative picture for central and southern Corsica,5 where the augment is written -isc-: prs.ind

impf.ind finiu

finiscu finisci finisci

fut

cond

pret finii

finii finia

finisciaraghju finisciarè finisciarà

finisciari finisciarii finisciaria

finimu finiti

finiami finiati

finisciaremi finisciareti

finisciariami finisciariati

finiimi finisti

finiscini

finiani

finisciarani

finisciariani

finini

5

finisti finì

ger

ptp

finendu

finitu

Some varieties variably show the augment in 1/2pl present indicative, and gerund.

123

Glimpsing the future

In Gallo-Romance, the same range of developments is found. The fact that stress fell on the root only in the singular and 3rd person plural forms of the present (plus some infinitives) commonly created an apophonic alternation between tonic and atonic outcomes of the same original vowel in open syllables, with the future and conditional exhibiting the expected atonic outcome, as can be seen e.g. from pouvoir ‘be able’, vouloir ‘want’, mourir ‘die’: inf.

prs.ind 1sg pouvoir peux (earlier puis)6

2sg 3sg peux peut

vouloir mourir

veux veut meurs meurt

veux (earlier vueil) meurs (earlier muir)

impf.ind 1pl 2pl 3pl pouvons pouvez peuvent pouvait voulons voulez veulent voulait mourons mourez meurent mourait

fut

cond

ger

pourra pourrait pouvant voudra voudrait voulant mourra mourrait mourant

As in Italo-Romance, the common alternative development is analogical levelling, although it is more normal in Gallo-Romance for the reflex of the original tonic vowel to be eliminated (compare il treuve > il trouve ‘he finds’ etc.); in only a minority of cases is it generalized throughout the paradigm (compare nous plourons > nous pleurons ‘we weep’; nous amons > nous aimons ‘we love’, etc.). Elimination inf (earlier) trouver (later) trouver

prs.ind 1sg 2sg

3sg

1pl

2pl

3pl

impf.ind fut

cond

ger

treuve

treuves treuve

trouvons trouvez treuvent trouvait

trouvera trouverait trouvant

trouve

trouves trouve

trouvons trouvez trouvent trouvait

trouvera trouverait trouvant

Generalization inf (earlier) plourer (later) pleurer

prs.ind 1sg 2sg

3sg

1pl

2pl

3pl

impf.ind fut

cond

ger

pleure

pleures pleure

plourons plourez pleurent plourait

plourera plourerait plourant

pleure

pleures pleure

pleurons pleurez pleurent pleurait

pleurera pleurerait pleurant

However, there are examples in both the standard language and various dialects of the originally tonic stem being extended into the future and conditional only (see Schwan and Behrens 1932: 219; Pope 1952: 367; Ewert 1953: 207–8; Nyrop 1960: 159, 160, 165; Price 1971: 202–3).

6 The earlier 1st person singulars represent the regular development of Latin forms containing a yod; they were subsequently subject to analogical remodelling.

124

Martin Maiden and John Charles Smith

Modern standard French inf boire tenir venir asseoir seoir mener lever

prs.ind3sg boit tient vient assied sied mène lève

prs.ind2pl buvez tenez venez asseyez seyez menez levez

impf.ind3sg buvait tenait venait asseyait seyait menait levait

fut.3sg boira tiendra viendra assiéra7 siéra mènera8 lèvera

cond.3sg boirait tiendrait viendrait assiérait siérait mènerait lèverait

ger buvant tenant venant asseyant seyant menant levant

ptp bu tenu venu assis sis mené levé

gloss ‘drink’ ‘hold’ ‘come’ ‘sit’ ‘suit’ ‘lead‘ ‘raise‘

cond.3sg vrå purå wårå

ger nalå peyå wadå

ptp — peyu wadè

gloss ‘go’ ‘be able’ ‘keep’

Ranrupt (Lorrrain) (Aub-Büscher 1962: 89–91) inf nalè peyu wadè

prs.ind3sg vè pu wå

prs.ind2pl nali peyi wadi

impf.ind3sg nalèy peyèy wadèy

fut.3sg vrè purè wårè

Similar patterns are noted for the Wallon variety of Namur (Niederländer 1900: 282–3) and the Francoprovençal variety of Vaux (Duraffour 1932: 60). Ronjat (1937: 208) observes this pattern in ‘most’ Aquitanian varieties of Occitan, especially western varieties, and Toscano (1998: 109–10, 111–12) makes similar observations for Niçois. Keller (1928: 97) reports the extension of the v- stem of the present into the future and conditional in the Savoyard varieties of Francoprovençal spoken in the localities of Savigny, Marthod, Annecy, Balme de Savoie, Rumilly, Leschaux, Montagny, La Thuile, and Ceresole, whilst ALF map 28 (toi, tu iras) records the extension of the v- stem of the present into the future in the Francoprovençal varieties of the north of the département of Savoie and in most of the eastern oïl varieties spoken in the départements of Haute-Saône, Vosges, Meurthe-et-Moselle, and Meuse, spilling over into Marne and Haute-Marne. In parts of this area, strikingly, the v- appears to be added to the ir- future stem found elsewhere (including in standard French), creating a ‘blend’. To conclude, we have seen that allomorphs originally restricted to the present tense (and sometimes also the infinitive) follow one of four lines of distributional development: continuation unchanged (e.g. It. morire; Fr. mourir), complete elimination (e.g. Fr. trouver), generalization throughout the paradigm (e.g. It. suonare; Fr. pleurer), or extension just into the future and conditional (the focus of this chapter). Numerous conceivable ‘intermediate’ patterns of extension (or elimination) are unattested in the Italo-Romance and Gallo-Romance data. For example, extension of the alternants to all and only singular and 3rd person forms in other tenses never seems to occur, nor does extension of the alternant to all and only present-tense

7 Alternative future/conditional forms are assoira/assoirait and asseyera/asseyerait, the latter found only rarely (Grevisse and Goosse 2011: }847b). 8 The future mènera and conditional mènerait are remodellings of earlier merra, merrait (see Pope 1952: }970).

Glimpsing the future

125

forms (with a couple of exceptions).9 The occasional sallies of present-tense alternants into the future/conditional constitute, then, a significant exception to the generalization that speakers do not tend to make partial extensions of such present tense alternants within the paradigm.

6.2 Semantic or phonological causation? How we are to explain the sporadic (yet paradoxically systematic) forays of presenttense morphology into the future and conditional? One possibility is that the change has a semantic motivation involving reanalysis of ‘present’ as ‘non-past’. On this basis, ‘present’ root allomorphs become potentially extensible to the future. That this is not a plausible explanation is shown by the fact that, for allomorphs associated with 1/2/3sg/3pl present, we would expect—contrary to fact—the extension into 1/2pl present as well (e.g. siediamo ‘we sit’, siedete ‘you sit’ as well as siederò ‘I will sit’).10 This hypothesis is also inadequate to explain the extension into the conditional as well as the future. If the conditional is (arguably) [—past] in modern Italian, it was certainly not so in the Middle Ages (cf. OIt. Dissero che verrebbe = modIt. Dissero che sarebbe venuta ‘They said she would come’; Maiden 1996a), and has never been so in Gallo-Romance (Fr. Ils dirent qu’elle viendrait). One would at any rate have to appeal, in addition, to the effect of the general pattern of identity between future and conditional stems observable in Italo-Romance and most of Gallo-Romance. Could the motivation be, instead, phonological in nature? Most of the vocalic alternations with which we have been concerned are the results of differentiation in

9 An ‘intermediate’ redistribution of the alternants arises in connection with the present subjunctive, where segmental and prosodic characteristics of stressed roots are often extended in central and northern Italo-Romance and beyond into the 1/2pl present subjunctive. Maiden (2010) reviews the analogical mechanism behind this change, which involves extension of a pattern of alternation initially characteristic of palatals and velars. In many varieties of Acadian French, the original tonic alternant is now limited to the singular forms of the present tense; in the case of some verbs which have a distinct present subjunctive stem in standard French, it is the original tonic variant which is also found in these forms in Acadian: compare standard French je peux (ind)/je puisse (sbjv) ‘I am able’ with Acadian French je peux (ind)/je peuve (sbjv) (see Péronnet 1990; Smith 2011). A similar extension is found in some varieties of colloquial Parisian French (Lodge 2004: 221). 10 In some varieties of French, we find exactly this type of extension in the verb boire ‘drink’— thus, nous boivons, vous boivez (see Lodge 2004: 221 for colloquial Parisian; Verrier and Onillon 1908: 24 for the Choletais; GDFC s.v. boivez for Québécois; Brasseur 2001 for Newfoundland French; and Beaulieu and Cichocki 2009: 213 for the Acadian French of northeastern New Brunswick). However, as will be noted from the table on modern standard French above, the vowel that spreads in this way is also the vowel of the infinitive, which may point to a more complex type of change. Similarly, in several Gallo-Romance varieties, the v- stem of the verb ‘go’ has been extended within the present tense, at least to the 1st person plural, although the 2nd-person plural form is often immune (see Duraffour 1932: 76 for the Francoprovençal dialect of Vaux, and ALF map 27 (nous allons), which shows that the phenomenon is common in Lorraine, widespread in Comtois and Francoprovençal, and frequent in the Occitan varieties of northern Auvergnat and northern Limousin, and in the Croissant area.

126

Martin Maiden and John Charles Smith

the nature of the vowel as a function of primary stress. The roots of future and conditional forms are not and never have been stressed in the varieties at issue, and so the conclusion must be that the presence there of vowels expected only in primary stressed syllables must be a matter of analogical extension. Some of the cases we have presented, however, are—in principle—analysable phonologically, where what is involved is the result of a phonological process affecting unstressed vowels. Now the root of the future/conditional characteristically bears ‘secondary stress’ (or ‘contraccento’—cf. Lepschy 1992; 1993),11 by virtue of non-adjacency to the stressed vowel (cf. It. "porte’ra ‘(s)he will carry’ "porte’rebbe ‘(s)he would carry’ vs por’tava ‘(s)he was carrying’). It is possible that secondary stress ‘protects’ pretonic vowels from those processes which only affect wholly ‘unstressed’ vowels. Many of the alternations presented above for Cascinagrossa may in fact reflect non-reduction of secondarily stressed vowels: "krasan’ra ‘(s)he will grow’ "krasan’raisa ‘(s)he would grow’ (vs. kar’siva ‘(s)he was growing’, etc.), "lezan’ra ‘(s)he will read’ "lezan’raisa ‘(s)he would read’ (vs. al’ziva ‘(s)he was reading’, etc.). In fact, it is hard to be sure whether the facts from Cascinagrossa are genuinely phonological, for appropriate controls from outside future/conditional verb forms are remarkably elusive. On the one hand Cascinagrossa has "aulan’tera ‘willingly’ and au’ra ‘will want’) with identical secondarily stressed and unstressed vowels, but on the other ’trai ‘three’ and "tradi’tʃezim ‘13th’ vs tar’zent ‘300’ (cf. AIS map 305), with the same vowel in stressed and secondarily stressed, but not in unstressed, syllables.

6.3 An accidental association? In our view, the cases we have reviewed are neither a matter of meaning nor (in the majority) attributable to any phonological process. They are a morphological accident, the result of a perceived formal identity between the stem of the future and conditional and certain whole word-forms of the present. To put it informally, the anomalous futures and conditionals arise because, quite often, the stem of the future and conditional ‘sounds like a present-tense form’. By ‘stem’ we mean the lexical root of the verb plus the following thematic vowel. The present-tense form which these stems sometimes ‘sound like’ is usually the 3rd person singular present indicative (although in the case of verbs with infinitives in -ire, the relevant present tense form would be the 2nd person singular indicative). Compare in Italian the future/conditional (here represented by 3sg future) and present forms of the verbs vendere ‘sell’, vincere ‘win’, correre ‘run’, partire ‘leave’, and mentire ‘lie’:

11

See also Schwan and Behrens (1932: 219) and Pope (1952: 366) for French.

Glimpsing the future

127

prs.ind fut.3sg 3sg

’vende

"vende’ra

3sg

’vintʃe

"vintʃe’ra

3sg

’korre

"korre’ra

2sg

’parti

"parti’ra

2sg

’menti

"menti’ra

To these examples we may add, for many central and other Italian dialects (although not, as it happens, for Florentine or Italian), forms of first conjugation verbs with infinitives in -are, e.g. fumare ‘smoke’, parlare ‘speak’: prs.ind fut.3sg 3sg

’fuma

"fuma’ra (It. fumerà)

3sg

’parla

"parla’ra (It. parlerà)

In these examples there is clearly segmental identity between the future/conditional stem (vende- etc.) and word-forms of the present (vende etc.). But this identity does not uniquely bind the present and the future/conditional, since the string vende (etc.) also occurs in other parts of the paradigm (e.g. imperfect indicative ven’deva ‘(s)he was selling’, imperfect subjunctive ven’desse ‘(s)he sold’). The additional factor that makes the stem of the future/conditional ‘sound like’ present tense verb-forms is prosodic: the very fact, as explained above, that the initial syllable is stressed. In the future/conditional it is secondarily stressed, and in the relevant forms of the present tense (and second person singular imperative) it is primarily stressed, but both share a prosodic contour ‘strong–weak’ which is generally not to be found in the stem elsewhere in the verb. Indeed, the utterance venderete /"vende’rete/ ‘you will sell’ is prosodically (and segmentally, allowing for the difference in the final vowel) identical to vende reti /’vende ’’reti/ ‘he sells nets’—where reti carries phrasal stress. In practically all other parts of the verb where the same lexical root occurs,12 it is wholly unstressed (e.g. the imperfect indicative and subjunctive forms just given, the past participle ven’duto ‘sold’, the gerund ven’dendo ‘selling’, the preterite ven’de ‘(s)he sold’).

12 In some varieties, including Italian, the "vende- stem also occurs in 1/2pl imperfect indicative ("vende’vamo, "vende’vate), but this is exceptional in Italo-Romance overall.

128

Martin Maiden and John Charles Smith

The position that we have here sketched from modern Italian is in general replicable for all dialects for which we have cited examples of the present stem in the future/conditional (and indeed far beyond). In northern Italo-Romance, indeed, the incidence of this prosodic contour in futures and conditionals seems always to have been much higher than historically predicted, because the intertonic vowels /e/ and /i/ tend not to be deleted in these verb forms. For medieval texts see Ghinassi (2006: 87, 89), Mussafia (1868: 33–4), Salvioni (1902), Vitale (1953: 93–4); for modern Milanese, see Salvioni (1975: 39–40). Our suggestion is that speakers’ perception of a frequent identity between present-tense forms and the future/conditional stem triggers a reanalysis such that futures and conditionals are assumed to contain present forms, leading to the introduction of a range of forms historically expected to occur only in the present.

6.4 The mechanism of the change We have made a suggestion regarding the basis of the occasional reanalysis of future/conditional stems as being those of the present, but we still have to address some other possibilities. The anomalous allomorphs of the future and conditional typically belong to a class otherwise correlated with rhizotony (stress falling on the root) in the Romance verb. To take again the example of Italian, these include not just vocalic allomorphy (muòre ‘(s)he dies’ vs imperfect morìva ‘(s)he was dying’, òde ‘(s)he hears’ vs imperfect udìva ‘(s)he was hearing’), but also suppletion (và ‘(s)he goes’ vs imperfect andàva ‘(s)he was going’, èsce ‘(s)he exits’ vs imperfect uscìva ‘(s)he was exiting’) and the distribution of the ‘augment’ -isc- ( finìsce ‘(s)he finishes’ vs imperfect finìva ‘(s)he was finishing’). While, as was said above, the anomaly cannot be the direct result of any phonological process, a phonological approach to the phenomenon still remains in principle possible. This involves appeal to ‘phonologically conditioned allomorphy’. A currently dominant type of view of the ‘stress-related’ alternants of the present tense (see Carstairs(-McCarthy) 1988: 17; 1990: 20; Krämer 2009: ch. 4.2; Carstairs-McCarthy 2010: 142–4; cf. also Anderson 2008; 2010 for similar and cognate phenomena in Romansh) assumes specification of the alternants in the lexical representation of the verb, and indexed as ‘stressed’ (or unstressed); the ‘stressed’ alternant is that selected where stress falls on the root, the ‘unstressed’ alternant being the default. As long as ‘stress’ means ‘primary stress’, our anomaly remains unexplained; but could we not still argue that the ‘anomalous’ development is the result of a reanalysis such that ‘primary stress’ is reanalysed simply as ‘stress’, thereby allowing for the relevant forms equally in secondarily and primarily stressed—but not in unstressed—roots? Such a putative reanalysis is, frankly, ad hoc, and it is also not wholly compatible with the observed facts. Varieties which extend the root va- ‘go’ just to the future and

Glimpsing the future

129

conditional cannot be doing so in response to ‘secondary stress’, since the result (like its predecessor) lacks the required stress contour: va’ra (and earlier i’ra) not **"va’ra (or **"i’ra). Exactly the same arguments will apply to Fr. viendra ‘(s)he will come’, tiendra ‘(s)he will hold’, siéra ‘it will suit/befit’, assiéra ‘(s)he will sit’ at the stage of the language when word stress was still found (see also Meyer-Lübke 1895: 396). Ledgeway (2009a: 60, quoted above) implies that the accento secondario is what might explain the presence of metaphony in the future/conditional root of old Neapolitan. But nowhere else does the accento secondario attract extension of diphthongizing metaphony, which remains firmly associated with the primarily stressed syllables in which it originated.13 Neapolitan metaphonic alternants are, and have been for centuries, very clearly morphologized markers of ‘2nd person singular present’ in the verb, so that what is involved is clearly a copying of a ‘(2nd person singular) present’ form into the future. As for the introduction of the ‘augment’ into the future and conditional in Lombardy, the phonological argument does not fit the morphological details. Let us assume (although the analysis is not self-evident) that the augment is an integral part of the root and that, since it is stressed in the present, the augmented form must be that allomorph of the root selected under stress. On this account, the reanalysis of the rule so that it applies to secondary as well as to primary stress leads us to predict the introduction into the future and conditional of the root of the present. Therefore, given an existing future like "kapi’ra ‘he will understand’ (*ka’pra is nowhere attested) the result of the reanalysis should be *ka"pisi’ra, but not the actually occurring ka"pisa’ra (or ka"pise’ra) (see also Meul 2010: 21, 26). What is actually involved is a change affecting not merely the lexical root but the entire future/ conditional ‘stem’, and grafting the whole 3sg present indicative word-form into the future/conditional stem (e.g. ka’pisa). In this connection, cf. also the type møjre’ra ‘(s)he will die’ rather than **møjri’ra in Realdo, where ’møjre is the 3sg present, or the old Tuscan example oderà ‘(s)he will hear’ for udirà (cf. 3sg present indicative ode). Meanwhile, a form such as Fr. mènera ‘(s)he will lead’ is highly unlikely to be the result of secondary stress. Such extensions of the originally tonic stem into the future and conditional of first-conjugation verbs are attested only in the later old French period (Pope 1952: }970), when mènera begins to compete with merra (which was still current as late as the 16th century), and therefore take place at a time when word stress had already begun to weaken and be lost from the language (Pope 1952: }}170, 223). Rather, the evidence points strongly to mènera and

13 In modern Neapolitan there are indeed forms such as viennarì ‘Friday’, miercurì ‘Wednesday’, or the 2nd-person singular imperative vieneténne ‘come’ (followed by both a 2nd person and a locative clitic), which present the diphthong in a position of secondary stress. But the crucial point is that in such cases the diphthong was originally in primary stressed position, the shift in stress being associated with compounding (ue´neris + dı´es > viennarì), or with the attachment of clitics which attract primary stress off the verb.

130

Martin Maiden and John Charles Smith

similar forms being cases of analogical remodelling (see also Nyrop 1960: 165; Price 1971: 202–3). A fortiori, the prefixing of the v- of the present stem of the verb ‘go’ to the future and conditional stem ir- in various Gallo-Romance dialects cannot possibly be due to stress and must be the result of blending. In contrast, the facts sit very easily with a morphological, indeed morphomic, account. Maiden (2004; 2006; 2007; 2011a; 2011b) has argued at length that the distribution of the phonologically very heterogeneous set of allomorphs, including suppletive and augmented forms, found in the ‘stressed’ roots of the present is not, has not been for centuries, and in some cases can never have been, conditioned by any phonological factor. Rather it is purely morphological. The present-tense allomorphy is sensitive to a heterogeneous but recurrently specified group of paradigm cells, comprising the singular and 3rd person forms of the present indicative and present subjunctive (the so-called ‘N-pattern’). Not the least reason for taking this position is that the distribution of root stress itself requires reference to the postulated morphomic structure, for it has to be specified as falling on the ‘singular and 3rd person forms of the present . . . etc.’. None of the evidence is incompatible with the notion that the change in the future/conditional is of a wholly morphological nature, arising from identification of the stem of these tense forms with forms otherwise indexed as belonging to a subset of present-tense cells. In other words, the set of paradigm cells with which the alternants are associated has been extended to include the future and conditional. What we have is a kind of analogical change prompted by a fortuitous occasional and partial ‘morphomic’ identity between the ‘stem’ of the future/conditional and present-tense word-forms. This, in turn, suggests that speakers analyse present-tense ‘N-pattern’ alternants as morphologically (not phonologically) distributed. In general, comparative-historical inspection of Romance verb morphology has indicated that morphomically distributed patterns of allomorphy are either robustly coherent or disappear completely. They do not usually shift their distribution in a gradual or piecemeal way. Maiden (2010) has suggested, in effect, that it ‘takes a morphome to shift a morphome’: ‘gradual’ readjustments in distribution occur in response to interference from other morphomic patterns. The behaviour of the future and conditional may be another case in point: an occasional, morphomic identity between the future/conditional stem and that of a subset of present-tense forms causes instability in the paradigmatic distribution of the ‘N-pattern’.

7 Person endings in the old Italian verb system L A U R A VA N E L L I

7.1 Introduction This chapter compares some aspects of old Italian verb morphology with those of the modern language. It grows out of a critical reflection on, and a deeper analysis of, the material contained in my chapter in Grammatica dell’italiano antico (Salvi and Renzi 2010), co-authored with Roberta Maschi.1 The intention is to provide a synchronic description of an aspect of the old Florentine verbal system as it was at the end of the 13th and beginning of the 14th centuries, that is to say in the first documented phase of what was to become standard Italian. My analysis, therefore, in line with the general goals of the Grammatica dell’italiano antico, is essentially contrastive. I will place side by side two synchronic stages of Italian in an attempt on the one hand to show the salient features of the old Italian system and on the other to identify the internal dynamics by which we can account for the changes that subsequently gave rise to the modern one. The specific focus will be on the system of personal endings, i.e. on that part of the verb-form which is primarily associated with expression of the category of person, although we will see how the ending can also convey tense and mood, categories usually associated with the verb stem. Since in the old Italian system, as indeed in the modern one, person is only expressed on finite verb-forms, I will have nothing to say about the non-finite forms of the verb.2 In identifying person endings one of the recurrent problems is the role of the so-called ‘thematic vowels’ (TVs), those elements which are part of the stem and 1 For the general principles and the specific methods adopted in that work, and for the theoretical foundations and details of the corpus on which that account is based, see the editors’ preface (Renzi and Salvi 2010: 7–19). 2 See Chapter 5 above, by Delia Bentley, for brief mention of dialects exhibiting non-finite forms with person marking, and see Pirrelli (2000) for the general frame of reference.

132

Laura Vanelli

which determine the assignment of verbs to inflectional classes or conjugations.3 It has often been observed that, despite the fact that TVs belong to the stem, their structural and linear adjacency to the ending means that in some cases it can be difficult to decide how to segment the TV and the ending. A typical instance, for which various solutions have been proposed in the relevant literature, concerns the -a of the 3sg of first conjugation verbs, for example in canta ‘(s)he sings’, which is sometimes analysed as the 3rd person suffix with deletion of the preceding TV and sometimes as the TV itself, which is then followed by a - suffix. Or consider the fact that in the 3pl of the same class of verbs the -a- of cantano ‘they sing’ is required to be treated as a TV, while in verbs of the second and third conjugations the -o- of mettono ‘they put’ and dormono ‘they sleep’ belongs to the ending.4 In fact, the analysis of the old Italian system of endings demonstrates that a sharp distinction between TV and person marker does not help if we want to explain the dynamics that are at work in the later innovations. We will see instead how the processes that give birth to new endings often involve the TV as well, which in this respect needs to be treated on a par with non-thematic vowels. Rather than drawing a dividing line between TV and ending, it makes more sense therefore to distinguish between:5 (a) simple endings, that is to say those consisting of a single syllable containing a single nuclear vowel such as the -o of the 1sg of the present indicative: as we will see, in some cases, in addition to marking person, the ending can also convey tense and mood. The simple endings are those of the three singular persons. (b) complex endings, or those which consist of at least two syllables within which we can distinguish two sub-parts: (i) primary (sub-)ending (PE): the portion of the ending directly associated with person, which appears identical in all the forms in the paradigm— this is the last part of the ending and is generally made up of a single syllable; (ii) secondary (sub-)ending (SE): the remainder of the ending, which provides information about tense and mood. (The vowels that are found in this portion of the ending may or may not be TVs; they will however be dealt with together because, as we have said, the processes that affect the endings do not appear to be sensitive to the distinction.) The endings of the three plural persons are complex. In old Italian these endings are: 1pl: -TVmo (prs.ind), -jamo (prs.ind/sbjv), -amo (impf.ind), -imo (impf.sbjv), -emo (fut), -TVmmo (prf.ind), -emmo (cond) 3 In what follows we make a notational distinction between the Roman numerals I, II, and III, which refer to conjugations, whereas the Arabic numerals 1, 2 and 3 refer to grammatical persons. 4 For a full discussion of the problems relating to the thematic vowels in Italian and a critical analysis of the relevant literature, see Da Tos (2011). 5 For more on these concepts, see Maschi (2005: 99–102).

Person endings in the old Italian verb system

133

2pl: -TVte (prs.ind), -ate (impf.ind), -ete (fut), -jate (prs.sbjv), -TVste (prf./impf.sbjv), -este (cond) 3pl: a) -ano (prs.ind i conjug., impf.ind, and prs.sbjv ii & iii conjug.), -ono (prs.ind II & III conjug., strong prf, impf.sbjv), -ino (prs.sbjv I conjug.), -anno (fut), -ebbono (cond) b) -TVro (weak prf), -ero (strong prf, impf.sbjv), -aro (prs.ind I conjug., impf.ind, prs.sbjv), -oro (weak prf, impf.sbjv), -ebbero (cond), -ebboro (cond) c) -TVrono (weak prf), -erono (impf.sbjv) Although the endings for each of these persons are distinct, a common ‘nucleus’ can be identified within each ending, here underlined, which represents the primary ending, namely: 1pl -mo, 2pl -te, 3pl -ro or -no (in the case of (c) both may coexist in the same ending). It is in this common nucleus that we can recognize the primary morphological value of person, respectively 1st, 2nd, and 3rd.

7.2 Verb endings in old Italian We will now review the endings for the finite forms of the verb as they emerge from our corpus of late 13th- and early 14th-century texts.6 The data should be exhaustive, since it was the intention of those who compiled the OVI corpus to include all available texts from the period under investigation, a period which is considered to be representative of the early stage of Italian. In the course of our analysis, from time to time we will also make reference to forms recorded up until the 15th century. In this case, the material will of necessity be partial and sporadic, and the aim is simply to suggest lines of analysis which will allow for the integration of data from the subsequent stages in the history of the language. Table 7.1 contains the endings which are documented in the texts in our corpus. Those in italics are to be found both in old and modern Italian, while those in bold are no longer present in the modern language. The underlined forms are ones that do not correspond to what on etymological grounds we would expect, again using italics and bold to distinguish those which do and do not survive into the modern language. The endings are set out according to the frequency with which they are documented, and those in parenthesis have only sporadic attestation. In all instances the TV is included within the ending.

6 The data used in this study were obtained by searching the corpus of the Opera del vocabolario italiano (OVI), a project financed by the Consiglio Nazionale delle Ricerche (CNR) with the aim of producing a historical dictionary of Italian. The OVI corpus is a database which is searchable via the internet. It includes vernacular texts from the whole of Italy and covers the period from the earliest attestations until 1375, the year of Boccaccio’s death.

134

Laura Vanelli

TABLE 7.1. old Italian verb endings Present Indicative Person

I

II

III

-o

1sg 2sg

-e / -i

-i

3sg

-a

-e

1pl

-iàmo / (-àmo)

-émo / -iàmo

-iamo / (-ìmo)

2pl

-àte

-éte

-ìte

3pl

-ano / -aro

-ono / (-oro) / (-eno)

Imperfect Indicative Person

I

II

1sg

-a (/ -o)

2sg

-i (/ -e)

3sg

-a

1pl

-àmo

2pl

-àte

3pl

-ano / -aro

III

Future Person

I

II

1sg

-ò

2sg

-ài

3sg

-à

1pl

-émo

2pl

-éte

3pl

-ànno

III

Person endings in the old Italian verb system

Weak Perfect Person

I

II

III

1sg

-ài

-éi / -étti / (-éo)

-ìi / (-ìo)

2sg

-àsti

-ésti

-ìsti

3sg

-ò

-é(o) / -étte

-ì(o)

1pl

-àmmo

-émmo

-ìmmo

2pl

-àste

-éste

-ìste

3pl

-àro / -àrono / -òrono

-éro / -éttero / -érono

-ìro / -ìrono

Strong Perfect Person 1sg

-i

3sg

-e

3pl

-ero / -ono / -oro

Imperative Person

I

II

III

2sg

-a

-i

-i

2pl

-àte

-éte

-ìte

Present Subjunctive Person

I

II

1sg

-i

-a

2sg

-i / -e

-e / -i (/ -a)

3sg

-i

-a

1pl

-iàmo

2pl

-iàte

3pl

-ino

III

-ano / -aro / (-ino)

135

136

Laura Vanelli

Imperfect Subjunctive Person

I

II

1sg

-e / -i

2sg

-i

3sg

-e / (-i)

1pl

-imo

2pl

-ste

3pl

-ero / -ono / -oro / (-erono) / (-ino)

III

Conditional Person

I

II

1sg

-éi

2sg

-ésti

3sg

-ébbe

1pl

-emmo

2pl

-este

3pl

-èbbero / -èbbono / -èbboro

III

A first inspection of Table 7.1 allows us to draw some preliminary general conclusions about the inflectional system of old Italian: (A) The system not only displays many allomorphs, something which is in general typical of verb morphology in the Romance languages, but also exhibits an especially high degree of allotropy,7 i.e. the presence of alternate forms within a single cell of the 7 I use the term ‘allotrope’, as in Pirrelli (2000), in the general sense as defined in the online Dizionario Treccani: ‘synchronic formal variant (or doublet) of another word (as for example malinconia beside melanconia) or of another morpheme (as for example in 19th-c. Italian io avevo beside io aveva), sometimes with a differing stylistic nuance’ (http://www.treccani.it/vocabolario/tag/allotropo/). The term should not therefore be understood in the other narrower sense of ‘etymological allotrope’, by which is meant ‘those words, whether of differing or kindred meaning, which go back by different routes to the same original form, e.g. the pairs vizio, vezzo; nitido, netto; collocare, coricare, in which the derivation from Latin vitium, nitidus, collocare was by an indirect or learnèd route for the first members of the pairs and by a direct or popular route for the second’ (once again from the Dizionario Treccani). Others, e.g. Thornton (2009), use the term ‘overabundance’ to indicate ‘the situation in which a cell in a paradigm is filled by two

Person endings in the old Italian verb system

137

paradigm. The presence of several alternate forms for the same ending does not necessarily mean that these can be used in free variation. Their distribution may be linked to textual or stylistic constraints, for example poetry vs prose or literary works vs private documents, or to sociolinguistic conditions. However, the present analysis is conceived in essentially linguistic or systemic terms; the aim is to tackle the question of the variation in the endings from a structural point of view. Seen in this light, what counts is that these forms should be purely Florentine and not ones that might have been imported from other dialectal varieties with which Florence was certainly in contact. Our confidence that these texts are reasonably reliable and representative of the language of the period is confirmed by the judgement of philologists such as Castellani (1952), Schiaffini (1954), and Nencioni ([1953–4] 1989), who have alluded on more than one occasion to the ‘polymorphy’ of Florentine verb inflection, in particular in the case of the 3pl, an ending which we will consider in detail in }7.2.2. Allotropy, and allomorphy in general, are distributed unevenly across the verbal paradigm. Consider first the two partial paradigms of the future and the conditional, which do not display any patterns of alternation in the endings, except for the special case of the 3pl of the conditional, to which we return below. The stability of these two tenses, which moreover carries over into the modern language, must in some way be connected to the fact that what lies behind them are not the corresponding forms of the Latin future or conditional—indeed, Latin did not even have a distinct conditional paradigm. Rather, they are Romance innovations which derive from an analytic structure made up of the infinitive plus the Romance continuers of the present and the perfect of habere ‘have’.8 This source has given rise to a set of endings with very specific properties, particularly for the future, which is the only circumstance in which the endings of the all the singulars and the 3pl are stressed. As for the conditional, we will see that the fact that it is made up of forms of the perfect of habere has led to its sharing properties with other 3rd person plural endings. However, these are not the only endings which are particularly stable and exhibit no variation across the whole paradigm. The most obvious instance is the 2pl, for which the endings (a) are the regular outcomes of the original Latin forms; (b) display no alternating forms; (c) have persisted without change into modern Italian.

synonymous forms which realize the same set of morphosyntactic properties (as in Italian devo and debbo ‘must.1sg.prs.ind’)’. 8 The roots of this Romance innovation are nonetheless in Latin, as many scholars have shown, not least because the Romance forms presuppose the normal Latin order infinitive þ auxiliary. See in this connection Thielmann (1885); Valesio (1968; 1969); Coleman (1971); Pinkster (1987); Adams (1991); La Fauci (2006); Bourova and Tasmowski (2007).

138

Laura Vanelli

Somewhat different is the case of the 1pl, where in the present indicative we see the coexistence of two complex endings: beside the innovative -jamo, on the origin and diffusion of which see Maiden (1995: }8.3.1) and more recently Meszler and Samu (2007), we find the etymologically expected endings -amo (< Lat -a¯mus) (only one occurrence), -emo (< Lat -e¯mus) (very frequent), and -imo (< Lat -ı¯mus) (few examples): (1)

I: pensamo (Brunetto Latini, Rettorica, p. 118, l. 19), otherwise always -iamo. II: credemo (Lettera di Consiglio de’ Cerchi I, p. 594, l. 6, but also crediamo 596.14), vedemo (Dante, Vn, 18.5, etc.), tacemo (Brunetto Latini, Rettorica, p.156, l. 3), scrivemo (Brunetto Latini, Rettorica, 157.5), but also sscriviamo (Lettera di Consiglio de’ Cerchi I, 596.10), semo (Novellino, 75, r. 44; Dante, Vn, 14.2; etc.), but more frequent is siamo (Bono Giamboni, Trattato 31.6, etc.); facemo (Brunetto Latini, Rettorica, 176.9, etc.), and also facciamo (Brunetto Latini, Rettorica,9 173.3; Bono Giamboni, Libro, 36.22, etc), volemo (Lettera di Consiglio de’ Cerchi, etc., I, 595.9; Dante, Vn, 25.4, etc), more rarely vogliamo (Bono Giamboni, Libro, 44.7; Fiore di rettorica, 82.101, 27, etc). III: venimo (Brunetto Latini, Rettorica, 172.8), gimo (Gatto lupesco, l. 23), etc., otherwise always -iamo.

(B) Our second observation concerns those endings which exhibit variation: when the variants alternate within the individual cells of the paradigm, we find at one and the same time both those which are the expected reflex of the Latin etymological source and those which develop subsequently and give rise to the modern system.10 In this sense, old Italian is revealed as an essentially conservative stage of the language, displaying a regular continuation of almost all the morphological forms inherited from Latin. At the same time—and this is perhaps the most striking feature—it includes all the innovations that will characterize the later development of the language. Thirteenth-century Florentine is thus the moment at which we can observe all the changes which underlie the future inflectional framework of Italian. Indeed, it is precisely through analysing this superabundance of innovative forms alternating with each other, some of which will in due course be abandoned, that we can shed new light on the dynamics of the systemic change which has affected the language.

9 It should be noted that the endings in -iamo are added to one of the stems of the present subjunctive (facciamo, vogliamo, etc), while the endings in -emo are added to the stem of the present indicative (facemo, volemo, etc), a fact which is in turn evidence that -iamo is in origin the ending of the subjunctive (< Lat -ea¯mus, -ia¯mus). 10 Since they are not relevant to the present argument, we will not address here issues relating to the etymological pathways which connect the Latin and the Italian endings, such as the formation of the 3pl ending in -no or the development of the 2sg ending in -e from Latin -as. See instead the discussion and references in Maiden (1995: }8).

Person endings in the old Italian verb system

139

In what follows we will pay particular attention to the non-etymological endings, and seek to explain how and why they have been introduced into the system. The sole exception will be the well-known and well-studied case of the 1pl ending in -iamo, to which we have already made reference in the foregoing discussion. The analysis falls into two parts: in }7.2.1 we will study the simple endings, that is to say those of the three singular persons, while }7.2.2 will be devoted to the analysis of the 3pl, which is where we find the most conspicuous variety of alternating forms. 7.2.1 Simple endings: 1st, 2nd, and 3rd persons singular 2.1.1 1sg The first two innovations which we will deal with, respectively the imperfect indicative and the weak perfect, relate to the extension of the -o of the present indicative, the regular reflex of Latin -o. These are analogical changes which have the effect of reducing allomorphy, and which were triggered by a general principle commonly seen in action in the formation of complex paradigmatic systems such as those operating in the verbal domain. This is the principle of ‘isomorphism’ or ‘uniform encoding’, as it is called in Natural Morphology (Mayerthaler 1987: 49), that is, a tendency for a bi-unique relation to be established between a signifier and its signified; this process introduces a degree of naturalness or transparency into the system, in the present case into the relation between the ending and the grammatical expression of person. The first of these changes, in the imperfect, has become firmly established in the modern system, while the change in the weak perfect has fallen away and has been without consequence in the modern language. The reason in the latter case is presumably to be found in the fact that, as we shall shortly see, the system became stabilized with two allomorphs for the 1st person: -o for the present and the imperfect indicative, and -i for those stems which formally have their origin in the Latin perfective subsystem: the perfect, the imperfect subjunctive (which derives from the Latin pluperfect subjunctive) and the conditional. (A) Imperfect indicative: as we can see, old Italian was characterized by the almost exclusive presence of the ending in -a, which is the regular outcome of the Latin -a(m). Thus: . . . ed io sorridendo li guardava . . . (Dante, Vn, 4.3); . . . com’io credeva (Novellino, 75, l. 40), etc). The modern system has replaced this -a with -o, on the analogy of the -o in the present. This modern ending is only sporadically documented in the first part of the 13th century, e.g. io ti serbavo . . . (Dino Compagni, Cronica (1312), p. 133, l. 1) or Io ti tenevo . . . (Paolino Pieri, Storia di Merlino (1330), p. 26, l. 38), becoming established later between the end of the 14th and the beginning of the 15th centuries. In the view of Manni (1979: 146), ‘it is the generations born after the middle of the 14th century which provide significant evidence of the transition from the older forms in -a to those in -o’. Petrarch (d. 1374) and Boccaccio (d. 1375), for example, still use the form in -a (Manni 2003: 195, 272).

140

Laura Vanelli

(B) Weak perfect: beside the regular forms in -i in the second and third conjugations, we also find forms ending in -o: poteo (Dante, Vn, 8.3); sentio (Dante, Vn, 24.1, etc), uscìo (Fiori e vita di filosafi, 28, l. 23), which thus come to be identical to the 3sg. This raises an obvious question: how can we be sure that in such cases we have to do with the extension of the -o of the 1sg and not simply with the 3sg which has sporadically extended its reference to include the 1st person, a phenomenon which is still today found in some southern varieties such as Procidano? The answer is that we cannot be absolutely sure, but an extension of the 3rd person to the 1st would have no internal motivation within the system. The third change which we will consider concerns the present and the imperfect subjunctives, forms in which the Latin endings in -em in words such as cantem (sing.1sg.prs.sbjv) and cantavissem (sing.1sg.prf.sbjv) and the like would lead us to expect -e but where we find instead -i (It. canti, cantassi): (C) Present and imperfect subjunctive: the ending -i is the only one documented for the first conjugation present, while in the imperfect -i alternates with the etymological -e:11 (2)

Vo’ tu che io li abandoni? (Novellino, 72, ll. 17–18), . . . non fa bisogno ch’io t’insegni andare alle Virtudi, né ch’io t’aconti co·lloro . . . (Bono Giamboni, Libro, 11.27).

(3)

a. io patisse, portasse, sostenesse (Bono Giamboni, Libro, 1.2), io mandasse (Dante, Vn, 41.1), parlasse (Dante, Vn, 35.4), io andasse, celasse (Gatto lupesco, ll. 51–2); unisse (Novellino, 65a, l.31) b. dicessi (Fiore di rettorica, ch. 37, p. 36, l. 9), volessi (Brunetto Latini, Tesoretto, l.543), perdessi, fossi, potessi (Dante, Vn, 15.2).

The origin and the establishment of these two new endings presents a number of problems. The presence of -i in the imperfect beside the etymological -e can be explained as part of the same process of reduction of allomorphy in the 1sg that we have just seen. The -i is in fact taken up as the 1sg ending of those tenses which have the ‘perfective’ stem, including of course the imperfect subjunctive, which the reader will recall derives from the Latin pluperfect subjunctive.12 However, an interpretation which invokes a tendency to extend an analogical process favouring isomorphism is harder to apply in the case of the present, since first we have here a stem which is not

According to Castellani (1952: 156), ‘the ending -e . . . remains the rule until the end of the 13th century’. 12 For a discussion of the various hypotheses that have been advanced in connection with this question, see Vanelli (in press), which argues for a ‘purely morphological’ motivation (in the sense of Aronoff ’s 1994 concept of the ‘morphome’) for the analogies that unite these three tenses in formal terms even though there is no semantic connection between them. 11

Person endings in the old Italian verb system

141

perfective, and second and more importantly, the adoption of -i only affects the first conjugation while in the other conjugations the regular ending retains the -a from Latin -a(m). Thirdly, we also find the same shift from -e to -i in the 3rd person singular, where we equally do not have allotropy insofar as we do not have endings in *-e coming regularly from -e(t), and in this case we cannot invoke analogical processes based on other tenses since there are no other 3rd person singulars in -i.13 For this reason we postpone further discussion and a possible solution until after we have dealt with the outcomes relating to the 2nd and the 3rd persons, where we will show that the processes in operation can be explained in the context of the partial paradigm made up of the singular persons of the subjunctive taken together. 2.1.2 2sg (A) The most obvious difference for this person between old and modern Italian is the fact that at the earlier stage the ending in -e is still documented while it is no longer present in the modern language, having instead been replaced by -i. The former is the regular outcome both of the Latin ending -as, which appears in the present indicative of the first conjugation, the present subjunctive of the other conjugations, and the imperfect indicative of all conjugations, and of the Latin ending -e of the second conjugation imperative. This -e has been gradually replaced by -i through an allomorphy-reducing analogical extension of the same type as we have seen at work in the 1st person. Recall that -i, the starting point of the process, is the regular outcome in the present indicative of the second conjugation, the present subjunctive of the first conjugation, the imperfect subjunctive (< Lat. -es), the present indicative of the third conjugation (< Lat. -is), and the third conjugation imperative (< Lat. -i). Its extension to the other tenses is already amply documented in old Italian but to differing degrees in the various tenses and moods. It is already complete in the imperative ( . . . questo consiglio, [ . . . ] intendi / [ . . . ] / lo male e ’l ben con ubidenza prendi, / [ . . . ] ti difendi, (Rinuccino, 9m, ll. 3, 5, 7–8)), and it is almost complete in the imperfect indicative, where only two residual instances of -e are recorded: mostrave (Brunetto Latini, Tesoretto, l. 2575) e pensave (Bono Giamboni, Libro, 60.4). In the first conjugation present indicative and the present subjunctive of the second and third conjugations, the analogical -i is by contrast still in variation with -e:

13 This oscillation between final -e and -i does not only affect verb endings, but is also to be found elsewhere in the Italian lexicon: see e.g. items like oggi ‘today’ (< hodie¯), domani ‘tomorrow’ (< de ma¯ne), dieci ‘ten’ (< decem), avanti ‘before’ (< ab ante), lungi ‘far’ (< longe¯) and the like, with -i instead of the etymological -e. In the past this state of affairs has led various authors (e.g. Meyer-Lübke 1890: i.262; Tekavčić 1980: i.8990; ii.279ff.; Rohlfs 1966: i.}142) to look for an explanation of these developments not (or not only) in terms of a reorganization of the inflectional system but rather by invoking an independent sound change of final [e] to [i]. For a discussion of this hypothesis and its attendant difficulties, see Vanelli (in press).

142

Laura Vanelli

(4)

Present indicative: ami (Novellino, 6, l. 37), canti (Guido Cavalcanti, 1, l. 14), pensi (Brunetto Latini, Tesoretto, l. 763), as opposed to: cante (Dante, Vn, 12, 10, l. 3), pense (Fiori e vita di filosafi, 24, l. 114).

(5)

Present subjunctive: . . . consiglioti per la fede, [ . . . ] che incontanente ti facci fedele de le Virtú ed entri di lor compagnia e prometti [ . . . ], e compî . . . (Bono Giamboni, Libro, 76.10), . . . voglio che tu dichi certe parole per rima . . . (Dante, Vn, 12.7), as opposed to: E tu vivi sie che tu non ti commette neuna cosa la quale tu non posse commettere al nemico tuo (Fiori e vita di filosafi, 24, ll. 178–9), . . . non vo’ che la dimette . . . (Brunetto Latini, Tesoretto, l. 1908).

(B) More problematic on the other hand is the presence of an innovative and nonetymological form which is neither -e nor -i, namely the -a of the 2sg of the present subjunctive outside the first conjugation and where the expected outcome would have been -e from the Latin ending -as. Examples are: (6)

. . . se tu ài presa alcuna volontade in mal dire, che tu la perda in mal udire . . . (Brunetto Latini, Rettorica, 192, ll. 16–17), Non vo’ ti faccia di ciò maraviglio . . . (Monte Andrea (ed. Menichetti), sonnet 104b, l. 35).

It is precisely this ending which will supplant all the others and become the regular ending in modern Italian. Since in this instance too the innovation is driven by mechanisms which involve the three singular persons of the subjunctive taken together, we will delay our discussion of it until the next paragraph. 2.1.3 3sg The ending of the 3sg displays less variation than do the other persons. In general, apart from one specific instance which we will come to shortly, the endings are the regular outcomes of the Latin inflections. The ending of the weak perfect of the second and third conjugations does, however, deserve particular comment. Here, beside the expected forms in -ìo (< Lat. -iu(it)) and -éo, we also find the endings with final stressed -é and -ì, which trigger consonant doubling (partíssi), and which have survived into the modern language.14 (7)

temeo (Novellino, 32, l. 24), poteo (Dante, Vn, 3.7), conbatteo (Cronica fiorentina, 146, l. 12), morio (Libro di Lapo Riccomanni, 542, l. 28), partio (Bono Giamboni, Libro, 7.1); but also abatté (Novellino, 99, l. 29), udì (Novellino, 99, l. 47).

In order to explain this change, Castellani (1952: 146) brings into the equation another phenomenon which affects the 3sg of first-conjugation perfects, namely the apparently similar alternation between forms in -ò (e.g. donò, Cronica fiorentina 111, l. 9) According to Castellani (1952: 144 ) there is an ‘increasingly rare use of -eo, -io in the generations born after the middle of the 13th century. The generations born after the end of the century have only -è, -i (and -ette).’ 14

Person endings in the old Italian verb system

143

and forms in -òe (e.g. donoe, Libro di Lapo Ricomanni 546, l. 26). However, this is a different phenomenon which does not concern verb morphology alone but is well attested across the whole lexicon: the addition of an epithetic or paragogic [e] as a way of avoiding words that end in a stressed vowel.15 On Castellani’s account, ‘the effect of these forms is that the -o of -eo, -io is considered to be epithetic, and its use becomes sporadic.’ One could perhaps reformulate Castellani’s explanation as a case of reanalysis: as a result of the comparison with the first-conjugation forms in -ò < Lat. -au(it), which are in alternation with -òe, the real unstressed ending -o is dropped and the stressed vowels -é and -ì, which were originally TVs, get reanalysed as endings. In this way, the second and third conjugations are also aligned with the first, and indeed we find forms in -ìe for conjugations other than the first: e.g. bandìe (Novellino, 75, l. 4) with an epithetic -e. However, the most interesting case concerns the endings of the first-conjugation present subjunctive, in which the 3sg undergoes the same changes as the 1st person (see }7.2.1.1). In both instances the regular ending -e < -e(m), -e(t) is no longer found, but only -i. Given that the Latin 2sg ending -e(s) also becomes -i by regular sound change, the effect of this innovation is to produce the levelling of the endings of the first three persons.16 One may then ask why this levelling should have generalized the ending in -i, which is etymological only in the 2nd person, and did not follow the opposite route of extending the -e of the 1st and 3rd persons into the 2nd. It is reasonable to think that -i prevailed through the convergence of a number of factors. In the first place, if the non-etymological -e had won out in the 2nd person, this would have been directly counter to the process in operation at that time whereby etymological -e was being replaced by -i (see }7.2.1.2). Second, the replacement of -e with -i in the 1st person is consistent with the general process of allomorphy reduction that we have already seen at work in the case of the imperfect subjunctive, where we also have -e > -i, albeit for different reasons. Whatever the precise motivation, the unstressed endings of the 1sg tend to polarize as either -o or -i, with the gradual elimination of the other possible variants, namely -a and -e, with -a remaining only in the present subjunctive, for reasons that we will shortly explore. Finally, the

15 Within the sphere of verb inflection, epithesis is also found in the future: e.g. canteroe (Novellino, 64, l. 73), ubidiroe (Novellino, 65b, l. 41), and in the final-stressed members of the paradigms of particular verbs: e.g. vae (Zucchero Bencivenni, Santà del corpo, 131, l. 34 [1310]), hae (Bono Giamboni, Libro, 7.10), òe (Libricciolo di Bene Bencivenni, II, p. 399, l. 27); èe (Compagnia di San Gilio, p. 38, l. 13). For a general account, see Larson (2010: 1518–19). 16 The same thing does not happen, however, in the 3sg of the imperfect subjunctive < -e(t) which in general keeps the ending -e (but see fn. 25). In this connection it is also relevant to note that, unlike modern Italian, old Italian was a partial null subject language in which main clauses allow a null subject and embedded clauses require an overt one, so that the ambiguity of the endings in the present subjunctive would have been compensated for by the presence of a subject pronoun. A residue of this state of affairs is to be seen in modern usage in the obligatory presence of the weak pronoun tu ‘you’ with the present subjunctive.

144

Laura Vanelli

shift from -e to -i is also independently motivated in the 3rd person: if the ending is -i, then there is a parallellism with the 3pl in -ino, in obedience to the principle of iconicity (for which see the next section), according to which the 3pl is commonly formed by affixing -no to the 3sg. Moreover, -ino is the regular outcome of Lat. -en(t), since a non-low front vowel regularly raises to [i] when it is post-tonic and followed by a nasal consonant: anima, uomini, pettine < anı˘ma(m), homı˘ne(s), pectı˘ne(m). A process similar to the one we have just described for the first conjugation present subjunctive also takes place in the second and third conjugations, which are likewise subject to an innovation whose effect is to produce levelling of the three person endings. In this instance, however, the innovating person, as we have seen, is the 2nd, where, beside the regular -e and the -i which is analogical on the other 2sg endings, we also find the variant in -a, which must be seen as an extension from the 1/3sg, where -a is the etymologically expected outcome. And it is indeed this form which comes to replace all the other variants and become the regular ending for the modern language. The levelling process for the three singular persons in -a only takes place later, between the 14th and 15th centuries (Castellani 1952: 71).17 These readjustments of the pattern of endings that lead to the modern system show that within the present subjunctive in the 13th century a process is already in operation which tends to unify the singular endings and to neutralize the formal difference between the persons. Such a process of syncretism seems to push in the opposite direction to the above-mentioned principle of isomorphism, in that the same realization, either -i or -a according to conjugation, comes to be associated with different persons. I suggest that the explanation for the fact that these processes seem to run counter to the principles of naturalness and transparency is to be found in the intrinsic morphological properties of the present subjunctive. In general, in a verb form the tense and mood values are expressed in the stem, while the ending is associated with person and number. However, in the present tense the stems of the indicative and subjunctive coincide, at least for the so-called regular verbs, so that at the level of the stem there is ambiguity in the expression of mood. In the subjunctive the functional load is therefore borne by the endings, which in consequence serve as the signifier not only for person but also for mood. The processes of levelling that take place within the system of the present subjunctive on the one hand neutralize the person values in the endings, but on the other serve to establish an unambiguous link between form and meaning. In other words, we observe here a different application of the principle of isomorphism, which in this instance affects the expression of mood rather than person: the endings -i/-a attached to the present stem denote ‘subjunctive’ for their respective conjugations.18 In Boccaccio we still see the alternation between the 2sg in -i and in -a (Manni 2003: 275). It should be noted that processes of the same kind have taken place in other Romance languages (although not in Romanian), where in general the present subjunctive exhibits the same unification of the 17 18

Person endings in the old Italian verb system

145

7.2.2 The complex ending of the third person plural In the analysis of the 3pl, which is characterized by the presence of complex endings throughout the paradigm, we need to deploy a further concept, one which also derives from so-called ‘Natural Morphology’ and which will help us to understand better the mechanisms which underlie the formation and distribution of new endings. This is the general principle of ‘iconicity’ (Mayerthaler 1977; 1987) which, together with the principle of uniform encoding or isomorphism, serves to yield naturalness and transparency, and therefore regularity, within systems. This principle, which originates with the American philosopher and logician Charles Sanders Peirce, covers those cases in which there is a structural analogy between a signans (or signifier) and a signatum (signified). The principle exploits the notion of ‘diagram’, by which is meant, according to the definition given in Dressler (1995: 21), a type of iconic sign in which we can ‘establish analogies between relations in the signans on the one hand, and relations of the signatum of the same sign on the other’. This principle is operative in particular in the relation between singular and plural. Thus, Thornton (1999) proposes, and backs up with a range of convincing arguments, a diagrammatic interpretation of the 3pl verb forms according to which ‘the 3pl is built as a genuine plural of the 3sg, with a plural morph (-no in most cases) added to the 3sg’ (p. 490).19 Bearing this principle in mind, we now proceed to the analysis of the old Italian endings. Here too we find a number of instances in which the 3pl can be analysed in iconic terms as the result of adding the (primary) ending of the plural to the 3sg. The primary endings in old Italian are moreover the same as in modern Italian, namely -no, -ro, -rono. We thus find forms similar those of the modern language in the following tenses: (8)

Present indicative, first conjugation: 3sg porta (Doc. fior., 1211, 36.21), 3pl portano (Bono Giamboni, Libro, 8.3).20 Present subjunctive: 3sg canti (Compagnia di San Gilio, 51, ll. 1–2), conosca (Fiorie vita di filosafi, 135.5), 3pl cantino (Stat. fior., 1280-98, par. 36, 62.23), conoscano (Bono Giamboni, Vegezio, L. 3, 12, 113.10). Imperfect indicative: 3sg dicea (Brunetto Latini, Retorica, 66.9), 3pl diceano (Dante, Vn, 22.6, l. 14). Future: 3sg conoscerà (Dante, Vn, 12.7, l. 8),

endings according to conjugation, or at least of the same suffixal vowel in those languages, e.g. Portuguese and Spanish, which maintain the final -s in the 2sg. In some languages, such as Catalan and Friulian, the suffixal vowel (-i) is the same across all the conjugations (Iliescu and Mourin 1991). 19

For more detailed discussion of this hypothesis, see Vanelli (2007). We also find cases like piaceno (Disciplina clericalis, 76, l. 4), puoteno (Tesoro volgarizzato (ed. Gaiter), L. 2, 44, p. a356, l. 2), poteno (Sommetta, 207, l. 21), in which the construction of the 3pl according to the formula [3sg + -no] is extended to other conjugations. It should be emphasized, however, that these are not pure Florentine forms but ones originating from western Tuscany where the phenomenon is much more widespread. See also fn. 21. 20

146

Laura Vanelli 3pl conosceranno (Compagnia di San Gilio, 44, l. 3) (with geminaton of the n). Weak perfect: 3sg fuggì (Novellino, 20, l. 2), 3pl fuggiro (Novellino, 18, l. 62), partì (Bono Giamboni, Libro, 43, rubr.), partirono (Cronica fiorentina, 127, l. 13). Strong perfect: 3sg vide (Dante, Vn, 9.11, l. 9, etc.), 3pl videro (Novellino, 46, l. 13). Imperfect subjunctive: 3sg credesse (Bono Giamboni, Libro, 75.3), 3pl credessero (Bono Giamboni, Libro, 41, 72.17). Conditional: 3sg perderebbe (Novellino, 13, l. 3), perderebbero (Novellino, 61, l. 45).

By comparison with modern Italian, there is a further case of diagrammaticity which concerns the weak perfect of the first conjugation: while in the modern language we have guardò vs guardarono ‘(s)he vs they watched’, in old Italian the 3pl also has forms like guardorono (Cronica fiorentina, 95, l. 13), costorono (Quaderno di Compagno Ricevuti, 571, l. 18), which are built on the singulars guardò, costò + -rono,21 as in the other conjugations. In the view of Manni (1979: 152) such cases reveal the influence of western Tuscan. However, in reality this is not the essential property of the 3pl endings. The most notable feature is the high degree of allotropy,22 as a result of which we have a system in which there is a plethora of coexisting endings. Indeed, this was something which Nencioni ([1953–4] 1989) had already commented on in discussing the ‘polymorphism’ of 3pl marking in old Italian. These endings are either different from those found in modern Italian or, more commonly, they are the same endings but with different distributions and with alternations within the same paradigmatic cell. We set out below the endings which are documented: (A) -ono, the ending of the present indicative of the second and third conjugations, and -oro, which is absent from the modern system, alternate with -ero, distributed as in the modern language, in the strong perfect, the imperfect subjunctive, and the conditional: Strong perfect (9) vollero (Compagnia di S. M. del Carmine, 63, l. 23), vollono (Bono Giamboni, Vegezio, L. 3, 1, 78, l. 14), and volloro (Brunetto Latini, Rettorica, 23, l. 2); ebbero (Bono Giamboni, Libro, 15, par. 13) and ebono (Libro di Ruggieri da Firenze, p. 479, ll. 8, 13, 22); dissero (Novellino, 42, l. 19) and dissono (Quaderno di Consiglio de’ Cerchi, p. 621, l. 14); caddero (Bono Giamboni, Libro, 5.4) and 21 We also find cases in which the primary ending -no is added directly to the 3sg of the perfect. This process applies both to the weak inflection in -ò, in which case the stressed final vowel triggers doubling of -n-, and to the strong ending: divoronno (Itinerario luoghi santi; 162.13), terminonno (Dante, Paradiso, 28.105 [no later than 1321]); volleno (Registro di S.M. di Cafaggio, 218, l. 12), funno (Garzo, S. Chiara, l. 308). This is presumably due to the influence of western Tuscan dialects (Castellani 1952: 152; Schiaffini 1954: xxi-xxiv). 22 In modern Italian the only instance of allotropy within the system of verb endings is the alternation in the second conjugation perfect between -ei, -è, -erono and -etti, -ette, -ettero as in temei, temé, temerono beside temetti, temette, temettero.

Person endings in the old Italian verb system

147

caddono (Bono Giamboni, Trattato, 32.3); fecero (Fiori e vita di filosafi, 13, l. 25), rifeciono (Quaderno di Consiglio de’ Cerchi, p. 621, l. 3), and fecior (Brunetto Latini, Rettorica, p. 34, l. 23); vennero (Brunetto Latini, Rettorica, p. 30, l. 14), vennoro (Registro di S.M. di Cafaggio, p. 188, l. 8); ricevettero (Cronica fiorentina, p. 141, l. 16) and ricevettono (Libro di Ruggieri da Firenze, p. 498, l. 1). Imperfect subjunctive (10) piangessero (Dante, Vn, 23.5), partissero (Brunetto Latini, Rettorica, p. 43, l. 8), morissero (Testamento di Beatice da Capraia, p. 239, l. 20); paghassono (Libro di Ruggieri da Firenze, p. 510, l. 5); conoscessono (Bono Giamboni, Trattato, 1. 1), dovessono (Cronica fiorentina, p. 140, l. 19);23 potessoro, facessoro (Compagnia della Madonna d’Orsammichele, p. 664, ll. 18 and 19), fossor (Rinuccino, 8a, l. 14), avessor (Brunetto Latini, Rettorica, p. 25, l. 10). Conditional (11) infamerebbero (Bono Giamboni, Libro, 12.4), porterebber (Forese Donati, p. 3, l. 8), perderebbero (Novellino, 61, l. 45); potrebbono (Brunetto Latini, Rettorica, p. 27, l. 5), vederebbono (Dante, Vn, 15. 8); offerrebboro (Brunetto Latini, Rettorica, p. 109, l. 17), portereboro (Pallamidesse Bellindote, l. 59). (B) In the weak perfect the alternation is between the etymological endings -aro, -ero, and -iro, which do not survive into modern Italian, and the corresponding -rono, in which the two primary endings are, so to speak, ‘added up’: (12)

domandaro (Disciplina clericalis, p. 77, l. 12) and insegnarono (Novellino, 54, l. 8); conbattero (Cronica fiorentina, p. 121, l. 22) and abacterono (Cronica fiorentina, p. 89, ll. 20 and 23); fuggiro (Novellino, 18, l. 62) and partirono (Cronica fiorentina, p. 127, l. 13).

(C) The ending in -ro is also found beside -no, which is the only option in modern Italian, in the present indicative of the first conjugation, in the imperfect and in the present subjunctive (-aro). Also attested, although only sporadically, are -oro for -ono in the present indicative of conjugations other than the first: (13)

a. Present indicative àmaro (Bono Giamboni, Libro, 11.11), raddóppiaro (Bono Giamboni, Libro, 4.12); mettoro (Chiaro Davanzati; 28.104) b. Imperfect indicative davaro (Libricciolo di Bene Bencivenni, II, p. 379, l. 4), stavaro (Gatto lupesco, l. 114), bagnavaro (Distruzione di Troia, p. 184, l. 18), dovearo (Libricciolo di Bene Bencivenni, I, p. 293, l. 22)

23

For the sporadic presence of the ending in -ino, see fn. 25.

148

Laura Vanelli c. Present subjunctive reddaro (Compagnia di San Gilio, p. 44., ll. 21–5), vadaro (Bono Giamboni, Libro, 10.4), vegnaro (Bono Giamboni, Libro, 14.6).

7.3 Concluding remarks Faced with this surprising superabundance of forms, we must now try to understand (a) how the new endings have come about, and (b) what the causes are of such a proliferation of endings. To answer the first of these questions, we set out what the endings would have been if they had followed the ‘etymological’ path: Present indicative: I conjug. -ano < -ant,24 II *-eno, III -ono < -unt Imperfect indicative: -ano < -(ab)ant Present subjunctive: I conjug. -ino < -ent, II/III -ano < -ant Weak perfect: -àro, -èro, -ìro < -arunt, -erunt, -irunt Strong perfect: -ero < -erunt Imperfect subjunctive: *-ino < -(ss)ent (not attested; always replaced by -ero) Conditional: -ebbero < ebbero (< habuerunt) When we come to reconstruct their Latin origins, we see that the forms in -ro go back to the perfect ending -runt and are therefore only etymologically justified in the perfect (cantaro, finiro, videro, fecero) and the conditional, since the latter is formed out of the infinitive + habuerunt, the perfect of habere ‘have’, whence canterebbero ‘they would sing’ and kindred forms. In all the other tenses and moods the expected etymological outcome is -no. In particular, the imperfect subjunctive derives from the Latin pluperfect subjunctive, which is built on the Latin perfect stem plus the ending -ssent, e.g. ama(ui)ssent, dormi(ui)ssent, fecissent, and should therefore have an ending in -ino as in the present subjunctive. Hence we would expect *cantassino beside cantino, but such forms are not—or not any longer—attested in 13th-century Florentine.25 The imperfect subjunctive ending in -ero (amassero, dormissero, facessero) has presumably been taken over from the conditional (Maiden 1995: 142). The 24 For the various hypotheses concerning the presence of the non-etymological final -o of this suffix, see Politzer (1958); Rohlfs (1968: 255); Tekavčić (1980: ii.284ff ); and Maiden (1995: 140–42). 25 Forms in -ino are in fact found but they are very rare. We have noted two, both from the year 1300: dovessino (Cronica fiorentina, p. 25, l. 28) e reducessino (Anonimo, Tesoro, p. D279, l. 4). In general the presence of these forms is linked to the corresponding presence of -i in the 3sg in place of the regular -e, and this too is very rare in the 13th century. The pattern -i, -ino, as Manni (1979: 159–61) has shown, is for the most part documented in the 14th and especially 15th centuries, although as she notes, ‘the third persons in -i, -ino develop through slow and complex stages and never become definitively established.’ In view of the origin and the development of these forms, they are best treated as innovations introduced by analogy with the present subjunctive, and not as relics of an older stage.

Person endings in the old Italian verb system

149

end result is the formal consistency in the expression of the perfect indicative, the conditional, and the imperfect subjunctive which was discussed in }7.2.1.1. The starting point for the system of 3pl endings is then the fact that the operation of the normal mechanisms of sound change has yielded two primary endings for this constellation of morphosyntactic features (-no and –ro) which are in complementary distribution within the paradigm and which give rise to various complex endings. What we see in 13th-century Florentine is that each of these two endings becomes extended through analogy into tenses and moods where it is not etymological. These analogies can in turn be attributed to the tendency towards uniform encoding, by which we mean the establishment of a bi-unique relation between the signifier (the ending) and the signified (3pl). The peculiar feature in this instance, unlike many others described in the literature, is that the analogical process does not affect just one of the primary endings, but both of them at the same time although in different ways and to differing degrees. The end product still respects the principle of ‘one form one meaning’ but with two competing forms rather than one. We thus end up with two parallel and alternative systems, one with -no and one with -ro, both of which respect the principle of uniform encoding. It is out of this situation that allotropy is born. The process is particularly evident with the more widespread primary endings in -no. These endings, which are the most frequent within the paradigm, must have been felt to be the unmarked ones, and in some cases they even stray outside the verb system altogether, as with the analogical plural pronouns ellino ‘they.m’ and elleno ‘they.f’.26 From this starting point, -no tends to be extended to those tenses where phonological changes have produced different endings, in particular the allomorphs in -ro. In consequence: (a) it is added to weak perfects in -ro which thus end up being doubly marked: this is the source of the modern Italian endings in -rono; (b) the complex ending -ono, characteristic of the second and third conjugations, comes to replace the complex ending -ero, which occurs in the other tenses, namely the strong perfect, the imperfect subjunctive, and the conditional. Underlying this extension is the parallellism at the paradigmatic level between the 3sg in -e of the present of the second and third conjugations, as in vede and sente, and the 3sg of the tenses which undergo replacement such as vide, sentisse, sentirebbe. Thus on the basis of the pattern vede : vedono we get similar patterns vide : vidono, vedesse : vedessono, vederebbe : vederebbono, and the like (Castellani 1953: 155). It is possible too that a contributory factor here was the fact that once the endings of the

26 It is also this ending which is found in those dialects which have developed inflected non-finite forms, e.g. ONap. avereno ‘to.give.3pl’, mostrandono ‘showing.3pl’, datono ‘given.3pl’ (Ledgeway 2009a: }14.5).

150

Laura Vanelli

weak perfect had been augmented to become -rono, the whole ending can be reanalysed and resegmented to yield the substring -ono, thereby increasing the transparency. However, beside this process which tends to extend -(o)no to all tenses and moods, there exists a kind of parallel counter-process which takes as its starting point the other primary ending -ro. The extension of the latter occurs in two directions: (a) as the simple replacement of -no in the present and imperfect indicatives and the present subjunctive and the creation of a new formation -aro which parallels -ano (see the examples in (13)); (b) through the new formation -oro, which is analogical on -ono (see the examples in (9)–(11)). That said, the power of -ro to extend is weaker than that of -no, and hence: (a) we do not find instances in which -ro has been added to -no, and we must assume that the ending *-noro was ungrammatical; (b) -oro only tends to be used with any degree of frequency in those tenses in which it alternates with -ero and -ono, while it is much rarer elsewhere, e.g. in the present indicative of the second and third conjugations. One instance, however, is mettoro (Chiaro Davanzati 28.104). From the analysis of the complex system of 3pl endings we can draw the following conclusions. First, the allotropy of the 3pl in old Italian is the result of a tension between the tendency towards uniform encoding and the presence in the system of historically determined allomorphy which forces into competition two primary endings, both of which, so to speak, aspire to assume the role of the sole exponent of 3rd person plural. Second, when it comes to determining which of our two principles of iconicity (or diagrammaticity) and isomorphism prevails, we see that in our data the tendency to isomorphism predominates. Only in the case of the substitution of -orono for -arono in the weak perfect—e.g. guardorono for guardarono ‘they watched’—does the innovation yield an iconic relation: 3pl guardorono = 3sg guardò + rono. In some instances the analogical extension maintains iconicity, as in the replacement of -no by -ro: ama, àmaro; dava, dàvaro; vegna, vègnaro. Diagrammaticity is on the other hand eliminated in all the cases in which -ono and -oro replace -ero: contrast volle, vollero; facesse, facessero; perderebbe, perderebbero with volle, vollono, volloro; facesse, facessono, facessoro; perderebbe, perderebbono, perderebbero. We seem therefore to have here processes which, while they do not systematically operate against diagrammaticity, nonetheless interfere with it. This perspective is consistent with Nencioni’s account of the Florentine changes:

Person endings in the old Italian verb system

151

To summarize then, when we compare in structuralist terms the morphological picture of the second half of the 14th century with that of the second half of the following century, it must be said that a vertical system of oppositions between the 3sg and the 3pl operative in the earliest stage of Florentine and hinged around the final vowel of the singular . . . was followed from the middle of the 14th century and after [though the data show that in reality the process is already active in the 13th century -L.V.] by a system of horizontal analogical levelling between the 3pl forms based on their final vowel o. (Nencioni [1953–4] 1989: 66).

Third, this tendency towards processes which promote uniform encoding has not prevailed. Modern Italian has neither accepted nor continued the innovations evidenced in the 13th and 14th centuries. Indeed, in a way it has retraced its steps. Allotropy has been almost completely eliminated, and even allomorphy has been reduced; the new endings -oro and -orono have not survived, and the system has been restructured in such a way that the tension between uniform encoding and diagrammaticity has been resolved in favour of the latter, even if only partially. Thus, the connections between the 3rd person singular and plural in the present indicative of the second and third conjugations still do not display any iconic relation; in this respect teme/temono and dorme/dormono differ from canta/cantano (see Thornton 1999 and the remarks in Vanelli 2007).

Part II (Pro)nominal Structures

8 The evolution of Italo-Romance clitic clusters Prosodic restructuring and morphological opacity* D I E G O PE S C A R I N I

8.1 Introduction This chapter focuses on the evolution of Romance clitic sequences with the intent to provide a principled analysis of the syntactic change reversing their internal order.1 In old Florentine,2 for instance, the order accusative > dative in (1a), found in the

* This paper is dedicated to Mair, for her kind support. The research was generously funded by the Italian Ministry of Education, Universities and Research as part of the FIRB project RBFR08KR5A, ‘A grammatical survey of Italian dialects: fieldwork, data management, and linguistic analysis’ (http://asit. maldura.unipd.it). I would like to thank audiences in Verona (IGG 39), Cedar City (LSRL 42), Goettingen, and Pavia for their very helpful feedback. I would particularly like to thank Mair Parry, Paola Benincà, Christina Tortora, and Silvia Rossi for having provided many helpful comments, which led to a much improved version. 1 For the sake of consistency, in what follows I will take into consideration data from Italian vernaculars, but the same analysis can hold for Gallo-Romance. 2 Following Renzi and Salvi (2010), I consider 13th–14thc. Florentine as the ancestor of modern Italian. Hence, in the remainder of the chapter, I will refer to this language as ‘old Italian’. The bibliographical details of the textual sources and their abbreviations cited in the examples below are as follows: Boccaccio, ChioseTes = Giovanni Boccaccio, Teseida delle nozze d’Emilia. Chiose, in Limentani (1964: 253–664); Boccaccio, Dec. = Giovanni Boccaccio, Il Decameron, Branca (1992). Boccaccio, Filocolo = Giovanni Boccaccio, Filocolo, in Quaglio (1967: 61–675); Dante (?), Fiore = Dante Alighieri, Detto d’amore, in Contini (1984: 485–512); Dante, Inf = Dante Alighieri, Divina Commedia, Petrocchi (1966); Dante, Vn = Dante Alighieri, Vita nuova, Barbi (1932); Dante, Rime = Dante Alighieri, Rime, Contini (1980); DialSGreg = Domenico Cavalca?, Dialogo de Sam Gregorio composito in vorgà, Porro (1979); LibrAmm = Libro d’amministrazione dell’eredità di Baldovino Iacopi Riccomanni, in Castellani (1982: 429–64); LibrCred = Libricciolo di crediti di Bene Bencivenni, in Castellani (1952: 363–458); LibroDare = Libro del dare e dell’avere di Noffo e Vese figli di Dego Genovesi, in Castellani (1952: 622–42); Novellino = Il Novellino, Favati (1970); PassGen = La Passione, in Parodi (1896); Tesoro volg. = Il tesoro di Brunetto Latini volgarizzato da Bono Giamboni (1878–83).

156

Diego Pescarini

earliest records, is progressively replaced by the order in (1b), attested since the end of the 13th century.3 (1)

a. che [ . . . ] voi la mi concediate (OFlo., Boccaccio, Filocolo 212) that [ . . . ] you.pl it.f= to.me= grant.sbjv ‘that you grant it to me’ b. Se Egli me la concede (OFlo., Boccaccio, Filocolo 72) if he to.me= it.f= grants ‘If He grants it to me’

As a consequence of this change, the order of clitic elements ends up mirroring—in Baker’s (1985) terms—that of their nominal counterparts. Building on Kayne (1994: 19–21; see also Cardinaletti 2008), I will argue that the change leading to the mirror order in (1b) is due to the left-adjunction of the dative clitic to the accusative one, as illustrated in (2b). (2)

a. [ la [ mi ]] b. [ me-la [ tme ]]

According to Kayne’s terminology, the resulting configuration in (2b) is a (true) cluster, while (2a) is a split sequence, where clitics occupy different, though adjacent, positions. On the basis of this hypothesis, we can therefore predict that the clusters with the mirror order (hence, true clusters) differ from the others under a series of syntactic and morphophonological aspects, which I will address in the second part of the chapter (}}8.3–8.6). The structure of the chapter is as follows: }8.2 introduces some data from medieval Florentine and provides a brief comparison with medieval and modern Ligurian varieties (Parry 1992; 1997a; 2005); }8.3 shows that true clusters are inseparable; }8.4 addresses a peculiar pattern of vowel alternation which targets Italian true clusters; after a brief analysis of the internal structure of clitic elements, }8.5 takes into consideration cases of morphological opacity targeting true clusters; in }8.6, I claim that the mirror order is due to a process of root incorporation; }8.7 focuses on modern Italian in order to account for an exceptional patterns of clitic doubling (Benincà 1988: 137).

8.2 The emergence of the mirror order When clitic elements co-occur, they are generally clustered together in a rigid order, which varies on a language-specific basis. In large part this synchronic variation results from a diachronic change, which, in some languages but not in others, made clitic combinations evolve from the archaic order accusative > dative to the mirror 3 Besides the order, (1b) differs from (1a) with respect to the vowel of the dative clitic (me vs mi). This alternation will be addressed later, in }8.4.

The evolution of Italo-Romance clitic clusters

157

one. As previously said, this change is documented in early Italian in combinations of 3rd person accusative and 1st/2nd person dative clitics (I repeat below the relevant examples). The earliest records exhibit the archaic order in (3a). In the first half of the 14th century, both orders were allowed, in apparent free variation.4 Later, however, the archaic order was progressively replaced by the innovative mirror order in (3b), which is the only possible order in present-day Italian. (3)

a. che [ . . . ] voi la mi concediate (OFlo., Boccaccio, Filocolo 212) that [ . . . ] you.pl it.f= to.me= grant.sbjv ‘that you grant it to me’ b. Se Egli me la concede (OFlo., Boccaccio, Filocolo 72) if he to.me= it.f= grants ‘If He grants it to me’

The same evolution can be observed when the dative clitic is the 3rd person reflexive si.5 (4)

a. lo 'mperadore lo si trasse di sotto (OFlo., Novellino 21, p. 180) the emperor it= himself= took.out from under ‘the emperor took it out from below himself ’ b. se lo levò in su il petto (OFlo., Boccaccio, ChioseTes 8.80, p. 523) himself= him= lifted in on the chest ‘he (Heracles) lifted him (Antaeus) up to his chest’

Traces of the same evolution are found also in clusters formed by a locative clitic and a 3rd person accusative one. In modern Italian, the locative clitic must occupy the leftmost position of the cluster, but in old Italian, there are a couple of attestations of the opposite order, with the locative clitic vi: (5)

S' alcun lo vi volesse aprossimare (OFlo., Dante?, Fiore, p. 60) If anybody it= there= would get.close.inf ‘If anybody would get close to there’

4 See Aski and Russi (2010) for a quantitative survey and a tentative pragmatic-based account of this alternation. 5 The ‘archaic’ sequence lo si is still in use in modern Italian, although it has a different interpretation. As shown in (i), when si follows the 3rd person accusative clitic, it is interpreted as an impersonal clitic:

(i) a. Se lo mangia (It., sirefl. > 3acc) himself/herself= it= eats ‘S/He eats it’ b. Lo si mangia (It., 3acc > siimprs) it= one= eats ‘One eats it’ Arguably, the orders se lo and lo si, which in the 13th/14th century were synonymous, started to be given different interpretations when the mirror order became the only possible one for combinations of dative and accusative clitics. Meanwhile, the impersonal si construction emerged (Salvi 2008b; Cennamo 1993a; 1993b; 1997) and, consequently, the archaic order lo si was given the new interpretation.

158

Diego Pescarini

When both clitics are 3rd person, the internal order of the combination cannot be established as the morphology of the cluster is not transparent.6 The accusative clitic shows no gender and number agreement and the cluster ends with an invariable -e, e.g. lile, glile, gliele: (6)

che gli le demo p(er) una inpossta that to.him= them= we.gave for a tax ‘that we gave them to him for a tax’

(OFlo., LibrAmm)

This opacity can be considered as a clue to an ongoing change from the order 3rd person accusative > 3rd person dative (undocumented) to the mirror one, which is the order of modern Italian. I will not speculate further on this, as at present the data do not allow any conclusion. Comparatively, it is worth noting that traces of the non-mirror order are found in Gallo-Romance and in several dialects of northwestern Liguria,7 like (7), while the other northern vernaculars have displayed the modern pattern since their earliest attestations. (7)

a. el u i duna (Olivetta San Michele, Manzini and Savoia 2004) he it/him= to.him= gives ‘He gives it/him to him’ b. el u mə duna (Olivetta San Michele, Manzini and Savoia 2004) he him/it= to.me= gives ‘He gives him/it to me’

A single case of the archaic order is found in an old Genoese text, which may indicate that the archaic order had been more widespread. (8)

E la Magdalena laor lo gue mostrà (OGen., PassGen, p. 36) And the Magdalena then him= to.her= showed ‘And then Magdalena showed him to her’

Let us turn to combinations containing the clitic ne.8 In both old and modern Italian, ne follows 1st/2nd person clitics (and the 3rd person reflexive se), e.g. me ne, te ne, se ne. Conversely, modern and old Italian differ with respect to sequences of ne 6 In 14th-century Florentine the distribution of the grapheme can be symptomatic of the position of the 3rd person dative clitic as is supposed to stand for a palatal lateral deriving from -ll- in front of /i/, cf. begli ‘beautiful (pl)’, capegli ‘hair (pl)’. Crucially, is always in the initial position of the cluster (or, in certain authors, both clitics are expressed by ). We can therefore advance the hypothesis that the order of 14th-century Italian was already 3rd person dative > 3rd person accusative. 7 Parry (2005: 268 n. 38); Borgogno (1972); Manzini and Savoia (2004). 8 In principle, different types of ne could occupy different syntactic positions and, consequently, might give rise to different orders once combined with other clitic material. To the best of my knowledge, however, the position of the clitic ne with respect to other clitics does not depend on the type of ne involved, but rather on the nature of the co-occurring clitic form.

The evolution of Italo-Romance clitic clusters

159

and 3rd person clitics: in modern Italian, ne must follow the dative clitic, while in old Italian ne can either follow or precede the dative clitic. This is illustrated in (10), with a partitive ne, and (11), with an ablative ne. (9)

a. e assai ne gli piacquero (OFlo., Boccaccio, Dec. II, 5, p. 98) and many of.them= to.him= pleased.3pl ‘and he liked many of them’ b. rimasero cimque fior. d’ oro, ed io gli ne rendei remain five florin of gold, and I to.him= of.them= gave.back quatro (OFlo., LibroDare, p. 633) four ‘there remained five florins and I gave him four (florins) back’

(10)

a. molti pericoli ne gli possono seguire (OFlo., Boccaccio, many problems from.there= to.him= can follow.inf Filocolo IV, 49) ‘many problems can result from it’ b. ché gli ne potrebbe troppo di mal seguire because to.him= from.it= could too.much of bad(luck) follow.inf (OFlo., Boccaccio, Dec. III, 3, p. 197) ‘because it could cause him too much misfortune’

A similar pattern of (apparent) free variation is attested in the Dialogo de Sam Gregorio, a 14th-century text written in a vernacular of the Ligurian/Piedmontese border (the alternation between the dative forms li/ge will be addressed in }8.5): (11) a. ne li avea daito a lor (OLig./Pie., DialSGreg 3.37, p. 211) of.it= to.them= has given to them ‘he has given them some oil’ b. una ge ne caìte de man (OLig./Pie., DialSGreg 1.7, p. 92) one to.him= of.them= falls from hand ‘one of them falls from his hand’ In modern Romance, the archaic order is found only in Sardinian dialects: (12) nde li dana (Srd., Manzini and Savoia 2005: ii.317–21) of.them= to.him/her= gives ‘S/He gives some of them to him/her’ The status of combinations of ne and a 3rd person accusative clitic is even more puzzling, as they are no longer allowed in modern Italian (contra Wanner 1977), while they were allowed in old Italian, with both orders:

160

Diego Pescarini

(13) *ne lo prendo, from.there= it= I.take ‘I take it from there’ (14)

il the

libro, book,

dalla from.the

libreria (mod. It.) shelf

a. appresso gir lo ne vedea piangendo (OFlo., Dante Vn 3, 12) further go him= there= I.saw crying ‘I saw him depart in tears’ b. Egli ne lo fa uscire He of.there= it= make exit.inf ‘He makes it get out of there’

fuori (OFlo., Tesoro volg. 160) out

The observed changes are recapitulated in the following table: (15) old Italian modern Italian 13th century

14th century

3p acc > 1/2p dat –

3p acc > 1/2p dat 1/2p dat > 3p acc

3p acc > 3p refl. dat –

3p acc > 3p refl. dat – 3p refl. dat > 3p acc 3p refl. dat > 3p acc

– 1/2p dat > 3p acc

3p acc > locative (vi) locative > 3p acc

– locative > 3p acc

ne > 3p dat 3p dat > ne

– 3p dat > ne

ne > 3p acc 3p acc > ne

– –

(15) illustrates a general trend towards establishing a rigid order in which the clitic elements mirror the order of arguments and adjuncts. To the best of my knowledge, the evolution is one-way: the mirror order replaces the archaic one in many Romance languages, while the opposite change is never attested.9 However, not all combinations were affected by a similar change, and consequently they have kept the ‘medieval’ order. This is particularly true for combinations of case-syncretic clitics (mi ti, gli si, mi ci, etc.), i.e. pronouns which can express either the direct or the indirect object. For reasons of space, I cannot elaborate on the relation between case syncretism and cluster formation. 9 In principle, the evolution might have been more complex and widespread than it is normally considered to have been on the basis of the attested patterns. In particular, the same change could have affected other types of combination or other languages (e.g. Ibero-Romance) in a previous, undocumented stage, as probably happened in northern Italian vernaculars.

The evolution of Italo-Romance clitic clusters

161

Rather, in the following sections, I will entertain the hypothesis that the evolution illustrated in (15) is due to the left-adjunction of one clitic to another, and that the resulting combinations correspond to single complex heads (true ‘clusters’, in Kayne’s 1994 terms). The other combinations, by contrast, are supposed to be split, with each clitic occupying a dedicated projection.

8.3 Separability In principle, we expect that split clitics can be separated: in restructuring contexts (Rizzi 1982), for instance, our prediction is that one clitic can attach proclitically, while the other occupies an enclitic position, e.g. cl–V–cl. By contrast, the same pattern is supposed to be ungrammatical with true clusters. As shown below, this prediction is borne out: the separation of true clusters, as in (16c,d) and (17c,d) gives rise to severe ungrammaticality, while split sequences, as in (18) and (19), can be separated in a colloquial register. (16) a. Te lo può portare (It.) to.you= it= can bring.inf b. Può portar=te=lo (It.) can bring.inf=to.you=it c. *Ti può portar=lo (It.) to.you= can bring.inf=it d. *Lo può portar=ti (It.) it= can bring.inf=to.you ‘S/He can bring it to you’ (17)

a. Te ne può portare (It.) to.you= of.it/them= can bring.inf b. Può can

portar=te=ne (It.) bring.inf=to.you=of.it/them

c. *ti to.you=

può can

portar=ne (It.) bring.inf= of.it/them

d. *ne può portar=ti (It.) of.it/them= can bring.inf=to.you ‘S/He can bring of.it/them to you’ (18)

a. Ti you= b. Può can

ci there=

può can

portare bring.inf

portar=ti=ci bring.inf=you=there

lui, he,

lui, he,

all’ to.the

all’ to.the

aeroporto (It.) airport

aeroporto (It.) airport

162

Diego Pescarini c. ?Ti you=

(19)

può can

portar=ci bring.inf=there

lui, he,

all’ to.the

aeroporto (coll. It.) airport

d. ?Ci può portar=ti lui, all’ there= can bring.inf=you he, to.the ‘He can bring you there (to the airport)’

aeroporto (coll. It.) airport

a. Ti si può portare (It.)10 You= one= can bring.inf b. Si può portar=ti (coll. It.) one= can bring.inf=you ‘One can bring you’

The contrast between (16,17) and (18,19) supports the hypothesis that the combinations that in the 14th century changed their order behave now as true clusters. By contrast, those combinations that have kept the original order—which I have argued correspond to a split configuration—are nowadays separable. Furthermore, this hypothesis entails that, before the change in (15), all combinations were split and hence every sequence could be separated. We therefore expect old Italian to exhibit combinations like (16c,d) and (17c,d). Even if traces of separated sequences are rather scarce, this prediction is confirmed by cases like (20), in which one clitic is proclitic to the finite verb and the other is enclitic to the lexical verb. It is worth recalling that, in modern Italian, these combinations are completely ungrammatical. (20)

a. Ma la cosa incredibile mi fece but the thing incredible me= made ‘But your plight, being incredible, made me Indur=lo ad ovra ch' a me stesso pesa (OFlo., Dante, Inf. 13: 50–51) induce=him to work that to my self weighs goad him to this deed that weighs on me’ b. se 'n tal maniera mi dovete dar=lo if in such way to.me= have.to give.inf=it (OFlo., Amico di Dante, Rime, Son. 44) ‘if you have to give it to me in this way’

The fact that old Italian clusters can be separated is consistent with the hypothesis that clitic combinations were originally split, and only later on began to form a single syntactic unit.

10 For orthogonal reasons, the impersonal si cannot occur enclitically: see Cardinaletti (2008). This is why I have reported only two combinations out of four.

The evolution of Italo-Romance clitic clusters

163

8.4 Allomorphy In Italian, some combinations are characterized by a change of the vowel of the leftmost clitic (either a dative or a locative), which ends with -e instead of the expected -i as shown in (21). Other sequences, like those in (22), do not show any alteration and the linking vowel is the expected -i. (21)

a. [me] lo porti (It., [*mi]) to.me= it= bring.you ‘You bring it to me’ b. [ʎe] ne porti to.him= of.them= bring.you ‘You bring him two of them’

(22)

due (It., [*ʎi]) two

a. [mi] ci porta Mario (It., [*me]) me= there= brings Mario ‘Mario brings me there’ b. [ʎi] si parla dopo (It., [*ʎe]) to.him= one= speaks later ‘We’ll speak to him later’

The situation is summarized in (23): the first column reports oblique forms in isolation (mi, ti, gli, etc.); in the second and third columns, the same forms are clustered with the 3rd person accusative lo (‘him, it’) and the partitive ne (‘of.it/them’); in the fourth and fifth columns, the same clitics appear before the 3rd person reflexive and impersonal clitic si (‘himself/herself/themselves/one’) and the locative clitic ci (‘there’). (23) 1sg 2sg 3sg.dat 1pl/loc etc.

(mi) (ti) (gli) (ci)

with the vowel -e-: me lo me ne te lo te ne glielo gliene ce lo ce ne

with the vowel -i-: mi si mi ci ti si ti ci gli si gli ci ci si *

In this section I argue that the -i/e- alternation is sensitive to the syntactic make-up of the combination. In fact, the sequences where the linking vowel turns into -ecorrespond to the class of true clusters (cf. (15)). Before addressing this point, however, I will revise briefly the existing literature on the topic, starting from D’Ovidio (1886: 71), who argues that -e- is a reflex of the etymological initial vowel of the rightmost clitic: ˇı-llum, ˇı-nde. The etymological vowel has been preserved in cluster-internal position, as shown below:

164 (24)

Diego Pescarini a. me ˇıllum

> m’ıˇllu

> me lo

‘it/him to me’

b. ˇılli ıˇnde

> ill’ıˇnde

> gliene

‘it/him to him/her’

Such a proposal, however, has two major drawbacks. First, as Parodi (1887: 189–90) pointed out, the 13th-century reflexes of ıˇlle, ıˇnde occupy the leftmost position of the sequence, as shown in }8.2. The fact that the archaic order of these clusters is accusative > dative contradicts D’Ovidio’s hypothesis that the linking vowel -e- is a reflex of ıˇ in cluster-internal position. Rather, Melander (1929) pointed out that -e should be considered a side-effect of the change that led to the mirror order: in fact, as soon as the dative clitics began to occupy the leftmost position of the cluster, their vowel suddenly changed into -e with very few exceptions.11 Second, if -e- were the reflex of ˇı, this would entail that the cluster originally included a disyllabic reflex of ıˇlle, ıˇnde. If so, the resulting cluster would show a geminate sonorant (cf. ıˇlle, ıˇnde > *ello, *enne) since in Florentine etymological geminates are normally maintained. The expected evolution would therefore be as follows: (25) a. me ıˇllum b. ıˇlli ıˇllum

> m’ıˇllu

> *mello

‘it/him to me’

> ıˇll’ıˇllum > *gliello

‘it/him to him/her’

It is worth noting that in Italian this gemination is shown in sequences of preposition + article (the so-called preposizioni articolate, lit. ‘articulated prepositions’), but not in pronominal sequences: (26)

a. de ˇıllum

> d’ıˇllu

> dello

‘of the’

b. in ıˇllum

> (i)n’ıˇllum > nello

‘in the’

In (26), the preservation of the disyllabic form of the determiner (ıˇllu > ello) provides a straightforward explanation for both the vowel -e- and the following gemination.12 On the contrary, the case of pronominal sequences calls for a different 11 Interestingly, these exceptions normally regard sequences containing a 3rd person clitic (li/gli), which tend to maintain the vowel -i- (cf. (29)). In these cases, we can suppose that the -i- of li/gli is a reflex of the Latin dative ending. 12 The conclusion that the -e- of pronominal clusters and the -e- of P+D sequences have a different nature is highly desirable because it prevents a possible paradox. In fact, Cardinaletti’s (2008) claim that true clusters ‘display the same vowel that is found in the combinations of preposition and determiner’ ends up contradicting Cardinaletti’s own analysis that sequences of pronominal clitics are true clusters in the sense of Kayne (1994). In fact, P+D sequences cannot be clusters à la Kayne (1994), as the linear order P > D cannot be due to movement of P
past D
. Hence, if we want to maintain the hypothesis that -e- is a clue of true clusters, we have to demonstrate that their -e-’s are different. Gemination provides proof that P+D sequences differ from pronominal clusters like me lo, gliela: the former, but not the latter, derive from the univerbation of the preposition and the following disyllabic determiner: e.g. de ıˇllum > d’ıˇllu > dello ‘of the’. The absence of gemination in pronominal clusters, by contrast, indicates that the rightmost element was not disyllabic and, consequently, that the linking vowel -e- cannot be a reflex of ıˇ- preserved in cluster-internal position.

The evolution of Italo-Romance clitic clusters

165

explanation, as the absence of gemination is not compatible with the same derivation.13 An alternative hypothesis is that the -e/i- alternation originates from a raising process, which in old Florentine turned final unstressed -e into -i (Rohlfs 1966: 178). (27) /e/

[i] /

# [-stress]

This rule targeted adverbs and semi-functional words (e.g. avante > avanti ‘before, in front of ’, diece > dieci ‘ten’, longe > lungi ‘far’) and personal pronouns, giving rise to the alternation between strong pronouns, which maintained the etymological vowel (e.g. Lat. me > me ‘me’), and clitic forms, which underwent raising (e.g. me > mi). Arguably, such evolution followed from the change that turned strong pronouns (i.e. XPs corresponding to fully-fledged PrWs) into clitic elements (i.e. X
s corresponding to bare syllables); see Cardinaletti and Starke (1999) and Egerland (2005) on the syntactic evolution, and Selkirk (1995) on phonology. This change, coupled with the rule in (27), led to the alternation between strong e-forms (me, te, se, etc.) and clitic i-forms (mi, ti, ci, etc.). (28) a. XP (1sg strong) | [me]PrW

b. X
(1sg clitic) | [mi]ó

Let us turn to the -e/i- alternation in clusters. At first sight, one might argue that, in cluster-initial position, the underlying vowel /e/ surfaces as a consequence of secondary stress, which blocks the rule in (27). This hypothesis, however, is contradicted by two pieces of evidence. First, take into consideration the 3rd person dative clitic gli (‘to him’) < Lat. ıˇlli. Since the original ending of the dative clitic is -i (and not -e), it is expected to show no -e/i- alternation as its underlying form /ʎi/ cannot undergo further raising. Old Italian was consistent with this prediction, as 3rd person dative clitics exhibited the etymological vowel -i even in cluster-initial position. (29)

a. che gli le demo p(er) una inpossta (OFlo., LibrAmm) that to.him= them= we.gave for a tax ‘that we gave them to him for a tax’ gli ne potrebbe troppo di mal seguire to.him= from.it= could too.much of bad(luck) follow.inf (OFlo., Boccaccio, Dec. III, 3, p. 197) ‘because it could cause him too much misfortune’

b. ché because

13 Formentin (1996) has pointed out that, at least in some cases, e.g. nello ‘in the’, the gemination must depend on the preservation of the disyllabic form of the determiner (in ıˇllu > (i)nello). In fact, ll in P+D sequences is attested also in the dialects in which articles are not subject to regular consonantal doubling (so-called raddoppiamento (fono)sintattico).

166

Diego Pescarini

Later on, however, the linking vowel of these clusters became -e-, which is the only possible form in modern Italian (I repeat below the relevant example). The phonological explanation cannot account for the evolution from (29) to (30), as in (30) the linking vowel -e- cannot be considered the underlying vowel, surfacing as a consequence of foot formation. (30) [ʎe] ne porti due [*ʎi] to.him= of.them= bring.you two ‘You bring him two of them’ Second, a pure phonological analysis cannot explain why the -e/i- alternation is not allowed in several clusters, where both clitics display the vowel -i (I repeat below the relevant examples): (31)

a. [mi] ci porta Mario [*me] me= there= brings Mario ‘Mario brings me there’ b. [ʎi] si parla dopo [*ʎe] to.him= one= speaks later ‘We’ll speak to him later’

In order to account for (31) under the hypothesis that -e- is a consequence of secondary stress, we should postulate that the combinations in (31) are sequences of extrametrical syllables rather than feet. However, such classification is not supported by any independent piece of phonological evidence. The alternative explanation is that the -e/i- alternation, although originating from a phonological process like (27), is syntax-driven—i.e. in synchrony, it is sensitive to the syntactic make-up of the cluster, rather than its prosodic structure. Let us assume that the original phonological rule—rewritten in (32a)—has been reanalysed as a morphological alternation, sensitive to the presence/absence of a morphosyntactic boundary, as represented in (32b). (32)

a. /e/

[i] / ____#

b. -i / ____# -e

This would explain why the -e/i- alternation correlates with the distinction between cluster and split combinations proposed in the previous sections. As illustrated in (33), split sequences are characterized by an internal morphosyntactic boundary (#), which triggers the insertion of the default final unstressed -i. By contrast, true clusters count as a single morphosyntactic unit and, consequently, the linking vowel is -e- because it is no longer analysed as a final vowel.

The evolution of Italo-Romance clitic clusters (33)

167

a. combinations with the vowel -i-: [cl1 [cl2 . . . ! mi#si, mi#ci, gli#si, etc. b. sequences with the vowel -e-: [cl1 + cl2 [ . . . ! me lo, me ne, gliene, etc.

I have argued that the split/cluster divide helps to account for a morphophonological puzzle which cannot be explained otherwise. I have claimed that the above alternation can be explained once previous phonological reconstructions—like those of Rohlfs—are revised in the light of a principled model of the syntactic make-up of clitic combinations.

8.5 Suppletion In various Italo-Romance dialects, the etymological form of the 3rd person dative clitic (e.g. li/le) has completely fallen out of use, and in modern dialects this pronoun is expressed by a suppletive exponent which normally coincides with the locative or the 3rd person reflexive clitic. In many dialects, such a suppletion is absolute (see Calabrese 1994; 2008; Loporcaro 1995; 2002), while in others it is context-determined: the 3rd person dative clitic is replaced by a suppletive exponent when it is clustered with another clitic element (see e.g. Pescarini 2010). As Cardinaletti (2008) pointed out, this normally happens in true clusters. This section aims to investigate further the relation between suppletion and cluster formation Before addressing the relevant data, some general remarks on the morphology of the 3rd person dative clitic are in order. In general, Romance clitics are formed by a person morpheme followed by a vowel (a ‘thematic vowel’, according to Harris 1994). However, it is worth distinguishing between two types of vowel: (i) agreement markers carrying morphosyntactic information about gender and number (as in the case of 3rd person non-reflexive pronouns, e.g. lo, la ‘it/him, her’); (ii) oblique endings without morphosyntactic value, as in the case of 1st/2nd person pronouns, e.g. me, te ‘me, you’. Following Kayne’s (2000) proposal, 3rd person clitics, which end with an agreement marker, are bimorphemic, while oblique clitics are monomorphemic:14

14 Kayne (2000) and Cardinaletti and Repetti (2008) argue for a more radical analysis by assuming that oblique endings are epenthetic, i.e. segments which are not part of the morpholexical representation of clitic elements. However, the epenthetic status of these final vowels remains rather obscure to me, in particular in the case of the Italo-Romance varieties which have never undergone a generalized and systematic loss of final unstressed vowels. In fact, in these varieties the default vowel normally coincides with the expected evolution of Lat. -e in final, unstressed position. Monomorphemic clitics can be therefore viewed as regular reflexes of the Latin forms me, te, se, ince, inde without postulating the intervention of epenthesis processes.

168

Diego Pescarini

(34)

a. Monomorphemic D°

b. Bimorphemic D°

√ mi ‘me’

√ l-

AGR–1 -o ‘it/him’

Things are a bit more complicated with the 3rd person dative clitic. In several Romance varieties, it derives from the Latin determiner illi (pl. illis) and, although it does not exhibit gender morphology, it can be analysed as a bimorphemic element (à la Kayne) formed by a root l- followed by an agreement marker. Frequently, however, the etymological form li/le has been replaced by locative or reflexive exponents (Calabrese 1994; Loporcaro 1995; 2002). Moreover, even those languages displaying a bimorphemic dative are subject to context-determined suppletion replacing the etymological form li/le with a monomorphemic item. We can distinguish at least three main patterns of substitution on the basis of the etymology of the replacing item: (i) Spurious se patterns, attested in Ibero-Romance and Campidanese Sardinian: in true clusters, the etymological 3rd person dative le/li < illi) is replaced by the 3rd person reflexive element (with a non-reflexive interpretation). (35)

a. ɖi pottu unu to.him= I.bring a ‘I bring him a book’

libru (Sarroch (CA), Sardinia) book

b. si/*ɖi ɖu pottu (Sarroch (CA), Sardinia) to.him= it= I.bring ‘I bring it to him’ (ii) Spurious locative patterns, attested in many Italo-Romance dialects, Logudorese Sardinian, and Catalan: in true clusters, the etymological 3rd person dative le/li is replaced by the locative clitic ci/bi/hi/y < ince, ibi. (36) bi/*li l’ appo to.him/her /them= it= have.I ‘I gave it to him/her/them’

datu given

(Log., Jones 1993: 220)

(iii) Spurious ne patterns (several southern Italian dialects): in true clusters, the etymological 3rd person dative le/li is replaced by the partitive element deriving from Lat. inde. (37) a. i da kkuist@ to.him/her/them= gives this (Rocca Imperiale (CS), Calabria, Manzini, and Savoia 2005: ii.291) ‘S/He gives this to him/her/them’

The evolution of Italo-Romance clitic clusters

169

b. n/*i u da to.him/her/them= it= gives (Rocca Imperiale (CS), Calabria, Manzini, and Savoia 2005: ii.291) ‘S/He gives it to him/her/them’ In the remainder of this section, I will argue that there is a tight link between the patterns in (i–iii), the internal structure of clitics, and the order of the cluster. In particular, it is worth noting that: (38)

suppletion consists in replacing a bimorphemic clitic (e.g. √l+i) with a monomorphemic one;

(39)

this happens in true clusters (i.e. in clusters with the mirror order).

The latter generalization is supported by patterns like (40): in these Sardinian dialects (Ittiri (SS), Padria (SS), Luras (OT), Siniscola (NU), Galtellì (NU), Bosa (OR); Manzini and Savoia 2005: ii.317–21), the etymological form li occurs in isolation (40a) or when it follows another clitic (40b). However, when it occupies the leftmost position of the cluster, as in (40c), it must be replaced by the ‘spurious’ exponent bi. (40)

a. li dana kustu (Srd.) to.him/her= gives this ‘S/He gives this to him/her’ b. nde li dana (Srd.) of.them= to.him/her= gives ‘S/He gives some of them to him/her’ c. bi/*li lu dana (Srd.) to.him/her= it= gives ‘S/He gives it to him/her’

The correlation between suppletion and the mirror order is found in medieval Italo-Romance as well, in the rare vernaculars in which etymological reflexes of Lat. illi are still attested.15 For instance, the Dialogo de Sam Gregorio, a text written in a vernacular of the Ligurian/Piedmontese border, displays three allomorphs of the 3rd person dative clitic: li, gl’, and ge. (41)

a. Li dise (OLig./Piem, DialSGreg 1. 4, p. 86) to.him= says ‘He says to him:’

In general, in northern Italo-Romance the 3rd person dative clitic is usually expressed by the same exponent as expresses the locative clitic since the earliest attestations and, in many cases, such syncretic items cannot be derived easily from Lat. illi (for a proposal, see Benincà 2007). 15

170

Diego Pescarini b. segundo according

che that

ello he

gl' to.him=

aveo inpromoso I.have promised (OLig./Piem, DialSGreg 2.25, p. 138) ‘according to what I have promised to him’

c. elo ge fu mostrao he to.him= was shown ‘he was shown to him’

(OLig./Piem, DialSGreg 1.4, p. 86)

Li is the regular reflex of illi, gl its prevocalic allomorph ( is always used to express the outcome of the palatalization of l in front of the glide j, e.g. filiu > figlo ‘son’), while the relation between li and ge is, at first sight, phonologically opaque (but see Benincà 2007). With respect to the l-/g- alternation, the important distinction between the two is that li can be analysed as a bimorphemic element l+i (where i < dat -i), while ge exhibits the same vowel of 1st/2nd person pronouns me, te, etc. and, in my opinion, counts as a monomorphemic element. As in modern Sardinian, the bimorphemic item li is used in isolation or in second position, as in (42a), while it is systematically replaced by the monomorphemic ge when the cluster exhibits the mirror order, like in (42b). (42)

a. ne li avea daito a lor (OLig./Piem, DialSGreg 3.37, p. 211) of.it = to.them= has given to them ‘he has given them some oil’ b. una ge ne caìte de man (OLig./Piem, DialSGreg 1. 7, p. 92) one to.him= of.them= fell from hand ‘one of them fell from his hands’

A similar pattern is found also in a Genoese text, the Passione edited by Parodi (1896), where three allomorphs of the 3rd person dative clitic occur in free variation: li, ge, and gue. (43)

a. li demandà (OGen., PassGen p.29) to.him= asked ‘he asked to him’ b. ge respoxe (OGen., PassGen p. 31) to.him= answered ‘he answered to him’ c. gue eram date (OGen., PassGen p. 38) to.him= were given ‘(they) were given to him’

As in the case with the aforementioned Dialogo, it is almost impossible to establish the phonological value of the element . Nowadays, the locative/dative clitic is ghe

The evolution of Italo-Romance clitic clusters

171

/ge/, but the pronunciation of its medieval counterpart is still a puzzle: in medieval texts, the grapheme was not used to indicate a velar pronunciation [g] in front of palatal vowels, and furthermore, was also used to indicate the palatalized reflex of lj, e.g. filium > figiu ‘son’, whose real phonetic value is still controversial ([j], [ʤ]?). Moreover, we still lack a plausible etymological solution for . These difficulties aside, it is worth noting that the etymological 3rd person dative clitic li never combines with other clitic pronouns. In clusters we found only ge/gue, which precedes the co-occurring clitic in accordance with the mirror order.16 Under the analysis suggested here, the above distribution is not at all surprising. As in Sardinia, the monomorphemic clitic ge/gue replaces the bimorphemic li whenever the dative clitic occupies the leftmost position of the cluster. Elsewhere, li can occur freely.

8.6 Root incorporation This section aims to account for the generalizations in (38) and (39) by establishing a link between suppletion and cluster formation. First, I assume, as in (34), that 3rd person clitics are bimorphemic, i.e. they are formed by a root expressing person features (say, √{P}) followed by an agreement marker expressing number and gender: (44)

[D
√{P} [Agr {G,N} ]]

On the basis of these features, vocabulary items are late-inserted at PF (Halle and Marantz 1993): (45) [D°√{P}[Agr {G,N} ]]

/l/

/e/

= It. le ‘to her’

My hypothesis goes that, to form a true cluster, only the root of the dative clitic undergoes incorporation, while its inflectional material is stranded. In this case, gender and number remain unpronounced (‘silent’, in Kayne’s 2000 terms), although in some languages we can observe it overtly (see below): (46) [

16

D° . . .

[D° √{P}

[Agr {G,N}

Save for a single case, in which the dative clitic gue follows the 3rd person accusative lo (cf. 8).

172

Diego Pescarini

As a consequence of this process, the moved (sub-)constituent √{P} can no longer trigger the insertion of the item l-, which must be followed by a proper inflectional ending. (47) [ √{P}  D°

∗l

[D° t√{P} [Agr {G,N}

...

lo

Since the agreement material of the dative clitic is stranded in a discontinuous position, a monomorphemic element must fill the head hosting √{P}. 1/2 clitics cannot be inserted instead of l, as they cannot match the {P} specification of the root. As a last resort, a dummy clitic—subject to cross-linguistic variation (see Pescarini 2010)—is inserted, as schematized below: (48)

[ √{P}  D°

...

[D° t√{P} [Agr {G,N}

gli

lo

modern Italian

bi

lu

Logudorese

si

lu

Campidanese

n

u

Rocca Imperiale, etc.

The above analysis is supported by the phenomenon of parasitic plural (Halle and Harris 2005; Kayne 2010b; Manzini and Savoia 2009), which is attested in those languages in which Number is expressed by the plural suffix -s.17 In these languages, 3rd person plural clitics exhibit a trimorphemic exponent, as schematized below: (49)

[D° √{P} [Agr {G} [Agr {N}

/l/

/o/

/s/ = Sp. los ‘them(m)’

The same analysis holds for the 3rd person dative clitic les/lis ‘to them’. When the plural dative occurs in true clusters (for instance, before the 3rd person accusative clitic lo ‘it/him’), it is replaced as usual by a dummy exponent (e.g. bi), but, crucially, its plural feature is expressed by the morpheme -s, which attaches to the right of the whole cluster, as shown in (50).

17

The phenomenon is mainly attested in South American and Sardinian varieties. Parasitic plural is pervasive in the Catalan dialect spoken in Barcelona (Bonet 1991). Traces of parasitic plural are found also in old French (Giampaolo Salvi’s p.c. reported in Benincà and Poletto 2005: n. 14).

The evolution of Italo-Romance clitic clusters (50)

173

Nara=bi=lo-s (Log., Jones 1993) tell=there=it-pl ‘Tell it to them’

Following the above analysis, the position of the plural suffix -s can be accounted for as an instance of stranding of the agreement features of the dative pronoun, whose root has incorporated into the accusative clitic lo: (51) [ √{P}  D° . . . bi

[D° t√{P} [Agr {G} [Agr {N}

lo

-s

Jones (1993), focusing on Logodurese Sardinian, reports also cases of parasitic gender, i.e. cases in which the rightmost vowel expresses the gender of the dative clitic rather than that of the accusative one: (52) Nara=bi=l-a-s (Log., Jones 1993) tell=there=3p-f-pl ‘Tell it to them.f ’ Again, the hypothesis of root incorporation provides a straightforward account of the pattern in (52), which is illustrated below: (53)

[ √{P}  D° . . . [D° t√{P} [Agr {G} [Agr {N}

bi

l

-a-

-s

8.7 Italian Italian exhibits a non-etymological 3rd person dative feminine form (le ‘to her’) in opposition to the masculine li/gli ‘to him’.18 Possibly, the emergence of a 3rd person feminine dative clitic is due to the analogy with 3rd person accusative plural clitics, as schematized below (the double asterisk marks the analogical output): (54)

liacc(< illi) : leacc(< illae) = lidat(< illi) : **ledat

The question concerning It. gli/le is whether they can be analysed as bimorphemic or monomorphemic elements, i.e. whether their endings count as agreement markers. Several pieces of evidence seem to confirm that gli counts as a monomorphemic

18 As far as I can see, the emergence of a dedicated dative feminine exponent is a peculiarity of Florentine, as in the rest of Romance, the 3rd person dative clitic is not inflected for gender.

174

Diego Pescarini

clitic, while the latter (le) is still synchronically analysed as a bimorphemic one (see also Cardinaletti 2008).19 First, the ending of gli coincides with the default vowel of monomorphemic clitics like mi, ti, ci, etc. Second, gli undergoes the same pattern of allomorphy of monomorphemic clitics (see }8.4): its vowel turns into -e- before a 3rd person accusative clitic or ne. Third, like monomorphemic clitics, gli is free to combine with other clitics, while le cannot occur in true clusters. In this case, le is always replaced by gli, although the 3rd person dative pronoun references a feminine individual: (55)

a. *Le / Glie =lo regalo to.her / to.him =it give.I ‘I have given it to her (Maria)’

a to

Maria (It.) Maria

b. *Le / Glie =ne regalo due to.her / to.him =of.them I.give two ‘I have given two of them to her (Maria)’

a to

Maria (It.) Maria

The pattern of substitution in (55) is therefore akin to the canonical cases of suppletion illustrated in }8.5, once it is assumed that gli is monomorphemic and le is bimorphemic. Further evidence in favour of this hypothesis comes from the analysis of certain peculiar cases of clitic doubling. Italian—unlike Spanish and other Romance languages—does not allow dative DPs/pronouns to be doubled by a coreferent clitic (needless to say, I am not dealing here with resumptive clitic and dislocated DPs): (56)

a. *Gli ho regalato il libro to.him= I.have given the book ‘I have given the book to him (Mario)’ b. *Gli ho regalato due libri to.him= I.have given two books ‘I have given two books to Mario’

a to

Mario (It.) Mario

a to

Mario (It.) Mario

However, Benincà (1988: 137) noticed that if the 3rd person dative clitic is part of a true cluster, doubling is allowed: (57) a. Glie =l’ ho regalato a to.him =it I.have given to ‘I have given it to him (Mario)’

Mario (It.) Mario

b. Glie =ne ho regalati due to.him =of.them/it I.have given two ‘I have given two of them to him (Mario)’

a to

Mario (It.) Mario

19 One can speculate whether the reanalysis of gli as a monomorphemic exponent might be triggered by palatalization.

The evolution of Italo-Romance clitic clusters

175

On the contrary, doubling is ungrammatical with split sequences: (58)

a. (*Gli) si regala il libro (to.him=) one= gives the book ‘One gives the book to him (Mario)’

a to

Mario (It.) Mario

b. (*Gli) si presenta Mario (It.) (*to.him=) himself/herself= introduces Mario ‘One will introduce Mario to himself/herself ’ c. (*Gli) ci attacco un cartello (*for.him=) there= I.hang a sign ‘I’ll hang a sign there for him (Mario)’

per for

Mario (It.) Mario

Intuitively, these data mean that the dative clitic embedded in a true cluster is not a genuine doubler (see also Benincà and Poletto 2005: 232). In the light of a root incorporation analysis, we can argue that, when it is inside a true cluster, the dative clitic is a root without agreement, which cannot compete with a co-occurring DP goal for checking against the same probe. On the contrary, when the same clitic gli occours in isolation (namely as a single clitic) or when it is in a split configuration with another clitic, as in (58), it counts as a fully-fledged pronoun with a complete set of phi specifications, and as such it competes with a dative lexical item in the same clause.

8.8 Conclusions This chapter has addressed the evolution of the internal order of Romance clitic combinations by hypothesizing the existence of two classes of clitic combinations in modern Italo-Romance. This led me to account for several collateral issues, including the separability of clitic cluster in restructuring contexts; a pattern of allomorphy found in Italian; a generalized pattern of suppletion attested in many Romance varieties when a 3rd person dative is part of a true cluster; parasitic plural; and cases of exceptional doubling. In the first part of the chapter I argued that many Romance languages have undergone a change reversing the order of certain clitic combinations (those containing a 3rd person accusative element or ne). Clear examples of this change are shown in 13/14th-century Florentine with sequences of a 1st/2nd person dative and a 3rd person accusative clitic: (59)

a. che [ . . . ] voi la mi concediate (OFlo., Boccaccio, Filocolo 212) that [ . . . ] you.pl it.f = to.me= grant.sbjv ‘that you grant it to me’ b. Se Egli me la if he to.me= it.f= ‘If He grants it to me’

concede grants

(OFlo., Boccaccio, Filocolo 72)

176

Diego Pescarini

According to my analysis, the innovative mirror order in (59b) is due to the leftadjunction of the dative clitic on to the accusative one. The resulting configuration counts as a single complex head (a true cluster, cf. Kayne 1994: 19–21), while the other combinations correspond to a split configuration where the clitics occupy different syntactic positions: (60)

a. [ cl1 [ cl2 ]]

(split)

b. [ cl2 + cl1 [tcl2 ]]

(cluster)

On the basis of this hypothesis, Italian clitic combinations can be therefore divided into two major classes:20 true clusters are those that have undergone a change like (59), while the others are considered to be split. (61) 1sg 2sg 3sg.dat 1pl/loc etc.

(mi) (ti) (gli) (ci)

with the vowel -e-: me lo me ne te ne te lo glielo gliene ce lo ce ne

with the vowel -i-: mi si mi ci ti si ti ci gli si gli ci ci si *

In the second part of the chapter I elaborated on the above hypothesis, reaching the following conclusions: i. I have shown that true clusters, unlike split sequences, cannot be separated in restructuring environment. ii. I have argued that the linking vowel -e- is an indicator of incorporation. iii. I have observed that true clusters are almost always subject to suppletion because bimorphemic datives are replaced by a dummy monomorphemic item. The hypothesis goes that the root of the dative clitic incorporates (while its agreement features remain stranded in their original position), preventing the insertion of bimorphemic exponents. iv. Under the previous hypothesis I have reconsidered patterns of ‘parasitic’ inflection as cases of stranding. As the formation of true clusters results from root incorporation, the remaining inflectional material can be expressed by means of a parasitic suffix, which ends up attaching to the right of the whole cluster. v. Lastly, the hypothesis of root incorporation can account for the acceptability of doubling in Italian when the 3rd person dative clitic is part of a true cluster. I have argued that, once incorporated, the dative clitic is no longer a real doubler.

20 Cardinaletti (2008) argues for a slightly different classification as she considers the combinations with the clitic ci as true clusters

9 Subject clitics and macroparameters* IAN ROBERTS

9.1 Introduction This chapter follows in the tradition of Rizzi (1986; 1987) in that it looks at what the variation in subject-clitic systems in the northern Italian dialects (NIDs henceforth) can tell us about the nature of the parameters of Universal Grammar (UG), taking as a starting point the conception of the organization of UG proposed in Chomsky (1981a). It is well known that the Italo-Romance varieties provide much apparent evidence for ‘microparametric’ variation, i.e. low-level, rather ‘surfacey’ variation of an intricate nature holding among closely related varieties (see Kayne 2005). The notion of microparameter, although intuitively rather clear, is not often given a precise characterization; here I will provide a characterization of macro- vs microparametric variation based on Holmberg and Roberts (2010) and Roberts (2012), relate this to the null-subject parameter, and then show how the variation in NIDs in relation to subject clitics, seen as an extreme of microvariation in Manzini and Savoia (2005), can actually be seen as the ‘realization’ form of microparameter. I will suggest that microparameters generally have to do with modes of PF-realization. In (1) I give some illustrative examples of the kind of variation one finds in the subject-clitic systems of the NIDs (see also Parry 1998b; 2000 on Piedmontese): (1)

i. a dOrm@, t@ dOrm@, i/al dOrm@, a durmiŋ, durmit@, i/al dOrm@n@ (Carrara (MS), northern Tuscany) ii. dOrmi, ta dOrmat, al/la dOrma, dOrmum, dOrmuf, dOrmaŋ (Como) iii. (a) dormi, a t dormi, a l/la dorma, a dormome, a dormiv, i dormen (Gainago (PR), Emilia-Romagna, Maria Pedretti, p.c.).

* The paper on which this chapter is based was given at the 3rd Cambridge Italian Dialect Syntax Meeting, held in Cambridge in June 2009. My thanks to Adam Ledgeway and Roberta D’Alessandro for the invitation to speak there, and to the audience for questions and comments. This work is now funded by the European Research Council Advanced Grant No. 269752, ‘Rethinking Comparative Syntax’.

178

Ian Roberts

The paradigms in (1.i) and (1.ii) are from Manzini and Savoia (2005). The three paradigms all represent the present-tense forms of ‘to sleep’ as they appear in declarative contexts (i.e. where there is no possibility of subject-clitic inversion). The variation in the patterns of syncretism is immediately apparent, and the question of what the subject clitics really are (pronouns, agreement markers, or something else) immediately arises. These questions have been discussed in the literature on NIDs at least since the pioneering work of Renzi and Vanelli (1983). As mentioned above, these questions have been related to the question of the status of the nullsubject parameter in NIDs since Rizzi (1986). Manzini and Savoia (2005: i.55) give the null-subject parameter as follows: (2)

D lexicalized by: i. Italian ii. NIDs iii. English

V D D'

‘D’ here refers to the nominal features, ç-features in the terminology of Chomsky (1995; 2000; 2007; 2008), associated with the clausal functional head T. The claim is then that the verbal inflection, combined with V-to-T movement, lexicalizes these features in standard Italian (see also Alexiadou and Anagnostopoulou 1998; Borer 1986; Barbosa 1995; 2009); the subject clitic, seen as a D-element since it is a kind of pronoun, plays this role in NIDs, while in non-null-subject languages such as English a DP plays this role by raising to the specifier of the head bearing the D-features requiring lexicalization (for independent reasons connected with locality of movement, this will always be the surface subject). Manzini and Savoia (2005: i.120) distance themselves from earlier formulations of the null-subject parameter by pointing out that, in their approach, ‘the parameter in [(2)] cannot be defined for the entire language, but must be applied to the individual forms of the paradigm’.1 The motivation for this is the variation in forms illustrated in (1): the 2pl form in Carrara and all the plural forms of the Como variety, lacking a subject clitic but showing differentiated inflectional endings, function like standard Italian. On the other hand, the singular forms of the Como variety show differentiated clitics but syncretic inflection: here the D-feature of the clitic lexicalizes D. Other cases are more delicate: the a clitic of the Gainago variety clearly does not differentiate the four or five persons it appears with; instead here the inflections do this. Hence the a clitic apparently falls outside the system of lexicalization of D (Poletto 2000 argues that these ‘vocalic’ clitics are assertion markers in the C-system). Manzini and Savoia (2005: i.357) further state that ‘the microparametric approach predicts that each parameter has strictly local consequences; these strictly local

1

All quoted matter in languages other than English has been translated by the author.

Subject clitics and macroparameters

179

parameters vary independently of one another and combine freely giving rise to a fine-grained surface parametrization.’ In other words, the microparametric approach they adopt provides a descriptively adequate tool for accounting for the observed variation involving subject clitics in NIDs, and of course can be extended to other properties (negation systems, past-participle agreement, and auxiliary selection in central-southern varieties, clitic-climbing, clausal complementation, etc.; see Manzini and Savoia 2005 for ample illustration of these and many other variant properties of Italo-Romance generally). However, one can raise various empirical and theoretical objections to this radically microparametric approach. The most important is that it takes us away from explanatory adequacy (in the classical sense of Chomsky 1964). Given its surface orientation and the fact that the microparameters are explicitly stated to be local and independent of one another, this conception of microparameter is no help with poverty-of-thestimulus problems: each value of each parameter must be independently learnt directly from the primary linguistic data (PLD). Setting the value of one parameter gives the learner no clue as to the value of the next one. So these parameters tell us nothing about UG or language acquisition beyond the simple observation that the realization of person–number marking varies unpredictably from one cell in the paradigm to the next, and that each must be independently learnt. As such, the notion of parameter does not take us beyond traditional or structuralist conceptions of how these phenomena are internalized; indeed, these differences are not really ‘parametric’ in any substantive sense, but rather represent unpredictable taxonomic variants. In fact, this approach worsens the learnability problem, since it multiplies the number of parametric systems dramatically. Instead of a single null-subject parameter, there are as many null-subject parameters as there are cells in the paradigms, i.e. six per tense. Moreover, if these parameters are rooted in UG, then we should not limit the number of potential person–number forms to six: we must take into account the possibility of dual and perhaps trial number, and 1pl inclusive/exclusive oppositions, etc. Furthermore, the number of possible tense paradigms must be maximal, with all departures from that the result of syncretism. It is easy to see, then, that there are hundreds of ‘null-subject’ parameters in NIDs, many of them showing large numbers of syncretisms (i.e. all dual forms are syncretic with plurals; all inclusive and exclusive forms are syncretic, etc.). But of course, these parameters are universal: they therefore hold in Mandarin Chinese and English (where, quite by coincidence, they all have the same value, i.e. (2.iii)) and in standard Italian (where again they have the same value throughout all paradigms (2.i), with the possible exception of 2sg of the present subjunctive). It seems clear that this approach is missing generalizations, and indeed (as a reviewer points out) cannot even look for generalizations as formulated, in addition to causing the number of parameters of UG to increase significantly, probably well into the thousands (and thereby making the number of possible grammatical systems hyperastronomical).

180

Ian Roberts

A further point is that, while, as (1) illustrates, many dialects oscillate between values (2.i) and (2.ii), it is less clear that there is similar oscillation regarding (2.iii). Why do we not find paradigms where some persons in some tenses require a strong pronoun, while others trigger V-to-T movement, or require a clitic?2 Why do we not find all three possibilities in a single paradigm? Again, it seems that the microparametric approach is missing higher-level generalizations. Third, we can ask why this variation only seems to affect subjects. Why do we not find the comparable variation with (indirect) object clitics and overt (indirect) objects? Although there is well-known variation regarding clitic doubling in these cases, I am not aware of any variety which lexicalizes object D/ç-features with a clitic in one person/number form, but does not allow this in another form.3 Instead, both direct- and indirect-object clitic paradigms seem quite uniform; again, this is purely accidental on the microparametric approach put forward by Manzini and Savoia. Despite these critical comments, we should recognise that Manzini and Savoia’s approach is successful in providing an accurate description of the facts concerning subject clitics in NIDs, and one which is grounded in a general conception of parameters as options for the lexicalization of features of functional heads which are universal, such as T’s D-feature in (2). But the approach is clearly problematic. In this chapter, I will try to reconcile the descriptive success of the microparametric approach of the Manzini and Savoia kind with the explanatory goals which motivated the postulation of parameters of UG in the first place (see again Chomsky 1981a). To do this, I proceed in three stages. First, I will present the view of macro- vs microparametric variation put forward in Holmberg and Roberts (2010) and Roberts (2012). Second, I present the view of the null-subject parameter put forward by those authors. Third, I will speculate, based on an idea in D’Alessandro and Roberts (2008), on why subject clitics in NIDs are special while their object clitics are not. 2 A reviewer points out that something like this may be partly the case with the 2sg of the standard Italian subjunctive which, in contrast to other persons and paradigms, obligatorily requires the presence of the (weak/clitic) pronoun tu (e.g. vogliamo che *(tu) ci scriva al più presto ‘we want you to write to us as soon as possible’). There are similar facts in some southeastern Occitan dialects which, in a parallel fashion, require expression of the subject in the 2sg of subjunctive clauses. These are fairly marginal cases, though, and do not involve variation in verb movement. 3 Paoli (2009) discusses the dialects of the Comelico area in the Dolomites of northeastern Italy, in two varieties of which 1pl accusative clitics are systematically absent and 3sg and 3pl inanimates are generally silent. The same is seen in three varieties with indirect-object clitics. Paoli also points out that the RætoRomansch of Ardez lacks dative clitics entirely. Finally, Brazilian Portuguese clitics generally are being lost, especially in the 3rd person. Paoli in fact suggests that the 1pl clitic never developed in these varieties; and it does seem that this person–number combination is unusual, in that in many varieties (including standard Italian) it has been replaced by a form derived from a locative. The case of Brazilian Portuguese is special, as the entire subject- and object-agreement system appears to be undergoing radical change. So, although there are certain exceptional cases, the observation in the text holds up to the extent that defective paradigms of subject clitics are much more readily found than defective paradigms of object clitics. Thanks to a reviewer for drawing my attention to Paoli’s work.

Subject clitics and macroparameters

181

In conclusion, I will make two tentative generalizations. First, that certain kinds of microparametric variation involve simple PF realization (see Berwick and Chomsky 2011, and largely in line with Manzini and Savoia’s position), while macroparametric variation concerns narrow syntax (pace Berwick and Chomsky 2011). Second, I will suggest that subjects and objects behave differently in relation to agreement-related parameters because of the Phase Impenetrability Condition (see D’Alessandro and Roberts 2008).

9.2 Micro- and macroparametric variation Over the thirty years since the idea of a parametrized UG was first proposed, there has been a steady move away from the original conception of a small number of highly abstract parameters with proliferating effects to a view of parameters as more local in their effects (see Kayne 2005). Manzini and Savoia’s view, described in }9.1, represents a radically microparametric approach. Baker (1996; 2008a; 2008b) has been a rare advocate of retaining at least some notion of macroparameters, and we will look at his reasons for this in some detail below. Here I want to sketch out the advantages and disadvantages of each kind of approach, beginning with microparameters. 9.2.1 In favour of microparameters The microparametric approach arguably originates in Borer’s (1984) proposal that parametric variation must be tightly linked to variation in inflectional systems, specified in the lexicon. Subsequently, this idea was taken up by Chomsky (1995). Baker (2008b: 353) refers to this as the ‘Borer–Chomsky conjecture’ (BCC), which he states as follows: (3) All parameters of variation are attributable to differences in the features of particular items (e.g. the functional heads) in the lexicon. More precisely, Chomsky (1995) restricts the possible variation to formal features of functional heads. These include at least the following: (4) Structural Case, ç (i.e. person, number, and gender features), categorial features (e.g. D, V); attraction features (e.g. EPP features, Edge features). This approach arguably has four main advantages. First, it imposes a strong limit on what can vary. If parametric variation is restricted to functional heads of the kind in (4), then it is easy to imagine possible parameters which are automatically excluded, e.g. those in (5): (5) a. ‘Arity’ of Merge b. existence of Agree c. mode/level of selection from lexicon

182

Ian Roberts

Concerning (5a), a central postulate of the theory of bare phrase structure is that Merge must be binary. It may not be entirely clear why this is so (see Watumull 2010 for a very attractive formal account of this), but it is clear that this cannot be parametrized, since no formal features are involved in the operation of Merge (this is not to say that formal features may not trigger Merge; they do, but they are not part of the operation itself). Hence we do not find variation across languages for binarybranching versus, for example, flat structures, but if parameters of the type in (5a) were allowed, flat, n-ary branching structures could exist. In fact, (5a) rules out ‘W*-languages’ of the kind discussed in Chomsky (1981b) in relation to the notion of non-configurationality. So this approach makes the substantive claim that nonconfigurational languages of the kind envisaged in the early 1980s cannot exist: all natural-language grammars are binary-branching, since the only operation which creates structure is unparametrized Merge. Concerning (5b), while it is clear that agreement systems vary very significantly (indeed, the microvariation in subject clitics in NIDs represents a case of this), the Agree operation is arguably operative in every language. To the extent that some subset of DP arguments must bear structural Case features, this is necessarily true, since those features make the DP arguments active Goals in need of a Probe to Agree with (although see Saito 2007). Again, since the presence of the operation Agree is not regulated by formal features (although it manipulates these features, through the sub-operation Match), the presence of Agree is not open to parametric variation, given (4). Concerning (5c), there is no consensus at present regarding the point in the derivation at which lexical insertion takes place, with advocates of Distributed Morphology arguing in favour of ‘late’ (post-Spell Out) insertion (see Halle and Marantz 1993), while Chomsky’s view has generally been that lexical insertion takes place in two steps: the creation of the Numeration from the lexicon, and then Merge of items from the Numeration to form the derivation. But the option that one language might use late insertion while another has insertion of lexical items from the Numeration has not been mooted, and it is difficult to see why it would be. Since lexical insertion involves more than purely formal features, again (4) rules this option out. So, we can see that (4) represents a desirably restrictive hypothesis regarding the workings of UG. The second desirable property of microparameters stems from language acquisition, as in fact originally argued by Borer, who pointed out that ‘associating parameter values with lexical entries reduces them to the one part of a language which clearly must be learned anyway: the lexicon’ (Borer 1984: 29). This seems clearly true; ultimately this view can relate parametric variation to Saussurean arbitrariness, which seems conceptually desirable. This applies in particular to ‘realization’

Subject clitics and macroparameters

183

parameters, since realization may take arbitrarily variant forms, including, most importantly, no overt form at all. Third, the microparametric view implies a restriction on the form of parameters, which we can formulate as follows: (6) For some formal feature F, P =
F. Here are some concrete, rather plausible, examples instantiating the schema in (6): (7)

a. Non-finite T is [
ç] b. N is
Num c. T is
EPP

(7a) captures the difference between a language in which infinitives inflect for person and number, such as old Neapolitan, and a language in which they do not, such as English, Italian, or the NIDs. This may in turn have consequences for the distribution of overt subjects of infinitives (see Vincent 1996; Ledgeway 1998; 2007; 2009a). (7b) captures the difference between a language in which number does not have to be marked on (count) nouns, such as Mandarin Chinese, and one in which it does, such as English; this difference may underlie much more subtle semantic and syntactic differences in nominals in the two types of system (see Chierchia 1998; Huang 2006). (7c) determines the position of the overt subject; in conjunction with V-to-T, a negative value of this parameter gives VSO word order, providing a minimal difference between, for example, Welsh and French. The simplicity of formulation in (6) in turn makes possible a statement of parametric variation at the UG level which relies on the logic of underspecification: (8)

a. For some formal feature F, F is the default value of P. b. P has þF when triggered (i.e. under specified conditions), F elsewhere. c. þF is the marked value of P.

As we will see below, there are good reasons to favour an ‘underspecification’ view of parameters. Fourth, the microparametric view allows us to put an upper bound on the set of grammars. Suppose we have two potential parameter values per formal feature (i.e. each feature offers a binary parametric choice), then we define a quantity n as follows: (9) n = |F|, the cardinality of the set of formal features. It then follows that the cardinality of the set of parameter values |P| is 2n and the cardinality of the set of grammatical systems |G| is 2n. So, if |F| = 30, then |P| = 60 and |G| = 230, or 1,073,741,824. More realistically, Kayne (2005: 14) concludes that |F| = 100, so |G| = ‘on the order of 1030’ (the actual number is 1,267,650,600,228,229,401,496,703,205,376). It is clear from the discussion in }9.1 that Manzini and Savoia must countenance at least that

184

Ian Roberts

many possible parameters and therefore a number of grammars of roughly the size just given, among which the acquirer has to select. In this connection, Kayne points out: ‘There is no problem here (except, perhaps, for those who think that linguists must study every possible language)’ (2005: 14). But it is here that we can see, as pointed out in }9.1, the possibly fatal weakness of the microparametric approach. To see this, consider two methodological principles of historical linguistics, which the field has relied on explicitly or implicitly since its inception in the early 19th century: (10) Uniformitarianism: ‘the languages of the past are not different in nature from those of the present’ (Croft 1994: 204). (11) Connectivity: ‘within a set of attested language states defined by a given typological classification, a language can . . . shift from any state to any other state’ (Croft 1994: 205). The formulations in (10) and (11), from Croft (1994), are general in nature. We can restate the same ideas in the terminology of principles and parameters as follows: (12) Uniformitarianism: the languages of the past conform to the same UG as those of the present. (13) Connectivity: a grammatical system can change into any other grammatical system given enough time (i.e. all parameters are equally variable, given enough time). Now, suppose that at present approximately 5,000 languages are spoken,4 and that this figure is constant throughout human history (back to the emergence of Homo sapiens). Suppose further that every language changes in at least one parameter value with every generation. Then, if we have a new generation every 25 years, we have 20,000 languages per century. Finally, suppose that modern humans have been around for about 100,000 years, i.e. 1,000 centuries. It then follows that 20,000,000 languages have been spoken in the whole of human history, i.e. 107  2. This number is 27 orders of magnitude smaller than Kayne’s estimate of the number of possible grammatical systems. So, if the parameter space is as large as Kayne suggests, there simply has not been enough time since the emergence of the species (and therefore of UG) for anything other than a tiny fraction of the total range of possibilities offered by UG to be realized. This conclusion effectively empties uniformitarianism and connectivity of content. In theory, we simply could not know whether a language of the past corresponded to the UG of the present or not, since the overwhelming likelihood is 4 For the initial formulation of this argument, I do not distinguish I-language from E-language; this does not affect the point being made here. See fn 6.

Subject clitics and macroparameters

185

that these languages could be typologically different from any language that existed before or since, perhaps radically so.5 In fact, even with a UG containing just 100 independent parameters, we should expect languages should appear to ‘differ from each other without limit and in unpredictable ways’, in the famous words of Joos (1957: 96).6 But of course, we can observe language types, and note diachronic drift from one type to another. Moreover, given what is currently known about historical relationships among languages, and about diachronic typological change, it does not seem plausible that the limited variation we can observe has a purely historical explanation. Arguably, the developments in principles and parameters theory reflect the familiar tension between the exigencies of empirical description—which lead us to postulate ever more entities, i.e. in this case ever more microparametric variation (and it is no accident that Manzini and Savoia, in their painstaking and monumental description of variation in Italo-Romance (2005), take an extreme view in this respect)—and the need for explanation, which requires us to eliminate as many entities as possible. In other words, microparametric descriptions, as pointed out in } 9.1, tend to sacrifice the explanatory power of the notion of parameter of UG in order to achieve a high level of descriptive adequacy. Despite the clear merits of the microparametric approach, it appears that a way must be found to lower the upper bound on the number of parameters, on a principled basis. In this connection, let us now turn to macroparameters. 9.2.2 In favour of macroparameters 9.2.2.1 Baker (2008b) Baker (2008b) makes an interesting and convincing argument in favour of macroparameters. According to him, the microparametric view predicts that ‘there should be many mixed languages of different kinds, and relatively few pure languages of one kind or the other’ (Baker 2008b: 360). On the other hand, the macroparametric view predicts, falsely, rigid division of all languages into clear types (OV vs VO etc.): every category in every language should pattern in one way or the other. Regarding this possibility, Baker comments (p. 359): ‘We now know beyond any reasonable doubt that this is not the true situation.’ Kayne (2000: 6–8) discusses the number of parameters and the number of grammatical systems, and makes an interesting and rather plausible case that there are at least as many grammatical systems in the world as there are people, i.e. upwards of 5 billion. This conclusion does not alter the point being made in the text: if there are so many grammatical systems, then vast numbers of them differ only slightly from one another. But we still need to allow for ‘macrovariation’ in properties such as basic word order, and so still need to allow in principle for a wide typological range. Essentially, Kayne’s argument leads one to the conclusion that there may be more different grammatical systems in the world than is usually thought, but they are all clustering around the same basins of attraction. This fact then requires explanation if Kayne is right. 6 See Biberauer (2008: 1) and the references given there for discussion of how the views of Joos, and more generally of American structuralists, of universals have been misrepresented. 5

186

Ian Roberts

Baker further observes that, combining the macroparameters and microparameters, we expect to find a bimodal distribution: languages should tend to cluster around one type or another, with a certain amount of noise and a few outliers from either one of the principal patterns. And, as he points out, this often appears to be the case, for example regarding the correlation originally proposed by Greenberg (1963) between verb–object order and preposition–object order. The figures from the most recent version of The World Atlas of Language Structures (WALS) are as follows (these figures leave aside a range of minority patterns such as ‘inpositions’, languages lacking adpositions, and the cases Dryer classifies as ‘no dominant order’ in either category): (14) OV and Po(stpositions) OV and Pr(epositions) VO and Po VO and Pr

472 14 42 456

(Dryer 2011a; 2011b)

It is very clear that here we see the kind of normal distribution predicted by a combination of macro- and microparameters. Baker therefore concludes that the theory of comparative syntax needs some notion of macroparameter alongside microparameters. 9.2.2.2 A diachronic (Italo-)Romance view Baker’s argument is convincing. Here I would like to add a similar diachronic argument, taken from Holmberg and Roberts (2010). As we have seen, the Italo-Romance dialects represent a canonical case of microvariation. Although the variation among these varieties is highly impressive (as the 2,500 pages of Manzini and Savoia 2005 attest), a large number of features remain constant: all Italian dialects are SVO, all are prepositional, none shows a systematic ergative case/agreement pattern (although some ‘split ergativity’ is attested), none is fully polysynthetic, none shows the Chinese value of Chierchia’s (1998) Nominal Mapping Parameter (i.e. in allowing a singular count noun to stand alone as an argument, giving I saw cat), all have definite and indefinite articles, all have moderately rich agreement systems, all (except a small number of Ræto-Romansch varieties—see Benincà and Poletto 2005) have complement clitics, none has a full morphological case system, etc. On the other hand, the microparametric variation involving the existence and behaviour of subject clitics, the expression of negation, the position of both finite and non-finite verbs in relation to subject and complement clitics and various classes of adverbs, the nature of object and subject agreement on past participles in compound tenses, the nature and choice of aspectual auxiliaries, the expression of various forms of finite and non-finite complementation, and a range of other properties, is extremely intricate. These are exactly the conditions

Subject clitics and macroparameters

187

which favour productive microparametric work, as Kayne has convincingly argued (see esp. Kayne 2005), and as Manzini and Savoia (2005) have demonstrated. But, one could ask, why are certain properties variable in Italo-Romance and others not? The microparametric answer, as it were, is that no theoretical significance should be attached to what varies and what does not in this particular synchronic geographically defined domain; this is attributable to a historical accident, in that the common features are due to a shared inheritance. But what exactly is this shared inheritance? If we look at Latin, we find OV order (alongside VO; see Adams 1976; Ledgeway 2012a: ch. 5), a full morphological case system, the complete absence of pronominal clitics, no active compound tenses (except with deponent verbs), and a system of complementation in which finite clausal subordination was a minority pattern. As has often been observed, the modern Romance languages (and the ItaloRomance subgroup) are more similar to one another than any of them are to their common ancestor, Latin. The microparametric explanation for this observation would presumably appeal to the accumulation of microparametric changes in the common ancestor language before it broke up into the dialects, i.e. in late or vulgar Latin. The question here, though, is to what extent vulgar Latin can be reasonably regarded as a single system; the term is generally used a cover term for the varieties of non-literary Latin spoken in Italy and elsewhere in the Roman empire, whose written records are somewhat uniform but have been argued to form a koiné (Palmer 1961: 223). In this connection, Clackson (2004: 790) says: ‘the construction of a uniform “vulgar Latin” probably oversimplifies a very complex linguistic situation. Different communities of speakers used different varieties.’ If there ever was a single ‘Proto (-Italo)-Romance’ variety, it would probably have to be dated rather early, as Hall (1950) suggests on phonological grounds (proposing 250–200 bc, exactly the period in which Roman rule was extended to the whole Italian peninsula). Although the Latin of this period is known to differ somewhat from Classical Latin, and to have certain vulgar features, it is highly unlikely that it had the syntactic characteristics of Romance rather than Classical Latin (OV alongside VO order etc). It seems then that the current microparametric variation either derives historically from an archaic, typologically distinct, single ancestor variety of Latin, or there is no single ancestor variety common to all the dialects.7 Either way, the major typological differences between Latin and (Italo)-Romance cannot be traced to a single microparametric change or series of microparametric changes in a single variety. The only explanation is then that there must have been typological drift across the varieties of 7

There is also the substratum theory, which starts from the observation that, apart from a small area in central Italy, all Romance varieties derive from a local form of Latin, whose characteristics may have been influenced by a language spoken earlier—or simultaneously and in many cases for more than one generation—in each locality (these languages would include Celtic in various forms, Venetic, Greek, Sabellian, Etruscan, etc.). Substratum effects can be thought of in various ways in relation to principlesand-parameters approaches; for discussion, see Roberts (2007: 389ff.).

188

Ian Roberts

vulgar Latin. This poses a problem for a purely microparametric approach: other things being equal, we might have expected some dialects to have retained a case system, or OV order, or synthetic passive forms, and not to have developed clitics, etc., others to have developed in the way we observe, and still others to have developed in a mixed fashion, preserving certain archaic features and innovating others. But what we observe, instead, is typological drift: from OV to VO, and in the general direction of greater analyticity (as elsewhere in Indo-European), allowing the development of a considerable amount of truly microparametric variation of the kind that we observe. The simplest account of this kind of parallel development involves distinguishing macroparametric from microparametric change: certain macroparameters changed in the transition from Latin to Romance, or perhaps earlier (as was probably the case with the OV>VO change; see Adams 1976; Ledgeway 2012a), while much of the synchronically observable variation among the Romance languages, and certainly among Italian dialects, could be seen as the result of microparametric change.8 9.2.3 Conclusion So let us conclude, with Baker, that macroparameters exist alongside microparameters. Then two related questions arise: (i) what are their properties? (ii) how are they distinguished from microparameters? Two rather unsatisfactory and partial answers to these questions are that macroparameters ought to be rather few in number, and they ought to be extremely pervasive in their influence on the 8 One could perhaps attempt a contact-based explanation for the parallel developments. The southern dialects were in contact with Greek and the other Italic varieties of Indo-European: Oscan, Umbrian, and related varieties, collectively known as Sabellian. These are broadly similar to Latin in typological terms, being predominantly SOV (Wallace 2004: 832), and so are unlikely to be responsible for the common development of the dialects. The northern varieties were in contact with forms of Celtic and Venetic. In Tuscany, and indeed in the early days of Rome itself, there was contact with Etruscan, a non-IndoEuropean SOV language (Rix 2004: 961). A very thorough study of contact between Latin and all of these languages, at various times and places, is Adams (2003). One possibility, which could have had far-reaching consequences, is that Latin demonstratives developed into articles partly due to contact with Greek, which had an article system. Adams (2003: 518) points to the use of demonstratives as articles in a passage of Plautus. This is significant because it can be traced to Greek influence and because of its early date. However, in order to understand the nature and indeed the possibility of the potentially ‘far-reaching consequences’ of this putative change, we would need effectively to treat the existence of an article system as something more than just the lexicalization of D, which is presumably how Manzini and Savoia would treat it. For a suggestion regarding the potential importance of the development of the article system, see Giusti (in preparation). It seems that the nature of the D-system is something which has proliferating effects elsewhere in the grammatical system, perhaps because D is the locus of interpretable formal features such as person, number, and Case. On the other hand, as a reviewer points out, the emergence of articles from demonstratives is a natural change (a form of grammaticalization; see Roberts and Roussou 2003: ch. 4) observed in various language families, but, aside from the Plautine cases just alluded to, articles do not clearly emerge in texts until centuries after the period of the most intense contact between Latin and Greek. Also unexplained, here or anywhere else to my knowledge, is why the Slavonic languages (with the exceptions of Bulgarian and Macedonian) did not develop articles, especially given their contact with Greek.

Subject clitics and macroparameters

189

grammatical system. The first point holds because, as Baker (2008b: 358) says, ‘[i]f there were many macroparameters and they interacted with one another in complex ways, then languages could differ crazily in ways that would be hard to pull apart’. But this does not seem to be the case in practice: universal properties and microparameters account for much that is shared and much that varies. The second point holds because macroparameters, perhaps by definition, can affect large-scale aspects of the grammar such as all headed phrases, or all instances of Agree. Section 9.3 sketches out conception of macro- and microparameters and relation between them first suggested by Holmberg and Roberts (2010) and developed further by Roberts (2012).

9.3 Macroparameters and markedness The central idea in Roberts (2012) (which also forms the central object of the ERCfunded Advanced Grant project mentioned in footnote 1) is: (15) Macroparameters are the result of aggregates of microparameters acting together, effectively as a single parameter.9 This idea reconciles Baker’s notion of macroparameter with the idea that parameters are specified as the formal features of functional categories, i.e. the BCC, by construing macroparameters as aggregates of microparameters. Apparent macroparametric variation appears when a group of functional heads is specified for the same properties: for example, if all heads implicated in determining word-order variation have the same word-order-related property, the system is harmonically head-initial or head-final. This aggregate behaviour is determined, not by UG, but by learning strategies (Feature Economy and Input Generalization: see below), hence the distinction between micro- and macroparameters is not part of UG, but is an emergent property of the interaction between the UG, the PLD, and optimization strategies in learning (note that these three factors correspond to the three factors of language design discussed in Chomsky 2005). In these terms, we can set up hierarchies of parameters in relation to different axes of variation (word order, null arguments, etc.—see below for examples). True macroparameters sit at the top of the hierarchy, as here all heads parametrized for the feature(s) in question behave as one. Moving down the hierarchy, parameters become more ‘micro’, behaving in a non-uniform, differentiated fashion which is inherently more complex than the systems defined higher in the tree. The options appear to move from applying to all relevant categories, to applying to natural

9 This idea is sketched in Baker (2008b: 354, n. 2), and has also been put forward informally by James Huang and Lisa Travis. Baker points out: ‘It would be worthwhile to articulate this intermediate position more fully, and consider its implications.’ This is exactly what Roberts (2012) attempts to do, as summarized here.

190

Ian Roberts

subsets of the entire set, to smaller classes of functional categories F, to increasingly context-sensitive environments, ultimately perhaps to single lexical items (on this last possibility, see Biberauer and Roberts forthcoming). In these terms, we can think of language acquisition as moving down the hierarchies, halting at the first point at which there is no disconfirming PLD (but never being able to backtrack). As such, each parameter hierarchy defines a learning path, much in the sense of Dresher (1999). The acquisition device searches the space by looking for the ‘easiest’ solution at each stage, where a solution is defined as a parameter-setting compatible with available PLD. The device moves from a relatively simple stage to the next most complex stage (‘down’ the hierarchy) only when forced to by PLD incompatible with the current setting. Regarding diachronic change, we can reason that, since acquirers, and therefore languages, favour higher positions in the network, we predict a general typological preference for the higher options. In the case of word order, it is well known that harmonic languages are more frequent, as indicated for one well-known dyad by the figures in (14). Second, if change is driven by acquisition, in line with the general view of Lightfoot (1979; 1991; 1999), then, all other things being equal, we expect languages to ‘drift upwards’ in hierarchies diachronically. Of course, other things usually are not equal, and the roles of contact and of grammaticalization (which can endogenously cause microparametric change and thereby destabilize an unmarked system) must be taken into account. It is not necessary to see the hierarchies as prespecified by UG. Instead, as already mentioned, they emerge from the interaction of (i) the fact the formal features of certain functional heads are underspecified by UG (UG makes available a certain set of features, but ‘doesn’t mind’ how or whether they are deployed in a given system); (ii) the PLD; (iii) general markedness conditions. The two principal general markedness conditions are as follows: (16)

(i) Feature Economy (FE) (Roberts and Roussou 2003: 275) Given two structural representations R and R' for a substring of input text S, R is less marked than R' iff R contains fewer formal features than R'. (ii) Input Generalization (IG) (Roberts 2007) If a functional head F sets parameter Pj to value vi then there is a preference for similar functional heads to set Pj to value vi.

(Input Generalization may well be a consequence of the acquirer’s initial ‘ignorance’ of category distinctions, with these coming ‘on-line’ as acquisition proceeds, through maturation or through the acquisition of ever-finer featural distinctions, or through a combination of the two; cf. Biberauer 2011; Branigan 2011; 2012.) The higher options, i.e. the minimal categorial distinctions and therefore the most general parametric options, are inherently preferred by the acquirer, because FE and IG favour the higher options in the absence of PLD regarding the full range of

Subject clitics and macroparameters

191

options (in other words, Input Generalization leads to overgeneralization in many cases). In this way, macroparametric effects in grammatical systems derive from markedness, which emerges from the computational conservativity of the learner. Furthermore, there is no need to formulate a difference between micro- and macroparameters. It emerges given our characterization of markedness; we are free to place the cut-off point anywhere on the hierarchy (and to add intermediate concepts such as ‘mesoparameters’ as appropriate; see Biberauer and Roberts forthcoming). In order to understand how these proposals really work, let us look at some examples of parametric hierarchies.

9.4 Parametric hierarchies Here I briefly illustrate the idea of parametric hierarchies with the hierarchy pertaining to word order/linearization. Then I turn to a more detailed discussion of the hierarchy most relevant to subject clitics in NIDs, namely the one concerning null arguments. 9.4.1 Word order The hierarchy in (17) illustrates word-order parameters: (17) Is the head-final feature present on all heads? Y: head-final (a)

N: present on no heads? Y: head-initial(b)

N: present on [+V] categories? Y: head-final in N: present the clause only(c) on . . . .

Languages of type (a) are Japanese, Korean, Dravidian, etc.: the harmonically, rigidly head-final systems. Type (b) includes the rigidly, harmonically head-initial Celtic and Romance languages. Type (c) features German and Dutch, to a close approximation, since they show head-final TP, vP, and VP but are (almost) headinitial in all other categories. Lower in the hierarchy, we find a series of increasingly specific microparameters, relating to the order inside specific categories (rather than classes of categories), ultimately perhaps to specific lexical items (e.g. English enough is the only degree modifier to follow its adjective). The ‘head-final feature’ alluded to here can be construed more technically in one of two ways. On the one hand, we may assume that the default linearization option is head-initial, with head-final order derived by marking the relevant heads in some way (e.g. for triggering movement of their complements, as in Kayne 1994). On the other hand, we could assume a PF Head Parameter as discussed by Richards (2004)

192

Ian Roberts

and Sheehan (forthcoming). For present purposes, it is not necessary to distinguish between these two alternatives. Note that the very highest option conforms fully to both FE and IG: it maximizes feature economy in not postulating any features and it generalizes this lack of features to all relevant heads. Assuming the head-final option to require a special feature (this follows readily assuming the approach in Kayne (1994) along with the idea that all movements must be triggered), the initial option is therefore harmonic head-initial word order. The second option is more costly (and therefore marked) in terms of FE in that a feature is postulated, but it is still optimal in terms of IG as that feature is assigned to all relevant heads; the result is harmonic head-final order. Further options are more marked in that the feature is always assumed, but its distribution becomes dependent on steadily more fine-grained categorial distinctions, making it more marked in terms of IG. Recall that we do not have to assume that the options are prespecified: parameters are created through interaction with the PLD in a fashion reminiscent of the Piagetian concept of epigenesis. Having seen how the parameter hierarchies are thought to work, we can now move on to the case of null arguments. This will bring us back to the null-subject parameter, and the question of microvariation in subject clitics in NIDs. 9.4.2 The null-argument hierarchy In order to understand how the null-argument hierarchy works, three preliminaries are necessary. The first preliminary is to define probe-goal Agree. Agree is a syntactic relation between two heads, a Probe P and a Goal G, where the following hold: (18) a. P asymmetrically c-commands G. b. P and G are in a local relation (undefined here). c. P has unvalued features (e.g. [Person:–]) and G has ‘corresponding’ valued features (e.g. [Person: 1]). d. Then G’s values may be copied into P’s attributes (Match). The second preliminary concerns the nature of the ‘radical’ or ‘discourse’ pro-drop characteristic of many East Asian languages. Here, for reasons that will become apparent I focus on Japanese. (19) gives examples of radical pro-drop affecting a variety of arguments in Japanese (from Neeleman and Szendrői 2007: 672–3): (19) a.  siken-ni otita ] exam-dat failed ‘pro failed the exam’ b. Bill-ga  settokusuru Bill-nom persuades ‘Bill persuades [pro]’

Subject clitics and macroparameters

193

c. [ mimi-ga ] nagai ear–nom long ‘[pro’s] ears are long’ The salient properties of radical pro-drop systems are as follows: (20) a. Arguments bearing any grammatical function may be freely non-realized, subject to discourse conditions. b. These arguments may receive any kind of pronominal interpretation (referential, bound, E-type, indefinite, etc.; Tomioka 2003). c. Radical pro-drop languages lack overt Ds (Tomioka 2003). d. There is no overt realization of agreement inflections (Huang 1984). Saito (2007) proposes a very interesting analysis of this phenomenon. He suggests that there is a single LF-copying mechanism which allows for radical pro-drop and argument ellipsis. This mechanism involves copying elements into argument positions at LF from a set of discourse-given entities. He then points out that it ‘is conceivable that pro, being a typical discourse entity, is always included in this set’ (p. 7). The precondition for this kind of ‘covert merge’ of arguments is that ç-features not be obligatory on probes; if they are, then the absence of a goal when Agree takes place in the pre-LF part of the derivation will mean that these features fail to be valued and the derivation will crash. If, on the other hand, these features may fail to be present, then arguments may fail to be present prior to LF. Presumably, ç-features may be absent from probes in systems where they completely fail to have any PF-realization: that is, whilst ç-features may lack overt morphophonological realization (as a matter of PF Spell Out), the absence of ç-features cannot give rise to person–number marking. In this way, the relation between radical pro-drop and absence of agreement marking is established (see also Kuroda 1988).10 The third preliminary concerns the proposals in Roberts (2010a) regarding defective goals: (21) a. A goal G is defective iff G’s features are a proper subset of those of G’s Probe P. b. Defective goals always delete/never have a PF-realization independently of their probe. The fourth, and last, preliminary concerns Müller’s (2005: 10) ‘pro generalization’: (22) An argumental pro DP cannot undergo Agree with a functional head Æ if Æ is subjected to a ç-feature neutralizing impoverishment. Note that we are not denying that Japanese has Agree. This arises when the ç-features are present on probes, and in Negative Concord (Watanabe 2004) and, if wh-in-situ involves Agree between C and D without movement, in wh-constructions. 10

194

Ian Roberts

Here the term ‘impoverishment’ refers to a concept in distributed morphology. Impoverishment is a deletion operation which affects the feature bundles created and manipulated by the syntax, taking place after syntax but before ‘vocabulary insertion’, the post-syntactic operation which pairs phonological and morphosyntactic features. Impoverishment rules ‘neutralize differences between syntactic contexts in morphology’ (Müller 2005: 3), creating ‘system-defining syncretisms’ (distinct from accidental homophony or gaps in a paradigm). Now consider the following postulate: (23) If a category has a D[efinite] feature, then all that category’s ç-features are specified. (Russell 1905) This postulate has a certain naturalness, in that it claims that for definite reference (or a definite description) to be established, information regarding at least person and number must be specified. It follows that a T with impoverished ç-features cannot bear a D-feature. Since subject pronouns (weak and strong) are DPs, given the definition of defective goal in (21) it follows that where T lacks a D-feature a subject pronoun is not a defective goal and therefore cannot be null. In this way, the link between ‘rich’ agreement and null subjects is established. One question which has been much debated in the literature on null subjects concerns what, if anything, is in SpecTP when we observe a null subject. More precisely, in Italian-style null-subject languages, is the subject position (SpecTP) occupied by an empty pronoun? Different authors have given different answers to this question: Barbosa (1995; 2009), Alexiadou and Anagnostopoulou (1998), and Manzini and Savoia (2005) have argued that there is no subject present, basically because it is not needed on interpretive grounds. On the other hand, Burzio (1986), Rizzi (1987), and Cardinaletti (1997) have argued, on distributional grounds, that a null pronoun is present. Roberts (2010b: 85) concludes: These arguments [those of Burzio, Rizzi and Cardinaletti, IR] are obviously compatible with pro’s absence at PF, and could be compatible with its absence at LF. So it is conceivable that null-subject deletion takes place in core syntax: the deleted pronoun satisfies the EPP and the absence of a subject in SpecTP has the interpretative effects Barbosa and others discuss at LF.

Pronoun deletion is subject to the condition on defective goals in (21). As Roberts (2010b) points out, the comparison of Saito’s analysis of radical prodrop in Japanese and his analysis of null subjects in Italian is revealing, as summarized in (24): (24) a. Japanese Radical pro-drop is possible iff ç-agreement is not obligatory (because probes won’t be looking for goals, and the latter can be inserted at LF subject to discourse conditions).

Subject clitics and macroparameters

195

b. Italian Consistent null subjects are possible iff there is no impoverishment of T’s ç-features (because T then has a D-feature, which makes it able to delete a pronominal DP which is its goal, i.e. the subject). The two systems are derivational mirror images of one another, and this is the direct consequence of the different status of ç-features on probes (fully optional vs obligatorily present and unimpoverished), which in turn is directly reflected in the agreement morphology (totally absent vs ‘richly’ realized). This observation leads to the postulation of the following hierarchy for null arguments: (25) a. Are uϕ-features obligatory on all probes? No: Radical pro-drop

Yes: b. Are uϕ-features fully specified on all probes? Yes: Null arguments

No: c. Are uϕ-features fully specified on some probes? No: Non-null-subject

Yes: d. Are the uϕ-features of T impoverished Italian, etc.

As indicated, the ‘No’ value in (25a) gives radical pro-drop. The positive value of (25b) may give rise to a ‘pronominal-argument’ system of the type described in Jelinek (1984). A positive value for (25c) gives a non-null-subject language like English. (25d) gives a consistent null-subject language like Italian. Lower in the hierarchy, the null-subject parameter starts to ‘break up’ into microparameters as individual probes are evaluated in relation to it. Having seen the general idea of parameter hierarchies and the specific proposal for null arguments, including null subjects in languages like standard Italian, and nonnull-subject languages like English, we can now ask where the familiar northern Italian subject-clitic systems fit in. This is the topic of the next section.

9.5 Northern Italian subject clitics and the null-argument hierarchy The first point to make in this connection comes essentially from Rizzi (1986): subject clitics are not equivalent to object clitics in NIDs, for three reasons. First, subject clitics can double a full DP subject with no trace of ‘Kayne’s generalization’, i.e. no requirement that a preposition or special marker of some kind for the putatively

196

Ian Roberts

doubled subject argument appear. On the other hand, those Italo-Romance varieties which allow object clitic doubling obey Kayne’s generalization (see e.g. Ledgeway 2000 on Neapolitan and related dialects). Second, the structural relations are different: preverbal subjects move over the cliticization site (T or part of a ‘split’ T) from their first-merged vP-internal position. This does not appear to be the case with object clitics. Third, subject clitics show paradigmatic gaps much more than object clitics, as can be seen from (1). Gaps of this kind in object-clitic paradigms are rare in Italo-Romance to my knowledge, but the paradigms in (1) are representative of subject-clitic paradigms. As Rizzi and many others have pointed out, these differences point in the direction of treating the subject clitics as agreement markers. Agreement markers normally ‘double’ subjects, in the sense that person and number can be encoded both in the agreement marking and on the subject DP (indeed, it is the Celtic-style pattern where this does not happen which is cross-linguistically rarer). Agreeing heads, i.e. probes, typically attract their goals to their specifiers; this is the basic role of EPP features in Chomsky (2001). Third, there is nothing unusual about gaps (or syncretisms) in agreement paradigms. By contrast, object clitics, at least in NIDs, show typical properties of pronouns in resisting doubling, not attracting goals (because they are not probes), and showing fully specified paradigms (but cf. Paoli 2009, discussed in footnote 5).11 The question is: if subject clitics are agreement markers, what is their relationship with the verbal agreement inflection? In order to start to answer this, let us take, as a purely descriptive device, the feature [
agr] to denote whether a clitic or agreement paradigm shows a ‘full’ set of morphological person–number distinctions, where ‘full’ means at least five distinct forms, and zero counts as distinct. We can then distinguish four morphological patterns involving subject clitics and verbal agreement inflection: (26) a. b. c. d.

SCL[+agr] SCL[+agr] SCL[–agr] SCL[–agr]

V[+agr] V[–agr] V[+agr] V[–agr]

‘fully redundant’, null-subject system non-null-subject system a non-redundant null-subject system (usually) a complementary system

We can find systems of all four types in NIDs and elsewhere in northwestern Romance. Consider first the following paradigm from Florentine (parl- ‘speak’), as described in Brandi and Cordin (1989): sg (27) (E) parlo Tu parli E/la parla

pl Si parla Vu parlate E/le parlano

11 I follow Kayne (1984) and Rizzi and Roberts (1989) in treating French subject pronouns as phonological, rather than syntactic, clitics. See Roberts (2010c) on varieties of French.

Subject clitics and macroparameters

197

Here we see one two-way syncretism in inflection (3sg, 1pl); one in the clitic paradigm (3sg.m, 3pl.m), taking 1sg to be distinct as it freely alternates with zero. This is then a fully redundant system, showing double agreement marking and clearly able to license null subjects by the criterion given in the preceding section. Compare now the paradigm for French dormir ‘to sleep’: sg (28) Je dors Tu dors Il/Elle dort

pl Nous dormons Vous dormez Ils/Elles dorment

Here there is a three-way syncretism in inflection (1sg, 2sg, 3sg), in that the orthographic distinction between final -s and final -t makes no difference to pronounciation,12 but none in the subject-clitic paradigm. The ç-features of French T are thus impoverished, and so T lacks a D-feature and the subject pronoun cannot be defective and delete in the syntax. In other words, French is not a null-subject language. Consider next the paradigm for dorm- ‘sleep’ in the variety of Como, already shown in (1): (29) ta al/la

sg dOrmi dOrmat dOrma

pl dOrmum dOrmuf dOrmaŋ

(Manzini and Savoia 2005: i.100)

Here there are six distinct inflectional forms, and four identical (zero) clitic forms (1sg, 1pl, 2pl, 3pl). Here, then, the verbal inflection is capable of licensing a null subject. This is a non-redundant (or partially redundant) system. Finally, the fourth possibility is illustrated by the Carrara variety (dVrm- ‘sleep’): (30) a tə i/al

sg dOrmə dOrmə dOrmə

a i/al

pl durmiŋ durmitə dOrmənə

(Manzini and Savoia 2005: i.80)

Here both paradigms are impoverished: there is a three-way syncretism in inflection (1sg, 2sg, 3sg), and a three two-way syncretisms in the clitic paradigm(1sg/1pl; 3sg.m/3pl.m; 3sg.f/3pl.f). Taking the two paradigms separately, null subjects should not be allowed in this variety, contrary to fact. However, if we take the two paradigms, clitic and agreement inflection, together, there are six distinct forms. These are boldfaced in (30). The /t/ that appears in the inverted 3sg form: dort-il (/dͻrtil/) ‘sleeps he?’, i.e. ‘does he sleep?’) is a case of the ‘epenthetic’ /t/ that appears with all verbs in this context, cf. parle-t-il (/parlәtil/ ‘speaks he?’ i.e. ‘does he speak?’). The orthographic conventions are misleading here. 12

198

Ian Roberts

How can we relate all this to the null-argument hierarchy in (37)? There are two options. The first is to posit some very deeply embedded right branch of (37) which offers the options in (26), presumably along the lines of ‘Is [Person] impoverished when Singular, when Plural, etc.?’ down to each person/number combination (when impoverishment and non-realisation become the same thing). This treats all these systems as highly marked, and is clearly very similar to Manzini and Savoia’s proposal, criticized in }9.1 (except that it does not treat these options as independent microparameters). But the paradigms in (27) and (29) show non-impoverished inflection, differing only in the realization of the subject clitic. Hence we have a second option: we can treat these as on a par with standard Italian: consistent nullsubject languages with slightly differing realizations of T’s ç-features. There is then no need in these cases for the extreme microparametric variation just described. This leaves (30), a very common pattern in NIDs, where both paradigms taken separately are defective, but which together give a full set of distinctions. We can treat these along the lines of (27) and (29) if we adopt a more sophisticated notion of realization. Here two further ideas from distributed morphology concerning the mapping from syntax to morphophonology are relevant: (31) a. Fission: ‘splitting off ’ of a subset of features of a head to form a separate morpheme. b. Copying: the features of a probe are copied on to the goal. In general, T’s features (Tense, ç, and, in a null-subject language, D) are copied onto V as a function of V-T Agree. This happens in all the cases under discussion here, giving rise to finite verbal inflection. So this is what happens in standard Italian: T’s (unimpoverished) features copied on to T. In Florentine (where we see agreement doubling), both copying and ç-feature fission take place, so that the subject çfeatures are realised on both T and V. In the Como variety, we see copying and ç-fission, along with impoverishment. But impoverishment cannot apply to the node containing D, i.e. T, or T won’t be able to treat the subject pronoun as a defective goal for the reason given above; hence V’s inflection remains fully specified while the subject clitics are impoverished. Finally, in the Carrara variety, fission of a subset of ç-features (here singular) precedes copying, and there is no impoverishment. None of these options involves impoverishment of the D-marked node, and so all can count as null-subject languages. As far as the hierarchy in (25) is concerned, they are all the same, all being like standard Italian. French, on the other hand, shows impoverishment before copying; hence D is not allowed (by (23)), and hence cannot be a null-subject language. In this context it is worth noting that not one of the 180 dialects reported by Manzini and Savoia has exactly the French pattern of partial syncretism in the verb endings and total differentiation of the subject pronouns (in proclisis, in non-V2 systems).

Subject clitics and macroparameters

199

So we see that, if we adopt a more sophisticated account of how ç-features may be realized (and some account of this kind is needed in any case, and would be if there were no parametric variation regarding the realization of subject arguments), we can assimilate most NIDs to the same position in the null-argument parameter hierarchy as standard Italian. They are null-subject systems in good standing (in contrast to French).

9.6 The difference between subjects and objects The above presents a view of what is going in NIDs regarding the null-subject parameter. However, although we have noted the differences between subject clitics and object clitics, we have not explained why object clitics/agreement do not behave in the same way as subjects (note that for Manzini and Savoia this is just due to a further arbitrary set of microparametric values; the situation could easily be reversed, in an Italo-Romance variety or elsewhere). In a different context, looking at past-participle agreement in standard Italian and some central-southern varieties, D’Alessandro and Roberts (2008: 482) propose the following condition in the realization of agreement marking: (32) Given an Agree relation A between probe P and goal G, morphophonological agreement between P and G is realized iff P and G are in the complement of the minimal phase head H. The condition in (32) essentially states that overt agreement will only appear when both the probe and goal of the abstract Agree relation are in the spell-out unit, the immediate complement of the same phase-head. Now consider the canonical environments for subject and object agreement: (33) a. Subject agreement: [CP C [TP Subject T . . . ]] b. Object agreement: [vP v [VP V Object ]] In (33a), the subject and T are in C’s complement, TP, so this probe–goal pair satisfies (32) and overt agreement can be realized. In (33b), on the other hand, the object and v are not both in v’s complement, VP (since the notion ‘complement’ is irreflexive). Thus this probe–goal pair fails to satisfy (32). The structural difference between the canonical positions of subjects and objects combined with (32) may underlie the following cross-linguistic observations: (34) a. Subject agreement is more common cross-linguistically than object agreement.

200

Ian Roberts b. (Agreement-licensed) null subjects are more common cross-linguistically than null objects.13 c. The realizational options listed above do not apply to object agreement with v, since v more rarely shows overt object agreement.

In the context of NIDs, (34c) is the relevant observation, explaining the major subject–object asymmetries we observe.

9.7 Conclusion Let us briefly reflect on some more general questions. Chomsky (2007; 2008) and Berwick and Chomsky (2011) both emphasize the ‘PF–LF’ asymmetry in language design, claiming that the Strong Minimalist Thesis (SMT) holds only of the mapping to the SEM interface, with mapping to the PHON interface, or externalization, seen as a secondary process. The following quotation sums up this position, and relates it to parametric variation: Parametrization and diversity, then, would be mostly—possibly entirely—restricted to externalization. That is pretty much what we seem to find: a computational system efficiently generating expressions interpretable at the semantic/pragmatic interface, with diversity resulting from complex and highly varied modes of externalization, which, furthermore, are readily susceptible to historical change. (Berwick and Chomsky 2011)

Here I have taken a different view. The SMT holds everywhere; in fact one can legitimately ask why it should not hold in the mapping to PF. Copying and fission are natural operations—in fact both can be seen as instances of Merge—and impoverishment is deletion; these are thus natural operations found elsewhere in the computational system. Moreover, Kayne (2009: 6) has suggested that ‘morphemes are derivationally built up out of “phonemes” via the application of Merge (with onsetnucleus-coda syllable structure asymmetrically mimicking spec-head-complement phrasal structure)’.14 Morphophonology makes use of combinatorial processes: either these are instances of Merge or they are not. If they are not, then natural language is ‘uglier’ than it needs to be, in conflict with the spirit (at least) of Minimalism. If they are, then the PF interface is less different from narrow syntax than has been claimed. The latter view seems both empirically and conceptually the more plausible one. In these terms, then, we can think of macroparametric variation as a property of syntactic derivations, relating to the presence of formal features (syntax), while 13

A reviewer reminds me that Latin had null objects, but no object agreement. And the general absence of branching onsets with non-branching nuclei mimicking the Final Over Final Constraint of Biberauer, Holmberg, and Roberts (2007; 2008), as an instance of *[n [Æn Æ ª ]  ]. See Biberauer, Holmberg, and Roberts (2011) on FOFC in phonology. 14

Subject clitics and macroparameters

201

microparametric variation relates to their realization (morphology); syntactic impoverishment rules are an ‘in-between’ case, since the distinction is not one of principle. So construed, macroparametric variation may still play its ‘traditional’ role in relation to the poverty of the stimulus/explanatory adequacy. This is perhaps the most important conclusion in the present context, and we have seen that the intricate variation observed in the subject-clitic systems of NIDs does not lead us to abandon this view.

10 Sicilian 1st and 2nd person oblique tonic pronouns A historical and comparative examination* ROSANNA SORNICOLA

10.1 Conservation and innovation in Romance pronominal systems The remarkable degree of synchronic and diachronic polymorphism displayed by Romance 1st and 2nd person tonic pronouns makes them an extremely interesting area of investigation both from a theoretical and a historical perspective. Despite some tendencies in specific areas towards paradigmatic levelling in favour of a single form,1 Romance pronominal systems, and particularly those associated with personal pronouns, have generally maintained allomorphic distinctions for case,2 in contrast to nominal systems where, since our earliest records, case inflections have been confined to just a few geographical areas and today are limited to Romanian.3

* I dedicate this study to Mair Parry, whom I hold in high regard and consider a dear friend. 1 For example, in Gascon and Occitan varieties the nominative form, especially in the 2nd person, has replaced the original oblique forms, and in Gallo-Romance varieties, including those spoken in Italy, the oblique forms are often employed in topic function. 2 Here we make no distinction between abstract Case (with an upper-case ‘C’) and morphological case (with a lower-case ‘c’). In what follows, we simply use the latter in all instances. 3 As is well known, both old French and old Occitan displayed a binary nominative~oblique case distinction, a system which is also visible in the earliest Ræto-Romance texts (see Meyer-Lübke 1890–1906 [1974]: ii. }3). Traces of this system can still be seen today in some northern Italian varieties (see Benincà and Vanelli 1978). The changes affecting the inflectional paradigms of the noun in the passage from Latin to Romance represent an extremely complex and interesting diachronic problem, and not without interest for the theoretical linguist. Among other things, they impinge on the difficult and controversial question of the formation of the Romance plural (for Italo-Romance, see Aebischer 1960; 1961; Sabatini 1965a; 1965b; Maiden 1996; 2000). The stages involved in this long evolutionary process might not have been as linear as is often assumed; problems of such an assumption are discussed with particular reference to Gallo-Romance in Sornicola (2011).

Sicilian 1st and 2nd person oblique tonic pronouns

203

Within Romance, binary systems that distinguish between pronominal forms for subject and non-subject functions are widespread, as well as ternary systems that distinguish subject, object, and indirect object functions. Also attested are four-way systems which increment the ternary system through the addition of a distinct pronominal form for the comitative function.4 It would however be rash, and indeed incorrect, to think that the case allomorphy displayed by Romance personal pronouns can simply be interpreted as a continuation of the corresponding Latin forms. Indeed, one of the most interesting aspects of Romance 1st and 2nd person pronouns regards the complex ways in which they simultaneously exemplify conservation and innovation. Although conservation can usually be readily identified, it can sometimes be concealed by phonetic changes which quite radically transform what were already quite ‘insubstantial’ mono- and bisyllabic pronominal forms (cf. Gallo-Romance forms such as mei, moi (< me ‘me’) tei, toi (< te ‘you.2sg’). Innovations, on the other hand, are either the product of analogical remodelling operating at the purely morphological level (cf. OSp. forms like myb(e) ‘me’ based on the model of the original 2sg dative tibi, just like the early Italo-Romance forms meve, mebe ‘me’),5 or the tendency of once quite restricted forms to extend their distribution to a larger number of syntactic contexts. Syntactic conditioning of this type is observable at different periods and in different areas, and with non-uniform outcomes. Exemplary in this respect is the extension of the Catalan ‘nominative’ forms jo ‘1sg’ and tu ‘2sg’ to oblique contexts, as well as the generalization, in various Gallo-Romance areas, of the original oblique 1/2sg forms moi and toi in topic function, as illustrated in (1a,b):6 (1)

a. Moi, je crois qu’ il s’ me.top I believe that he= self= ‘I believe that he’s made a mistake’ b. Toi, tu dors you.top you sleep ‘You always sleep’

est is

trompé (Fr.) mistaken

toujours (Fr.) always

Also to be included among such cases are much older processes which, in certain respects, prove less easy to reconstruct. For example, the Ibero-Romance 1sg oblique form mí, which is used after all prepositions except con ‘with’, might represent the generalization of the Latin dative pronoun mihi to prepositional contexts, perhaps initially limited to the preposition ad ‘to(wards), at’ following the merger of the

4

For an overview of the various types of Italo-Romance case systems, see Loporcaro (2008). The representation of the vowels in these pronominal forms is discussed below. It is not possible to do justice here to the enormous bibliography on the relevant diachronic processes involved and the morphological structures that these have produced. For a more detailed description of the facts with a partial overview of the relevant literature, see Sornicola (2013). 5 6

204

Rosanna Sornicola

syntactic marking of direct and indirect object functions, whereas the 2sg form ti might represent an analogical formation on the 1sg pronoun, itself based on the dative form. (Documentary evidence from both late antiquity and the early medieval period shows unmistakable signs of an early process of Erstarrung (‘crystallization’) of the originally dative 1/2sg forms, even in contexts where Classical Latin would have required the accusative; cf. Sornicola 2000; 2013.) These examples highlight the strict interrelation of morphological, syntactic, and phonetic processes in the development of the Romance pronominal paradigms, arguably a fertile test-bed for studying the polymorphic processes involved in diachronic change and for developing theories on the multiple causes involved in the formation of paradigms and the changes affecting them.7 We must also recognize that it is not always easy to identify the relevant Latin (or Romance) underlying forms for the pronouns in question, such that the polymorphic diachronic processes involved also often remain unclear. It is entirely plausible that the Latin personal pronouns were characterized by a high degree of polymorphism, for which there are indeed various pieces of evidence. In particular, diatopic, diastratic, and (by all accounts) diaphasic variation must have been such that different pronominal forms were commonly in use. Indeed, ‘light’ forms such as those which characterize the pronominal paradigm are particularly prone to variations in prosody, syntax, and style and register. In what follows we shall examine in greater detail the principles and peculiarities outlined above through a particularly interesting case study from Sicilian. Both for modern and early varieties of Sicilian it is possible to recognize, for the 1st and 2nd persons singular, binary paradigms which contrast the outcomes of ego (1sg) and tu (2sg) used in subject function with oblique forms associated with the functions of specifier, direct object, and indirect object. In the earliest literary Sicilian texts dating from the 14th and 15th centuries mi and ti are by far the most common forms used in all functions other than that of subject, but alongside these we also find the forms me, te and mia, tia, which enjoy a more restricted intertextual distribution and are overall the least frequent. The picture in modern Sicilian varieties, by contrast, is radically different. The forms mi, ti and me, te have almost disappeared entirely, except in a few Gallo-Italic dialects (which are, however, often known to differ from the indigenous Sicilian varieties),8 whereas the forms mia and tia now predominate almost

7 The problem of multiple causation in the formation and transformation of paradigms is expertly tackled in various studies by Malkiel (e.g. 1967; 1969). 8 Map I, 54 of the AIS reveals that the form mia is only absent in Sicily in Mascalucia (p. 859, where the 1sg form is me) and in the Gallo-Italic varieties spoken in San Fratello (p. 817, where the 1sg form is yía), Sperlinga (p. 836, where the 1sg form is mi), and Aidone (p. 865, where the 1sg form is me). A substantially identical areal distribution is found with the forms of the 2sg tonic pronoun, with te in Mascalucia and the nominative tu extended to oblique functions in San Fratello and Sperlinga (AIS 4, 701). Calascibetta (p. 845) has the forms mi@ and ti@ with a final centralized vowel.

Sicilian 1st and 2nd person oblique tonic pronouns

205

everywhere (see AIS I, 54; AIS 4, 701). It would however be rash, not to mention ingenuous, to conclude from this that the forms mia and tia only became established in recent times, or at least more recently than mi and ti. An etymological examination of the various pronominal pairs, although not without its difficulties, coupled with an awareness of their different distributions across early Sicilian, would seem to provide some useful clues for understanding some of the diachronic developments involved in the competition of these pronominal variants. Furthermore, a study of early Sicilian polymorphism can perhaps also offer some additional evidence in reconstructing the developments affecting Romance 1st and 2nd person singular pronouns.

10.2 1sg and 2sg oblique tonic pronouns in early and modern Sicilian: Romance equivalents and etymological hypotheses We now turn to examine the Romance equivalents of the Sicilian forms and the diachronic factors which might have contributed to their formation. As a preliminary observation we may note that the forms mia and tia are today only found in Calabrian, where they are more or less generalized, and in some Salentino varieties (cf. AIS I, 54; AIS 4, 701). Within these same areas we also find the forms mie and tie,9 which have formal equivalents also in Sardinian (cf. Blasco Ferrer 1988: 74; Virdis 1988: 909) and Romanian (Iordan, Guţu Romalo, and Niculescu 1967: 121–2), although in these varieties they assume different functions within the pronominal paradigm. In the relevant Sardinian dialects, for example, mie and tie represent the dative forms of a ternary paradigm which formally distinguishes case forms for nominative, dative, and accusative; identical considerations hold for the Romanian dative forms mie and ţie (tie) (Table 10.1). TABLE 10.1. Sardinian and Romanian 1/2sg pronominal forms Sardinian

Romanian

Dative

a mie

a tie

mie

ţie (tie)

Accusative

a mé

a té

pe mine

pe tine

Furthermore, it should be noted that the Sardinian forms mie and tie can only occur after a preposition, just like the Sicilian, Calabrian, and Salentino forms 9 See points 745, 761, and 771 of AIS I, 54. The form recorded at point 745 is mmy@. The same AIS map also records mie for the Salento at Vernole and Salve (points 733, 749). Cf. also Rohlfs (1977: 415), where the distribution of the form mie in Calabria is said to be limited to a few areas in the provinces of Cosenza and Catanzaro. Rohlfs (1976: 342) only records the form mia for one locality in the province of Brindisi.

206

Rosanna Sornicola

mia and tia,10 whereas in Romanian mie and ţie (tie) occur without a preposition insofar as they continue the original bare dative forms. This latter observation underlines the conservative nature of these pronouns, which is further supported by their distribution in so-called archaic areas such as Sardinia and Romania. Yet the etymology of the pronominal variants mia, tia, mie, and tie and the various relationships between them raise a number of problems. A comparative analysis might lead us to suppose that they can all be variously traced back to Latin dative forms. However, this hypothesis is only plausible for the 1sg pronoun which can be retraced to the dative form mı¯hı¯ (mı¯hı˘), which since early times seems to have been subject to oscillations typically affecting ¯ı, sometimes realized as e¯, sometimes as the diphthong ei: witness the oscillations in such spellings as mihiii and mihei found in the Pompei Inscriptions (cf. Väänänen 1966: 22–3). In such oscillations it is perhaps possible to recognize rustic Latin pronunciations, ultimately due to the influence of the surrounding Italic dialects in which the diphthong ei, which corresponds to varying degrees to Classical Latin ¯ı, tended to be realized somewhat like e¯.11 At the same time we cannot fail to recognize that the form mie might equally originate from a form with a short final vowel mı¯hı˘ or mı˘hı˘ (this latter form being frequent in poets such as Catullus, Virgil, and Martial; cf. ThLL 5, 2, 255, 19ff.), especially as confusion between final ˘ı and e¯ is known to be particularly frequent from the 3rd century ad (cf. Väänänen 1981: }55). In the case of the 2sg pronoun, by contrast, we are forced to assume that it was analogically modelled on the forms of the 1sg pronoun. We must therefore make sure to distinguish in our diachronic account between the 1st and 2nd person pronouns, even if the development of both forms must at a certain point have become indistinguishable. If in terms of relative chronology the type mie has to be considered earlier, we still need to establish how recent analogical formations like tie are. There are also many other questions that need to be asked. Do the Sicilian, Calabrian, and Salentino forms mia/mie, tia/tie and the Sardinian and Romanian forms mie and tie/ţie really represent the outcomes of the same diachronic development? What is the relationship between the southern Italo-Romance forms mie, tie and mia, tia? While both sets of variants are still found today in Calabria and Salento, Sicilian varieties have apparently only ever known the forms mia and tia. Regarding the first question, I would be inclined to think that, at least initially, there was substantial unity in the diachronic developments which brought about the formation of these Romance pronominal types. Consequently, Meyer-Lübke’s ([1890–1906] 1974: ii.}75)

10 In Sardinian mie and tie are restricted to occurring after the preposition a ‘to’, while in Sicilian, Calabrian, and Salentino the corresponding forms can follow any preposition. 11 On the various diachronic relationships holding between ei and ī in Latin, see Väänänen (1966: 22–3). Interesting in this respect are the pronunciations spe¯ca for spı¯ca ‘ear of corn’ and ue¯lla for uı¯lla ‘farm, estate’ attested in Festus and Varro (see Väänänen 1966: 23).

Sicilian 1st and 2nd person oblique tonic pronouns

207

claim that the Romanian forms can be explained diachronically as the agglutination of the preposition ad ‘to(wards), at’ to the end of the relevant pronouns, with the subsequent development a > e, strikes me as unconvincing. Although in the case of the personal pronouns we cannot a priori exclude the possibility that different diachronic processes might have produced similar forms, the extent of the distribution of the types mie and tie across the Romània, coupled with a historical-geographical examination of the areas involved, would seem to support the hypothesis that the genesis of these pronominal forms broadly represents a unitary development. It appears that the existence of a formally plausible base form, together with its possible phonetic outcomes for which we have considerable documentation, represent structural data which can readily be integrated with the available diatopic and diastratic evidence. Indeed, the areas which have preserved the types mie, mia, tie, and tia could be described as ‘southeastern’ and ‘insular’ parts of the Romània. It is these same parts of the Romània which have also preserved and developed other Latin vulgarisms of various periods and which, it must not be forgotten, were for a long time drawn into the Byzantine world. The fact that the whole area under discussion displays a structural isomorphism between the 1st and 2nd person pronouns might lead us to suspect that the paradigm was formed at an early date. The scenario which we have proposed in our answer to the first question also has implications for the second. If we take the base forms to be mı¯hı¯ with its variants mihı¯, mihei, mihe¯, and mı˘hı˘ with its later variant mihe˘, the form mie appears to be the closest and most ‘natural’ outcome: the development from a base form in final ¯ı forces us to hypothesize an intermediate stage in which -ei undergoes monophthongization to -e¯, while a more straightforward development would be from a base form with final ˘ı on account of its frequent confusion with e˘. This hypothesis is in line with the observation that the form mie enjoys a greater areal distribution than mia. Furthermore, within Italo-Romance traces of mie are also to be found in 13th- and 14th-century Bolognese and Venetan texts,12 whereas the type tia (apart from the Ritmo Cassinese, to which we will return below) only occurs outside Sicily (and even there appears relatively late) in a handful of early literary texts from continental southern Italy that show deliberate Sicilian influence. Exemplary in this respect is the scribe’s insertion of tia in the second redaction of Boccaccio’s Neapolitan Epistola (Bolimo buffeniare ’na picca con tia, se chiace a tia ‘we want to jest a little with you, if it pleases [to] you’; cf. Sabatini [1983] 1966), as well as its appearance in a 14th-/early 15th-century siciliana, a popular verse form ([ . . . ] e’ morirò per tia [ . . . ] Stèssima alle

12 The oldest literary attestation is to be found in Guido Fava’s Parlamenti ed Epistole (23 (84)) from about 1243, namely no diamo salute a tie ‘we do not greet you’ (lit. ‘we not give salutation to you’). The TLIO corpus provides another 42 occurrences, all from the 13th to 14th centuries: 3 in Giovanni da Vignano, 2 from the Leggenda di Santa Caterina, 8 from the Commento to Jacopo della Lana’s Commedia, 21 from the Laudario dei Battuti di Modena, and another 8 from texts from the Emilia–Veneto area.

208

Rosanna Sornicola

càmari, entra mi e tia! ‘I will die for you [ . . . ] If we were [together] in the bedchamber, [what would happen] between me and you!’ cf. Coluccia 1975: 126). But what of the form mia? It is only in Calabria and Salento that both mie and mia are found, whereas in Sicily, as already noted, only the latter form is found, at least judging by the available evidence. One possible explanation would be to assume a phonetic development mie > mia. As observed by Rohlfs (1966: }144), in southern Calabria (Catanzaro, Cerva, Pentone, Satriano, Centranche, Simbario, Stilo) unstressed final -e regularly becomes -a (fama ‘hunger’, barcuna ‘balcony’, vida ‘sees’, studenta ‘student’, noma ‘name’, virda ‘green’, patra ‘father’, jocara ‘to play’, nenta ‘nothing’, cocira ‘to cook’, ava ‘has’, òja ‘today’, nda ‘therefrom’). He also notes that in northern Calabria (Cosenza and Acri) final unstressed -e only becomes -a when it occurs in hiatus following another vowel: tria for trie ‘three’, sia for sie ‘six’, nua for nue ‘we’, pua for pue ‘then’, vua for vue ‘you’, i vua for i vue ‘the oxen’.13 Now, this phonetic context would provide the perfect environment for the development of the Sicilian, Calabrian, and Salentino pronominal forms mia and tia, if it were not for the fact that this phonetic development is not found outside Calabria except in Umbria (Gubbio, Todi), the Marches (Fano), Cortona, and in a few early Romanesco texts (e.g. doa ‘two’, noa ‘we’ in Cola di Rienzo; cf. Rohlfs 1966: }144). Furthermore, the distribution of the types mia and tia in Calabria extends well beyond the northern area (Cosenza and Acri) in which final unstressed -e regularly becomes -a in hiatus with a following vowel. This observation undoubtedly forces us to reflect upon the partial equivalence in the conditioning effects of hiatus observed between northern Calabria (province of Cosenza) on the one hand and Sicily and Salento on the other, where the relevant change is limited to the 1st and 2nd person pronouns. Are we to conclude from this that the relevant phonetic development was found in a much wider area in the past than today? If this were so, then we should expect to find more systematic evidence of its presence in contexts other than the simple 1st (and 2nd) tonic person pronoun(s). Another peculiarity concerns the fact that in Sicilian, unlike in Calabrian and Salentino, there never appears to have been a form mie from which the attested form mia could have developed.

13

Interesting confirmation of the early date of this change in Calabrian comes from the results of a recent study by Ledgeway (in press c) on the inflectional morphology of the imperative and the distribution of metaphony in the dialects of southern Italy. In Cosentino final -e regularly becomes -a in the 3sg present tense of all verbs, in the 2sg imperative of verbs derived from the Latin II, III, and IV conjugations, as well as in the paroxytonic infinitival forms of these same conjugations. The fact that the 2sg imperatival forms, including those derived from the Latin IV conjugation, do not display metaphony provides significant proof that the change from -e to -a cannot be a recent one (I thank Adam Ledgeway for bringing these facts to my attention and for pointing out their relevance for the present discussion). This evidence and the diachronic inference that can be drawn from it fit perfectly with the data from Latin epigraphic sources discussed below, as well as with the results of some analyses carried out on the 9th-century documents of the Codex Cavensis (Sornicola et al. in preparation).

Sicilian 1st and 2nd person oblique tonic pronouns

209

A quite different explanation is offered by D’Ovidio (1886), who argues that both mie and mia are derived from mi through the adjunction of a paragogic vowel. While such developments are certainly not unusual in the dialects (cf. Rohlfs 1968: }335), the range of cases involved are not necessarily all of the same type. This tendency has been observable since early times, as evidenced by medieval manuscripts (e.g. OTsc. piúe ‘more’, fúe ‘was’ (3sg), formòe ‘formed’ (3sg), cosíe ‘thus’, and especially important for our purposes the pronominal variants mèe and mee in Guittone and Dante), and is found across a number of areas (e.g. Lmb. cantòe ‘sang’ (3sg), clamòe ‘called’ (3sg), intròe ‘entered’ (3sg); Flo. noe ‘no’, síe ‘yes’, cosíe ‘thus’, piúe ‘more’, peròe ‘but’, túe ‘you’, làe ‘there’, dureràe ‘it will last’; Sic. cchiùi ‘more’, and the less frequent ɖɖài ‘there’, aitài ‘age’, nicissitài ‘necessity’). Many of these forms, as Rohlfs himself observes, are presumably the result of an etymological vocale di ripercussione (‘fallout vowel’) in such pairs as fu/fúe < fuit ‘was’ (3sg) and ha/hae < habet ‘has’ (3sg). In cases like mée ‘me’ and túe ‘you’, by contrast, no such explanation is available, and the presence of the final unstressed vowel can only be explained in terms of the prosodic structure of the words involved. The diffusion of such final unstressed vowels for purely metrical reasons must be considered a spontaneous phonological process, observable across all periods and subject to a degree of irregularity. In many respects it resembles other phenomena which, although phonetically and structurally quite different, serve the same function of syllable lengthening or syllable adjunction in oxytones. These include (a) processes of lengthening and diphthongization of the stressed final vowel (as in the following forms from the Campi Flegrei area of western Naples [p@k´kUi] ‘because, why’, [vu´lUi] (< vulé) ‘to want’, and [mUi] ‘me’; see Sornicola 2003); (b) the adjunction of fricative segments in such words as [meiç], [meʝ] ‘me’, [p@k'keiç] ‘because, why’, [akkus'siç] ‘thus’, typically occurring at the end of the intonational phrase in contexts of strong emphasis and often found in the spontaneous use of dialect and many modern varieties of Italian;14 (c) the adjunction of a syllable with material which is not the result of a phonetic process affecting the tonic syllable, such as the paragogic element -ne found in various Italian dialects in words like pecchéne ‘why’, réne ‘king’, sìne ‘yes’, nòne (cf. Rohlfs 1968: }336). It is worth stressing that these phenomena are anything but regular: they are present only in the speech of some speakers and even then are not systematic. Types (a) and (b) could both be described as an ‘internal extension’ of the syllable, as opposed to type (c), which involves an ‘external extension’ of the word as a whole. Unlike the dissimilatory phonetic development e > a considered above, D’Ovidio’s hypothesis has the advantage that it involves a process of prosodic morphology operative across all periods, but for which there is no necessary expectation that it This lengthening through the addition of a fricative is discussed in Retaro (2005; 2009) for the dialects spoken in the Campi Flegrei and the varieties to the east of the Vesuvius. This same lengthening process is also well documented in various styles of spoken French. 14

210

Rosanna Sornicola

should apply in all possible contexts. It is also supported by the important consideration that early Sicilian texts, on a par with their Calabrian and Salentino counterparts, present a high incidence of the 1sg and 2sg oblique tonic pronominal forms mi and ti. In view of the Sicilian tonic vowel system, it is not clear however whether the forms mi and ti in the early texts of these areas can legitimately be considered the outcome of the dative mı¯hı¯ (with analogical formation of ti on the 1sg form),15 subsequently generalized to all oblique functions on a par with what is observed in the Iberian peninsula, or whether they simply represent the phonetic outcome of the accusative or ablative forms me¯, te¯, and me¯(d), te¯(d), respectively. However, even D’Ovidio’s hypothesis is, in my opinion, not without its problems. In modern Sicilian we would expect an internally extended unstressed vowel to surface as a high front vowel, in line with what we know about the processes of spontaneous diphthongization and unstressed vowels in the dialects of the island. While it would be difficult enough to justify a high-mid front vowel (/e/), it would prove even more difficult to justify a central mid-low vowel (/a/), which never seems to be involved in such extension processes in Italo-Romance.

10.3 Textual evidence from early literary Sicilian An examination of the evidence provided by early Sicilian literary texts, and a comparison with early Calabrian and Salentino literary texts, can provide further interesting clues worth reflecting on. We begin by summarizing the absolute and relative frequencies of the 1st and 2nd person pronominal variants mi, me, mia, ti, te, and tia recorded in the early Sicilian ARTESIA Corpus (Table 10.2).16 TABLE 10.2. Distribution of 1st and 2nd person pronouns in early Sicilian ARTESIA corpus 1st person

2nd person

mi

459 (79.55%)

ti

311 (73.7%)

me

62 (10.74%)

te

45 (10.66%)

mia

56 ( 9.71%)

tia

66 (15.64%)

Total

577

Total

422

15 A development tibi > tiâi > ti cannot be assumed for the areas in question. On the problem of the generalization of the dative forms in personal pronouns, see Sornicola (2013). 16 The corpus is part of the Opera del Vocabolario Italiano database (available at www.ovi.cnr.it). Also to be considered here is the study by Pagano (2003) which highlights the problem of the differential distribution of mi, te and mia, tia in 14th- and 15th-century Sicilian texts, including a discussion of its implications for the concept of ‘14th–15th-century Sicilian’. Pagano’s study is principally based on an analysis of the distribution of oblique pronominal forms (including the comitative forms micu ‘with me’

Sicilian 1st and 2nd person oblique tonic pronouns

211

As can be seen, both mi and ti account for about three-quarters of all occurrences of the 1st and 2nd person oblique tonic pronouns in their respective individual samples (with a slightly higher frequency for mi than ti). The individual percentages for me and te are also very similar, each accounting for about a tenth of all tokens in the sample. Slightly different is the picture for the forms mia and tia which make up just over a tenth and a little less than a seventh, respectively, of all occurrences of the 1st and 2nd person pronouns. Also worthy of note is the distribution of the individual pronominal forms according to text type. Mi represents practically the only form found in historiographic texts (Valeriu Maximu, Rebellamentu di Sichilia, Istoria di Eneas), and also predominates in religious and hagiographic texts: there is not one occurrence of mia in the Libru de li vitii et di li virtuti and just one in the Sposizione del Vangelo, in contrast to 121 occurrences of mi. The distribution of this latter form is almost absolute in epistolary and legal texts.17 By contrast, the form mia enjoys a somewhat greater distribution in a number of hagiographic texts from the 15th century: witness the absolute frequencies for mia and mi given in Table 10.3. TABLE 10.3. Distribution of mia and mi in some 15th-century Sicilian hagiographic texts Text

mia

mi

Alfabetin

21

–

Munti della santissima oracioni

1

7

Regula di santu Benedittu abbati

4

–

Translacioni di S. Agata

14

6

Libru di lu transitu et vita di misser sanctu Iheronimu

3

30

1

5

3

–

Quaedam profetia

a

Raxunamentu di l’abbati Moises a

The only example of mia in Quaedam profetia occurs in close proximity to an example of mi and is employed solely for metrical reasons: Scriptura, tu non tolli a mia stu tali laczu; / anti a mi cumbogli gruppu cun grandi maczu ‘[ . . . ] you do not remove from me [ . . . ] before [to] me you cover [ . . . ]’ (v. 125, p. 27, r. 5).

and ticu ‘with you’ in addition to mi, ti, mia, tia) in the two versions, from the 14th and 15th centuries, of the Life of Sant’Onofrio. Pagano also undertakes a preliminary examination of the relative frequencies of mi, ti, and mia, tia in a more extensive corpus of early Sicilian texts, the statistical results of which coincide entirely with those of the present study based on the entire ARTESIA corpus (itself put together by the Catanese team led by Pagano). 17 There is just one occurrence of mia in a letter of 1402 by Giovanni Abbatelli (albeit alongside a single occurrence of mi in the same text).

212

Rosanna Sornicola

As can be seen, there is just one text, the Alfabetin, a Judaeo-Sicilian translation, in which mia represents the only attested pronominal form. This text is certainly very different both in genre and linguistic register with respect to the literary prose texts, and might therefore lead us to hypothesize that the sole use of mia and tia in this text reflects spoken usage (or at least those sociolinguistic registers which are closer to the spoken language).18 However, this hypothesis, although not without some appeal, must be treated with caution, inasmuch as the tradition surrounding the text and its composition present some complex and controversial cultural-historical problems (for discussion, see Sermoneta 1994: xvii–xxx). Furthermore, the Alfabetin contains a number of hybrid features which make it difficult to locate the origin of the text, although those that have considered the question acknowledge, from their different positions, that the text can be generically considered Sicilian.19 As Sermoneta (1994: lvi, my translation) observes in the introduction to his edition: Our text—just like other Judaeo-Italian texts—is characterized by a typical dual character: on the one hand it reflects a certain degree of artificiality on account of its scholastic origins and its status as a ‘calque’ language, and on the other it adapts, or better absorbs at the phoneticolexical level the basic linguistic features of the period and the geographical area in which the text itself was translated. These two facets merge together, mutually adapt to each other, and give rise to a specific language which differs from the other literary texts of the period produced outside of the Jewish environment in which the author lived.

The me variant occurs sporadically in manuscript A of the Istoria di Eneas (with 5 occurrences, all in contexts such as oy me ‘poor me!’), but proves very frequent in manuscript B (33 occurrences, 4 after oy ‘poor’ and the others all after a preposition). The Sposizione del Vangelo della passione secondo Matteo also shows a handful of occurrences of me in constructions which seem to be directly related to a Latinate syntax or to have been influenced by the fact that they occur in portions of the text written in Latin. In fact, of the 4 occurrences in this text, 3 (cf. 2a–c) involve the pronoun me as the object of the verb in contexts with word order patterns in which the pronoun seems to retain full syntactic and prosodic autonomy, as witnessed by the fact that in the second and third examples (cf. 2b,c) the pronoun precedes the verb but is separated from the latter by an intervening negation in accordance with a structural pattern known as ‘interpolation’ in Romance.20 18 The sole use of mia and tia in the Alfabetin is also highlighted by Pagano (2003: 169), who links the phenomenon to the fact that it is ‘a text which is undoubtedly more open to the influence of the spoken language’ (cf. also Di Girolamo 1977). 19 cf. Sermoneta (1994: xxx), who also acknowledges the Sicilian origin of the text, and Minervini (in press), who, although more cautious in her approach, also arrives at the same conclusion following a thorough and meticulous linguistic analysis of the text. 20 It must be pointed out in this regard that in various contexts in which the personal pronoun does not follow a preposition, it is difficult to establish whether the pronoun should be considered tonic or unstressed.

Sicilian 1st and 2nd person oblique tonic pronouns (2)

213

a. per firmari me in la santa Cruchi (Sposizione I, p. 12, r. 6) for bind.inf me in the holy cross ‘in order to bind me to the holy Cross’ b. et me non aviriti (Sposizione I, p. 66, r. 3) and me not you.will.have ‘and you shall not have me’ c. ma me non aviriti sempri (Sposizione I, p. 71, r. 8) but me not you.will.have always ‘but you shall not always have me’

The fourth occurrence is the translation of an immediately preceding Latin passage: (3)

Apersiru supra me la bucca they.opened above me the mouth ‘They gaped upon me with their mouths’

loru (Sposizione 2, p. 81, r. 4) their

All in all, an examination of the intertextual distribution of me and of the syntactic and pragmatico-textual structures in which it occurs lead us to suppose that it should be considered a learnèd Latinism. However, we cannot entirely exclude the possibility that its presence in early Sicilian texts might be an indication of the fact that the current Sicilian vowel system had not yet become universally established across the island. The intertextual distribution of the forms of the 2nd person pronouns broadly corresponds to that witnessed for the 1st person. In fact, the distribution of ti is similar to that of mi, both typically surfacing in texts of a historiographic, hagiographic, and legal nature. In a similar fashion, tia, just like mia, predominates in the Translacioni di S. Agata, the Regula di Santu Benedittu, the Libru di lu transitu di misser sanctu Iheronimu, and above all in the Alfabetin and the Raxunamentu di l’abbati Moises, to which we can also add the Istoria di S. Ursula (see Table 10.4).

TABLE 10.4. Distribution of tia and ti in some 15th-century Sicilian hagiographic texts Text

mia

mi

Alfabetin

21

–

Regula di santu Benedittu abbati

8

1

Istoria di S. Ursula

7

1

Translacioni di S. Agata

7

–

Libru di lu transitu et vita di misser sanctu Iheronimu

5

15

Raxunamentu di l’abbati Moises

7

–

214

Rosanna Sornicola

The form te displays the same intertextual and syntactic and pragmatico-stylistic distribution as that noted above for me. More generally, an examination of the syntactic distribution of the variant pronominal forms highlights some further areas on which it is worth reflecting, although the results are at times contradictory. We begin by observing that over 90% of all oblique tonic forms found in the corpus occur after a preposition, namely in ‘configurationally’ similar syntagmatic structures.21 However, we must also analyse the positions of the variants mi/ti and mia/tia within the clause in order to assess whether their syntactico-prosodic position within a larger syntagmatic unit has any influence over the distribution of the (monosyllabic) short and the (bisyllabic) long forms. Indeed, we might expect the clause to be subject to a well-known principle of prosodic morphology, which ultimately aligns the distribution of long forms with focal syntactic positions (such as the clause-final position) and the short forms with non-focal positions (typically clause-medial positions).22 The null hypothesis could then be formulated as follows: in early Sicilian literary texts the bisyllabic variant mia/tia occurs above all in clause-final position, with mi/ti in clause-medial position. If this hypothesis is borne out, the synchronic conditioning in terms of prosodic morphology sketched above might be taken as proof that the competing pronominal forms do not occur in free variation. In this regard, we have identified the distributions given in Table 10.5. TABLE 10.5. Distribution of mia/tia and mi/ti in clause-initial (CI), clause-medial (CM), and clause-final (CF) positions Position

mia

mi

tia

ti

CI

3 (5.46%)

52 (11.28%)

3 (4.55%)

34 (12.23%)

CM

10 (18.18%)

235 (50.98%)

20 (30.30%)

92 (33.09%)

CF

42 (76.36%)

174 (37.74%)

43 (65.15%)

152 (54.68%)

While the distributional frequency of mia and tia in both positions appears to substantiate our hypothesis regarding their preference for clause-final position (albeit with some not insignificant differences in the overall frequency between the two), the results for the distribution of mi and ti are less straightforward: in the case of mi about a third of forms occur in clause-final position, with just under half in the case 21 Note that in conjunction with the intensifying adjective midemmi ‘-self ’, we only find the forms mi and ti. 22 Among clause-final positions I have also included here cases in which the pronoun occurs immediately before parenthetical material or an apposition.

Sicilian 1st and 2nd person oblique tonic pronouns

215

of ti. This leads us to suppose that mi and ti enjoy greater distributional freedom in the wider syntactico-prosodic context of the clause.

10.4 Textual data, areal data, and diachronic developments What interest, then, can the textual data examined hitherto have for our understanding of the relevant diachronic developments? The fact that the forms mia and tia prove overwhelmingly less frequent in early Sicilian literary texts, and then are only documented with any real frequency in 15th-century texts, should not lead us to assume that they are necessarily to be considered more recent that the corresponding forms mi and ti. The intertextual distributions examined above could simply indicate that they were considered at the time less learnèd, perhaps more ‘popular’ or ‘colloquial’, such that 14th-century texts, which were still bound up in the long and complex processes of symbiosis with Latin and stylization of the vernacular according to models that reflected the personal culture of the author or of the scriptorium, might simply have filtered them out. It is interesting in this regard to note that even the earliest Calabrian and Salentino texts do not show any trace of the bisyllabic variants, which today are so well represented in the local dialects; a similar situation has also been observed in relation to the earliest Sardinian texts.23 Finally, we must also note that the poets of the Sicilian School employ the forms meve and teve (which presuppose the etyma mı˘bı˘/mı˘be˘ and tı˘bı˘/tı˘be˘) but not, apparently, the forms mia and tia. In order to resolve the diachronic problem regarding the formation of the Sicilian forms mia and tia, we must therefore turn to other types of data and employ deductive methods. In fact, I believe there are grounds to suppose that the forms mia and tia, even if not as old as the Sardinian, Romanian, Calabrian, and Salentino forms mie and tie, nonetheless still represent very early formations. If, as seems plausible, we accept that mia and tia developed from mie and tie following a change in the final vowel, we are left with two possible hypotheses: (a) we are dealing with a polygenetic development in the extreme south; (b) the relevant change in the final vowel represents an early development which, ultimately, can be traced back to Latin ‘vulgarisms’ variously scattered across different areas of the Romània. Now, a late polygenetic development in the relevant areas is, I believe, to be ruled out since the development -e > -a, still productive today in various Calabrian dialects, albeit under differing conditions in different varieties, has never been attested for Salentino or Sicilian. This leads us therefore to suppose that the Sicilian forms represent a morphophonological archaism, ultimately to be associated with a more general series 23 In the Calabrian merchant’s letters from the end of the 14th century studied by Librandi (2006: 39), the oblique tonic form is me; similarly, the forms me and te are the sole forms found in the Salentino Sydrac (Sgrilli 1983: 11). In the earliest Sardinian texts, we find alongside the form mi also the reinforced form mimi (see Blasco Ferrer 1984: 95; 2003: 238; Merci 1992: 231).

216

Rosanna Sornicola

of vowel changes in the passage from Latin to Romance which in Sicily, unlike in other parts of Italy, never took root. Indeed, already in documents of the early late Latin period (e.g. Christian epigraphs, papyri, manuscripts of literary and legal texts) studied by Schuchardt (1866–8), there is evidence for the change -e > -a. Although the relevant examples are by no means numerous, they are nonetheless significant, and prove highly revealing (Schuchardt 1866–8: i.216): mesa (for classical menses ‘months’), in baca (in bacai) (for classical in pace ‘in peace’), libertabusqua (for classical libertabusque ‘freedwomen.dat/abl=and’), nequa (for classical neque ‘nor’), quinqua (for classical quinque ‘five’), paena (for classical paene ‘almost’), saepa (for classical saepe ‘often’).24 As Schuchardt (1866–8: i.222) observes, a number of these spellings are undoubtedly due to late scribal errors, but many must reflect the actual linguistic usage of lower diastratic levels. Thus, although we are dealing with non-systematic fluctuations, they do nonetheless highlight a tendency (or potential) ab antiquo for the phonetic change -e > -a that we have been assuming to have given rise to the forms mia and tia. One final piece of evidence for the hypothesis that mia and tia are to be considered early forms comes from the observation that the pronoun tia is also used in the Ritmo Cassinese, namely ca multu fora colejusu/tïa fabellare ad usu ‘for it would be highly desirable/to speak [to] you as a friend’ (vv. 46–7, Contini 1960).25 This is an interesting attestation which, together with the relevant etymological evidence, forces us to recognize that the chronological distribution of the pronominal forms in early Sicilian cannot be readily projected onto a straightforward linear diachronic development. Rather, this evidence could reinforce the hypothesis, already advanced above, that the most frequent occurrences of mia and tia in a number of 15th-century Sicilian texts are due to the deliberate choice of pre-existing variants, albeit most probably diastratically marked, belonging to a lower stylistic and more colloquial register.

10.5 Conclusions We are now in a position to make some general observations. The first concerns the representativity of the texts considered in relation to the hypothesized linguistic developments. Our examination has shown that at times etymological and intertextual investigations cannot necessarily be readily married together. The difficult 24 In late imperial epigraphic texts pre- and post-tonic ˘ı is also sometimes represented graphically by a, a phenomenon perhaps related to the frequent confusion between ˘ı and e˘ (v. Schuchardt 1866–88: i.217, 219, 220). 25 Contini (1960) integrates this pronominal form in his edition of the text, but observes that the correction ticu ‘with you’ suggested by D’Ovidio is most probably acceptable, also for prosodic reasons. While it is true, as he points out, that the 2nd person tonic oblique pronoun appears as tebe elsewhere in the text, this in itself does not seem a decisive argument, since in early Romance texts, especially those of the earliest period, the presence of competing morphological variants is anything but rare.

Sicilian 1st and 2nd person oblique tonic pronouns

217

integration of these two levels of analysis forces us to relativize to a certain extent the premises, methods, and conclusions which can be legitimately arrived at by one or the other of these two approaches. At the same time, however, both prove indispensable for the purposes of a diachronic investigation. Other observations relate directly to the peculiar development of the tonic personal pronouns. The Erstarrung (‘crystallization’) of originally dative 1/2sg pronouns represents a diachronic case study of considerable importance, for at least two reasons. First, the phenomena under examination suggest that the development of these personal pronouns followed, within certain limits, quite specific pathways, distinct not only from those of other nominal forms (which, I believe, show a much more complex evolution that cannot be simply reduced to the phenomenon of the ‘Extended Accusative’) but also from the 3rd person pronominal forms. Secondly, the Erstarrung (‘crystallization’), and more generally the case study outlined here, also allows us to make some theoretical observations. The passage from Latin to Romance has seen developments in pronominal (and, of course, nominal) morphology,26 which cannot be unilaterally accounted for in terms of the morphosyntactic properties of case assignment. Just as with examples of analogical levelling (cf. such cases as Sp. ti ‘you’ modelled on mí ‘me’, and OSp. mibi ‘me’ modelled on tibi ‘you’), the role of other autonomous morphological processes has not been insignificant, such as the generalization of a given form at the expense of all others, independently of the original conditioning morphosyntactic context. Such processes seem to have played a very important role in the decomposition and transformation of the tonic pronominal paradigms, although in no way exclusive to such paradigms. They obey general principles, the validity of which can however only be assessed on a case-by-case basis, such that there remain at times some grey areas and the evidence is not always decisive. As already observed, the etymology of the pronominal forms which we have been considering here is, to some extent, still open to question, and can perhaps be best assessed in relation to the rules of the game that hold for diachronic linguistics more generally; one which (let us not forget) often turns out to be nothing more than a problematic exploration of possibilities rather than certainties.

26

For a discussion of cases of nominal Erstarrung (‘crystallization’), see Sornicola (2012).

11 Patterns of variation and diachronic change in Piedmontese object clitic syntax* C H RI S T I N A T O R T OR A

11.1 Introduction Historically, object clitics in Piedmontese compound tenses changed from preauxiliary position to post-participle position. The discussion in the literature regarding this syntactic change is in part concerned with the change’s impetus, i.e. how and why the change started in the first place; a related question which arises is whether the change started with one clitic type (e.g. 3rd person singular accusative lo) and gradually spread to other clitic forms over time, or whether it initially struck all clitic forms simultaneously. However, as the work of Mair Parry has shown (e.g. 1991; 1995; 2005), such hypotheses can only remain pure speculations, as the historical texts themselves do not provide evidence one way or the other. Regardless of the lack of textual evidence to inform the question, Parry’s work has contributed many observations which, taken together, contain stimulating clues for further inquiry. The purpose of this chapter is to take Parry’s work on the subject as the most promising point of departure to revisit the issue, and to pursue an analysis driven by a novel perspective. The inspiration for this chapter thus comes directly from Parry’s research, which I believe still represents the most thorough and considered of any thought that has been given to the issue addressed here, as it combines * I would like to dedicate this work to Mair Parry. Mair’s previous work on this topic has done more than any other literature in the field to make me think about object clitic syntax and the relationship between synchronic variation and diachronic change. I am also grateful for her individual style of doing research, which has always reminded me that the most important contribution you can make as a scholar is to allow yourself to be uniquely you. This one truth, which is sometimes lost on us, might be the single most important thing that leads to progress in science, and as a community of scholars, we are fortunate that Mair has always been herself.

Patterns in Piedmontese object clitic syntax

219

a careful analysis of the extant historical texts with an in-depth understanding of the relevant synchronic variability in object clitic (OCL) syntax in the dialects of Cairo Montenotte and Torino. Her approach thus opens a novel and promising window on to what the historic patterns of change in Piedmontese OCL syntax might have been. Specifically, Parry’s precise description of the patterns of variability found in these present-day varieties has blazed a path for connecting what I will argue here to be relatable changes in OCL syntax in other Romance varieties, in completely different syntactic contexts. This chapter is organized as follows: in }11.2 I summarize the details of the historic change and synchronic variability in OCL syntax in Piedmontese compound tense clauses, as described in Parry (1991; 1995; 2005). As we shall see, although the historic texts do not reveal whether the syntactic change in question took place one OCL type at a time, the synchronic variability tells a revealing story. The conclusion that the clitic form se has not completed the change from proclisis to enclisis in Torinese sets the stage for discussion of the behaviour of se in Fassano (Ladin) and in Spanish (}11.3). Here I articulate an account which holds that the different behaviours of the different OCL forms in syntactic variation and change in these varieties is best understood if we take OCL placement to respect the rigid ordering of the functional projections of the ‘clitic placement domain’. This Functional Hierarchy Hypothesis (FHH) is then applied to the case of Piedmontese in }11.4. Here I show that once a few independently motivated theories are put in place, the facts of Piedmontese se fall out directly.

11.2 Piedmontese object clitic syntax 11.2.1 The diachronic change in Piedmontese object clitic syntax Let us begin with a summary of the story of diachronic change in Piedmontese object clitic syntax. As already noted in the Introduction, OCLs in Piedmontese compound tenses shifted from pre-auxiliary position to post-participle position.1 As Parry’s work

1 As Parry and others have explicitly shown and discussed extensively, this change took place in modal +inf and other complex predicate structures, as well (see the above Parry citations, and also Benucci 1993). For the purposes of the discussion of Piedmontese, I will focus principally on the compound tenses, though I do draw on examples of modal+inf structures. It should be noted, however, that the enclisis facts of the non-Piedmontese northern varieties complicate the question of the Piedmontese syntactic change described in the text. Specifically, as pointed out by Paola Benincà, the other northern dialects—like Piedmontese—robustly exhibit enclisis on the infinitive verb in modal+inf constructions. If this enclisis in modern northern Italian ultimately resulted from a syntactic change like the one described for Piedmontese compound tenses and modal+inf structures, the question arises as to why this change did not occur in compound tense constructions, in the other northern varieties. While I leave this question open, in }11.5 I revisit the question in light of the analysis provided in this chapter.

220

Christina Tortora

illustrates, this trajectory of change in Piedmontese OCL syntax is witnessed by the textual evidence. Specifically, the texts show that Piedmontese varieties, like other Romance varieties, started out with proclisis of the OCL on the inflected verb in the compound tenses, and stayed this way up through the 1300s. This is illustrated in example (1), which is a compound tense structure with the 1st person singular accusative clitic ma proclitic on the auxiliary (which seems to be orthographically integrated into the past participle presenta in this particular example): From La sentenza di Rivalta (1446) (as quoted in Parry 1991, taken from Brero and Gandolfo 1967) (1) . . . ma apresenta . . . (OPie.; cf. It.: mi ha presentato) ocl has.presented ‘he has presented me’ However, texts from a slightly later period, starting in the 15th century, reveal an emerging change from sole proclisis of the OCL on the inflected auxiliary verb (as in (1)), to a concomitant proclisis on the auxiliary and enclisis on the non-finite verb. Consider in this regard the example in (2): From Il Conte Pioletto (1784) (as quoted in Parry 1991, taken from Tana, ’Lcont Piolett, 1784) (2) L ’eu smentia-lo (OPie., cf. It. l’ho dimenticato) ocl= I.have forgotten=ocl ‘I have forgotten it’ Here we see that the change in OCL syntax seemed to initially involve an ‘intermediate’ state of affairs. This configuration, which I will call ‘clitic repetition’, is actually still variably present in Cairese, a Piedmontese dialect spoken in the Ligurian town of Cairo Montenotte (and also, in a limited way, in Torinese), as I shall discuss immediately below, in }11.2.2. As for what we could call the ‘final’ stage of change, let us consider the two examples in (3), both from modern varieties, where there is sole enclisis of the OCL on the participle: Torinese (example from ASIt) (3) a. L’ hai vist-la jer (= It.: L’hai vista ieri) scl= you=have seen=ocl yesterday ‘You have seen her yesterday’ Biella (example from ASIt): b. Antè ca l’ à where that scl= has ‘Where has he put it?’

büta-lu? (= It.: Dove l’ha messo?) put=ocl

Patterns in Piedmontese object clitic syntax

221

We can characterize the stages of change in Piedmontese OCL syntax in the compound tenses as follows. Piedmontese historically started out like Italian, in that the OCL was to the left of the auxiliary verb, and, as far as the written texts go, it seemed to be just like Italian until about the 15th century; by around the 19th century, however, most Piedmontese dialects completed the change in OCL syntax, such that we now have strict enclisis on the participle for most varieties. On the way to changing to this post-participial position, there was an intermediate stage for object clitic placement, the ‘clitic repetition’ stage, whereby the clitic could appear simultaneously both to the left of the auxiliary and to the right of the participle. Many scholars have discussed the nature and possible impetus for this syntactic change (e.g. Meyer-Lübke 1900; Salvioni 1903; Wanner 1983; Albin 1984; Tuttle 1992; Benucci 1993). Regarding the question of impetus, Meyer-Lübke (1900) suggests, for example, that the pre-auxiliary position of the 3rd person singular ‘l’ object clitic in a subject clitic language such as Piedmontese created a configuration whereby the adjacency of the object clitic l and the 3rd person subject clitic (SCL) l—the latter of which can be seen in the examples in (3)—made the two formally identical morphemes too close for comfort, causing speakers to displace the former to the postparticiple position. This explanation entails that the move towards enclisis would have happened one clitic-type at a time, with the 3rd person accusative forms appearing in post-participle position before others, such as the 1st and 2nd person forms, or datives and obliques. Arguments in the literature regarding the (lack of ) validity of this proposal (and others) abound. However, as Parry’s meticulous exploration of historical texts has shown (e.g. Parry 1991; 1995; 2005), hypotheses which suggest that the change affected the different clitic forms at different stages cannot be directly verified. This is because, by the time the first texts started appearing (in the 14th century), they were already exhibiting the innovative post-participle syntax for all clitic forms, as the examples from Parry (1991) show. As can be seen in this earliest text, clitics as different in type as se, gli, and lo all appear in this new syntactic configuration:2 Ordinamenti dei Disciplinati e dei Raccomandati di Dronero (Parry 1991)3 (4) a. e se no volesa confesar-se and if neg wanted confess.inf=se

se

2 As we shall see in }11.3, the forms se (reflexive) gli (dative) and lo (3rd singular accusative masculine) are treated differentially in other Romance varieties which are in the process of exhibiting a change in object clitic syntax. 3 The examples in (4) are all from modal+inf constructions; given the historic parallelism between these structures and the compound tenses in Piedmontese, I shall assume that the pattern seen in (4) is valid for aux+past participle structures as well. The comments in fn 1, of course, raise the question of whether it is correct to assume that a pattern found for modal+inf structures also held for compound tense constructions. The important point here is that there is no textual evidence that shows that the change started with one particular clitic type, for either the compound tenses or modal+inf structures. (cont.)

222

Christina Tortora b. e and

lo the

c. gli the

fregl deben aver misericordia e secorrer-lo brothers must have.inf mercy and help.inf=lo

prior prior

debia must

acordar-gli concede.inf=gli

da in

piaton secret

gli lo

So even if it were the case that the change in OCL syntax in Piedmontese gradually obtained one clitic-type at a time, this gradual change would have already completed itself before the first appearance of Piedmontese written texts; thus, the textual evidence can reveal nothing regarding the hypothesis that the change in OCL syntax hit different forms at different stages. 11.2.2 Synchronic variability in Piedmontese object clitic syntax Regardless of the lack of textual evidence to confirm or disconfirm the hypotheses in the literature, I would nevertheless like to explore three independent facts which strongly suggest that this diachronic change did indeed selectively affect different clitic forms over time. I discuss the first of these here; the second two are addressed in }11.3 (the ‘empirical consideration’ referenced in the final paragraph of this section). To understand the nature of the evidence, let us begin by recalling (example (2)) that the process of change from (A) sole proclisis on the auxiliary (beginning of change), to (C) sole enclisis on the participle (end of change), involved an intermediate stage, namely, (B) the ‘clitic repetition’ stage, whereby the object clitic could appear both proclitic on the inflected verb and enclitic on the participle. I would like to take this fact seriously, and give these three stages of change a theoretical status that can be tapped into. Specifically, I would like to capitalize on the intermediate stage (B). As Parry has shown, the clitic-repetition configuration still exists in some dialects today; it is most notably exhibited in modern Cairese, and—in a revealingly limited way (as we shall see shortly)—it is also still exhibited in modern Torinese. In both varieties, this ‘intermediate’ configuration (stage B) exists alongside the ‘final’

Potentially related to the question of these two different complex predicate types is an observation made by Paola Benincà (p.c.): it has been shown that in Italian and related languages, the infinitive verb moves higher than the participle. This might in turn suggest that, in general, the relative position of the clitic with regard to the infinitive vs the participle must be taken into account, such that the relative position of the clitic vis-à-vis the verb may have different origins, depending on verb form (infinitive or participle). However, in Tortora (2010; in press) I give arguments to support the idea that infinitves and participles embedded under modals and auxiliaries (respectively), as above, exhibit a different syntactic behaviour (e.g. with respect to adverbs) from the same verb forms when they are apparently unselected (e.g. such verbs as subject infinitives or absolute participles). Since the latter contexts are those used to show relative distance of movement of the verb forms, I do not believe that conclusions drawn regarding verb movement in these cases can be readily extended to the complex predicate constructions under investigation here.

Patterns in Piedmontese object clitic syntax

223

configuration (stage C); the two possibilities for Cairese are exhibited in (5a,b) and (5c), respectively:4 Cairese (Parry 2005: 178–9; 229–30) (5) a. A ’m sun fò-me in fazin scl= ocl= I.am made=cl a focaccia ‘I made myself a focaccia’ b. I l’ an catò-le scl ocl they-have bought=cl ‘They bought it’ c. I an rangiò-la scl they-have fixed=ocl ‘They fixed it’ I interpret this syntactic variability as follows. In Cairo Montenotte, there are two different grammars of object clitic syntax possible for speakers: the clitic-repetition grammar and the sole enclisis grammar; thus, the older grammar (B) coexists with the newer grammar (C). A speaker who exhibits both (5b) and (5c) would thus be said to possess two different grammars (in the sense of Kroch 1994); for our purposes, I put aside the question of how the two grammars are sociolinguistically governed (see Parry 2005). The variability between grammar B and grammar C is also exhibited in Torinese, but interestingly, as Parry (1995) illustrates, it is only found with the clitic se in this dialect; this is in contrast with Cairese, where the variability is exhibited with all clitic types. She thus characterizes a modern Torinese structure such as that in (6) as a ‘residual instance’ of the clitic-repetition construction: mod. Tor. (Parry 1995: 138) (6) . . . a s’ é trova-sse ant . . . scl se is found=se in ‘he found himself in an avenue’

na an

leja avenue

Parry’s characterization of the configuration in Torinese (6) as ‘residual’ leads us to the following conclusion: in Torinese, the form se is the only clitic ‘lagging behind’ in the process of change from proclisis to enclisis.

4 The complete Cairese picture, described in Parry (2005), is that all three configurations (A, B, and C) are possible for Cairese speakers. The basis of configuration A in modern Cairese (i.e. sole proclisis on the tensed auxiliary), however, is not clear: it could either be taken to represent the most archaic state of affairs or, alternatively, it could be taken to represent a grammar influenced by contact with Italian and/or other neighbouring regional varieties.

224

Christina Tortora

In an effort to unify these synchronic facts of Torinese with some apparently independent facts of change in OCL syntax in two unrelated Romance varieties (which I consider below, in }11.3), I would like to take this conclusion one step further. Specifically, I would like to suggest that the reason why the form se is lagging behind in this syntactic change in Torinese is because it got a late start. Put differently, we could say that the reason why the other clitic forms only occur in the sole enclisis configuration (stage C) in modern Torinese is that they got a head-start in the process of change; they thus completed the change sooner than se did. In most Piedmontese varieties, se has completed the change, so that the potentially different historic behaviours of the different OCL forms is completely obscured. However, the synchronic facts of Torinese, which is somewhat more conservative in this regard (though not as conservative as Cairese), give us a glimpse into the potentially complex history of OCL syntactic change in Piedmontese.5 Taken by themselves, this interpretation of the facts and the conclusions drawn are admittedly on very shaky ground, especially if we consider the Piedmontese situation in total isolation. However, in the remainder of this chapter I would like to turn to two independent considerations, one empirical and one theoretical, which I believe will make my suggestions here feel less tenuous. In }11.3, I turn to the empirical consideration, followed by the theoretical one. Specifically, a striking fact which cannot be ignored is that Piedmontese is not the only Romance variety in which the form se is an outlier in the game of variation and change in OCL syntax. As we shall see, both Fassano (a Ladin variety) and Spanish—two unrelated Romance varieties—exhibit variation in the placement of OCLs, each in a distinct kind of syntactic construction, and each in a way that implicates the various OCL forms differently. Importantly, despite the fact that we are dealing with unrelated varieties, and also with different kinds of syntactic configurations, the OCL se stands apart from the other OCLs in these cases, just as in Piedmontese. As we shall see, the Fassano facts will call for the kind of syntactic analysis offered by Kayne (2010c) for Spanish; this in turn will set the stage for a fresh approach to the Piedmontese situation, which I return to in }11.4.

5 Cairese is clearly the most conservative of the Piedmontese varieties, in that the change is still in progress with all clitic forms, as is witnessed by the variability described in Parry (2005). The approach offered here thus calls for a systematic, statistical study of Parry’s corpus of Cairese, which could potentially reveal that certain object clitic forms occur in the clitic-repetition configuration more frequently than others. It is possible that the less frequently a particular form appears in the repetition structure (and the more frequently it appears in the sole enclisis structure), the closer it is to completing the change to pure enclisis. Conversely, the more frequently a particular form appears in the repetition structure (and the less frequently it appears in the sole enclisis structure), the further away it is from completing the change.

Patterns in Piedmontese object clitic syntax

225

11.3 Relatable facts in Fassano and Spanish 11.3.1 The case of Fassano infinitive+OCL structures 11.3.1.1 Pattern of variation and change in Fassano Fassano is a group of RætoRomance varieties spoken in the Val di Fassa, in the Trentino-Alto Adige region of northern Italy. This group of dialects exhibits variation and change in OCL placement, in a syntactic context entirely different from that seen for Piedmontese; as such, it stands as a useful point of comparison. My entire description of the facts comes from Rasom (2006), as does the inspiration for the specific theoretical analysis I pursue. As Rasom (2006) notes, in contrast with languages like Italian, the relative order of the OCL and an infinitival verb (which is either a ‘subject infinitive’ or an object of a preposition) in Fassano has traditionally been OCL > infinitive.6 This can be seen in (7), from the Cazet variety of Fassano, spoken in the northernmost part of the valley (the relevant strings are underlined; all examples are taken from Rasom 2006): Cazet (conservative) (7) a. L’ à beù l lat zenza l sciudèr scl= has drunk the milk without ocl= warm.inf ‘He drank the milk without warming it’ b. I se à pissà de ve i manèr per posta, chi

documents scl= self= has thought of ocl= ocl= send.inf by post, those documents ‘Those documents, they thought of sending them to you by mail’

There is, however, evidence for a change in progress, whereby the order infinitive > OCL in certain Fassano varieties is beginning to emerge. This is exemplified in (8), from the Brach variety, spoken in the central and southern part of the valley: Brach (innovative) (8) L’ é miec bever-lo sobito chel cafè lo, 3rd accusative scl= is better drink.inf=ocl immediately that coffee ‘That coffee, it’s better to drink it immediately’ Two facts suggest that the innovative word order seen in (8) represents a change in progress. First, as Rasom notes, this word order is found more frequently among younger speakers than it is among older ones; second, the change is occurring in different Fassano varieties at different rates, such that ‘northern’ varieties are more conservative than ‘southern’ varieties.

6

This word order does not obtain with infinitives that form part of a complex predicate structure (i.e., modal+inf structures). The reader should thus keep in mind that we are dealing with an entirely different kind of structure here.

226

Christina Tortora

Of most interest to us here, however, is Rasom’s observation that this change in progress is not affecting all OCLs at once. Specifically, the OCLs most likely to be enclitic on the infinitive are the 3rd person accusative forms (as in (8)); the next most likely to participate in this change are the 1st and 2nd person accusative forms, illustrated in (9): Brach (9) Asto vedù a scutar-me me! you.have seen to listen.inf=ocl me ‘Have you seen (what happens) listening to me?’

me, 1st acc.

In contrast, obliques and clusters occur very rarely in this ‘new’ configuration; this is illustrated in (10): Brach (10) a. No se pel te domanar neg self= can ocl= ask.inf ‘You can’t be asked anything!’ a.' *?No se pel neg self= can

domanar-te ask.inf= ocl

nia! nothing *?dative

nia! nothing

b. Recordete de ge telefonar per temp. remember=you to ocl= phone.inf for time ‘Remember to telephone him in time.’ b.' *?Recordete de telefonar-ge per temp remember=you of phone.inf=ocl for time

*?dative

Thus, if we compare (10b) with (10b'), for example, we see that speakers are more likely to judge the order infinitive > OCL to be infelicitous with a dative clitic; as I understand Rasom’s (2006) description, however, this order is not completely rejected by all speakers, so it is not impossible in the most innovative of contexts. This contrasts with the final class of OCLs, namely, reflexives: as Rasom notes, reflexive OCLs are never enclitic—not even in the most innovative of varieties. As the example in (11) shows, they always appear in the conservative OCL > infinitive order: Brach (11) Dant de marena se cogn se lavar before of lunch is necessary ocl(self)= wash.inf ‘Before lunch you have to wash your hands’ (11') *Dant de marena se cogn lavar-se le man. before of lunch is necessary to.wash=ocl(self ) the hands

le the

man. hands *reflexive

Patterns in Piedmontese object clitic syntax

227

To summarize: in the change in progress in the Fassano varieties, the first OCL to change from proclisis to enclisis is the 3rd person accusative (lo, la, li, les); the next most likely is the 1st and 2nd person accusative (me); the next most likely are datives (ge, me, te), and the one OCL which has yet to become enclitic in Fassano is the reflexive (se), dative or accusative. This hierarchy is laid out in (12): Hierarchy of change in progress in Fassano varieties7 (12) 1. 3rd accusative (lo, la, li, les) 2. 1st/2nd accusative (me, te) 3. dative (ge, me, te) 4. reflexive (se) 11.3.1.2 Accounting for the OCL hierarchy in Fassano variation and change Given the above description of the facts, two questions arise. First, what underlies this pattern of variation and change in OCL syntax in Fassano? The second (related) question is: is the Fassano phenomenon relatable to other cases of variation and change in Romance OCL syntax? I argue that the key to understanding the pattern of change in Fassano finds itself (a) in Rasom’s own characterization of the phenomenon, and (b) in Kayne’s (2010c) analysis of an apparently unrelated phenomenon in some Spanish dialects. Let us begin with the first question. Rasom (2006: 112) characterizes the syntactic change in Fassano in terms of verb movement; specifically, she takes the move towards enclisis of the OCL to reflect infinitival verb movement past the OCL. This idea is sketched in (13): (13)

ocl

verbinf

But how does this idea help us make sense of the complex array of facts? The idea that the verb moves past some OCLs but not others suggests the following: the different clitic forms occupy distinct functional heads within the rigidly ordered functional hierarchy of the clause, within the stretch of functional architecture which we could term the ‘clitic placement domain’.8 This hypothesis, which I will call the Functional Hierarchy Hypothesis (FHH) here (and which has been independently argued for by

7 As pointed out by Adam Ledgeway (p.c.), a question arises as to whether the loss of object clitics in Dolomitic varieties, as investigated in Paoli (2009), follows the same hierarchy. While I leave this question as a matter for future research, it is interesting to note (as discussed in Tortora in press) that Pons (1990) observes loss of the OCL lò (3rd sing. fem.) in imperatives in Valdese Waldensian (an Occitan variety spoken in Valdese, North Carolina). 8 It is important to note here that I take the ‘clause’ in these Fassano cases to be the immediate series of functional projections projected by the embedded infinitival verb. Thus, (11) for example would have two series of clausal projections: those related to the main clause (clause1: se cogn se lavar le man), and those related to the embedded infinitive (clause2: se lavar le man):

(i)

[clause1 se cogn [clause2 se lavar le man] clause1]

228

Christina Tortora

e.g. Manzini and Savoia 2005), gives rise to a situation in which the morphological forms themselves are arranged hierarchically within the syntactic structure.9 If we adopt this hypothesis, the facts of variation and change in Fassano specifically suggest the hierarchy in (14): (14)

reflexive (se) > dative (ge) > 1st/2nd accusative (me, te) > 3rd acc (lo, la, li, les)

This is essentially the reverse of the hierarchy in (12), with 3rd accusative OCLs in the lowest position within the clitic placement domain, and the reflexive clitic in the highest position. The syntactic order in (14) follows from the idea that verb movement is upward, and that the 3rd person accusative forms are the ‘first to be jumped over’ by the verb (Rasom 2006: 112). And since reflexives never exhibit enclisis in this context, we must conclude that reflexive OCLs occupy the highest head—a head which the verb never reaches. This successive series of potential (or impossible) movements of the infinitival verb, as they are realized (or not) in the Fassano varieties, is illustrated in (15): (15)

se

>

∗

ge

>

me/te >

3rd acc (lo, la, li, les)

VERBinf

∗?

This illustration reiterates the syntactic hierarchy in (14), but also shows the licitness (or lack thereof) of the successive movements of the verb upward. The movements past (a) the lowest set of OCL forms (the 3rd accusative OCLs) and (b) the secondlowest set of OCL forms (me/te) are permitted in the most innovative varieties. The *? in (15) indicates that the third move upward is less licit, while the * indicates that no Fassano grammar has yet allowed for the fourth step upward, past the form se. A few comments are in order before I summarize and turn to the next subsection. First, note that if the verb moves as a head, the idea that the Fassano infinitival verb can move successive cyclically through the clitic placement domain (as in (15)) suggests that there is a different head position between each of the head positions occupied by the different OCL forms. In other words, for every FP harbouring an OCL, there should be another FP immediately dominating it, which can host the verb. Needless to say, this represents a proliferation of functional projections within the clitic placement domain which I do not otherwise have evidence for. A possible alternative analysis is to take verb movement to be phrasal, whereby the mobile XP dominating the verb (and pied-piping it along) moves to the specifier positions of the

9

This does not preclude the possibility that one and the same form (e.g. te) would occupy distinct positions, depending on its function (e.g. dative or accusative). See Benincà and Tortora (2010) and Tortora (in press) for further discussion of this idea.

Patterns in Piedmontese object clitic syntax

229

FPs projected by the heads hosting the OCLs. Another possibility (pointed out by Adam Ledgeway, p.c.) could be that the verb itself left adjoins to the to the OCL. I will leave this question open. Second, the above analysis predicts that ‘split clitics’ are possible in the innovative Fassano varieties. In other words, in a structure where both dative me and accusative lo (for example) are present, we predict that the first singlestep movement of the verb in diachronic change should give rise to the word order me > verbinf > lo. Problematically, this prediction does not seem to be borne out. If the FHH is on the right track, then, we must find some independent reason why split-clitic situations of the kind just illustrated are not possible. Again, I leave this matter open, though see Pescarini (Chapter 8 in this volume) for further discussion. Despite the problems noted above, the present proposal has strong theoretical appeal, in that it allows us to make precise predictions which entail a correlation between the variation and change in progress seen with Fassano OCL syntax on the one hand and an independently established OCL hierarchy on the other; the theory thus finds its strength in its falsifiability. Regarding these predictions, the question thus arises as to whether there is independent evidence for the hierarchy hypothesized in (14). That is, if this hierarchy—which was suggested by the variation and change facts described in }11.3.1.1—is indeed correct, then we expect to find only certain kinds of clitic orderings in clitic cluster structures (i.e. structures in which there is more than one clitic). For example, 3rd person dative ge should precede 1st or 2nd person accusative me/te when both are present in the same structure. While a full paradigm of the clustering possibilities has yet to be established for Fassano, there are two observations we can make which serve as preliminary evidence that the ordering facts in clusters will independently yield the hierarchy established in (14). First, we have already seen in example (7b) that the dative (ve ‘you.pl’) precedes the 3rd person plural accusative clitic i, which is consistent with our expectations. Second, as we shall see in }11.3.2, there is a striking similarity between the Fassano hierarchy established in (14) and the hierarchy independently established in the literature for Spanish. 11.3.2 Evidence of an OCL hierarchy: variation in Spanish dialect OCL syntax Although a full paradigm of the clustering possibilities has yet to be established for Fassano, it is encouraging to note that the hypothesis for Fassano is readily relatable to Kayne’s (2010c) analysis of an apparently independent phenomenon found in some Spanish dialects, which was described and analysed in detail by Harris and Halle (2005), within a Distributed Morphology framework. As discussed extensively in Harris and Halle (2005) and Kayne (2010c), many Spanish dialects (represented both in Spain and Latin America, as well as by JudaeoSpanish) exhibit a non-standard behaviour of the 3rd person plural inflectional suffix -n in imperatives (which are formally 3rd plural). To understand the non-standard

230

Christina Tortora

form, let us first consider the imperative in the standard variety (which again is formally 3rd person plural, but interpreted as 2nd person plural): (16)

¡Hagan el trabajo! (st. Sp.) do.3pl the work ‘(you.pl) Do the work!’

While the form in (16) is interpreted as a 2nd person (plural), the verb is analysable as a 3rd person plural form, with the inflectional suffix -n taken to be the morpheme instantiating the features 3rd person and plural. Expectedly (for the standard variety), when a complement clitic is present in an imperative of this type, it follows the -n ending: (17)

¡Sirvan-se! (st. Sp.) serve.3pl=ocl ‘(you.pl) Serve yourselves!’

It turns out, however, that in many Spanish dialects, the OCL can precede the inflectional suffix [-n] in this type of imperative, as follows: (18)

¡Sirva-se-n! (Sp. dialects) serve=ocl=3pl ‘(you.pl) Serve yourselves!’

Let us put aside here the numerous additional interesting facts and variations on the theme seen in (18) (see Harris and Halle 2005 for complete discussion), and turn directly to the fact that is of prime importance to Kayne (2010c): of all the varieties which allow this non-standard placement of the OCL to the left of the inflectional suffix -n, most only allow the phenomenon with the reflexive OCL se. Less common are the speakers/varieties that allow this construction also with 1st and 2nd person clitics, and less common still are those which allow it also with datives; finally, it is most rare with 3rd person accusative clitics, as in (19b): (19)

a. ¡Vendan-lo! (st. Sp.) sell.3pl=ocl ‘(you.pl) Sell it!’ b. ¡Venda-lo-n! (Sp. dialects) sell=ocl=3pl ‘(you.pl) Sell it!’

Importantly, then, there is an entailment whereby if a variety allows this construction with a 3rd person accusative clitic, as in (19b), then it necessarily allows it with all the other OCL forms. Likewise, if a variety allows the construction with dative forms, it does not necessarily follow that 3rd accusative OCLs are licit, but it does necessarily

Patterns in Piedmontese object clitic syntax

231

follow that 1st and 2nd and reflexive forms are licit. This hierarchy of entailments gives rise to four different possible grammars, as in (20) (modified from Harris and Halle 2005: 210, ex. (25)): Spanish (20) a. b. c. d.

se+n se, me+n se, me, le+n se, me, ledat, lo, la+n

Grammar A Grammar B Grammar C Grammar D

We can make two observations about (20): first, note that the sequence of OCLs in Grammar D is highly reminiscent of the Fassano OCL hierarchy in (14), repeated here as (21):10 (21)

Fassano se > gedat > me > lo / la

Second, and just as importantly, the hierarchy in Grammar D in (20) is exactly that which is independently established by examining clitic ordering in Spanish (cf. Perlmutter 1978). In fact, this second observation suggests to Kayne (2010c) a syntactic analysis of the phenomenon seen in (18) and (19b), which involves a hypothesis that is exactly along the lines of the one I proposed in }11.3.1.1 for Fassano, whereby the FHH is claimed to be at work. Specifically, following independent observations made by Parry (1997) and Zanuttini (1997) regarding the relative ordering of OCLs and preverbal negative markers in some northern Italian dialects, Kayne proposes that the reflexive OCL occupies the highest head within the rigidly ordered functional hierarchy of the clitic placement domain, and the 3acc OCL occupies the lowest. He thus interprets Grammar D in (20) as revealing an underlying syntactic hierarchy in Spanish, as follows: (22)

Spanish (cf. (21)) se > me > ledat > lo / la

Kayne then exploits this hypothesized hierarchy to account for why reflexive se would be the OCL most likely to appear in the non-standard imperative construction in Spanish. Specifically, he further assumes (a) that the 3rd person plural suffix -n resides in an independent syntactic head, and (b) that the OCL se is the most likely to appear to its left because it is the highest in the OCL hierarchy. The analysis is depicted in (23):

10

The one (obvious) difference between the Spanish hierarchy and the Fassano hierarchy is that in the former, the 1st person clitic precedes the dative (le), while in the latter, the 1st person clitic follows the dative (ge). I put this difference aside.

232

Christina Tortora

(23)

FP Spec

F' F sirva-

GP Spec

G' G se

HP Spec

H' H n

...

I refer the reader to Kayne (2010c) for further details of his analysis of the Spanish imperative, and move on to a summary of the main point of this section. We have seen that a completely independent fact of cross-dialectal variation in complement clitic syntax in Spanish varieties reveals a hierarchy of behaviour among the different morphological forms which looks strikingly like the hierarchy of behaviour among the different morphological forms in the Fassano varieties. Given that the variation and change in the two different Romance families (Spanish and Fassano) clearly cannot be attributed to any contact between the two communities, and given that the variation and change in the two families implicates two completely different construction types, I am led to conclude that something deeper is at play here, namely, an underlying universal fact about grammar. Specifically, these two apparently independent phenomena are arguably the consequence of a single property of the syntax—the rigid functional hierarchy of the clitic placement domain, i.e. the FHH.

11.4 Diachronic change in Piedmontese object clitic syntax revisited: the Functional Hierarchy Hypothesis This brings us back to Piedmontese, and the discussion from }11.2. Now that we have seen (a) that the form se is the odd man out in variation and change in OCL syntax in Fassano and Spanish (two different Romance types)—in two different construction types—and (b) that the FHH makes solid predictions, and captures the facts of variation and change, we must ask ourselves whether this can be brought to bear on the Piedmontese facts. In this section, I discuss how it can, by presenting a series of independent but interconnectable hypotheses, some already previously established in the literature. I consider this part of the chapter to stand as more of a set-up for

Patterns in Piedmontese object clitic syntax

233

future research, because the discussion and analyses I offer raise questions that would have yet to be answered through further exploration. I nevertheless offer these observations and analyses with the conviction that this is a promising avenue of inquiry for Piedmontese, especially since, as we shall see, the hypotheses I adopt make very clear predictions. 11.4.1 OCL ordering in Piedmontese Let us begin with the assumption that the FHH is valid for all Romance languages; in addition to Kayne (2010c), see e.g. Manzini and Savoia (2005) for other Italian dialects, and Săvescu (2007) for Romanian (and cf. Poletto 2000 for this perspective with respect to subject clitic ordering). Thus, by well-supported hypothesis, the ordering of Piedmontese OCLs should reflect the FHH. Let us further assume that the ordering of the functional heads in the clitic placement domain is consistent across Romance varieties (with the caveat given in fn 10). Thus, we can take the ordering in Piedmontese to be along the lines of that in (24):11 Hypothesized OCL ordering in Piedmontese (24) reflexive (se) > dative > 1st/2nd accusative

> 3rd accusative

The question which now arises is how the ordering in (24) can help us understand the behaviour of the form se as an ‘outlier’, in the historic change in Piedmontese OCL syntax discussed in }11.2. Before I address this, I must first address the apparently independent question of what accounts for the difference between Italian-type languages on the one hand (proclisis on the auxiliary), and modern Piedmontese languages on the other (enclisis on the participle), in the compound tenses. Once we accomplish this in }11.4.2, in }11.4.3 we can return to the question of how the FHH in (24) helps us with Piedmontese. 11.4.2 The structure of the compound tense in Romance and post-participial OCL placement in modern Piedmontese 11.4.2.1 The compound tense in Romance In order to get closer to an understanding of the behaviour of OCLs in the Piedmontese compound tense, we must first establish a reasonably well-motivated structure for the compound tense. Here, I shall adopt the hypothesis of ‘light bi-clausality’ offered in Tortora (2010; in press), whereby auxiliary + past participle structures in Romance involve two sets of independently projected FPs—one associated with the matrix auxiliary, and the other associated with the embedded participle. This hypothesis for Romance is illustrated in (25), where the X,

11

This calls for independent verification of this ordering by looking at the patterns of OCL cooccurrence, à la Perlmutter (1978). Note too that I am assuming the same ordering for all Piedmontese varieties.

234

Christina Tortora

Y, X, etc. serve as shorthand for the FPs projected independently by the auxiliary on the one hand, and the past participle on the other; the bolded X, Y, Z, W, U in (25) represent the FPs projected by the embedded participle (PtP):12 (25) [Clause1 . . . Auxk . . . X Y Z W U [VP tk [Clause2 . . . X Y Z W U [VP PtP OCL]]]] 11.4.2.2 The post-participial placement of the OCL in modern Piedmontese In the structure in (25), I have also depicted the merge site of the OCL. As argued in Tortora (2010; in press), in modern Piedmontese varieties (where the OCL is enclitic on the participle in compound tense structures), the OCL only moves as far as the participial Z head (a functional head which hosts the clitic). This movement is depicted in (26), where the OCL adjoins to Z (and the PtP adjoins to the preceding Y head); this is the structure which underlies example (3) in }11.2:13 (26)

[Clause1 . . . Auxk . . . X Y Z W U [VP tk [Clause2 . . . X PtPk+Y CLj+Z W U [VP tk tj ]]]]

As already established earlier in this chapter, this OCL placement in modern Piedmontese contrasts with that of Italian (and also with ‘stage A’ of historic Piedmontese), where the OCL is proclitic on the matrix auxiliary, as in (1) in }11.2. A question which arises then is what accounts for these two different OCL placement strategies in Italian vs modern Piedmontese (i.e. pre-auxiliary vs post-participial). In Tortora (in press), this is but one question which forms part of the more overarching inquiry into what regulates the many different OCL placement patterns seen across Romance and across different construction types. Here I shall adopt the system offered in that work, as not only does it account for the many OCL placement patterns seen across Romance, as well as for an important cross-linguistic entailment, but (as we shall see shortly) it also makes precise predictions regarding the historic change in Piedmontese OCL placement, discussed in }11.2. And as we shall further see, the predictions are consistent with what we have seen regarding the behaviour of the form se. For reasons of space, I very briefly summarize the essentials of the system here (for full disclosure of all of the issues involved, and how the theory relates to earlier theories of clitic placement, see Tortora in press). The theory of OCL placement I adopt here takes as fundamental a hypothesized mechanism of ‘feature harmony’ within the clause, which is claimed to obtain independently of the question of clitic syntax.14 Here I focus just on the concept of 12 See Tortora (2010; in press) for detailed empirical arguments supporting this hypothesis, and for fleshed-out discussion of the numerous empirical facts that this, and the hypotheses to follow in this section, capture across Romance; unfortunately, I must put all of this aside here for reasons of space. 13 I ask the reader to momentarily put aside the fact that (26) depicts only one generic adjunction site for the OCL, namely, the participial Z head. As we shall see in }11.4.3, this conflated Z position will be expanded into the articulated clitic placement domain seen in (24). 14 See Blanchette (2012; 2013) for a neg Feature Spreading account of Negative Concord in English (i), which also allow for double negation readings (as in (ii)) (a fact of English which is otherwise overlooked and unaccounted for in the syntax literature):

Patterns in Piedmontese object clitic syntax

235

harmony of the feature [finite]. The idea is that all of the functional projections of a clause can potentially (under the right conditions) acquire the feature [finite]. The clause’s FPs acquire this feature by virtue of a mechanism of ‘feature spreading’, whereby the T-head in the Inflectional domain copies the feature [finite] to the next lower head, F1, and then F1 copies this feature to the next lower head F2, and so on. This spreading of the [finite] feature across all functional heads of a compound tense clause in Italian is illustrated in (27): (27)

Italian

[cp1 [tp T[finite] [fp1 F1[finite] ... [xp X[finite] ... [zp Z[finite] ... [vp MATRIX CLAUSE

[Clause2 X[finite] Y[finite] Z[finite] ... [vp ...OCL PARTICIPIAL CLAUSE

(The long rightward pointing arrow is intended to illustrate the spread of the feature [finite], from its source (the T head), all the way down the structure. The relevance of the circled Z head in the participial clause will become apparent immediately.) Although this feature-spreading mechanism is claimed to obtain independently of the mechanism of OCL placement, it turns out that it has a fortuitous effect on OCL syntax. Specifically: by hypothesis, an OCL cannot adjoin to a head that has the feature [finite] associated with it. Thus, although the participial Z head circled in (27) is a ‘potential’ OCL host, it cannot serve as the host to any OCL moving up the structure in search of a hospitable functional head to adjoin to. As such, in a language like Italian, the OCL must continue to move up the structure until it finds a suitable host head. The first functional head it finds which is not laden with the [finite] feature is the one immediately c-commanding the T head (not depicted in (27)). This gives rise to the effect of proclisis on the auxiliary in the compound tenses. Under the Tortora (in press) system, the difference between Italian-type languages on the one hand and modern Piedmontese-type languages on the other is as follows: by hypothesis, the left edge of the participial clause in Piedmontese acts as a ‘barrier’ to feature spreading.15 This is depicted in (28), where the vertical line is intended to illustrate this ‘barrierhood‘ of the left edge of the participial clause:

(i) He didn’t have no ax. Negative Concord ‘He didn’t have an ax.’ (i) He didn’t feed the baby with no hair. Double Negation ‘The baby with no hair went unfed.’ 15 The word ‘barrier’ here does not have the same meaning as it does in Chomsky (1986); nevertheless, it is meant to informally capture the notion of local domain. See Tortora (in press) for discussion of the ways in which the proposed system is not a direct translation of earlier ‘clause union’ or ‘restructuring’ theories, which also had as a goal the explanation of ‘transparency effects’ such as clitic climbing. As discussed in that work, there is a range of facts which are captured by the feature-spreading hypothesis, which are not captured by earlier clause union/restructuring theories.

236

Christina Tortora

(28)

Piedmontese

[cp1 [tp T[finite] [fp1 F1[finite] ... [xp X[finite] ... [zp Z[finite] ... [vp

[Clause2 X[...] Y[...] Z[...] ... [vp ...OCL

MATRIX CLAUSE

PARTICIPIAL CLAUSE

Recall that—also by hypothesis—an OCL cannot adjoin to a head that has the feature [finite] associated with it. However, as can be seen in (28), our ‘potential clitic host’, i.e. the circled participial Z in (28), does not contain the feature [finite] (as this feature never spread into the participial clause to begin with). As such, this head is able to host the OCL; so when the OCL moves from its base position, it finds the first Z head to be a suitable host, and stays there. This gives rise to the effect of enclisis on the participle in the compound tenses in modern Piedmontese, as in (3) in }11.2. Before moving on to the final section, let us recall that the analysis of the Italian seen in (27) would also be the analysis of historical Piedmontese, given that, as we saw in }11.2, Piedmontese started out like Italian. This would mean that in the diachronic change described in }11.2, Piedmontese went from the configuration in (27) to the configuration in (28). Let us characterize this diachronic change from (27) to (28) as the retraction of feature spreading, or, ‘feature-spreading retraction’. As such, (27) and (28) should be understood as old Piedmontese and modern Piedmontese, respectively, with (28) representing the result of feature spreading retraction.16 In the next section, I show how this feature spreading retraction, combined with the FHH, allows the Piedmontese facts to fall out. 11.4.3 Diachronic change in Piedmontese OCL syntax and the case of se Now that our very abridged recounting of the Tortora (in press) account of Italiantype (= old Piedmontese-type) vs modern Piedmontese-type OCL syntax is behind us, let us return to the question of how we might use the FHH to account for the diachronic change in Piedmontese OCL syntax discussed in }11.2. The idea here is to show how the Piedmontese facts fall out from a combination of the FHH and the independent proposal summarized in }11.4.2. To do this, let us recall the hypothesized OCL hierarchy in Piedmontese in (24): Hypothesized OCL ordering in Piedmontese (24) reflexive (se) > dative > 1st/2nd accusative

> 3rd accusative

16 ‘Feature-spreading retraction’ should be understood in terms of a change in status of the left edge of the participial clause; i.e. in old Piedmontese in (27), the left edge of Clause2 was not a barrier to feature spreading, while in modern Piedmontese in (28), it is. I leave as a matter for future research the question of how this change in status of the left edge would come about, or of what other syntactic phenomena it might be related to (e.g. participial verb movement). See however the discussion in }11.4.3.

Patterns in Piedmontese object clitic syntax

237

Given the FHH, we must now revisit the independent proposal in }11.4.2 with an eye towards refining our notion of ‘OCL placement head’. That is, the participial Z head referenced in }11.4.2.2 should now be taken as a shorthand for a more articulated clitic placement domain, in which different heads specialize for different clitic forms, where the heads reflect the hierarchy in (24). For the purposes of exposition, let us simplify the matter, so that we only concern ourselves with three functional heads in the clitic placement domain, as follows: (29)

Z3 (head for se)

>

Z2 (head for me)

>

Z1 (head for lo)

This articulated representation, which takes participial Z in (25) to really be a field of specialized OCL placement heads, forces us to recast (25) as follows (with the relevant string of OCL placement heads highlighted in the box): (30)

[Clause1...AUXk...XYZWU [vp tk [Clause2...XY Z3 Z2 Z1 W U[vpPtP OCL]]]

With the structure in (30) in place, let us recall the observation made at the end of }11.4.2.2, namely, that the Italian state of affairs seen in (27) would also be the analysis of the historical Piedmontese state of affairs, given that Piedmontese started out like Italian. Let us thus repeat (27) and (28) here, in slightly revamped form, as old Piedmontese (31) and modern Piedmontese (32), and also with the articulated clitic placement domain boxed off as Z3–Z2–Z1 seen in (30): (31)

old Piedmontese

[cp1 [TP T[finite] [FP1 F1[finite] ... [VP MATRIX CLAUSE

(32)

[Clause2 X[finite] Y[finite] Z3[finite] Z2[finite] Z1[finite] ... [vp ...OCL PARTICIPIAL CLAUSE

modern Piedmontese

[cp1 [tp T[finite] [fp1 F1[finite] ... [vp [Clause2 X[...] Y[...] Z3[...]Z2[...] Z1[...] ----------------------------> MATRIX CLAUSE PARTICIPIAL CLAUSE

... [vp ...OCL

Now, returning to the idea at the end of }11.4.2.2 (and in fn 16), where modern Piedmontese (32) is taken to represent the completion of feature spreading retraction: let us imagine that feature spreading retraction is gradual, so that it happens in stages. We could imagine (31) as representing Stage 0, so that the first real stage of change, Stage 1, would have the ‘barrier’ to feature spreading appear at the left edge of Z1, as follows: (33)

Stage 1 of change in Piedmontese OCL syntax

[cp1 [tp T[finite] [fp1 F1[finite] ... [vp [Clause2 Z3[finite] Z2[finite] Z1[...] ... [vp

238

Christina Tortora

In Stage 1 of the change from proclisis to enclisis in Piedmontese, the potential OCL placement head for lo, Z1, fails to acquire the feature [finite], as a result of the first step of feature spreading retraction. Consequently, Z1 can host lo (by hypothesis; recall }11.4.2.2). However, Z2 and Z3 are still unavailable for OCL placement of me and se, respectively (recall (29)). This theory thus predicts that lo would have been the first OCL to be enclitic on the participle in Piedmontese. Let us put aside the fact that we have no textual evidence that reveals whether this was actually the case. For now it suffices to say that this is a precise prediction. Assuming that feature spreading retraction continues over time, our next stage of change, Stage 2, would be as follows: (34)

Stage 2 of change in Piedmontese OCL syntax

[cp1 [tp T[finite] [fp1 F1[finite] ... [vp [Clause2 Z3[finite]

Z2[...] Z1[...] ... [vp

In (34), we see that in the next stage of change from proclisis to enclisis in Piedmontese, the potential OCL placement head for me, Z2, fails to acquire the feature [finite], as a result of further feature spreading retraction. Consequently, Z2 is able to host me. However, Z3 is still unavailable for OCL placement. The theory thus predicts that me would have been the second OCL to be enclitic in Piedmontese; it also predicts that enclisis of me would have entailed enclisis of lo. Again, let us put aside the fact that we have no textual evidence that reveals whether this was actually the case. For now, it suffices to say that this is a precise prediction. Finally, still assuming that feature spreading retraction continues over time, our next (and final) stage of change, Stage 3, would be as follows: (35)

Stage 3 of change in Piedmontese OCL syntax

[cp1 [tp T[finite] [fp1 F1[finite]... [vp

[Clause2 Z3[...] Z2[...]Z1[...] ... [VP

In this hypothesized final stage of the change from proclisis to enclisis in Piedmontese (which corresponds to (32)), the potential OCL placement head for se, Z3, fails to acquire the feature [finite], as a result of further feature spreading retraction. Consequently, Z3 is able to host se. This gives rise to the effect of se being the last OCL to be enclitic in Piedmontese, with the enclisis of se entailing the enclisis of me and lo.

Patterns in Piedmontese object clitic syntax

239

11.5 Conclusions As discussed in }11.2, we have no textual evidence which reveals whether it was actually the case that se was the last OCL to be enclitic in Piedmontese. However, thanks to Mair Parry’s approach to the problem, we do have evidence from synchronic variability in modern Torinese which supports this conclusion. Specifically, we saw in }11.2 that the form se is the one OCL form which can occur in the earlier, clitic-repetition configuration in modern Torinese. Recall that this fact led us to conclude that se must have been the last OCL form to appear enclitic on the participle in compound tense structures. It is encouraging that the confluence of independently motivated hypotheses discussed in this section makes precisely this prediction. There are, of course, many outstanding issues which this chapter leaves open for future work (including those already mentioned). First, the theory of clitic placement I have proposed is intended to account for the variation between proclisis on the auxiliary (high placement) vs enclisis on the non-finite verb (low placement, within the participle’s own clausal domain). I have not discussed, however, how the clitic-repetition structure—the ‘intermediate stage’ of change—fits in with this theory. If nothing else is stated, the clitic-repetition structure should in fact not be possible. Here I offer some preliminary thoughts on the issue, which admittedly have yet to be worked out. In particular, I take the clitic-repetition structure to be a surface reflex of low clitic placement, i.e. adjunction of the OCL to the participial Z head. Thus, like the final stage of evolution (where there is sole enclisis on the participle), feature spreading has retracted, such that the low Z head does not acquire the feature [finite], thus allowing for adjunction of the OCL to it. The question which arises under this view is the nature of the clitic in pre-auxiliary position. Here, I would suggest that, in contrast with the ‘first stage’ configuration (with sole appearance of the OCL to the left of the auxiliary), this pre-auxiliary clitic in the clitic-repetition stage is not moved. Rather, I take it to be merged in this position. This means that this higher instance of the clitic is not the argument itself, in this structure. Clearly, this turns on its head another possible analysis of these structures (one which I do not adopt), namely, an analysis which takes the lower clitic to be a ‘copy’ of the higher (moved) clitic. These different approaches need to be compared, and their predictions need to be tested. Second, as is surely clear to the reader at this point, I have not established any principle which predicts when a particular ‘juncture’ in the clause will act as a barrier to [finite] feature spreading (the left-edge of the lower functional field? the left edge of the participial clause?). I would like to suggest, however, that it is possible that the ‘barrierhood’ status of particular junctures in the clause could be derivable from other independent syntactic phenomena, which do or do not render left edges of certain clausal domains ‘phase edges’. Verb movement is one place to look for this (as

240

Christina Tortora

the Romance languages notoriously differ with respect to how high verbs move, as noted in fn 3). As such, the present proposal has the promise of reducing OCL syntax to more general, and independent, principles of the grammar, and as noted in fn 14, these independent principles give rise to other apparently unrelated syntactic phenomena, such as Negative Concord. Another issue which needs to be explored is how the different OCL forms behave, in their different functions. As suggested in fn 9, a more refined examination of the facts is required, in order to determine whether the FHH can/must capture not only the form of the OCL but also its particular function. In this regard, more needs to be said about, for example, the form se, which can function not only as a reflexive, but also as an impersonal. As discussed in Tortora (in press), there are reasons to believe that this form, in its different functions, may occupy distinct functional heads, in different Romance varieties. If this is true, this may have consequences for the stages of change in the syntax of these different forms, the facts of which need to be better understood.

12 Gender assignment and pluralization in Italian and the Veneto* JOHN B. TRUMPER

12.1 Recent work on the theme One of the aims of our current research is to look for adequate means of capturing generalizations about gender assignment and assignment by default in modern Tuscan-based Italian and Italo-Romance dialects. The underlying question for such research is what kind of gender assignment model would be needed if one were to seek to describe more than just gender assignment processes in modern Italian. One obviously starts with a more modest goal, that of analysing present-day and historical Tuscan and neo-Venetian assignment. Maiden (1995: 116–17), addressing specifically Italian problems, had exhumed the arbitrariness hypothesis on gender assignment to nouns, in other words Italian continues Latin arbitrary gender assignment. At the same time he underlined an important principle, that of ‘assignment by default’, according to which there exists an unmarked gender to which loans, neologisms, nominalized non-noun categories, and the like, are automatically assigned. There has also been more general research seeking to identify principles which have cross-linguistic applicability. Such work may be summarized as follows. Corbett and Fraser (2000) theorize that assignment by semantic rule takes precedence over morphological and phonological classes, though most concede that this theory is too strong, and one would need, in any case, a sort of Semantic Override Principle for framing assignment principles. On the other hand, Dewaele and Véronique (2001) take up the cudgels against the hypothesis that rules are entirely semantic, as they do against hypotheses in favour of the completely idiosyncratic nature of gender assignment (p. 276). They emphasize (1) that, following Comrie (1999), criteria must be both semantic and formal; and (2) that incorrect gender assignment is extremely rare * Anrheg i Mair deg, am y gwaith a gysegrodd i dafodieithoedd yr Eidal. ‘A gift for fair Mair, for the work which she dedicated to the dialects of Italy.’

242

John B. Trumper

in the case of natives but frequent in non-natives and learners.1 The problem comes down to how rule-oriented assignment is labelled for storage. Experimental reaction times for speakers’ responses to assignment points to the exclusion of purely phonological influences, because gender selection is already present in lexical lemmatization and can be immediately retrieved before phonological processes intervene. They conclude that default assignment must be included in morphology, following Fernández González (1985), Maiden (1995), Dressler and Thornton (1996), and Comrie (1999), and that the default gender is the masculine, which overrides both feminine and neuter according to a scheme Masculine
Feminine
Neuter. Thornton (2001: 479) is also convinced that gender assignment is rule-governed but language-specific, and takes up the class idea she had previously formulated, showing that default assignment (namely, masculine overrides other genders) will account for four subclasses: (1) truncation, (2) clipping, (3) the nominalization of non-nouns (viz. of non-nominal grammatical categories), (4) purely semantic assignment as in feminine la jeep ‘jeep’, la Panda ‘Fiat Panda car’ vs masculine il suv (< Eng. sport utility vehicle) ‘four-wheel drive’, il panda ‘panda bear’. One might object that case (4) is not really semantically based, but involves assignment based on the gender of a deleted head. For cars with the understood head macchina (f) ‘car’, this seems to work, not only in cases such as la jeep, la fuoriserie ‘custom-built car’, la Panda, and la Lancia, but even in cases such as la Cinquecento, la Fiat, la Punto, la Tipo. With the masculine il suv, il fuoristrada ‘fourwheel drive’, il BMW, and il Mercedes there would seem to be an underlying macchinone (m) ‘large car’ or something similar involving size and/or complication (Marta Maddalon, p. c.). Similarly, orsetto (m) ‘little bear’ might be underlyingly present in il (orsetto) panda. These patterns would thus enter into the typology of cases (1) þ (2) (þ (4)), insofar as they involve an abbreviated noun phrase rather than an abbreviated noun. That lexico-semantic classes do exist is shown by the fact that: (a) rivers are masculine (il Brenta ‘the Brenta’, with deleted head fiume (m) ‘river’);2 and (b) towns and cities are feminine (Torino, Varese, Lodi, Ancona all require 1

In Italian, even on formal occasions, wrong assignment may occur in statements by native speakers: during the 20.45 RAI News, TG2 of 26.12.2010, a TV announcer came up with ‘la [fsg] proverbiale aplomb dei britannici’ for ‘il [msg] proverbiale aplomb dei britannici’ (‘the [fsg/msg] proverbial British aplomb’). Frequent cases occur in regional broadcasts and news programmes, whether by RAI or commercial broadcasters. It seems to occur most frequently in borrowings or acronyms with an initial vowel—thus a well-known political party L’Udc (= L’Unione dei democratici cristiani e di centro ‘the Christian Democrats and Centre Union’) may be either masculine or feminine despite the feminine gender of the head noun (unione). L’aplomb may well be ambiguous, notwithstanding standard dictionaries. See also Conzett (2006: 237), who notes that ‘[u]ncertainty and instability as to the gender of new nouns are richly attested in the literature’. 2 In Neo-Venetian, Venetan dialects, and regional Italian they are, however, all feminine if major rivers (La Brenta, La Sile, La Piave), whereas derivatives in -one (Il Bacchiglione), secondary rivulets, and minor streams (Il Tergolino), and canals (Il Piovego) may be masculine. This produces intraregional variation in Veneto regional Italian.

Gender assignment and pluralization in Italian and the Veneto

243

feminine morphological agreement with the deleted feminine head città ‘city, town’). However, football teams, like motorized vehicles, are ambiguous, apparently depending on the deleted head (cf. club (m), squadra (f) ‘team’), e.g. il Milan (m) vs la Lazio (f). Administrative regions take the gender of their endings, hence il Veneto (msg), la Lombardia (fsg) ‘Lombardy’, the now obsolete gli Abruzzi (mpl) ‘Abruzzi (= Abruzzo and Molise)’, le Marche (fpl) ‘the Marches’, while those in -e (Piemonte ‘Piedmont’, Molise) or -i (Friuli) take masculine gender by default. Borrowings are considered to be essentially genderless, so that their gender assignment is possibly decided: (a) on phonological grounds, i.e. nouns in -a are feminine, those in -o or -i are masculine; (b) by default in accordance with the aforementioned scheme Masculine
Feminine
Neuter. Such criteria may explain why in Italian we have feminine la sauna, la maratona ‘marathon’, l’orchestra, la cifra ‘number, figure’, l’asma ‘asthma’, la sentinella ‘sentry’, la spia ‘spy’, la guardia ‘guard’, la zebra, la dacia ‘dacha’, but masculine il batterio ‘bacterium’, l’ago ‘needle’ (formerly f), il dazio ‘tax, duty’ (< Lat. datio (f)), l’euro, il metodo ‘method’, il periodo ‘period’ (the latter two are both feminine in Greek and remain feminine in French la méthode and la période, where the phonological principle would not apply). This may explain why the mass media have unwittingly created a new female saint, Santa Foca, replacing the male San Foca (22nd September). However, in a contrary spirit to such reassignment, il mammasantissima ‘mafioso boss’ resists, governed by natural gender (male sex of the heads of organized crime). Notwithstanding the medieval Latin trend set by datio (f) > il dazio ‘tax, duty’, technical terminology may retain feminine gender, as with captatio benevolentiæ, constructio ad sensum, editio princeps as a special class when used in learnèd discourse. Furthermore, it is not always true, as Thornton claims, that nouns in -i are masculine (cf. alibi, bikini, karakiri ‘harakiri’, yeti, safari), since there are feminine nouns of Greek origin in -isi (crisi ‘crisis’), -esi (tesi ‘thesis’), -asi (parafrasi ‘paraphrasis’), -osi (artrosi ‘arthrosis’). Other cases which are not truncations, clippings, or otherwise abbreviated forms all seem to be, however, masculine, some even showing reintegration into the masculine gender of original feminine nouns: la propolis > la propoli > il propoli, l’abbiccì (f > m) ‘alphabet’. Many of these principles and class formations figure in D’Achille and Thornton (2003), where historical cases are discussed at length. The morphological classes of Dressler and Thornton (1996) and Thornton (2001) are considered central in the historical development of gender assignment and reassignment, as well as conditioning by endings which indicate morphological classes. This would explain why Latin neuter plurals in -a tend to become feminine in Romance. These authors insisted on the interplay between morphological class (primary) and semantic categorization (secondary), using as their starting point a six-class system based on the thematic vowels (1) -a (pl -e), (2) -o (pl -i), (3) -e (pl -i), (4) -a (pl -i), and two invariable classes (5 þ 6). Reassignment into different morphological classes explains a large

244

John B. Trumper

number of Italian historical gender shifts, with respect both to Latin and Græco-Latin origins and to earlier Tuscan forms. This is the case with neuters in -[m]a which become feminine such as asma ‘asthma’, calma ‘calmness’, and cresima ‘confirmation’. We are left with a small number of late, learnèd nouns (Renaissance borrowings) which, to maintain their masculine gender, have undergone class reassignment, sometimes reverting to the original -a and following the sequence -a > -o > -a over the whole of their history, and yielding irregular masculines such as the documented cases clima > climo > clima ‘climate’, gramma > grammo ‘gram’, and pachiderma > pachidermo > pachiderma ‘pachyderm’. On the other hand, we also have Latin -us feminine stems which should have given Italian feminine -o, but which either give feminine -a, as in nuora ‘daughter-in-law’ and suocera ‘mother-inlaw’ (cf. suora ‘sister, nun’ < Lat. soror (f) beside the old Italian form suoro)— outcomes common to most Italo-Romance dialects—or masculine -o for fruit trees, as in pero ‘pear tree’, ciliegio ‘cherry tree’, melo ‘apple tree’, in contrast to their respective fruits which have aligned with feminine gender in -a (< npl as in pirum ‘pear’ > pira ‘pears’, but also arancio together with arancia ‘orange’), a morphological shift paralleled in Spanish. Greek borrowings with masculine themes in -a, in old Tuscan a seemingly invariable class, which now either unusually display plurals in -i or switch class to -o (with plural in -i), are treated in terms of class change; no consideration is given to their status as the medieval products of linguistic contact between Venice, southern Italy, and the Byzantine world. Elements such as Gk. peiratés ‘pirate’ (Kahane and Kahane 1981: 46, 72; 1986: 278) gave a series of invariable nouns in Venetian, which, though still invariable in Venetan dialects, shift to masculine plurals in -i in Italian. In old Tuscan they remained invariable as in their Venetian source (il pirata (sg) > i pirata (pl)), while modern Tuscan from the 16th century on merges this subclass with the one with plurals in -i (il pirata > i pirati). The class in which a head which has undergone ellipsis assigns gender has been discussed in greater detail than these latter cases: for example la lampo ‘zip’ (< la chiusura (f) lampo lit. ‘the closure lightning’), la squillo ‘the call girl’ (< la ragazza (f) squillo lit. ‘the girl ring’) vs masculine il lampo ‘the lightning’, lo squillo (del telefono) ‘the ring(ing of the telephone)’, as well as various other classes in which assignment is rule-governed. Holmes and Segui (2004) would seem to be rather fuzzy on specific languages, though their important Connectionist Theory needs to be brought to the fore as a relevant innovation on storage and retrieval by rule. They agree that gender assignment is by cues which may be either lexical (lexico-semantic) or sublexical (by morphological markers), but are concerned with how such information is stored in the brain, concluding that storage is independent of both phonological and semantic cues (p. 426), a rather doubtful position.3 Their experimental findings

3

Morphological classes and semantic cues ought perhaps to be combined and not dismissed one at a time.

Gender assignment and pluralization in Italian and the Veneto

245

(pp. 428, 436) and those of other researchers in terms of speaker reaction times to cues suggest that native speakers assign gender more accurately in the case of gender-typical endings than in that of gender-neutral ones, so lexical identification is matched in some way with gender assignment, requiring a Connectionist Theory that relies on morphological cues and lexical frequency.4 They conclude that ‘people base gender decisions on both lexical and sublexical cues’, depending, in potentially ambiguous gender-neutral cases, on the frequency, thus availability in mental storage, of given lexical sets and classes. These ideas are original but require more research. Nor do Holmes and Segui insist on default assignment, which represents another weakness of their approach. Default seems a valid gender assigner in Romance, given that, out of about 200 gender-neutral foreign loans examined in a pilot study on Italian, only eight (4%) are assigned feminine gender, namely la box[e] ‘boxing’, la browning ‘revolver’ (perhaps sc. rivoltella (f) ‘revolver’), l’iris ‘iris’, la parur[e] ‘parure, set’, la password, la reclam ‘advert’, la Royal Canin ‘pet food’, la star ‘star, celebrity’ (perhaps sc. stella (f) del cinema ‘star of the cinema’), whereas the remaining 96% take default masculine gender. Conzett (2006) accepts the regularity of processes such as gender assignment based on a combination of semantic and morphological criteria, but concludes that a similar regularity must have its ultimate source in a ‘strongest candidate’ theory based not on semantic but on grammatical features, since, according to the author, gender is not encoded in, nor derivable from, conceptual representations. In fact, his linguistic model is hierarchically arranged in the rank order: (1) syntactic category, (2) gender, (3) semantic characteristics, (4) phonology, (5) inflectional class. While not separating phonology into two layers—the first dealing with what might once have been called abstract ‘morphonological alternations’, the second with phonemic adjustment—Conzett divides essentially a unique morphology into two layers (2) and (5) as above. He addresses two essential questions (p. 236): how gender is stored in the model and ‘how nouns are assigned gender that did not previously exist in the language of a community’. He distinguishes between rule generalizations across the ‘indigenous’ lexicon and productive gender assignment rules for newly integrated nouns, proposing a network model that relates the two sets of rules. In the first case gender is lexically stored, thus arbitrarily assigned to nouns, e.g. il tavolo (m) ‘table’ but la sedia (f) ‘chair’; in the second case three principles operate: (a) inflectional class criteria (il tango (m) ‘tango’, but la sauna (f) ‘sauna’); (b) determination in terms of a deleted head (il panda (m) ‘panda bear’ vs la Panda (f) ‘Fiat Panda car’) or of the full form of an abbreviation or clipping (il cinema < il cinematografo (m) ‘cinema(tograph)’, la foto < la fotografia (f) ‘photo(graph)’) or produced by the Holmes and Segui (2004: 447) observe that ‘French speakers rely on gender-typical endings and availability of dominant gender-marked lexical associates’. This may explain why Italian TV announcers vary gender-neutral aplomb and similar borrowings, as well as political acronyms. 4

246

John B. Trumper

nominalization of a different part of speech; (c) default assignment where conditions (a) and (b) do not hold: masculine gender for nouns in -e (e.g. il coyote),5 masculine gender in gender-neutral cases like l’alias, l’aplomb, il Basic ‘Basic programming language’, il camion ‘lorry’. This overlooks some rule-governed processes, apart from the fact that in Italian, nouns ending in stressed -ì, -é, -è, and -ù have to be given gender lexically.6 The first involves the opposition between feminine fruit and masculine trees (la pera (f) ‘pear’ vs il pero (m) ‘pear tree’), the second the supposed ‘large’ (f) vs ‘small’ or ‘usual size’ (m) opposition. This latter case is problematic and may be a question of [ metaphorical extension], e.g. tavolo (m) ‘table’ may be used metaphorically (il tavolo delle trattative ‘negotiating table’), but its feminine counterpart tavola cannot, as may buco ‘hole’ (m) (e.g. un buco nelle finanze ‘a hole in one’s finances’), but not its feminine counterpart buca. A further complication is that in the unmarked case augmentatives are masculine by default, whereas the feminine gender indicates something different, semantically extendible, complex. Rules must also be formalized for such cases, not just for productive gender assignment. In the same period Nesset (2006) replied affirmatively to the question of whether there were hierarchies amongst default rules and whether they might be considered global or language-specific, concluding in favour of a default hierarchy Masculine
Feminine
Neuter, as in the work of the 1990s discussed above. Insistence is placed, contrary to classical studies and Maiden (1995), on the ‘privileged position of sexbased rules in gender assignment’ (Nesset 2006: 1385), according to a semantic Override Principle based on prototypes (‘prototypes take precedence’). For example, masculine il killer would be unmarked, like il capotreno ‘station master’, il presidente ‘president’, and il prete ‘priest’, which would seem to be indexical of unusual female participation in mafia-type assassinations, as in the running of the national railways or a corporation or celebrating liturgies: until quite recently feminine la killer, la capotreno, la presidente would be marked, while feminine la presbìtera is marked lexically and in terms of gender in Italy as the wife of an Orthodox or Uniate married priest, il prete; and feminine sacerdotessa < masculine sacerdote ‘priest’ marks neopagan or voodoo celebrations, if it is not a reference to historical pre-Christian origins or metaphorical extension. Rice (2006) claims that there is no explicit typeranking, i.e. semantics does not have any priority over formal word-shape and vice 5

These appear to have regular plurals in -i in the modern mass media, namely i coyòti. cf. il pipì (m) ‘willy (male genitals)’ vs la pipì ‘wee, urine’, l’abregé (m) ‘compendium’ vs la koiné (f) ‘koine’, il tabù (m) ‘taboo’ vs la tribù (f) ‘tribe’. However, abstract nouns in -(i)tù are are always feminine, nouns with final stressed -ò, -à are masculine by default unless abstracts or collectives in -(i)tà, as are names of musical notes in -è, -ò, or -à, but feminine when a deleted head is feminine (e.g. agorà ‘agora’). Sometimes even native speakers vary in the attribution of gender in cases with final stressed -à (e.g. masculine il ratafià ‘ratafia liquor’ became feminine la ratafià for three characters in a Carlo Fruttero RAI1 thriller on the evening of 11.4.2011, unless this was a case of the ironical portrayal of hypercharacterized southern Italian police officers). 6

Gender assignment and pluralization in Italian and the Veneto

247

versa, but when conflict arises as to gender assignment the theory must explicate therapeutic cure or mediate between conflicting choices.7 However, Rice does come back to the point of type-ranking when he states (pp. 1401–2.) that ‘semantic principles outrank morphological ones’ in an increasing order that would seem to be structured as follows: (1)

a. Morphological principles (e.g. in French -in, -ier, -ard are masculine morphs). b. Semantic principles (e.g. sex-based gender, ‘path/ road’ words in French are feminine; the principle is abrogated in other Romance languages, while Rice states that masculine boulevard, chemin ‘way, path’, and sentier ‘path’ vs feminine allée ‘path, drive(way)’, avenue, chaussée ‘roadway’, poste ‘post’, route ‘route, way’, rue ‘road, street’, and voie ‘way, lane’ show that in this particular case principle 1 outranks 2).8 c. Class inclusion may outrank class membership or other semantically based rules (the example given is Ger. Ge- as a collective or class prefix), which may take us back to the idea of the ‘best example’ or prototype in ethnoscience.

Where such criteria balance out or are not sufficiently decisive, then markedness or default will take over according to the much-discussed Masculine
Feminine
Neuter default order. Moving on from similar positions, we note that Marchal et al. (2008) state that gender assignment is implemented by a combination of: (a) lexical information (cooccurring elements and lexico-semantic classes); (b) sublexical information (noun endings, morphological classes); (c) the application of an Optimality Theory which gives decisions in terms of markedness theory (default), simplicity, shortness and neatness of the final explanation and decision. Rule formation of the type ‘x ! f except in the cases a, b, c . . . j’, ‘y! z except in the cases k, l, m . . . t’, becomes central to their discussion. An important conclusion of their account is that assignment must be implemented in terms of the ‘best predicted’: in other words we hark back to ‘best examples’.9 7 Rice (2006: 1404) notes that ‘any theory of gender assignment must explicitly formulate a strategy for mediating conflicts’, where mediation is to be understood in terms of the ‘least marked’. 8 A more coherent Romance example would be the historical situation where egg-laying and milkproducing farmyard animals are unmarked feminines, while all other farmyard animals are unmarked masculines. French changes historical gender on brebis ‘ewe’ (le brebis > la brebis) to conform to this principle (following vache ‘cow’, chèvre ‘goat’), though today the principle only works in French in the case of milk-producers, the egg-laying gender assignment feature having been lost in the shift from old and middle French. This marks French as innovative vis-à-vis Italian. 9 To state, however, that morphological prefixes or suffixes are not good predictors (cf. Marchal et al. 2008: 80): ‘gender knowledge is not attached to morphological units such as suffix but is rather a knowledge associated with the smaller ending segment that best predicts gender’), though probably true for Romance, underestimates the predictability associated with, say, the Ger. Ge- collective prefix. The counterclaim may be true for gender assignment in many language groups other than Romance.

248

John B. Trumper

D’Achille and Thornton (2008), returning to their discussion of gender assignment in Italian illustrated in earlier studies, take as their starting point Durante’s (1981: 268–9) observations that the original Tuscan class of feminines in -o (OTsc. aco ‘needle’, caligo ‘soot’, mano ‘hand’, noro ‘daughter-in-law’, sinodo ‘synod’, suoro ‘sister’), which had been reduced to a single member by the 1800s (viz. mano), now presents either regularization in masculine -o (viz. ago, caligo, sinodo),10 or change of class to feminine -a in the other cases (nuora, suora). This class expanded in the late 19th and early 20th centuries through inclusion of clipped nouns and NPs with deleted heads (e.g. biro ‘biro pen’, dinamo ‘dynamo’, flebo ‘(medical) drip’, foto ‘photo’). The authors start from the premise that most nouns already have a relatively arbitrary lexical gender assignment, with gender classes associated with precise morphological rules. However, apparent exceptions not conforming to similar morphological markers and processes undergo corrective therapy in many languages such as Italian, where even more cases than those contemplated by Durante have swollen the feminine -o class in the 19th and 20th centuries, including: (2) a. Grammatical, historical or medical terms which keep their Latinate form and

gender (captatio, consecutio, condicio, damnatio, editio, fellatio, lectio, quaestio, ratio, reductio, variatio). b. Twentieth-century psychological or scientific terms (immago, libido, magnitudo, virago) which likewise keep their Latin gender. Invariables, however, undergo treatment by rule, namely: (2) c. Clippings and abbreviations take the gender of the full form.

d. Endocentric compounds take the gender of the head (cf. our discussion of cars above). e. Exocentric compounds are assigned gender by semantic criteria (sex or underlying ‘classemes’). f. Nominalized non-nominal forms take unmarked gender according to the rule gender [–marked] ! Masculine
Feminine. g. Invariable, potentially genderless forms assigned gender as by rule [z]. Criterion (2e) functions in the case of N þ N compounds but not in the case of V þ N ones. There may, however, be a solution in terms of lexical marking or assignment in the few feminine V þ N or V þ V cases, namely:

10

Given the presence in southern Tuscan of feminine aco, aca, and generally of caligine, one might hypothesize that It. ago (m), already present in Dante, and caligo (m), were Venetian loans. Indeed, DEI (i.683) claimed that caligo was from the ‘Venetian, Emilian and Lombard-Alpine area’.

Gender assignment and pluralization in Italian and the Veneto (2)

249

h. V þ N / V þ V: giravolta ‘(about-)turn’, lavapiatti ‘dishwasher (appliance)’, lavastoviglie ‘dishwasher (appliance)’, portaèrei ‘aircarft carrier’, etc. ! f, other cases by (2f); i. All lexically unmarked V þ V, V þ N compounds (hundreds) are assigned gender by (2f) and (2g):11 il lavapiatti (m) is a human who washes up in restaurants, but la lavapiatti (f) a machine which washes dishes.

The last study examined here, Washington (2008), makes a single valid point that it is not universally true that masculine gender indicates largeness, while feminine indicates smallness, but that a polar opposition m $ f may indicate extendibility, metaphorical, or even size differences, so Sp. madero (m) ‘board’ (specific instance, but adaptable to metaphorical use) vs madera (f) ‘wood’ is similar to the difference between It. masculine tavolo ‘table’, buco ‘hole’, or bilancio ‘scales; balance’ (metaphorically extendible) and feminine tavola, buca, bilancia ‘scales’ (the purely material instance or a large non-metaphorical instance). The whole question requires indepth study, as do the gender vagaries of outcomes of the Latin neuter when such semantic concerns are involved. The point is made that gender assignment is not just a question of arbitrary lexical marking, and a serious grounding is established for future analysis. Rules can be set up, and we will try to examine the reasons for historical dialect gender assignment and change in the next section. It remains the case however that the network model will have to be worked out more thoroughly, but the premises already set out in Conzett (2006: 237) hold good.12 The morphological component linked or associated with gender will have rules like those suggested above in (2a–i).

12.2 Pluralization, gender assignment, and shift in the Veneto (Neo-Venetian) Spoken urban Venetian, the Laguna and Chioggia varieties, as well as the PaduaVicenza-Rovigo central-southern group of dialects, show a modern scheme of pluralization as in Table 12.1.13 11 A notable difference between French and Italian is that the former has no lexical assignment as in (2h), but only the operation of a general rule like (2i). 12 cf. ‘the information about a given noun’s morphological structure [ . . . ] is not apparent from the stored item itself, but is provided by a set of symbolic derivational rules [ . . . ] In a network model [ . . . ] the morphological structure of lexical items is apparent from the way they are associated with each other [ . . . ] It may well be the case that only the strongest associations are at work in the beginning’ (Conzett 2006: 237). 13 Examples: el moléta–i moléta ‘knife-grinder(s)’, la gióva–le gióve ‘beam(s) (of scales)’, el soldá–i soldái/i soldá ‘soldier(s)’, stornitá–le stornitáe/le stornitá ‘vertigo’, l’árzare–i árzari ‘embankment(s)’, l’arte– le arte ‘profession(s)’, la mare–le mare ‘mother(s)’, el bignè –i bignè ‘choux pastry(-ies)’, la tési–le tési ‘thesis (-es)’, el dí–i dí ‘day(s)’, el tóso–i tósi ‘boy(s)’, la biro–le biro biro pen(s)’, el filò–i filò ‘meeting(s) of women’, la virtù–le virtù ‘virtue(s)’, el can–i cani ‘dog(s)’, la man–le man ‘hand(s)’, Vnz. el canal–i canái ‘canal(s)’, el

250

John B. Trumper

TABLE 12.1. Patterns of nominal pluralization in the Veneto -a

-á

-e

mpl - -/ -ái -i fpl -e

-é, -è

-i

-í

-o -ò

- - - -i [-éri, –èri] [-iri]

-/-áe - -

- -

-ú

- [-òri]

-n (-l) (-r)

. . . . . . . -ni (-li) (-ri)

- . . . . . . . -

- -

-

The scheme is slightly different from those proposed in Lepschy (1963) and Zamboni (1974), although it obviously takes into account Zamboni’s observations (pp. 33–4) on typical Laguna plurals sofá-sofari ‘sofa(s)’, paletò-paletòri ‘overcoat (s)’, canepè-canepèri ‘settee(s)’, cafè-cafèri ‘café(s)’, etc., which contrast with the invariable plural forms found in all other dialects in such cases. In other words, el poeta–*i poeti ‘poet(s)’, with similar examples, all admitted in Lepschy (1963), are taken out of the dialect model, as being at odds with strict dialect morphology with its el poéta–i poéta ‘poet(s)’ pattern following the model of el bandéta–i bandéta ‘bandit(s)’, el bòcia–i bòcia ‘kid(s)’, and kindred formations.14 Although such forms constitute a legitimate part of modern dialect, which has undergone interference from the standard language in a way which reflects sociolinguistic reality and complicates the analysis, they certainly will not help us to recreate the historical dialect situation. In such instances we need to be able to construct the two morphological systems which interact historically, eventually comparing the dialect patterns of nominal pluralization outlined in Table 12.1 with those outlined in Table 12.2 for Italian.

TABLE 12.2. Patterns of nominal pluralization in Italian -a

-á

-e

-è, -é

-i

-í

-o

-ò

-ú

-C

mpl

-i

-

-i

-

-

-

-i

-

-

-

fpl

-e

-

-i

-

-

-

-i

-

-

-

mural–i murái ‘mainbeam(s)’, la moral–le moral ‘moral(s)’. Slavic borrowings, more usual in the Veneto than in other northern dialects, are integrated as feminines in -a if they belong to the -a declension, e.g. brítva > brítola ‘pocket-knife’. Other types integrate as -o masculines with word-final (con)sonants (e.g. komát > comácio ‘horse-collar’), but as - masculines after final nasals (hrèn > crén ‘horse-radish’). 14 This class of nouns, with its invariable dialect plural, is treated at greater length in Trumper and Vigolo (1995: 11–14) and Trumper (2011: 796–8).

Gender assignment and pluralization in Italian and the Veneto

251

Whereas the data in Table 12.1 appear to be the product of a rule such as Rule 1, where x may be null, the data in Table 12.2 seem to be the product of a rule such as Rule 2, where x is definitely non-null. When a masculine noun ends in a consonant it will take a  plural morph by default (e.g. il revolver–i revolver ‘revolver(s)’), as is also the case with feminine nouns ending in a consonant or -o (e.g. la star–le star ‘star(s), celebrity (-ies)’), with the exception (outside specific central Italian areas with le mano ‘the hands’) of la mano–le mani ‘hand(s)’, which will have to be specified for its plural in the lexicon of most Italians.15 Rule (1)

M + Thematic Vowel -x PL

F + Thematic Vowel -A

-I →

+

-E

Otherwise

-

M + Thematic Vowel -x F + Thematic Vowel -E

-I

Rule (2)

PL

F + Thematic Vowel -A Otherwise

→

+

-E -

Taking into account Ineichen (1957; 1966) and our knowledge of texts from the central Veneto from the late 12th century onwards, we can conclude that the historical dialect situation, from which the modern one evolved, possessed a rule along the lines of Rule 3 below: where historically most - cases derive from a sigmatic plural in -s, e.g. dentem ‘tooth’ > dénte, dentes ‘teeth’ > *déntes > [i] dénte (PD, VI until 1700–1800) > modern dénti/dinti (see Sabatini 1965a for evidence that northern dialects preserved two mpl desinences and corresponding declensional classes in -i and in -s).16 At the same time a large number of historical feminine nouns in -e (e.g. nóse ‘walnut’, raìse ‘root’ in the Libro agregà of Serapion or 15 Types such as il braccio (msg) - le braccia (fpl) ‘(anatomical) arm(s)’ vs il braccio (msg) - i bracci (mpl) ‘(metaphorical) arm, branch, inlet’ will be treated in future work. Veneto dialects lost this class of masculine singulars with feminine plurals (e.g. uovo - uove ‘egg(s)’, osso - osse ‘bone(s)’, braço-braçe ‘arm(s)’) in the course of the 14th–15th centuries. 16 In chancellery prose at the beginning of the 15th century, although sg -e > pl -e is usual in both feminine and masculine nouns and adjectives, there is a slight tendency to remorphologize masculines as sg -e > pl -i, which may reflect the first signs of pressure exercised by contemporary Venetian prose, in turn influenced by Tuscan norms.

252

John B. Trumper

the Bibbia istoriata padovana, very early 14th century) change their thematic vowel into -a in the early 1500s. Rule (3)

M + Thematic Vowel -O PL

F + Thematic Vowel -A Otherwise

-I →

+

-E -

Thematic vowel changes have been noted in 35 cases so far analysed, the development being of the type ape(m)/apes ‘bee/s’ > ave, with new singular ava ‘bee’. With this shift of thematic vowel (sg/pl -e / -e > sg -a, pl -e), there are 30 counterexamples which retain -e with feminine gender (e.g. carnem ‘meat’ > carne, lucem ‘light’ > luse, matrem ‘mother’ > mare, noctem ‘night’ > nòte). Rural feminine (la) late ‘milk’, (la) sale ‘salt’, and (la) miéle ‘honey’ ought to be added to these, which would make a total of 33 cases of conservation.17 There would seem, then, to have been a movement towards thematic vowel substitution (-e > -a) in feminine nouns. More cases are documented in the early 14th-century Serapion than in modern dialect, including not only forms such as fievra/ fivra ‘fever’, vesta ‘dress, coat’, but also carna ‘meat’ (Serapion 121v17) and pella ‘skin’ (Serapion 216v22), which are nowadays, perhaps under Tuscan influence, la carne and la pèle. This does not seem to have been the case with masculine nouns, where only 15 cases of thematic-vowel reassignment seem to be known, including abietem > albéo ‘birch’, uesperos > brèspi ‘vespers’, laborem ‘work’ > laóro ‘work’, nepotem ‘nephew; grandchild’ > nevódo. At least 50–60 or more cases keep as masculines an -e thematic vowel. In other words, -a was historically becoming indexical of feminine gender, except for a small class of a masculines, all agents or metaphorical extensions in origin. By contrast, there was nothing particularly indexical of masculine gender (-e, -o, -i), which, one might argue, has tended over time to be the default gender with respect to the feminine. This produced numerous cases of indecision as to gender assignment in the period from the 13th to the 15th centuries in dialect texts; for example, in the Bibbia istoriata we find la lome/el lome ‘name’ (cf. also lume ‘lamp’ which was feminine until recently), el late/la late ‘milk’, el tribù/la tribù tribe’, el fiore/la fiore ‘flower’. There seems to be a movement towards the feminine as a ‘large’ version of the basic, unmarked form, e.g. baíle > baíla ‘large shovel’, buso > busa ‘huge, natural hole’ (also in place-names such as Busa di Vigonza), brénto ‘water-jar’ > brénta ‘liquid measure of 70–80 litres’, although the converse is not true, e.g. tòla ‘table’ cannot be

17

These examples contrast with urban and neo-rural masculine (el) late, (el) sale, (el) miéle.

Gender assignment and pluralization in Italian and the Veneto

253

made masculine for metaphorical extensions, unlike the Italian case examined above. However, there are marginal cases of metaphorical use with f -a > m -a, which are invariable in the plural, e.g. sg la móna ‘cunnus’, pl le móne > sg el móna ‘idiot’, with plural i móna. Modern gender assignment in dialect suggests, as do the Italian facts, that the masculine is the default gender (cf. discussion in }12.1 above). For example, all nominalized infinitives are by default masculine (el córare ‘the running’), as are nouns so derived,18 as well as unmarked exocentric compounds, usually of the V þ N type, but also including some of the V þ V type (e.g. el cónsa-pignate (lit. ‘repairsaucepans’) ‘tinker’, el fa-pónte (lit. ‘make-points’) ‘pencil-sharpener’, el magna-carte (lit. ‘eat-papers’) ‘low-grade clerk’). Exceptions are la/el tira-mòla (lit. ‘pull-release’) ‘type of sweet’ and la bati-sésola (lit. ‘beat-sickle’) ‘firefly’, which will have to be lexically marked. Obviously, when the referent is a human female, then the noun assumes feminine gender; otherwise it is masculine (except in marked cases) by default. Copulative V þ V compounds are either masculine, as in córi-córi (lit. ‘runrun’) ‘sudden flight’, or feminine na gira-vòlta (lit. ‘turn-turn’) ‘about-turn’, so gender assignment by rule is not possible in such cases. Endocentric compounds take the gender of the head, as in the bird-names (la) bóca-larga (f), lit. ‘mouth-wide’ (‘Caprimulgus europaeus’), (el) còlo-stòrto (m) lit. ‘neck-twisted’ (‘Jynx torquilla’). Pluralization produces dialect variation in the case of endocentric N þ de þ N compounds, since pómo-d’òro (lit. ‘apple-of-gold’) or pomodòro ‘tomato’ pluralizes rurally as pumi-d’òro, in the same way that òcio-de-bò (lit. ‘eye-of-bull’) ‘bull’s eye’ pluralizes i òci-de-bò, though the plural pumidòri is not unknown. Urban dialect usually pluralizes i pómidòro, though variation as in rural dialect is not unusual. All nominalized adjectives are assigned masculine gender (e.g. el dólse ‘sweet’, distinct from la dólse/dólsa ‘black pudding’), as are nominalized adverbs (el parché ‘the wherefore’), as well as foreign loans ending in consonants, which, if not a sonant, show epenthesis with -e and take the gender of a suppressed head (jeep > la gip, SUV > el suv), which is also true of acronyms (La CGL = la cigiele < Confederazione (f) generale del lavoro ‘General Labour Confederation’). In the case of final nasals, such loans present consonantal deletion in final stressed or unstressed syllables, with final ‘r’ or ‘l’ as exceptions (lichen > el liche, but film > el fil). A small number of loans have feminine gender assigned, e.g. password > la pássuor (a new website entry in Veneto, sometimes spelled paswor). As in Italian, almost all foreign loans are assigned masculine gender by default; where feminine gender is assigned, there seems to be gender agreement with the suppressed head of an endocentric compound (in passuor < password it may be the underlying parola (f) ‘word’).

18

For example, andare ‘attitude’, new regular pl andari, which has existed since the 16th and 17th cc.

254

John B. Trumper

There appear to be some, even if relatively few, categories where automatic gender assignment occurs, e.g. river names are always feminine (e.g. la Sile), except for rivers in -ón (el Bachijón), streams in -ín (el Tergolìn ‘Tergolino’), and canal names (el Pióvego). In a similar fashion, all collectives in -ime (bèco ‘beak’ > el bechime ‘birdseed’), -ume (fióre flower’ > el fiorùme, fierùme, farùme ‘freshly cut hay’), -ame (bèstie ‘beast’ > el bestiame ‘cattle’),19 all diminutives in -in (frésco ‘fresh’ > freschìn ‘smell of rotting fish’), and augmentatives in -ón, are all masculine,20 whereas derivatives in -(V)ssión are feminine (e.g. orassión ‘prayer(s)’), those in -ale are feminine if abstract (la morale ‘morals’), masculine if material or living (el bancale ‘windowsill’, el dentale ‘plough-share; dentex (sea-fish)’). Furthermore, all nouns in -ile and -ule are masculine (el copile ‘central chain of an ox-yoke’, el baúle ‘trunk (suitcase)’), as are all nouns in -ore.21 Apart from three exceptions, which will have to be lexicalized—nósa/nogara ‘walnut/walnut tree’, sarésa/saresara ‘cherry/cherry tree’, brómba/brombara ‘plum/plum tree’, all fruits present the derived tree name or large bush name as a masculine in -aro, whether the fruit be masculine (e.g. pómo/ pomaro ‘apple/apple tree’) or feminine (e.g. frágola/fragolaro ‘strawberry/strawberry plant’). Farmyard animals which are milk-producers and egg-laying farmyard fowl are still unmarked feminines (piégora ‘sheep’, galina ‘hen’), the males being marked (moltón ‘ram’, galo ‘rooster’), while with other mammal names the converse seems to be true (e.g. can ‘dog’ > cagna ‘bitch’, cavalo ‘horse’ > cavala ‘mare’, musso ‘donkey’ > mussa ‘jenny’). While nouns ending in stressed -í and -á may be either masculine or feminine (arbitrary historical assignment), dialect nouns which end in -ú are always feminine (e.g. la grú ‘crane’, la virtú ‘virtue’), those which terminate in stressed -ò are always masculine (el bacò ‘southern Italian grape’/wine produced from this grape, un bagolò ‘pastime; amusement’, el filò ‘evening story-telling’). Modern nouns in -é or -è have 19 In such cases, whenever there is coincidence with Tuscan derivational patterns (e.g. in the case of bechime, bestiame), it is difficult to unravel the knotty problem of whether such processes or items were introduced from Venetian into Tuscan or vice versa, or indeed whether what we see here are independent parallel evolutions. Without clear textual, documentary evidence, or clear phonological and morphological clues, the question is unsolvable. When such evidence is available, the borrowing direction is quite clearcut. In the case of -ime, -ume, and -ame, the suffixes are common to both subsystems, so no decisive criteria are available in the absence of textual evidence. 20 There are historical cases in -ón which represent early Romance diminutives, e.g. sardón ‘anchovy’ (cf. sarda/sardèla ‘sardine/pilchard’), though these are no longer part of general derivational morphology, but are independent lexical items, formally unrelated to sarda/sardèla, even if there is an underlying semantic class relationship of inclusion, i.e. semantically sarda  sardón ‘anchovy’ þ sardèla ‘pilchard; sardine; gilt sardine’. Such derivatives are always masculine. 21 The majority of masculine nouns in -ore such as brusóre ‘burning’, dolóre ‘pain’, saóre ‘taste’, süóre ‘sweat’, together with agents/actors in -ore (dotóre ‘doctor’, magnaóre ‘eater’, pescaóre ‘fisherman’, pitóre ‘painter’), seem to have attracted the gender of all such nouns, which have accordingly changed gender over the centuries, e.g. OVen. la fióre ‘flower’ (Serapion, Bibbia istoriata, etc.) > modern el fióre, the exact opposite of the French situation where feminines in -eur such as la fleur ‘flower’ have attracted erstwhile masculines which are now assigned feminine gender (Fr. la douleur ‘pain’, la sueur ‘sweat’).

Gender assignment and pluralization in Italian and the Veneto

255

lexicalized gender: the child-language form la/le pepè ‘shoe(s)’, archaic la ré ‘net’ (= la réde), must be marked as feminine in the lexicon, whereas all others are masculine (e.g. el cabarè ‘cabaret’, el cabarnè ‘type of wine’, el cafè ‘coffee’, el ré ‘king’). Having evaluated more closely the dialect cases reviewed above, our provisional conclusion is that the classes set up in D’Achille and Thornton (2008)—see subclasses (2a–i) in the previous section—will have to be slightly enlarged and modified, if one is to get into the nitty-gritty of gender attribution and pluralization in both dialect and Italian. Apart from the general default principle which is adequate for all the ‘Other’ cases, which may be expressed as Masculine
Feminine, the first thing we note is that dialect does not possess the (2a,b) cases,22 while condition (2c)— clippings and abbreviations take the gender of the full form —can be left as it stands (e.g. Vnz. na fòto/It. una foto, which take their feminine gender from the full form fotografia). As amply demonstrated above, condition (2d) is specifiable as in Italian (namely, endocentric compounds take the gender of the head). A number of morphologically determined or semantically sensitive subclasses, where gender assignment would seem completely predetermined and not lexical, may be set up as follows: (3)

(a) -ú (stem-final stressed -U); (b) suffixal [-V]ssión; (c) suffixal -ale [þmoral, þevaluative]; (d) noun [þanimal, –human, þdomesticated, þegg-laying/ þmilk-producing] ! Feminine

Male animals in the last subcondition are derived items. Note that suffixal -ore, -ile, -ule (together with the -ale cases not specified in (2e)), -ón, -ín, -ime, -ume, -ame, as well as the phonological subclasses of nouns which have stem-final stressed -ò, now all have gender assigned by the general default principle as above. We should note en passant that gender cannot be assigned by rule to copulative compounds, but has to be specified in the lexicon, whereas it can be assigned by the following loose rule in the case of exocentric compounds: (4)

Exocentric compounds are assigned gender in terms of a suppressed head or class term or at least in terms of clear-cut semantic criteria on the element being deleted, wherever possible. Where such criteria cannot be evinced, gender is assigned by the default principle.

This leaves us with two subclasses corresponding to the previous (2f) and (2g) cases studied above, in almost perfect correspondence with the same categories as in Italian where gender is also assigned by the general default rule: (5)

a. Gender is assigned by default to all nominalized non-nominals. b. Gender is assigned by default to invariable, naturally genderless nominals.

22 All direct medieval Latin learnèd items in -io (f) become masculine in dialect, witness datio ‘tax, duty’ > el dássio (m), passio (Christi) ‘Passion of Christ’ > el Pássio (m) ‘Passion Gospel’.

256

John B. Trumper

12.3 Conclusion Historical problems remain above and beyond those dealt with here, to which we have dedicated but little attention—problems such as the ambiguous historical treatment of original Latin neuters, the singularization of more Latin neuter plurals than in Tuscan, and the dialect maintenance of an original gender vis-à-vis Tuscan. Also of interest is the variable treatment of the Latinized Greek lexicon ending in -ma. With respect to Tuscan, the Veneto dialects present asma (f) ‘asthma’ (though until the 17th century Venetian showed variation between asma (f) and as[e]mo (m)), crésema (f) ‘confirmation’ (probably conditioned by ecclesiastical use), but gramo (m) ‘gram’, batézo/batéso (m) ‘baptism’, crime (m) ‘inclination; character’, spásemo (m) ‘spasm’. For the small subclass of Venetan masculine nouns with feminine plurals between the 12th and 14th centuries (cf. It. braccio–braccia ‘arm(s)’), the pluralization rule will have to be more complex for masculine nouns in -o in its 14th-century Veneto version than is the modern rule. We are currently working on a general model with rules for gender assignment and pluralization not just for Italian and Veneto dialects but extended to southern Italian dialects, though a great deal of work still remains to be done. In short, problems increase as geographical variation and historical depth are added.

13 Kind-defining relative clauses in the diachrony of Italian* PA O L A BEN IN CÀ A N D G UGLI ELM O CIN Q U E

13.1 Introduction Within the larger class of relative clauses, we will isolate a specific subclass, first discussed in Benincà (2003; 2012a): that of relative clauses modifying a nominal predicate, exemplified in (1) and (2): (1) Gianni è un uomo che non si Gianni is a man that not self= ‘Gianni is a man that never surrenders’

dà gives

mai never

(2) Mario è un ragazzo che puoi convincere Mario is a boy that you.can convince.inf ‘Mario is a boy that you can easily convince’

per for

vinto (It.) won

facilmente (It.) easily

We will show that such relative clauses belong to a larger class of kind-defining relatives, in which other kinds of verb incorporate a stage-level predicate. We will consider their properties in varieties of Italian and English, and their evolution in the history of Italian. The copular structures whose predicate can be the antecedent of a kind-defining relative belong to the predicational (canonical, extensional) type, while the identificational (inverse, intensional) ones are excluded.

* Part of the matter discussed in this chapter was presented in Bristol in 2003 at the annual ‘Incontro di Dialettologia Italiana’, organized by Mair Parry; she found the proposal interesting, and soon afterwards used the new classification for her analysis of Ligurian and Piedmontese texts (Parry 2007a). This is just an example of Mair’s attitude in her research: she is always open-minded and ready to see with insightful intelligence which tools are more productive to explore the nature of every detailed linguistic phenomenon. We dedicate this work to Mair, the best colleague and fellow-researcher, the most generous and lively friend. For data, judgements, and discussions we thank Mariachiara Berizzi, Terry Freedman, Adam Ledgeway, Elena Perna, and Christina Tortora.

258

Paola Benincà and Guglielmo Cinque

The relatives that appear in this context, namely those modifying the nominal predicate of a copular sentence, display a number of special properties that set them apart from ordinary restrictive relative clauses and render them partly similar to non-restrictive relatives. We will try to relate these special properties to the fact that such relatives concur to define a kind, and more specifically to the fact that (in opposition to restrictives) the relative clause content is not presupposed as true, and the DP in predicate position is not truly referential. This class of relative clauses can be identified comparing old Florentine with modern Italian on the one hand and with other medieval varieties on the other. It will appear that diachronic change is very limited; the distinctive features that set this class apart are present in all of these linguistic areas. What has changed is the stylistic or social level in which these features appear on the surface; what was found in old Florentine and early Italian vernaculars is visible today only in dialects or in informal colloquial Italian. This aspect of grammatical change should also be taken into account, and find its place in the reflection on diachronic phenomena. 13.1.1 Relative clauses in Italian Modern standard Italian has a system of relativization which uses lexical wh-pronouns or abstract wh-elements; wh-movement in restrictive relatives leaves a gap that cannot be bound by a clitic. In modern colloquial standard and regional Italian, as well as in the dialects, headed relative clauses, when a DP is relativized, are introduced by a bare complementizer, che, co-occurring in some cases with a resumptive clitic expressing the grammatical function of the wh:1 (3)

a. Il ragazzo che gli ho dato il libro si chiama Mario (coll. It.) the boy that to.him= I.have given the book self= calls Mario ‘The boy to who I gave the book is called Mario’ b. Mario, che nessuno lo sopporta, non sarà invitato (coll. It.) Mario, that nobody him= tolerates, not will.be invited ‘Mario, who nobody can stand, will not be invited’

In particular, in relatives on a PP wh-complement, it appears that a strategy in fact resembling modern English P-stranding (allowing for the presence of clitic pronouns in Italian but not in English) is used, with a complementizer, and either a gap or a clitic pronoun in the body of the clause (perhaps with a silent P, represented below in capitals).

1

We do not here consider the possibility that che, rather than a complementizer, is a (weak) relative pronoun, as Sportiche (2008) and Kayne (2010a) have argued for French que and English that, respectively. This proposal is irrelevant with respect to our issue.

Kind-defining relative clauses in the diachrony of Italian (4)

259

a. Hai conosciuto il bambino che gli hanno assegnato you.have met the boy that to.him= they.have assigned il premio? (coll. It.) the prize ‘Did you meet the child who they gave the prize TO?’ b. La ragazza che ci esce assieme da un anno si the girl that there= he.exits together WITH since a year self= chiama Carla (coll. It.) calls Carla ‘The girl who he’s been going out WITH for a year is called Carla’ c. Quello è il ristorante che abbiamo mangiato that is the restaurant that we.have eaten ‘That is the restaurant which we ate IN yesterday’

ieri (coll. It.) yesterday

Studying headed relative clauses in old Florentine, as well as in other regional early languages of Italy demonstrates that the wh-pronoun paradigm appears still well preserved; relative clauses show very clear evidence of wh-movement of a whpronoun, which is normally not accompanied by any clitic pronoun. In old Florentine, the paradigm of relative pronouns is still fairly well preserved, and sentences of the kind in (3) are not expected (Benincà and Cinque 2010). Nevertheless, some rare examples are found where a relative appears with a clitic corresponding to the grammatical function of the wh-. Noordhof (1937) and Fiorentino (1999), for old Florentine, and Parry (2007a; 2007b), for old northwestern vernaculars, in particular, have paid special attention to this kind of exception;2 their careful scrutiny reveals some cases of clitics copying the wh- in relative clauses in the early texts. One can suppose that these cases show the emergence of a new grammar, competing with the old one, in which the properties of wh-pronouns are weakened and on their way out. This possibility does not prevent us from further enquiring whether cases of relatives with a clitic copy of the wh-pronoun are not random examples of the influence of a new emerging grammar, but, rather, share some particular characteristics that constitute a cluster of phenomena.3 2 It is interesting that the phenomenon is so unexpected to a modern reader that in many cases the editor has put an accent on the complementizer che ‘that’ to yield ché ‘for’, thereby transforming the relative clause into a causal clause, without argumental gaps or wh-movement. 3 Noordhof (1937) puts together the cases of relatives with resumptive clitics in old Florentine and compares them with modern Italian examples taken from novels written in colloquial style, such as the following from Collodi’s Pinocchio:

(i) un conto facilissimo, un conto che lo puoi fare sulla punta delle dita (It.) a calculation very.easy a calculation that it= you.can do.inf on.the end of.the fingers ‘a very easy calculation, a calculation which you can do on the fingers of your hand’ Even though some of the examples can have a causal or final interpretation, Noordhof maintains that nevertheless they are syntactically relative clauses, modifying a noun head.

260

Paola Benincà and Guglielmo Cinque

We will see that in fact they correspond to what we find in modern standard Italian, where these cases are strictly limited to the colloquial style, while in old Florentine they appear in all types of text, even in formal treatises. One subset of examples corresponds in fact to modern colloquial cases: they are found in non-restrictive relatives; another subset, also corresponding to modern phenomena, is apparent restrictive relative clauses which display some peculiar semantic characteristics and behave in a consistent way. These apparently restrictive relatives, which we will call kind-defining, characteristically modify a postcopular DP. It is this second construction (which has already been examined in Benincà 2003, Benincà and Cinque 2010, and Benincà 2012a) that we will study in more detail here, with the aim of providing an in-depth analysis of their syntactic and functional properties. 13.1.2 Relatives with resumptive clitics in modern colloquial Italian Modern relative clauses only have copies of the wh- in certain cases, the more systematic of which is that of non-restrictive relatives. See (5a), which contrasts with (5b), containing a restrictive relative: (5)

a. Mario, che non lo conosci ancora, verrà stasera (coll. It.) Mario that not him= you.know yet will.come tonight ‘Mario, who you don’t know yet, will be coming this evening’ b. Il ragazzo che non (*lo) conosci ancora verrà stasera (coll. It.) the boy that not him= you.know yet will.come tonight ‘The boy that you don’t know yet will be coming this evening’

In northern Italian dialects we can find examples of non-restrictive relatives on the subject with a clitic copy, which can be observed neither in standard nor in colloquial Italian (which lack subject clitics). See (6), from Paduan: (6)

a. Mario, che el gaveva dedicà la vita ala scola, se gà ritirà (Pad., Mario that scl= had devoted the life to.the school, self= has retired non-restrictive) Mario, who had dedicated his life to the school, has retired’ b. Mario ze un professore ch el gà dedicà la so vita ala scola (Pad., Mario is a teacher that scl= has devoted the his life to school kind-defining) ‘Mario is a teacher who has dedicated his life to the school’ c. Un professore che (*el) gaveva dedicà la vita ala scola se gà ritirà a teacher that scl= had devoted the life to.the school self= has retired (Pad., restrictive) ‘A teacher who had dedicated his life to the school has retired’

Kind-defining relative clauses in the diachrony of Italian

261

13.1.3 Early Italian relatives with resumptive clitics In early Italian (Florentine) as well, the few examples of clitic pronouns resuming a relativized argument are in non-restrictive relatives, and furthermore in a kind of relative clause that can be interpreted as including a kind-defining post-copular relative: the sentences in (7) are non-restrictives; those in (8) are apparently restrictive, but have, as we will show, a different semantic interpretation and different syntactic properties: (7)

a. Quando i figliuoli del re Priamo ebbero rifatta Troia, che when the sons of.the king Priam had rebuilt Troy that l’ aveano i Greci disfatta . . . (OFlo., Novellino, 91) it= they.had the Greeks destroyed . . . ‘When the sons of King Priam had rebuilt Troy, which they the Greeks had destroyed . . . ’ b. cioè monsignore messer Lancialotto del Lac, che già that.is my.lord sir Lancelot du Lac that already tanto pregare d’ amore ch’ elli avesse di no ’l so.much pray.inf of love that he had of not him= seppi me mercede (OFlo., Novellino, 82) I.knew me mercy ‘I mean my lord Sir Lancelot du Lac, whom I could not sufficiently beseech with love that he should have mercy on me’

(8)

a. acciocchè [ . . . ] neuna cosa possa loro incontrare che non l’ abbiano in.order.that no thing may them meet.inf that not it= they.have provata (OFlo., Bono Giamboni, Arte della guerra di Vegezio 3, 2, p. 81) tried ‘in order that nothing meet them which they had not experienced’ b. darà lo volgare dono non dimandato, che non l’ averebbe will.give the vulgar gift not requested that not it= would.have dato lo latino (OFlo., Dante Alighieri, Convivio I, 9, p. 39) given the Latin ‘[the] vulgar [language] will give an unsolicited gift, which Latin would not have given’

These apparent restrictives do not have the function of narrowing down the reference of the Head noun, which can remain undetermined, but that of giving the features which characterize the class they belong to. Parry (2007a; 2007b), analysing relative clauses from northwestern medieval vernaculars, documents the contexts that characterize the presence of a clitic copying the grammatical function of the wh-pronoun, comparing the description of early Florentine of Benincà (2003); as in Florentine, the clitic copy is only found in

262

Paola Benincà and Guglielmo Cinque

non-restrictive and kind-defining relatives. We choose the following examples, all from 14th-century Ligurian (Parry 2007a: }2.3.4). The first relative (9a) is nonrestrictive, introduced by the wh-pronoun lo qual ‘the which’ (which is typical of non-restrictive relatives in modern and old Italian; see Benincà and Cinque 2010): (9)

a. messer Orcham [ . . . ], lo quar lo Segnor Dee lo mantegna in sir Orcham the which the Lord God him= maintain in grande honor et possanza (OLig.) great honour and power ‘Sir Orcham, whom may God keep in great honour and power’

The second type of relative could be kind-defining, but could also be a sort of reduced-purpose clause: (9)

b. Noi sì te avemo menao questo homo mal fator e we so to.you= we.have led this man evil doer and peccaor che tu lo debi zuigar e occir (OLig.) sinner that you him= must judge.inf and kill.inf ‘We have brought you this evildoer and sinner, so that you judge and kill him’

The third is a kind-defining relative, providing the features that mark a semantic class which the head belongs to: (9)

c. ché quele cose che tu non ne parli e manifesti sum for those things that you not of.it= speak and show are a lo tuto ascose da lo nostro conoxemento (OLig.) at the all hidden from the our knowledge ‘because those things which you neither speak of nor express are entirely hidden from our understanding’

The syntax of relative clauses in Ligurian medieval vernaculars, as illustrated in Parry (2007a; 2007b), reflects then the syntax of medieval Florentine.

13.2 Kind-defining relatives The relative clauses with a clitic copy which we have seen so far—non-restrictive and kind-defining relatives clauses—have in common the fact that they do not narrow down the reference of the Head noun, unlike restrictive relative clauses with gaps. This happens for completely different semantic properties:4 while the Head of 4 This class of relatives can be clearly identified in ancient Greek (see Monteil 1963; Perna 2013): they use the pronoun hostis, hetis, hoti (corresponding to Latin quisquis, quidquid, an indefinite relative pronoun), which is never used in genuine restrictives; it appears in very few cases in non-restrictives, and (as Perna 2013 points out) in kind-defining relative clauses (typical structures are existentials such as ‘there is somebody that . . . ’), generally with the subjunctive mood accompanied by the ‘irrealis’ particle an.

Kind-defining relative clauses in the diachrony of Italian

263

non-restrictive relatives is already identified (as in (10a)), the antecedent of (postcopular) kind-defining relative clauses is indefinite and can remain so, because the aim of the relative is only that of defining it as a kind (as in (10b)); the restrictive relative clause, instead, univocally contributes, with the Head, to the identification of the referent of the DP (as in (10c)): (10)

a. Maria, che sa cucinare Maria that knows.ind cook.inf ‘Maria, who can cook lasagna . . . ’ b. una ragazza che sa(ppia) a girl who knows.ind(sbjv) ‘a girl who can(/could) cook lasagna . . . ’

le the

lasagne . . . (It.) lasagna

cucinare cook.inf

c. la ragazza che sa cucinare the gril that knows.ind cook.inf ‘the girl who can cook lasagna . . . ’

le the

le the

lasagne . . . (It.) lasagna

lasagne . . . (It.)5 lasagna

13.2.1 Kind-defining contact relatives in English The kind-defining relative, exemplified so far from Italian and from old and modern Italian vernaculars, bears some resemblance to the so-called ‘contact relative’ on the subject of regional and local varieties of English, in particular as analysed in Hermann (2003) (see (11a,b).6 In the county of Greater Manchester (England, see (11c), from Berizzi 2001) and in the Appalachian region (USA, (11d), from Tortora 2006; 2008), the structure appears also in the predicate of a locative copular sentence: (11)

a. b. c. d.

Leck is a young boy Ø was coming home from school That’s another place Ø is called after the Isle, (Ilminster, you see). There’s a train Ø goes without stopping (Greater Manchester) There was a snake Ø come down the road (Appalachian County)

5 The subjunctive mood induces the kind-defining interpretation, which otherwise remains ambiguous (esp. in those varieties of Italian where the subjunctive mood is less used):

(i) a. (Vorrei sposare) la ragazza che sa cucinare le lasagne (It.) I.would.like marry.inf the girl that knows.ind cook.inf the lasagna ‘(I would like to marry) the girl who can cook lasagna’ ([+def., +spec.] > restrictive) b. Una ragazza che sa cucinare le lasagne (It.) a girl that knows.ind cook.inf the lasagna ‘A girl who can cook lasagna (ambiguous: [–def., +/– spec.] > restrictive/kind-defining) c. Una ragazza che sappia cucinare le lasagne (It.) a girl that knows.sbjv cook.inf the lasagna ‘A girl who can cook lasagna’ ([–def., –spec.] > kind defining) 6

For an early discussion of this type of relatives, see Harris and Vincent (1980).

264

Paola Benincà and Guglielmo Cinque

Other structures have to be interpreted as hidden locative predications, as the following (from Berizzi 2001; see also Hermann 2003):7 (12)

a. We had this French girl Ø came to stay b. My friend’s got a brother Ø used to be in the school

Hermann (2003: 35–6) records other relative structures (see (13)), which he interprets as containing a locative-existential predicate (to know, to see = [there in the mind/in the sight]] is [DP [relative clause]]:8 (13)

a. I know two or three Ø went over (‘I know two or three persons that . . . ’) b. I seen a chap at Broughton Moor Ø got his leg took off

An apparently different kind of structure is exemplified by Hermann, who characterizes them as having a ‘lexically empty’ antecedent: (14)

a. Anybody Ø used to come in our house on a weekend, they always thought there was a party b. Leonard up here would be the only person Ø goes to the bog

We have elicited interesting data on colloquial English from a north-of-England speaker. She accepts the structures above as belonging to normal spoken northern English, and judges the following examples, which involve unequivocal restrictive relatives, as unacceptable: (15)

a. b. c. d.

*I met the boy has arrived yesterday *I will read the book has arrived yesterday *The man will arrive today is my brother *The book will arrive today is interesting

13.2.2 Kind-defining headless relatives in Italian Headless relatives have severe restrictions in modern Italian: they can only be introduced by a bare wh- (subject or object of the relative clause); a prepositional wh-argument is not admitted:9 7 For arguments for interpreting have as corresponding to a locative (or, more abstractly, a preposition + be, and evidence of the multiple functions of locatives to express predicates, see e.g. Freeze (1992) and Kayne (1993). 8 It is interesting to compare the interpretation of the verb ‘know’ as involving a locative in its semantic composition, with the fact that in some central and southern Italian dialects the verb ‘know’ is a pronominal verb, always accompanied by a clitic cluster locative+object ce lo so che . . . ‘there.it=I.know that . . . ’ As a reviewer observes, possibly in a few cases these sentences could be interpreted as cases of asyndetic coordination with an anaphoric null subject in the second clause; it seems to us (as, evidently, to Herrmann, and to Harris and Vincent quoted above) that at least some cases are not open to this possible interpretation. 9 Sentences like the following are grammatical because, despite superficial appearences of surface structure, the wh- relative pronoun with an empty antecedent is not a PP but a bare wh-:

Kind-defining relative clauses in the diachrony of Italian (16) *Ho invitato [DP Ø [PP a chi avevi I.have invited to who you.had ‘I have invited who you had spoken to’

265

parlato]] (It.) spoken

The only systematic exception is that of infinitival relatives in the predicate of existential ‘be’, ‘have’, or ‘find’:10 (17)

a. Ho [DP Ø [PP con chi parlare]] (It.) I.have with who talk.inf ‘I have somebody to talk with’ b. Non c’ è [DP Ø [PP a chi chiedere informazioni]] (It.) not there= is to who ask.inf informations ‘There’s nobody who you can ask for information’ c. Non trovo [DP Ø [PP a chi dare l’ incarico]] (It.) not I.find to who give.inf the duty ‘I can’t find anyone to whom I can assign this duty’

We will concentrate, however, on those kind-defining relatives that modify a postcopular DP (as in (1), (2), (6b) and (11a,b)), taking them as the prototype of kinddefining relatives, even though these relatives are also found modifying a DP in the predicate position of existential sentences (cf. (11 c,d)), and in the apparent object position of a class of verbs (‘have’, ‘meet’, ‘know’, ‘look for’, etc.: cf. (12) and (13)), which may involve a hidden predicative structure (possibly ‘look (for)’ [whether there is] DP, ‘know’ [that there is] DP, etc.).

13.3 Properties of post-copular kind-defining relatives Properties differentiating post-copular kind-defining relative clauses from ordinary restrictive relative clauses that we discuss are as follows:

(i)

Aveva ballato [DP con [DP Ø chi aveva she.had danced with who she.had ‘She had danced with whom she had invited’

invitato]] (It.) invited

The pronoun chi ‘who(m)’ is the direct object of the verb of the relative; the preposition con is selected by the main verb and governs the empty Head of the headless relative clause. The sentence in (i) can be compared with the following sentence, which is ungrammatical because the wh- with empty antecedent is a PP selected by the verb of the relative clause: (ii)

chi aveva *Aveva invitato [DP Ø [PP con she.had invited with who she.had ‘She had invited with whom she had danced’

ballato]] (It.) danced

In some contexts (like the complement position of verbs like ‘find (out)’) it may appear difficult to tell apart headless relatives from embedded interrogatives given that they utilize the same set of pronouns, but see Benincà (2012b) for diagnostics that clearly distinguish them. 10

266

Paola Benincà and Guglielmo Cinque

(i) The use of presuppositional negative adverbs such as mica The postverbal negative polarity element mica, which is possible in non-restrictives (cf. 18), renders restrictive relatives ungrammatical, as in (19) (see Cinque 1976): (18)

a. Ti regalerò questi libri, che non ho mica ancora letto (It.) you= I.will.give these books that non I.have neg yet read ‘I’ll give you these books, which I haven’t read yet’ b. Ho invitato Mario, che non conosci I.have invited Mario that not you.know ‘I’ve invited Mario, who you don’t know’

(19)

mica (It.) neg

a. *Ti regalerò i libri che non you= I.will.give the books that not ‘I’ll give you the books which I haven’t read’

ho I.have

mica neg

b. *Ho invitato il ragazzo che non I.have invited the boy that not ‘I’ve invited the boy that you don’t know’

conosci you.know

letto (It.) read

mica (It.) neg

Interestingly, kind-defining relatives behave like non-restrictives (see (20)): (20)

a. Questi sono i libri che non ho mica ancora letto (It.) these are the books that not I.have neg yet read ‘These are the books which I haven’t yet read’ b. Mario è un uomo che non esita mica a rischiare (It.) Mario is a man that not hesitates neg to risk.inf ‘Mario is a man who doesn’t hesitate to take risks’

(ii) The possibility of using a non-realis mood The use of subjunctive is impossible in both genuine restrictives (21a) and nonrestrictives (21b); it is instead possible in kind-defining relatives, in both early and modern Italian ((21c) and (21d), respectively): (21)

a. *Ho incontrato il ragazzo che abbia risolto il problema (It.) I.have met the boy that has.sbjv solved the problem ‘I’ve met the boy who’s solved the problem’ b. *Mario, che abbia risolto il problema, verrà domani (It.) Mario that has.sbjv solved the problem will.come tomorrow ‘Mario, who has solved the problem, will come tomorrow’ c. Moises fu il primo uomo a cui Iddio desse la legge Moses was the first man to who God gave.sbjv the law (OTsc., Tesoro volg., 1, 17) ‘Moses was the first man to whom God revealed the law’

Kind-defining relative clauses in the diachrony of Italian

267

d. Mario è l’ unico che abbia risolto il problema (It.) Mario is the only.one that has.sbjv solved the problem ‘Mario is the only one to have solved the problem’ The subjunctive mood in (21a,b) forces the kind-defining interpretation, and is incompatible with the referentiality of the head, deriving in (21a)—a genuine restrictive—from the realis mood of the main clause and the definite article, and in (21b)—a non-restrictive—from the proper name antecedent (see above, fn. 5). The grammatical (21c,d) are not restrictive but kind-defining. (iii) The possibility of using a modal in genuinely epistemic readings, which in Italian is admitted in non-restrictive but not in restrictive relatives (22)

a. Il ragazzo che deve partire presto la mattina arriva the boy that must leave.inf early the morning arrives sempre stanco (It.) always tired ‘The boy who must leave early in the morning always arrives tired’ (restrictive: only deontic interpretation) b. Un ragazzo che deve partire presto la mattina arriva a boy that must leave.inf early the morning arrives sempre stanco (It.) always tired ‘A boy who has to leave early in the morning always arrives tired’ (restrictive: deontic interpretation only) c. Mario, che deve partire presto la mattina, arriva sempre Mario that must leave.inf early the morning arrives always stanco (It.) tired ‘Mario, who has to leave early in the morning, always arrives tired’ (nonrestrictive: both epistemic and deontic interpretation)

In the restrictives (22a,b), independently of the definite or indefinite article, the epistemic interpretation of the modal deve ‘must’ is blocked; it is instead possible in the non-restrictive (22c). If we insert an apparent restrictive clause in a DP which is the predicate of a copular structure (as in 22d), the epistemic interpretation becomes possible (or, in other cases, obligatory, for semantic reasons, as in (22e)): (22)

d. Mario è un uomo che deve alzarsi molto presto la mattina (It.) Mario is a man that must raise.inf=self very early the morning ‘Mario is a man who must get up very early in the morning’ (epistemic or deontic interpretation)

268

Paola Benincà and Guglielmo Cinque e. Mario è un uomo che deve amare molto i suoi figli (It.) Mario is a man that must love.inf much the his children ‘Mario is a man > who must love his children very much’ (only epistemic interpretation)

(iv) The illocutionary independence of the relative clause (only in colloquial style) Example (23a), a kind-defining relative clause, and (23b), non-restrictive, can both have the interrogative illocutionary force which is not admitted in a genuine restrictive relative clause such as (23c): (23)

a. È un uomo al quale avresti mai pensato di he.is a man to.the which you.would.have ever thought of doverti rivolgere un giorno? (It.) must.inf=you address.inf a day ‘Is he a man to whom would you would have expected to have to speak one day?’ b. Lo puoi chiedere a Mario, al quale avresti mai it= you.can ask.inf to Mario to.the which you.would.have ever pensato di doverti rivolgere un giorno? (It.) thought of must.inf=you address.inf a day ‘You can ask Mario, to whom would you have expected to have to speak one day?’ c. *Oggi hai conosciuto l’ uomo al quale avresti mai today you.have met the man to.the which you.would.have ever pensato di doverti rivolgere un giorno? (It.) thought of must.inf=you address.inf a day ‘Today you met the man to whom would you have expected to have to speak one day?

(v) Kind-defining relatives occurring necessarily after ordinary restrictives and before ordinary non-restrictives The contrast (24a, b) shows that the kind-defining relative clause (in which we have inserted the negative polarity item mica to force the desired interpretation) is obligatorily ordered in the DP after a genuine restrictive: (24)

a. Quello è un ragazzo che conosco che non esita mica a rischiare (It.) that is a boy that I.know that non hesitates neg to risk.inf ‘That is a young man that I know who doesn’t hesitate at all in taking risks’ b. *Quello è un ragazzo che non esita mica a rischiare che conosco (It.) that is a boy that not hesitates neg to risk.inf that I.know ‘That is a young man who does not hesitate at all in taking risks whom I know’

Kind-defining relative clauses in the diachrony of Italian

269

The contrast (25a, b) shows that a kind-defining relative (with mica) is ordered before a non-restrictive relative clause: (25)

a. Quello è un ragazzo coraggioso che non esita mica a rischiare, le that is a boy brave that not hesitates neg to risk.inf the azioni del quale, tra parentesi, mi hanno sempre colpito (It.) actions of.the which, between brackets, me= they.have always struck ‘That is a brave young man who does not hesitate in taking risks, whose actions, incidentally, always struck me’ b. *Quello è un ragazzo coraggioso, le azioni del that is a boy brave the actions of.the tra parentesi, mi hanno sempre colpito, che non esita between brackets, me= have always struck that not hesitates rischiare (It.) risk.inf ‘That is a brave young man, whose actions, incidentally, always struck me, who does not hesitate to take risks’

quale, which mica a neg to

(vi) The impossibility of stacking Stacking is the occurrence of two relative clauses, one nested within the other, sharing a unique antecedent. It is possible to have stacking of two restrictive relative clauses, with the second one restricting the set of referents specified by the first (e.g. i ragazzi padovani che conosco il cui nome comincia per z sono due ‘Paduan boys that I know whose name begins with a z are two’). Non-restrictive relative clauses, as expected from their semantic characteristics, cannot be stacked (apparent examples are in fact cases of coordination). With respect to the possibility of stacking, kind-defining relatives once again appear to pattern with non-restrictives rather than with restrictives: (26) a. *Quello è un ragazzo che deve essere sempre stato coraggioso that is a boy that must be.inf always been brave che non esita mica a rischiare (It.) that not hesitates neg to risk.inf ‘That is a young man who must have always been brave who does not hesitate to take risks’ b. *Quello è un ragazzo che deve essere sempre stato that is a boy that must be.inf always been coraggioso che non deve spaventarsi mai (It.) brave that not must frighten.inf=self never ‘That is a young man who must have always been brave who must never get scared’

270

Paola Benincà and Guglielmo Cinque

(vii) The use of resumptive pronouns (see (27) and (6)) (27) a. È una cosa che non si può facilmente farne a meno (coll. It.) it.is a thing that not self= can easily do.inf=of.it to less ‘This is something that we can’t easily do without’ b. Gianni è uno che non gli si può dire di no (coll. It.) Gianni is one that not to.him= self= can say.inf of no ‘Gianni is someone to whom you can’t say no’ c. Questo è un problema che non lo risolvi facilmente (coll. It.) this is a problem that not it= you.solve easily ‘This is a problem that you can’t solve easily’ This property distinguishes genuine restrictive from kind-defining relatives, which behave in this respect like non-restrictives. (viii) The possibility of retaining the internal Head As pointed out to us by Diego Pescarini, kind-defining relatives once again appear to pattern with non-restrictives and to differ from genuine restrictives in allowing the retention of the internal Head. See (28a,b) vs (28c); Cinque 1978: 88–9). (28)

a. Non hanno ancora trovato una sostanza dalla quale not they.have yet found a substance from.the which sostanza ricavare un rimedio contro l’ epilessia (It.) substance obtain.inf a remedy against the epilepsy ‘They have not yet found a substance from which substance they can obtain a remedy against epilepsy’ (kind-defining) b. Se non parlano più di questa sostanza, dalla quale sostanza if not they.speak anymore of this substance from.the which substance speravano di ricavare una cura per l’ epilessia, una ragione c’è (It.) they.hoped of obtain.inf a cure for the epilepsy a reason there=is ‘If they no longer talk about this substance, from which substance they had hoped to obtain a cure for epilepsy, there must be a reason’ (non-restrictive) c. *Non parlano più della sostanza dalla quale sostanza not they.speak anymore of.the substance from.the which substance speravano di ricavare una cura per l’ epilessia (It.) they.hoped of obtain.inf a cure for the epilepsy ‘They no longer talk about the substance from which substance they had hoped to obtain a cure for epilepsy’ (restrictive)

13.4 Deriving the properties of kind-defining relatives We take the above properties characterizing kind-defining relatives to be dependent on two facts: (1) that (as with non-restrictives, and unlike restrictives) the proposition

Kind-defining relative clauses in the diachrony of Italian

271

expressed by the relative is not presupposed to be true; (2) that, unlike restrictive and non-restrictives, the Head of the relative is non-referential (since it is a predicate). (i) The use of presuppositional negative adverbs such as mica The postverbal negative polarity element mica is ungrammatical in restrictive relatives, among other contexts (Cinque 1976). Its ungrammaticality appears to depend on the existence of conflicting presuppositions; restrictive relatives presuppose the truth of the proposition conveyed by the relative (such a proposition is presented as shared knowledge on the part of both speaker and addressee), while mica presupposes that the addressee believes the opposite to be true. Non-restrictives and kinddefining relatives do not carry such a presupposition (the proposition is asserted), and indeed both admit mica. (ii) The possibility of using a non-realis mood The possibility of using a non-realis mood (e.g. the subjunctive) may seem a characteristic of restrictive relatives, as in (29): (29)

Cercava una segretaria che conoscesse he.looked.for a secretary that knew.sbjv ‘He was looking for a secretary who knew Chinese’

il the

cinese (It.) Chinese

In fact, this possibility is only open to apparent restrictives that modify, in modal contexts, indefinite objects of verbs like ‘look for’, ‘know’, etc. (see the last paragraph of }13.2.2), which, as noted, may involve a hidden predicative structure. This means that these are not genuine restrictives, but incorporate a kind-defining relative clause (as briefly pointed out in the discussion around example (10) and in fn. 5). Infinitival relatives—which also involve a type of non-realis mood—appear to occur in largely the same contexts as subjunctive relatives (for some discussion of the distribution of infinitival relatives, see Cinque 1988b): (30)

una segretaria a cui Cercava /non trova/ non c’è secretrary to whom he.looked.for /not he.finds/ not there= is a affidare la traduzione (It.) entrust.inf the translation ‘He was looking for/he can’t find/there isn’t a secretary to entrust the translation with’

(iii) The possibility of using a modal in genuinely epistemic readings, which in Italian is allowed in non-restrictive but not in restrictive relatives The fact that a modal verb can, in addition to root interpretations, have an epistemic interpretation in both non-restrictive and kind-defining relatives seems to show that the proposition expressed by these relatives (unlike that of restrictive relatives) is not presupposed as true. If it were, no epistemic interpretation should be possible, as the

272

Paola Benincà and Guglielmo Cinque

epistemic modal introduces an element of non-commitment about the truth of the proposition. (iv) The illocutionary independence of the relative clause (only in colloquial style) The illocutionary independence which kind-defining relative clauses share with (discourse grammar) non-restrictives (see Cinque 2008a) is another property showing that the proposition expressed by these relatives is asserted rather than presupposed to be true. If it were, the mood would necessarily be declarative. (v) Their occurring necessarily after ordinary restrictives and before ordinary nonrestrictives The fact that kind-defining relative clauses are ordered after ordinary restrictives and before non-restrictives also suggests that they form a separate kind of relative, which has a dedicated Merge position within the extended projection of the nominal phrase, and cannot be assimilated to ordinary restrictive relative clauses. (vi) The impossibility of stacking The impossibility of stacking may also be related, it seems, to the fact that kinddefining relative clauses are not genuinely restrictive, which is a prerequisite for admitting stacking. This depends, as noted, on the facts that the Head is not genuinely referential (it is a predicate), and that it is not presupposed to be true. (vii) The use of resumptive pronouns The possible presence of resumptive pronouns, which kind-defining relatives share with non-restrictive relatives, appears not to follow in any simple way from the properties of kind-defining relatives mentioned above (the non-presupposed status of the relative clause and the non-referential status of the predicate DP containing the kind-defining relative), nor in fact from the characteristics of non-restrictive relatives. We conclude that both non-restrictive and kind-defining relatives, for different reasons, can use a strategy without wh-movement; a clitic appears in the relative clause, which is connected (through a rule of predication—Chomsky 1977: 41) to an argument in TopP.11 We leave this as an unexplained property of both kinds of relative. (viii) The possibility of retaining the internal Head This property, opposing genuine restrictive relatives to both non-restrictive relatives and kind-defining relatives, also remains to be understood.

11

Non-restrictive relatives are the typical context where the relative pronoun il quale is employed, which Cinque (1978) analysed as generated in most cases in TopP position, or moved there as a leftdislocated argument, which has then to be linked to a clitic in the clause.

Kind-defining relative clauses in the diachrony of Italian

273

13.5 Post- and pre-copular position (canonical and inverse predication) So far we have considered only post-copular kind-defining relatives which appear to instantiate a canonical predicative relation (cf. 31a), but they can also occupy a precopular position (cf. 31b), which instantiate an inverse (or specificational) predicative relation (Moro 1997): (31)

a. Gianni è un uomo che non si dà mai per vinto (It.) Gianni is a man that not self= gives never for won ‘Gianni is a man that never gives up’ (canonical) b. Un uomo che non si dà mai per vinto è Gianni (It.) a man that not self= gives never for won is Gianni ‘A man that never gives up is Gianni’ (inverse)

Following Moro (1997) (see also den Dikken 2006), the two types of predication (the canonical and the inverse) can be derived from a small clause embedded under a copula (32a) by moving either the subject of predication (32b) or the predicate (31c), before the copula, as informally represented in the following derivational stages: (32)

a. [ È [[Gianni] [un uomo che non si dà mai per vinto]] (It. = base structure) b. [[Gianni]i è [ti [un uomo che non si dà mai per vinto] (It. = canonical predication) b'. [Un uomo che non si dà mai per vinto]i è [[Gianni] ti] (It. = inverse predication)12

That [Gianni] is the subject of predication and [un uomo che non si dà mai per vinto] the predicate can be seen from the contrast between (33a) and (33b), where the two orders are embedded under a small clause taking a verb like ritenere ‘consider/ believe’. As Moro (1997) shows, canonical, but not inverse, predications are possible in such environments in Italian: (33)

a. Ritenevamo [Gianni un uomo che non si dà mai per vinto] (It.) we.considered Gianni a man that not self= gives never for won b. *Ritenevamo [un uomo che non si dà mai per vinto Gianni] (It.) we.considered a man that not self= gives never for won Gianni ‘We considered Gianni a man who never gives up’

It is interesting to consider copular sentences with an ambiguous interpretation such as the following one, which can either be an intensional description (e.g. the answer to a question ‘Who is this John you are talking about?’, interpretation A), or an extensional description attributing John to a kind (interpretation B). With 12

Or ‘predicate inversion’ in den Dikken’s (2006) terminology.

274

Paola Benincà and Guglielmo Cinque

interpretation A, we have a restrictive post-copular relative; with interpretation B we have a kind-defining post-copular relative clause: (34)

Mario è un uomo che incontri la mattina presto al bar (It.) Mario is a man that you.meet the morning early at.the bar A: ‘Mario is (precisely) that man that you meet early in the morning in the bar’ (identificational > restrictive) B: ‘Mario is the (kind of) man that you meet early in the morning in the bar’ (predicational > kind-defining)

If a clitic pronoun is inserted, doubling (or in place of) the wh-, the sentence ceases to be ambiguous and the relative has only a kind-defining interpretation:13 (35)

Mario è un uomo che lo incontri la mattina presto al bar (coll. It.) Mario is a man that him= you.meet the morning early at.the bar

13.6 Raising and matching derivations of kind-defining relatives To judge from the availability of sentences like those in (36) and (37), kind-defining relatives appear to allow, like restrictives (and unlike non-restrictives: see Bianchi 1999), both raising and matching derivations. (36a,b) involve the relativization of an idiom chunk, which requires a raising derivation (cf. Kayne 1994 and references cited there). The sentence in (37), on the other hand, appears to require a matching derivation, as no reconstruction of the Head inside the relative clause is involved (reconstructing unico fan di Mario inside the relative clause in a position c-commanded by lui would cause a violation of Principle C of the Binding Theory, which prohibits a pronoun to be c-commanded by its antecedents; cf. Sauerland 1999; 2003; Cinque 2008b). (36)

13

a. Questi sono conti che non tornano these are calculations that not return ‘These are things that simply don’t add up’

mica (It.) neg

b. La parte che deve aver avuto the part that he.must have.inf had ‘The role that he must have had is this’

questa (It.) this

è is

Another interesting consequence is that in the kind-defining version of the relative, the interpretation of the subject of the 2sg verb cannot be a deictic 2sg, but a generic impersonal. This is consistent, in some way, with the modalities that can appear in this kind of sentence.

Kind-defining relative clauses in the diachrony of Italian (37)

275

Giorgio è l’ unico fan di Marioi che luii deve ritenere Giorgio is the only fan of Mario that he must consider.inf degno dei suoi favori (It.) worthy of.the his favours ‘Giorgio is the only fan of Mario that he must consider worthy of his favours’

13.7 Post-copular relatives and agreement All of the properties we have seen so far are found to be possible with non-restrictive and post-copular kind-defining relative clauses, but are banned from ordinary restrictive relative clauses. With respect to these properties, modern standard Italian only differs from early Florentine and northern Italian dialects with respect to stylistic level: the clitic copy of the wh- is only possible in a colloquial style in modern Italian, and avoided in formal, written language.14 On the other hand, old Italian (early Florentine) post-copular kind-defining relatives presented a further property which has since been lost, namely, the fact that the verb in the relative clause agrees with the subject of the copular sentence rather than with the relative clause Head (Noordhof 1937). This can only be observed—for independent reasons—if the copula is in the 1st or 2nd person, as in the following examples: (38)

a. Or se’ tu quel Virgilio e quella fonte che spandi now are you that Virgil and that source that you.spread di parlar sì largo fiume? (OFlo., Dante Alighieri, Inf., 1, 79–80) of speak.inf so wide river ‘Are you that Virgil and that source from which sheds so wide a river of speech?’ b. Io son colei che sì importuna e fera I am she that so tiresome and fierce da voi . . . (OFlo., Petrarch, Trionfi, T. Mortis) by you ‘I am the one that you call so annoying and fierce’

chiamata called

son I.am

The same is found in other Italian vernaculars (see (38a)) and in modern French, as pointed out to us by Adam Ledgeway (cf. 38b):

14 As a reviewer observes, this is true of the other colloquial modern Romance languages, with the exception of Romanian, where copying of the wh- is part of the standard language, even in restrictive relative clauses; the exception of Romanian is possibly only apparent, if the wh-pronoun care in certain contexts has in fact the status of a complementizer, as argued in Cinque (2008a: 126–7).

276 (39)

Paola Benincà and Guglielmo Cinque a. he’ som quella che lo portay nove meysi e che lo I am that.one that him= I.bore nine months and that him= norigay com lo me’ layte . . . (OLig., quoted in Parry 2007a: 26 fn. 16) I.nourished with the my milk ‘it is I who carried him for nine months and nourished him with my milk’ b. C’ est moi qui suis / *?est venu te chercher ce matin (Fr.) it is me who am is come you= search.inf this morning ‘it’s me who came to look for you this morning’

In modern Italian, this characteristic is not totally unknown, but is strictly banned in formal styles, both written or spoken; a residue of this pattern is perhaps to be observed in infinitival relatives (Cinque 1988b; 2010a), in which a reflexive pronoun can agree in person with the subject of the copular sentence: (40)

a. Sono stato il solo ad accorgermi/-si di questo (It.)15 I.am been the only to realize.inf=myself/=himself of this ‘I was the only one to realize this’ b. *?Sono stato il solo che mi sono accorto di questo (coll. It.) I.am been the only that me= I.am realized of this ‘I was the only one that I realized this’ c. Sono stato il solo che si I.am been the only that himself= ‘I was the only one that realized this’

è is

accorto realized

di of

questo (It.) this

We have very little to say about this property, apart from the intuition that it might possibly be related to the lack of referential features in the endowment of the syntactic head of the relative clause.16 15

For some speakers this agreement scheme is the preferred one. The agreement scheme observed in the relative structures exemplified above is also found in another peculiar structure of early Florentine, which may support the idea that referentiality of the antecedent is relevant: 16

(i) a. tu, sì come colui che le mi desti, atar mi déi da lei you so as the.one that them= me= you.gave help.inf me= you.must from her (OFlo., Frescobaldi, Dino. Rime, 22, p. 418) ‘you, as the person that gave them to me, you must help me with her’ b. tu vorresti che fuoco venisse da ciel [ . . . ], sì come colui che se’ you would.like that fire come from heaven so as the.one that you.are così vago di noi come il can delle mazze (OFlo., Boccaccio, Dec. V, 10) so longing of us like the dog of.the cudgels ‘you’d wish for fire to come from heaven . . . being one that is so fond of us as a dog is of cudgels’ c. poche orazioni ho per le mani, si come colui che mi vivo all’ antica few orations I.have for the hands so as the.one that me= live to.the ancient (OFlo., Boccaccio, Dec. II, 2) ‘few prayers have I at my command, being one that lives in the good old way’ Ulleland (1961; 1963) develops a very penetrating interpretation, showing that the pronoun colui is not referential. This construction deserves further attention.

Kind-defining relative clauses in the diachrony of Italian

277

13.8 Conclusions and speculations We have collected evidence identifying a new type of relative clause, which we have called kind-defining, and have shown that its grammar and the features that characterize it have not changed from the first medieval records to the present time. A kind-defining relative can use a strategy that probably involves not wh-movement, but a resumptive clitic: the verb of the kind-defining relative on the subject does not agree with the wh-subject of the relative, but with the subject of the main clause. Only the register in which the surfacing of the distinctive features is permitted has changed: what was possible even in very formal texts in the 13th and 14th centuries is only permitted in the colloquial register in present-day Italian. Two grammars have coexisted in Italian for centuries with respect to this construction, changing only their stylistic collocation. Other aspects of grammar have changed dramatically in Italian, and in Romance more generally. V2 syntax, which was present in every language of the Romance area, has disappeared since the end of the 14th century, leaving traces in marginal areas such as Ræto-Romance and Portuguese (‘lateral areas’ in terms of Bartoli’s theory). This change affected the syntax of verb movement, and a continuation of the same process can still be seen at work in modern Romance varieties of northern Italy and in French, where the inflected verb in the interrogatives of some varieties still moves to a low Head in C, while in other varieties it no longer moves overtly but stays in Infl. Other, probably related, changes affected N(P)-movement in the DP (see Giusti 2006; Poletto 2006). These aspects can be classified in traditional terms as Head Movement: Head movement seems to progressively weaken through time. Another part of grammar that has been affected by a similar change, even though in more limited areas than Head movement, regards functional lexemes: in Benincà and Poletto (2005) changes affecting wh-pronouns, personal pronouns, and negation in Romance are analysed and described as a drift that allows speakers to avoid moving functional elements in the functional structure: negation disaggregates and, in the initial stage, only a subpart of the feature complex is moved (as in standard French ne . . . pas); the high element weakens and finally disappears, leaving only the postverbal element in a very low position (as in spoken French, where only pas now appears); wh- items and personal pronouns underwent a parallel evolution. Changes of this kind appear to reproduce the same scheme as the ‘cycle’ that Jespersen discovered in the history of negation.17 But other parts of syntax have remained the same from the Middle Ages. Since Romance languages developed clitic pronouns, they have come to realize Clitic Left We refer the reader to van Gelderen (2011) for the properties characterizing the ‘linguistic cycle’, and to Parry (1996; 1997) for the reconstruction of the negation cycle in northwestern Italy (for the cycle in Milanese, see also Vai 1996). 17

278

Paola Benincà and Guglielmo Cinque

Dislocation; from the High Middle Ages a preposed direct object with a clitic copy is unmistakably to be interpreted in Romance as a (Left Dislocated or Hanging) Topic (see Benincà 2006). Other constructions have only changed their stylistic collocation. Hanging Topic in particular is very similar in its evolution to kind-defining relatives. In Italian, for example, it is still as it was in the 14th century—a bare NP in the left periphery and an obligatory copy (direct or indirect object, depending on its grammatical function) in the sentential core. The only difference is that in the 14th century it was appropriate in any style, while today it is restricted to the very informal spoken language, as is the case of kind-defining relatives with a clitic copy. Further reflections should be devoted to syntactic properties or constructions that do not change for centuries, and other properties that are apparently undergoing slow continuous modifications, often with a precise direction; we should characterize and classify these different grammatical aspects in order to gain new insights into the theory of morphosyntactic change.

14 Synchronic and diachronic clues on the internal structure of ‘where’ in Italo-Romance* NI C O L A M U N A R O A ND C E C I L I A P OL E T T O

14.1 Introduction Classical and recent work on the syntax of constituent interrogatives has shown that there are different types of wh-items with different internal structures: the traditional distinction between bare wh-words and wh-phrases containing an overt lexical restrictor has been well known since Pesetsky (1987) and Reinhart (1998), among others. Investigating the internal structure of interrogative wh-words has proven particularly difficult, for the obvious reason that it can only be done in an indirect way, looking at the peculiar properties that they display in sentence structure. Capitalizing on Katz and Postal’s (1964) proposal that wh-words are construed with a feature on top of the nominal structure where the variable is inserted, we would like to explore the idea not only that the wh-item meaning ‘where’ is made up of a wh-component (corresponding to some sort of operator projection) and by a null lexical restrictor, but that its internal structure is more complex. We will adopt a line of reasoning that has been around for over twenty years, namely the original observation made by Baker (1988) that morphology and syntax mirror each other. This view has been systematically employed for verbal morphology since Belletti’s (1990) work on verb movement to T and Agr heads;1 it has also been employed to * This chapter is dedicated to Mair, a cherished colleague, a good friend, and an outstanding dialectologist. Although this work has been written together, Nicola Munaro takes responsibility for sections 14.1–14.3 and Cecilia Poletto for sections 14.4–14.6. We warmly thank two anonymous reviewers, Paola Benincà, Jacopo Garzonio, and Rachel Nye for constructive comments on previous versions. 1 Starting from Baker’s (1985) Mirror Principle, which states that morphology and syntax mirror each other, Belletti proposes that the order of the functional projections in the IP layer is not TP higher than AgrSP, as Pollock (1989) originally proposed, but AgrSP higher than TP. This explains the order of the functional morphemes on the verb, whose order is T followed by Agr, and not the opposite one, as predicted by Pollock.

280

Nicola Munaro and Cecilia Poletto

explain noun movement internal to the DP (see Cinque 1994), but has been only sporadically applied to the internal structure of PPs, and only in cases where a very complex structure is morphologically visible (Koopman 2010). Here we would like to extend this view to locative PPs in general, even those represented by a single word (as is the case for bare wh-words).2 Moreover, note that our extension goes beyond inflectional morphology as in the approach initiated by Baker, since it considers word formation, which is more akin to derivational morphology. In the spirit of Kayne’s (2007) intuition about English where (which he analyses as containing an r-pronoun preceded by a wh-form and with a third formative -e- that joins the two),3 we will consider the morphological formatives found in different Italian dialects and provide a first overview of the possible morphemes that can contribute to the compounding of ‘where’. The reason why we will take the wh-item ‘where’ as the first example of this methodology of investigation is that it displays a wide range of lexical realizations in different dialects, allowing us to make a typology of possible formatives. Moreover, we observe that the same lexical formatives occur in rather distant areas, which suggests that the ‘morphological pieces’ with which the wh-item is construed are recurrent across varieties. Our account is based on the (non-trivial) underlying assumption that each morphological formative corresponds to a functional projection, and that it is possible to reconstruct the internal layering of wh-words starting from their morphological composition. According to our comparative approach, the difference between, say, central Ræto-Romance ulà and Friulian dulà or the difference between Valfurva indoe and Chioggia indove is due to the presence of an additional formative in one variety that is not realized in the other; this additional formative is taken to correspond to a formal feature encoded by an independent functional projection.4 In order to find out what the morphological components are that make up the complex word, we will analyse a number of different but closely related dialects like those of northern Italy, so that the morphological formatives can be more easily identified on the basis of a comparative procedure. We will also occasionally make reference to the etymological source of those formatives, though our analysis should not be read as a proposal on the etymology of the various forms. Our view is that etymology is tangential to the synchronic analysis made by the speaker when s/he 2 The extension to derivational morphology of the idea that morphology mirrors syntax has already been applied in Cinque (2007) to the structure of diminutives in Italian. 3 According to Kayne (2007) r-pronouns are locative pronouns which contain an r-formative, namely ‘here’, ‘there’, and ‘where’. 4 It is clearly necessary to exclude the possibility that the changes in the form of the wh-item are due to phonological rules. We will address this issue later, at least for some cases. It is evidently impossible to undertake a detailed analysis of the morphophonology of all the dialects involved here; for our argument to go through, we only need to show that the alternation is not phonological for one pair of dialects. Actually, in the dialect of Valfurva there is no generalized rule according to which the voiced fricative [v] is deleted in intervocalic contexts, as shown by a form like rivaré ‘you will arrive’.

The internal structure of ‘where’ in Italo-Romance

281

acquires the properties of the functional items of her/his language, although in some cases etymology and the synchronic analysis of the speaker may coincide. The speaker has no knowledge of the previous stages of his/her language; s/he only tries to maximize the number of formatives to have them match the formal features and to cover the whole functional layering that s/he independently has as part of her/his endowment (for the idea that syntactic structure is part of Universal Grammar, see Cinque 1999; 2006). This is evidently not always possible, since, as we will see, several projections can be left unexpressed. Our aim is to find out these formatives, thus retracing the path the speaker has to follow to learn the properties of wh-items in her/ his language. However, given that even etymology is often uncertain in the case of functional items like wh-pronouns, we will exploit only the clearest etymological analyses to strengthen our point when necessary. This approach will enable us to determine which internal projections the wh-word shares with non-interrogative locative PPs and which ones are peculiar to wh-items. The comparison will lead us to conclude that wh-words contain an existential component in addition to an operator and a lexical restrictor. The chapter is structured as follows: in }14.2 we set the theoretical background and review previous work on the internal structure of interrogative wh-items. In }14.3 we show the lexical formatives of the words for ‘where’ in northern Italian dialects on the basis of chart 821 in vol. IV (Dove vai? ‘Where are you going?’) of the AIS and of the ASIt database for interrogative wh-items, and set out the attested forms and their morphological decomposition. In }14.4 we compare the internal structure of interrogative wh-items with the structure of PPs. In }14.5 we examine additional usages of the wh-item ‘where’, which can appear in relative, adversative, and result clauses and can take on the function of a preposition. Section 14.6 summarizes the main findings and points to further possible developments.

14.2 Theoretical background Since the work by Katz and Postal (1964), wh-items have been generally assumed to contain a nominal category which restricts the range of the possible values of the variable: according to Katz and Postal wh-items formed by single words (like ‘what’, ‘who’, ‘where’, ‘how’, ‘when’, and ‘why’) stand for the combination of the wh-item with an indefinite article and a lexical restrictor. The distinction between wh-items containing ‘which’ and those containing ‘what’ is due to the fact that with ‘which’ we have a definite article, with ‘what’ an indefinite one. Katz and Postal do not provide morphological evidence for their claim, although in Romance reflexes of the form qualis are actually analysed as containing an ‘l’ formative which could be synchronically related to the definite article (see Munaro 2001). As for the type of wh-items we investigate, we will see that there is no empirical evidence for postulating the actual presence of an indefinite article occurring within the forms corresponding to ‘where’.

282

Nicola Munaro and Cecilia Poletto

We will see, however, that Katz and Postal’s intuition with respect to the presence of an existential element is fundamentally correct (see }14.4). More recently, Vangsnes (2008) has used the same procedure we adopt here in looking at the morphology of several Scandinavian varieties in order to investigate the formatives present inside the wh-item ‘how’.5 Vangsnes’ work highlights the links between the morphology of different wh-items. We reproduce here (Table 14.1) his Table 8 as an example of the type of reasoning he follows. The table shows that a given form can translate one or more wh-items, which is interpreted as a consequence of the fact that the internal structure of the wh-items listed in the first line have at least one internal feature (i.e. internal projection) in common. TABLE 14.1. Manner how vs degree how, where, and what across Scandinavian varieties (and English) (Vangsnes 2008: Table 8)

English Swedish Faroese Danish/Bokmǻl Norwegian Nynorsk Norwegian Sogn Norwegian Eastern Norwegian Icelandic

Wh- manner

Wh- degree

Wh- locative

Wh- pronoun

how hur hvussu hvor-dan kor-leis ke-lais ǻ-ssen hvern-ig

how hur hvussu hvor kor ke ǻ hvað

where var hvar hvor kor ke-st ǻ . . . hen hvar . . .

what vad hvat hva(d) kva ke ǻ . . . (for no) hvað . . .

Vangsnes concludes that degree ‘how’ is always an integral part of manner ‘how’ across Scandinavian varieties, which can be morphologically more complex; the morphological complexity is provided by incorporation of an N and a case marker or of a past participle of the verb ‘do’. Complex manner ‘how’ can use the forms for ‘where’ and ‘what’, showing that there is a relation between them. Vangsnes’ inspiring work does not attempt to translate these observations into a fine internal structure of the various wh-items. We will try to do so, but only for the wh-item corresponding to ‘where’. We will assume that the internal structure of the wh-item ‘where’ is at least partially similar to the internal layering of locative PPs as proposed in Cinque (2010b). We will see that only some of the internal projections postulated by Cinque for locative PPs are made visible in ‘where’ through the insertion of lexical formatives in northern Italian dialects. 5 Vangsnes’ work strongly recalls the nanosyntactic approach developed in Starke (2009) and much subsequent work in its attempt to see beyond the word level into the internal formatives of syntactic terminals. We will adopt a similar approach, with the addition that we hypothesize that each morphological formative composing the word ‘where’ lexicalizes just one functional projection, as it expresses an independent formal feature.

The internal structure of ‘where’ in Italo-Romance

283

14.3 The formatives of ‘where’ in northern Italian dialects In this section we present the lexical formatives of the wh-item corresponding to ‘where’ in northern Italian dialects. Although our analysis is a synchronic and comparative one, some diachronic remarks are in order. According to Lichtenhahn (1951), Latin had three items for expressing ‘where’ (in a way similar to German wohin ‘whereto’, wo ‘where’, and woher ‘from where’), corresponding to quo ‘whereto’, ubi ‘where’, and unde ‘from where’, with ubi and unde being originally relative pronouns. The form expressing movement towards a goal is quo, which is not attested in any of the dialects we have investigated, the only possible exception being ku/inko in Alpine Lombard dialects such as those of Isone and Semogo.6 Lichtenhahn (1951) notes that quo was already substituted by the stative form ubi in late Latin; the nonexistence of forms derived from quo is therefore not surprising.7 In what follows we provide a list of the synchronic formatives that can be combined to form where, derived from an investigation that we carried out on the chart 821 vol. IV of the AIS and a search in the ASIt data base on the basis of all main interrogatives contained therein: (1)

a. b. c. d. e. f.

preposition in (as in Florentine indove) preposition de (as in Samolaco indoa) o/u formative (as in Campitello di Fassa olà) locative tonic pronoun là (as in Claut dulà)8 locative formative v(e) (as in Claut vi compared to Veneto dove) locative formative /n(d)/ (as in Novi Ligure ndonde)

Each of the formatives presented in the list above can be justified on the basis of minimal pairs found in different dialects, which clearly show the presence or absence of the formatives in (1).9 In our survey we have observed that several of the formatives may be missing in a lot of dialects, as Tables 14.2–14.8 show. Whether they are absent because they have been deleted by a phonological rule but are still present in the syntax, or whether they are totally absent already in the syntax, is a thorny problem that we are unable to solve for the moment. Notice that in a cartographic approach, all functional projections are always present, even when they are not lexically filled (see Cinque 1999).

A similar form, ko, is also attested in old Logudorese, as pointed out to us by a reviewer (REW 6969). According to Lichtenhahn (1951), the formatives used in Italian dialects derive from a combination of some of the following: ubi, unde, the locative demonstrative element illac, and the prepositions in and/or de. 8 Friulian has no clitic locative form and uses a full form, while other dialects use the same formative found in the locative clitic vi. 9 In quoting the forms, we follow the standard tradition and provide the point number for the examples taken from the AIS, while we quote the name of the village for the data taken from the ASIt. 6 7

284

Nicola Munaro and Cecilia Poletto

TABLE 14.2. Cross-linguistic evidence for the formative ina Venice 376: indove Parma 423: indo Aldon (VR) 372: ando Grado (UD) 367: indola Poschiavo (Swl.) 58: indond Bologna 456: indum Mathon (Swl.) 15: anté Osco (Swl.) 31: indé Soglio (Swl.) 45: andu Pejo (TN) 320: endo Pietraporzio (CN) 170: ente Corio (TO) 144: nté Galliate (NO) 139: nunda

Teolo (PD) 374: dove Fiume (HR) 379: do Fiume (HR) 379: do Ruda (UD) 359: dola Brescia 256: dund Bologna 456: dum Montanaro (TO) 146: te Isola S. Antonio (AL) 159: dé Pramollo (TO) 152: an Fiume (HR) 379: do Montanaro (TO) 146: te Montanaro (TO) 146: te Rochemolles (TO) 140: unta

a Notice that the form of the preposition can change the colour of the vowel, which can be [a] or [e] or [i] as in standard Italian. This could be explained by assuming that the vowel is actually an epenthetic vowel; and the same could be true of the final segment, which is also realized by different vowels or not realized at all. An anonymous reviewer points out that there might have been a previous stage where the nasal was syllabic. In this perspective, an could be the outcome of the preposition in or of (grammaticalized) unde, through a stage with no vowel and a syllabic nasal.

Hence, we assume that also in our case the projections internal to the wh-item are present, whether filled or not. Tables 14.2–14.8 display precisely this contrast between dialects that realize the formatives and dialects that do not. In Table 14.2 we show that there are dialects where the initial formative in, corresponding to a preposition, is realized while in other dialects it is not. Moreover, this formative can occur alone, as witnessed by the fact that an is the form for ‘where’ in the dialect corresponding to Pramollo (TO), point 152 of the AIS. In Table 14.3 we present minimal pairs of dialects that have the d- formative corresponding to the initial one of standard Italian dove (in the first column) and dialects that do not realize it (in the second column). In Table 14.4 we take into account dialects that display the vocalic formative u/o and dialects that do not. Moreover, this formative can also appear alone, as u is the form for ‘where’ in the dialect of Brusson (AO), Point 123 of the AIS (cf. also standard French). In Table 14.5 we present cases of dialects with and without the deictic locative formative là. The fact that là can occur alone as the form for ‘where’ also shows that this is an independent formative. In Table 14.6 we provide a list of dialects where the formative -v- is realized, contrasting them with others where it is not. This formative can also be found alone meaning ‘where’ in Zoagli (GE), point 187 of the AIS. There is no indication that this dialect has a regular rule deleting the

The internal structure of ‘where’ in Italo-Romance

285

TABLE 14.3. Cross-linguistic evidence for the formative d Isola (GE) 179: dande Mathon (Swl.) 15: ante Osco (Swl.) 31: indé Soglio (Swl.) 45: andu Malesco (NO) 118: du Tramonti di Sotto (UD) 328: dulà Calizzano (GE) 184: dan Ruda (UD) 359: dola Airole (IM) 190: dunda Brescia 256: dund

Sassello (GE) 177: ande Montanaro (TO) 146: te Isola S. Antonio (AL) 159: dé Ruata (TO) 152: an10 Brusson (AO) 123: u Dignano (HR) 398: ulà Ruata (TO) 152: an S. Vigilio di Marebbe (BZ) 305: olà Rochemolles (TO) 140: unta Taggia (IM): und

TABLE 14.4. Cross-linguistic evidence for the formative u/o S. Vigilio (BZ) 305: olà Soglio (Swl.) 45: andu Malesco (NO) 118: du

Forni Avoltri (UD) 318: là Fiamenga (CN) 175: and Brusson (AO) 123: u

TABLE 14.5. Cross-linguistic evidence for the formative là Dignano (HR) 398: ulà Tramonti di Sotto (UD) 328: dulà Ruda (UD) 359: dola

Forni Avoltri (UD) 318: là Malesco (NO) 118: du Fiume (HR) 379: do

TABLE 14.6. Cross-linguistic evidence for the formative v Venezia 386: dove Cozzo (PV) 270: duva Rivolta d’Adda (CR) 263 nduva Carpaneto (PC) 412: iduv Taglio di Po (RO): dov Comacchio (FE) 439: indov

Bienate (MI) 250: doe Albogno (NO) 109: dua Alba (CN) : indua Comano (Swl.): indu Fiume (HR) 379: do Livigno (SO): indo

10 Notice that this pair might not be a minimal one, as an could either be the high preposition formative or a low formative (see Table 14.8 below). This could also be the case for other formatives such as the ones in Table 14.3: te, de are in principle ambiguous between a higher and a lower formative. We report them here nonetheless for completeness.

286

Nicola Munaro and Cecilia Poletto

voiced fricative [v] in intervocalic position. However, even though we can exclude a regular phonological rule, there might be rules only applying to a restricted set of elements. Notice, however, that, as we see in Table 14.7, the consonant [v] never cooccurs in any dialect with the locative formative la.11 Clearly, this cannot be a phonological phenomenon, as not all dialects have a phonological rule which deletes the intervocalic [v]. In Table 14.7 we illustrate cases of dialects that realize the formative n(d). In several dialects a vowel [a] or [e] is added after the lower formatives v or d. We do not consider this vowel as an independent formative but as an epenthetic vowel (see fn. 11), for the following reasons: first, the vowel is not always the same; second, it generally corresponds to the epenthetic vowel of the dialect (for instance [e] of indove in Venetian is the ordinary epenthetic vowel in this dialect). Moreover, it seems to be found in the dialects where a word-final fricative is not allowed. Notice that this does not apply to any vowel ending the word, but only to the vowel that comes after the formatives v or the lower d. Before presenting all the forms we have come across in the northern Italian domain in a comparative survey through the AIS maps and the ASIt database, we would like to concentrate on some of the most complex forms which show rather clearly how the various formatives can combine. The forms we have chosen as a paradigmatic example are Tonezza (VI) andove, Grado (UD) indola, and Poschiavo (CH) indond. These forms share the initial component, in our view corresponding to the preposition ‘in’: in some varieties this formative is only expressed by the nasal component, in others it displays the initial /i/ segment, others display a different vowel (in the case of andove /a/). The second formative found is a d- formative, which we interpret also as the residue of a prepositional element (corresponding to either da or di in Italian). As we will see below, both elements correspond to the highest TABLE 14.7. Cross-linguistic evidence for the formative n(d) Bardi (PC) 432: dun Sommaino (Swl.) 58: indond Brescia 256: dund Gromo (BG) 237: ndund

Malesco (NO) 118: du Antronapiana (NO) 115: indun Bardi (PC) 432: dun Samolaco (SO): indu

11 As we might expect to find dialects where the two items combine and create forms like dovela or dolave, the fact that we never find them shows us that -v- occupies a precise position and is indeed an independent formative and not simply a segment. This also rules out the problem of possible insertion of -v- only for phonological reasons. More generally, we are always confronted with the problem that the forms might result from phonological rules of deletion or insertion. As for insertion, we do not think there are other possible cases beside the one of -v- just discussed. As for deletion, in this case the formative is present but deleted, hence the projection we claim to be there must be there anyhow.

The internal structure of ‘where’ in Italo-Romance

287

positions within the structure of locative PPs. The three examples chosen all present a vocalic formative -o- after the prepositions; some varieties also realize this as -u- or as -w-. In }14.4 we propose that this corresponds to the wh-morpheme of the complex form. The last formative is different in the three forms chosen: one presents -v-, another -la, and the third -nd. Nevertheless, we believe that these three formatives realize one and the same projection, as they never co-occur, as we see in Table 14.8. The last element—which we think is an epenthetic vowel, which can be absent in several dialects—is either /e/, /i/, or /a/. As anticipated above, the main reason why we think that this is only an epenthetic vowel is the instability of its realization across dialects.12 On the basis of the sequence of the relevant formatives that we have established, we interpret all the other forms we have found, where some of the formatives have no phonological realization. Table 14.8 provides a tentative decomposition of the 56 forms attested in the AIS atlas for northern Italy. The sample we use is most probably not complete for the whole geographical domain, but nonetheless provides some insights into the TABLE 14.8. Decomposition of the forms for ‘where’ attested in northern Italy in am an (i)n in

d d d d d

n (i)n i in

d d d d

in in n

d(a) d d d t

n (n) n in

d d d g

o w o o o u u o/u o/u u o o u u u u u u u o u w

nd (a)nd

là/v v v

nd nd nd v v v là nd nd n n m n nt nd

Epenthetic V a e e a/i a e/a

e/a e (a)

a

w

a a a e

Complete form amuanda* andove (i)ndove indond dunda/dundi nunda dove/duve/duva (i)ndov iduv indola onde/a dande dun(a) indun indum ntun unta dund ndua (n)doa nuwa* ingwe (continued)

12 One might assume that this is also a formative, most likely—according to our hypothesis that locative wh-items replicate at least in part the structure of locative PPs—the low locative preposition a ‘to/at’. However, in this case we would expect more stability of the vowel, which we do not find.

288

Nicola Munaro and Cecilia Poletto

TABLE 14.8. Continued in en an (i)n in an an en

ne/a (i)n an n (n sillabica) n an (a)n in

g d d d g d d t d d d d

e o u o/u o/u o

o/u u/o u u w o/u w

e/a e e/a là n m a a a

t g d t è (copula) d è (copula) d da

u/o o/u o

là n d v

n agn

e/i

o o t

t

a

e/a lo

d

d

e/a

u an là

inge* endo andu (i)ndo/ndu ingo/ingu* ando ande/anda ente doe/dua dulà/dola dun dum newa/nawa* (i)noa/nua anwa nt ng and (a)ntè indè ulà/olà do/du dan* od ve, vi no agno* te/ta* alo13 de/da* u an là

distribution of the formatives. It shows which formatives can co-occur with each other and which cannot.14 13 We follow here the etymology proposed by Lichtenhahn, who proposes that this form might be a residue of deictic forms like illò found in Bonvesin de la Riva (for details, see Lichtenhahn 1951). 14 The forms marked with an asterisk are ambiguous forms where a formative could be interpreted in two distinct positions (as the case of -nd-, which can either be the two high prepositions or the lower -ndformative derived from inde), or cases where the occurrence of a vowel could simply be due to phonological diphthongization or epenthesis. We are aware of the fact that we may have misanalysed some of the attested forms; however, we think that gathering all the forms we have come across can provide a sound

The internal structure of ‘where’ in Italo-Romance

289

On the basis of this schema, we can make the following generalizations concerning the compatibility of the formatives: (2)

a. The two formatives -là and -v never co-occur. b. The formative -nd is incompatible both with -là and with -v. c. Each formative alone can exhaustively represent the whole wh-form meaning ‘where’.

We interpret the generalizations in (2a) and (2b) to mean that these three morphemes, -là, -v, and -nd, express one and the same value, namely locative deixis, as the etymology of là as illac suggests (see below for a more theoretical discussion). The last generalization is interesting from the syntactic point of view, as we will discuss in detail.

14.4 Matching the structure of locative PPs The decomposition of the various formatives of the wh-item ‘where’ has revealed that it is made up of three types of element: a prepositional formative, which can correspond to the preposition in ‘in’, di ‘of ’, or da ‘from’, or a combination of them; a vocalic formative o/u; and a deictic locative element derived from various sources -nd, -v, or -la. The internal structure of wh-items is generally assumed to at least partially reflect the structure of the corresponding element they question: hence something like ‘who’ or ‘what’ will have a nominal structure underneath the operator layer (as already proposed by Katz and Postal 1964). Given that ‘where’ expresses a locative wh-item, this has to be compared with the structure of locative PPs in order to establish which projections the two categories share. We adopt here Cinque’s (2010b) proposal for the internal structure of locative PPs (but see also a very similar proposal by Svenonius 2010): (3)

[PPDirSource from [PPDirGoal to [PPDirPath across [PPStat AT [DegreeP two miles [ModeDir diagonally [AbsViewP north [RelViewP in [DeicticP here [AxPartP under [PP [P ] [NPplace the mountain [NP PLACE]]]]]]]]]]]]]

If this general view of wh-items is correct, then the null hypothesis is that the locative wh-item realizes at least some of the syntactic projections internal to a PP, and hence there should be a partial correspondence between (3) and the internal structure of ‘where’. An independent difference between complex locative PPs and the wh-item is the fact that it is possible to sub-extract the internal projections in a PP, but the empirical basis for future research in this field. Therefore we decided to keep them all in light of their documentary value, as they witness the wide variability found in this domain.

290

Nicola Munaro and Cecilia Poletto

formatives in the wh-item always come in the same order; this amounts to saying that while it is possible to move in the syntax, the morphological component admits no movements of any sort.15 In other words, the line we take is the same as that usually adopted in the Minimalist Program, where complex morphological forms are not created in the syntax but are already present in the lexicon, and each internal morpheme checks the features of a FP. Although the various morphemes correspond each to a FP, they cannot move independently, because they are not independent words—which amounts to saying that movement of morphemes does not exist in the syntax, as standardly assumed in the Minimalist framework. Therefore, we take the formatives in (4) to be unmovable in the extended projection of the wh-item. This means that the order of the functional projections can be directly read off the order of the formatives. This assumption depends on the fact that each formative only expresses one functional feature, which corresponds to a single syntactic projection and therefore cannot be moved through feature-driven movement.16 Inserting the formatives of the wh-item inside the structure of the PP in (3), we obtain the structure in (4): (4)

[PPDirSource da [PPDirGoal in [PPDirPath [whP o/u [StatP [DegreeP [ModeDirP [AbsViewP [RelViewP [DeicticP là/v/nd [AxPartP [PP [P ] [NPplace e [PLACE]]]]]]]]]]]]]]

This structure contains a set of prepositions, which can be realized in different dialects. In (4), the element o/u corresponds to the wh-operator feature, as we are led to interpret considering its etymology: the form -u is one of the two possible realizations of the original Indo-European wh-formative qw, which then was simplified in -u through the loss of the original velar (cf. Pokorny 1930: 1099). We will adopt the version of the internal structure of wh-items proposed in Poletto and Pollock (2009): (5) [DisjP [ExistP [RestrictorP ]]] If -u is a wh-formative, we can hypothesize that it corresponds to the DisjunctionP in (5). If we further compare the structure in (4) with the one in (5), we can reinterpret the locative formative as marking the existential feature in (5), while the restrictor position is occupied in the wh-item ‘where’ by a null PLACE as indicated in (4).

15 Notice that our approach is incompatible with Distributed Morphology, as the lexical formatives have to be already present in the syntactic component. 16 As for non feature-driven movements at PF, we also assume that our formatives cannot be targeted by this type of movement, since at PF the form of the entire word is already established and cannot be scrambled.

The internal structure of ‘where’ in Italo-Romance

291

The fact that in many languages the form of wh-items is identical to existential quantifiers has led several authors (see Cole et al. 2001 and references therein) to conclude that wh-items are not operators but bound variables. In languages that display in situ wh-forms with question interpretation, this results from these variables being bound by null or overt question operators. Indefinite interpretations result from these same variables being bound by other null or overt indefinite operators, or by existential closure. Here we do not provide a semantic account but rather concentrate on the morphological component in a wide range of dialects: from what emerges, we can conclude that wh-items include an existential/indefinite component (though not in the form of an indefinite article, as originally proposed by Katz and Postal 1964); but this is not the only element present in the internal structure of the wh-item. In accordance with Cole et al. (2001), it might be the case that in those languages, like Chinese or Japanese or even German (at least for those forms like was ‘what/ something’ and wer ‘who/someone’), where the form of the wh-item is identical to the existential, the higher projections internal to the wh-item are not morphologically realized, though they are present; the choice between the two alternatives in (6a) and (6b) would be the source of the ambiguity between an interrogative and an existential reading of the actual morphological forms: (6) a. [DisjP [ExistP was [RestrictorP ]]] b. [ExistP was [RestrictorP ]] Our investigation has shown that there are other projections on top of the existential one, and that they can be filled or not depending on the dialect, though they are always projected, as these dialects only use these forms for wh-items and not for indefinites. This suggests that, more generally, wh-items are not indefinites/ existentials per se although they always contain an indefinite component, which is morphologically spelled out in several dialects, but not in all. Notice furthermore that (6), since it is a structure which is valid for all wh-items, does not contain a prepositional XP; however, the data presented in Table 14.8 show that prepositions are possible formatives for the wh-item ‘where’, and in some cases they can also be the only one (see forms like de/da, te/ta, an).17 17 An indirect argument that SourceP is the most external projection, as it is in regular PPs, comes from the data adduced in Lichtenhahn (1951) in the course of her discussion of the development of prepositions expressing source: while Latin unde already expressed source, in Renaissance Italian the form is strengthened by the preposition di yielding donde or even didonde, where the preposition has been incorporated into the wh-form and reinserted once more: in La vita by Benvenuto Cellini the forms onde, d’onde, donde, di donde all alternate (apparently) freely in addition to di dove. Incidentally, this shows that there is a grammaticalization process of the most external preposition expressing source which becomes part of the wh-item (for a detailed analysis of the grammaticalization cycle of prepositions, see van Gelderen 2011). Notice that the grammaticalization process can also target the more internal preposition in/di, as shown by Table 14.8.

292

Nicola Munaro and Cecilia Poletto

If we integrate the layering of wh-items in (5) into structure (4), we obtain a final internal structure of the wh-item ‘where’ as follows: (7)

[PPDirSource da/di [PPDirGoal in [PPDirPath d [DisjP o/u [StatP [DegreeP [ModeDirP [AbsViewP [RelViewP [DeicticP/ExistP là/v/nd [AxPartP [PP [P ][NPplace/Restrictor e [PLACE]]]]]]]]]]]]]]

Comparing (7) to the structure of a regular PP (see (3)), we see that only four of the potential functional positions are relevant for the wh-item: the two highest PPs expressing Source and Goal, the DisjunctiveP (only typical of wh-items and not of the corresponding PP) and the DeicticP/ExistentialP, in addition to the lexical projection hosting the restrictor.18 As discussed above, we assume that the whole extended projection in (7) is active even when a single lexically realized morpheme is present, irrespective of whether it occupies a high or low position.19

14.5 Ubiquitous ‘where’: relative, interrogative, and prepositional forms In this section we will describe the distribution of the various forms across interrogatives and relatives in order to highlight the interplay between the structure of the two clause types and the wh-item used. 14.5.1 Relative vs interrogative clauses It is well known that the wh-item ‘where’ can be used in relative clauses as well as in interrogative clauses, and several (though not all) dialects display such a convergence. Among the dialects that have two different forms for interrogative and relative ‘where’, the first distinction concerns those varieties that have a clitic form generally used in main interrogatives (and in some varieties also in embedded ones) but not in relative clauses. For instance the Friulian variety of S. Michele al Tagliamento displays two forms, one clitic (do) and one tonic (dulà). The clitic form can only occur close to the inflected verb; hence it is ungrammatical in embedded interrogatives and in contexts where the wh-item is split from the inflected verb by intervening material, as is also the case in relative clauses, where there is no V-to-C movement:

18 Notice that in Italian there is a difference between the wh-item and the quantifiers expressing location, which always contain a lexically realized noun parte ‘part’ or luogo ‘place’, as with in nessun luogo/da nessuna parte ‘nowhere’ or in qualche luogo/da qualche parte ‘somewhere’. 19 Many problems remain to be solved, some of them at the border between morphology and phonology. For instance, we should check whether the alternation between /u/ and /o/ as a formative is a result of regular phonological rules of a single dialect. Another problem concerns the final vowel of forms like dove, which sometimes occurs as /a/, sometimes as /e/, and sometimes is not present (see Lichtenhahn 1951: 59ff.). This should also be related to phonological properties of the different dialects, and if this turned out not to be the case, we would have to look for a morphosyntactic explanation for its presence.

The internal structure of ‘where’ in Italo-Romance (8)

Do (*a) van-u? where a go=they? ‘Where are they going?’

(9)

a. A mi an domandat dulà ch al era sut scl= me= have asked where that he was gone ‘They asked me where he had gone’

293

b. *A mi an domandat do ch al era sut scl= me= have asked where that he was gone ‘They asked me where he had gone’ c. Dulà ch’ al vedi mitut chel libri? where that he has.sbjv put that book? ‘Where could he have put that book?’ The standard tests for clitic-hood (coordination, modification, use in isolation) show that the shorter form is actually a syntactic clitic (for a detailed discussion, see Poletto 2000: 25): (10)

a. *Do e quant van-u? where and when go=they? ‘Where and when do they go?’ b. Dulà/*Do? where? c. Di dulà al vegna?/ *Di do al vegna? from where scl come? ‘Where does he come from?’ d. I so-tu zut dulà? / *I so-tu zut do? scl are=you gone where? ‘Where have you gone?’

The eastern Lombard variety of Monno also displays two forms: the tonic one is located in situ, the clitic one at the beginning of the clause. They can also double, and the clitic form is also used in embedded clauses (but only if the complementizer is not realized: see Munaro 1999: 192): (11)

a. Ngo fet majà ngont? where do=you eat where? b. Ngo fet majà? where do=you eat? c. Fet majà ngont? do=you eat where? ‘Where will you eat?’

294

Nicola Munaro and Cecilia Poletto

(12) I m’ à domandà ’ngo la fus ’ndada la Maria they me= have asked where she was gone the Mary ‘They asked me where Maria had gone’ There are dialects where the distinction between interrogatives and relatives is not related to a clitic form occurring only in interrogatives: the dialect of Grosio (Alpine Lombard) presents the form inghé for the relative and the form ingù for the interrogative (in addition to indoe, which is possible in both relatives and interrogatives). The dialect of Semogo (Alpine Lombard) displays the form ando only in main interrogatives, the form indo in embedded interrogatives and relatives, and the form inco only in relative clauses. Another interesting distinction has to do with the realization of the complementizer, which, as noted by Lichtenhahn (1951), is much more frequent in relative clauses than in interrogatives (see e.g. the Nònese dialects, where the complementizer occurs in relative clauses but not in embedded interrogatives; the same is true of some RætoRomance dialects such as Corvara in the Badia valley): (13)

El paés en dóche me sen encontradi la prima bota el se ciama Mirano the village in where.that we are met the first time it is called Mirano ‘The village where we first met is called Mirano’

There are also dialects where the complementizer is obligatory in both relatives and interrogatives (as in Venetian): (14)

a. No so dove che el ze ndà not know where that he is gone ‘I do not know where he went’ b. El paese dove che se gavemo conossuo se ciama Miran the village where that ourselves have met is called Mirano ‘The village where we first met is called Mirano’

However, the complementizer alone can substitute for the wh-item in relative clauses, but not in interrogatives (see maps 1535–6 in vol. VIII of the AIS), as in the dialect of S. Stino di Livenza: (15)

La cuzia che ti te cuze ades she sewed that you sew now ‘She used to sew where you are sewing now’

(16)

Al posct ch ’em s’è incontrei la prim’ olta, al se ciama Mirano the place that scl=ourselves=have met the first time is called Mirano ‘The village where we first met is called Mirano’

The internal structure of ‘where’ in Italo-Romance

295

These distinctions point towards an internal difference between the interrogative and the relative form. However, given that the data are relatively sparse and nonhomogeneous, we leave this topic to future research. 14.5.2 Dove inside locative expressions Lichtenhahn (1951) reports cases of dove used as a preposition in Renaissance Italian (15th c.): (17)

Mi usciva la voglia di non mai tornare dove lui me= exited the lust of not never come.back.inf where him ‘I felt like not going back to him ever again’ (Benvenuto Cellini, La vita, p. 19)

This phenomenon is also noted by Rohlfs (1969: 223), who provides several examples of an ambiguous form used for the wh-item ‘where’ as well as for a locative preposition in various southern Italian dialects. The phenomenon is attested in the Calabria, Campania, and Basilicata areas: (18)

a. Iamu duve u miedicu (Cal.) we.go where the doctor ‘We are going to the doctor’s’ b. Vaju duve u cummientu (Cal.) I.go where the monastery ‘I go to the monastery’ c. Ti spettu undi mia (S.Cal.) you I.wait where me ‘I’ll wait for you at my place’ d. È gghiuto addò la sora (Nap.) is gone where the sister ‘He has gone to his sister’s’

From our point of view, these cases are expected if we think that the distinction between the internal structure of the wh-item and the internal structure of the PP is related to the activation of the wh-feature and to the low NP position, which contains a null variable in the case of wh-items and a lexically realized nominal expression in cases like (18). Deactivating the wh-feature and filling the NP position, we obtain a normal PP, as illustrated by the contrast between (19a) and (19b): (19) a. [PPDirSource da/di [PPDirGoal in [PPDirPath d [DisjP o/u [StatP [DegreeP [ModeDir [AbsViewP [RelViewP [DeicticP/ExistP là/v/nd [AxPartP [PP[P ] [NPplace/Restrictor e [PLACE]]]]]]]]]]]]]]

296

Nicola Munaro and Cecilia Poletto

b. [PPDirSource da/di [PPDirGoal in [PPDirPath d [DisjP [StatP [DegreeP [ModeDir [AbsViewP [RelViewP [DeicticP/ExistP là/v/nd [AxPartP [PP[P ] [NPplace/Restrictor [PLACE] DP ]]]]]]]]]]]]] As for the deictic element, it probably takes on a neutral reference, and is not related to the speaker as in the case of wh-items. Under an alternative analysis, these structures could be interpreted as reduced relative clauses meaning ‘he went to the PLACE where the doctor IS’;20 in this case the null PLACE component might have been extracted from within the wh-structure as in head-internal relative clauses.21 Cinque’s (2010b) structure, which we assume here, is in accordance with Kayne’s (2007) interpretation of the wh-item ‘where’, which he takes to be paired with a null noun PLACE. Kayne (2007) suggests that nouns can only be null when they are located at the edge of a phase. In structures like those in (19) the null PLACE would be located at the edge of the DP phase.22 If our hunch about prepositional ‘where’ being a reduced relative clause is correct, we would expect PLACE to raise from the edge of the DP to the edge of the whole PP. The reduced relative clause would thus have a null PLACE in its highest position, in a way which strongly recalls one of the possible analyses of free relative clauses, where a null noun precedes the wh-item. In Kayne’s analysis the wh-item ‘where’ can also contain other null nouns like TIME, REASON, WAY.23 This predicts that there exist other usages of ‘where’ related to these null nouns. In the next section we will see that this prediction is borne out. 14.5.3 Other usages Like other elements introducing an adverbial clause, ‘where’ can be extended to other usages, in a way parallel to the element ‘while’, which has a core temporal meaning and a peripheral adversative one (for a detailed analysis of these two types of clauses, see Haegeman 2010). We will list here some of the usages that the item corresponding to ‘where’ can acquire in older or modern varieties of Italian, and we will try to sketch a possible diachronic path through which the locative item might have developed these additional usages. 20 An anonymous reviewer points out that this is true historically for Campanian varieties, where you can, indeed, find in old texts the copula after ‘where’, which through time seems to have been eliminated. 21 An argument in favour of this hypothesis are the data coming from Sardinian varieties, where a form of the verb ‘be’ is realized after the wh-element in the AIS maps IV 821 and VIII 1637; ‘at the doctor’s’ is translated as und è lu dutturi, lit. ‘where is the doctor’ (AIS 916) and ‘by us’ is translated as aumbe semu nuois (AIS 938), lit. ‘where are we’. 22 Evidence supporting this is the overt realization of the null nouns in some dialects which translate ‘where do you come from’ as da ke banda/part, literally ‘from which part’ (see AIS, map II, donde vieni? ‘where do you come from?’). 23 Kayne (2007) discusses cases of other null nouns and concludes that ‘where’ is never paired with the null noun THING. We follow here his suggestion, although we have no empirical evidence in favour of or against it.

The internal structure of ‘where’ in Italo-Romance

297

In old Italian, ove can also take on a temporal meaning, as in this example from Dante:24 (20)

come si frange il sonno, ove di butto nova luce percuote il viso chiuso as self= breaks the sleep where abruptly new light hits the face closed ‘as sleep is broken when new light strikes suddenly on the closed eyes’ (Dante, Purgatorio XVII l. 40, quoted by Lichtenhahn 1951: 4)

This type of construction might have developed in a way similar to the one we saw above for the prepositional usage of the wh-form meaning ‘where’, i.e. starting from a relative clause where ove is the relative pronoun which refers to a null restrictor noun functioning as the head of the relative clause. In the case of a temporal meaning, the null restrictor would be something like TIME.25 The item ove/dove/laddove is also used as an adversative conjunction, similar to English ‘whereas’. Old Italian presents cases of ove (and the modern Florentine variety still tolerates this form): (21)

Lagrime triste e voi tutte le notti m’accompagnate, ov’ io vorrei tears sad and you all the night me accompany, where I would.like star solo to stay alone ‘Sad tears and you all the nights accompany me, whereas I would like to be alone’ (Petrarca, Canzoniere, XLIX, 10)

On the other hand, modern Italian uses either dove, or in a higher register laddove (lit. ‘there.where’): (22)

Ha deciso di parlargliene, laddove io non lo avrei fatto. he decided of talk=to.him=about.that, whereas I not it= would.have done. ‘He decided to talk to him about it, whereas I would not have done it’

It is also interesting to notice that in these cases there is an additional deictic locative element la- ‘there’ at the beginning, which might support the tentative idea that these extensions of meaning are all derived through the same mechanism exploiting a relative structure whose head is extracted from within the wh-item: in

24 It is interesting to note that the form of the wh-item lacks the prepositional formative -d-, and this might not be due to chance. 25 That this might be a more general process is confirmed by the fact that analogous extensions from the locative to the temporal dimension are observed with several locative prepositions. The prepositions a, in, per all have both usages; see e.g. a casa ‘at home’, alle tre ‘at three o’clock’, in casa ‘at home, in the house’, in tre minuti ‘within three minutes’, per terra ‘on the floor’, per tre minuti ‘for three minutes’. As suggested to us by E. Brandner (p.c.), the link between the locative and the temporal value of ‘where’ seems to be rather systematic across languages, as the cases of Ger. wo, Fr. où, and Ro. unde show. We exemplify the link with Romanian examples:

(i) o vreme unde justiţia se va face (Ro.) a time where justice=the self= will make ‘a time when justice will be done’

298

Nicola Munaro and Cecilia Poletto

this case what is extracted is the locative component la- (probably plus the null restrictor PLACE). A similar link between a temporal and an adversative interpretation is also found with the element ‘while’ (It. mentre). The etymology of forms like Italian invece or English instead, which derive from the merger of the preposition ‘in’ with a noun (and of transparent forms like It. al posto tuo = Eng. in your stead/place), suggests that we might be on the right track in assuming a null PLACE (or a null TIME) in these constructions. The reason why the deictic la- is also realized is the presence of a deictic component in these cases, realized by your/tuo in cases like in your stead/al posto tuo. A conditional usage is also found in old Italian (cf. 23) and preserved in modern Italian only at a very high stylistic level (cf. 24); again the item does not include the formative -d: (23)

che senza fallo, ov’ egli avvenga che io scampi, io lo serverò that without fail, where it happens that I escape, I it= will.preserve fermamente firmly ‘that without fail, if I happen to escape, I will preserve it firmly.’ (Boccaccio, Decamerone, Giornata III, novella VII)

(24)

a. Ove tu volessi, potresti farlo where you wanted, could do.inf=it ‘In case you wanted, you could do it’ b. ??Dove tu volessi, potresti farlo where you wanted, could do.inf=it

In old Italian the form onde is also found in result clauses (25) and reason clauses (26): (25)

Ma ben veggio or sì come al popol tutto favola fui gran tempo, but well see.I now so as to.the people all tale was long time, onde sovente di me medesmo meco mi vergogno so that often of me self me.with me ashame ‘But now I see very well how everybody was talking about me, so that I feel ashamed of myself ’ (Petrarca, Canzoniere, I)

(26) La gola e ’l sonno e l’ oziose piume hanno del mondo ogni the greed and the sleep and the idle feathers have of.the world every vertù sbandita, ond’è dal corso suo quasi smarrita nostra natura virtue banned, so.that from.the course its almost lost our nature vinta dal costume. defeated by.the habit ‘The greed, the sleep and the idle feathers have banned every virtue of the world, so that our nature, defeated by the habit, has almost lost its course.’ (Petrarca, Canzoniere, VII)

The internal structure of ‘where’ in Italo-Romance

299

In modern Italian ‘where’ also occurs in some fixed expressions where it is used as a nominal element meaning ‘place’, like per/in ogni dove ‘anywhere’ (lit. ‘through/in any where’). Also in these cases we are led to assume that a null PLACE is associated with the construction, as Kayne proposes for English. A further fixed expression is the one found in cases like averne ben donde (lit. ‘to have good wherefrom’), which means ‘to have good reason’.26 Kayne (2007) also proposes that ‘where’ can be associated to the null noun REASON. If this is the case here too, then the null noun must have derived from a process that identifies REASON with ORIGIN in the sense of the locative source, as the English translation ‘wherefrom’ clearly shows.

14.6 Concluding remarks From the analysis of the formatives presented in Table 14.8 we can draw the following generalizations: (a) A single formative is enough to realize and activate the whole set of projections of a wh-item (see e.g. the case of ‘u’, which realizes only the wh-formative). (b) The relevant formative can be of any type (see the case of an which only realizes a prepositional formative and the case of là which only realizes the deictic formative). (c) The occurrence of a single formative is not related to any other formative; they can all occur independently from each other, and hence all combinations are possible. (d) The only formatives that are in complementary distribution are those realizing the same feature (e.g. the locative). We have seen that lexical formatives can be of three types: a preposition, a whformative, and a deictic locative. We have argued that in addition to these three visible categories there is a fourth null one, namely the null restrictor, which is a classifier-like noun. As for the idea that wh-items are bound variables, we have observed that the whitem ‘where’ does not provide any morphological evidence for the occurrence of an indefinite article inside the structure of the wh-item, as one might expect on the basis of the original proposal by Katz and Postal (1964). However, the locative deictic formative found in several dialects can indeed be interpreted as standing for an

26 A reviewer suggests to us that donde in this case means something like di cui ‘of which’, as witnessed by the fact that de unde is also the origin of Fr. dont ‘whose’. Notice, however, that if we extend the idea that these are cases of reduced relative clauses with an empty verb meaning ‘to complain’, the empty noun REASON must still be present in the structure of the reduced relative, as TIME is in temporal clauses and PLACE is in cases in which ‘where’ is apparently used as a preposition (see }14.5.2).

300

Nicola Munaro and Cecilia Poletto

existential element. Hence, under our analysis, wh-items contain an existential component in addition to the wh-operator and the lexical restrictor. These findings could be applied to languages where wh-items actually have the form of indefinites although they can behave as wh-items. In our view, wh-items are never really only indefinites: in some languages the operator component is simply not morphologically spelled out, as is the case in several of our dialects, where even a simple preposition can stand for the whole wh-item. That morphological forms can be ambiguous is also witnessed by the fact that the wh-item ‘where’ can be used as a preposition, a temporal, a reason, a causal, or an adversative element. To interpret these cases we have adopted Kayne’s (2007) proposal that the form ‘where’ is associated with different types of null nouns (or maybe null classifiers like TIME, REASON, and PLACE) which trigger the different interpretations. Both the prepositional usage and the relative usage can be captured under the assumption that they possess a structure similar to free relative clauses, in that they contain a null noun on the left of the wh-item.

References Abeillé, Anne, and Godard, Danièle (2000). Varieties of esse in Romance languages. In Dan Flickinger and Andreas Kathol (eds), Proceedings of the 7th International Conference on Head-Driven Phrase Structure Grammar. Stanford, Calif.: CSLI. http://cslipublications. stanford.edu/HPSG/1/hpsg00.html Abeillé, Anne, and Godard, Danièle (2010). Complex predicates in the Romance languages. In Danièle Godard (ed.), Fundamental Issues in the Romance Languages. Stanford, Calif.: CSLI, 107–70. Adams, James (1976). A typological approach to Latin word order. Indogermanische Forschungen 81: 70–99. Adams, James (1991). Some neglected evidence for Latin habeo with infinitive: the order of the constituents. Transactions of the Philological Society 89: 131–96. Adams, James (2003). Bilingualism and the Latin Language. Cambridge: Cambridge University Press. Adams, James (2013) Social Variation and the Latin Language. Cambridge: Cambridge University Press. Aebischer, Paul (1960). La finale -e du féminin pluriel italien: étude de stratigraphie linguistique. Studi linguistici italiani 1: 5–48. Aebischer, Paul (1961). La finale -i des pluriels italiens et ses origines. Studi linguistici italiani 2: 73–111. Alberti, Roberto (1993). Die Mundart von Gavardo, Provinz Brescia. Geneva: Droz. Albin, Thomas James (1984). Syntactic peculiarities of Piedmontese. Doctoral thesis, University of Michigan. Alboiu, Gabriela, and Motapanyane, Virginia (eds), Comparative Studies in Romanian Syntax. Amsterdam: Elsevier. Alexiadou, Artemis, and Anagnostopoulou, Elena (1998). Parametrizing AGR: word order, verb-movement and EPP checking. Natural Language and Linguistic Theory 16: 491–539. Anderson, Stephen (2008). Phonologically conditioned allomorphy in the morphology of Surmiran (Rumantsch). Word Structure 1: 109–34. Anderson, Stephen (2010). Failing one’s obligations: defectiveness in Rumantsch reflexes of debere. In Matthew Baerman, Greville Corbett, and Duncan Brown (eds), Defective Paradigms. London: British Academy and Oxford University Press, 19–34. Aranovich, Raúl (2003). The semantics of auxiliary selection in Old Spanish. Studies in Language 27: 1–37. Aronoff, Mark (1994). Morphology by Itself: Stems and Inflectional Classes. Cambridge, Mass.: MIT Press. Aski, Janice, and Russi, Cinzia (2010). The pragmatic functionality of atonic double object clitic clusters in 14th-century Florentine. Folia Linguistica Historica 31: 47–96.

302

References

Aub-Büscher, Gertrud (1962). Le parler rural de Ranrupt (Bas-Rhin): essai de dialectologie vosgienne. Paris: Klincksieck. Baker, Mark (1985). The mirror principle and morphosyntactic explanation. Linguistic Inquiry 16: 373–416. Baker, Mark (1988). Incorporation: A Theory of Grammatical Function Changing. Chicago: University of Chicago Press. Baker, Mark (1996). The Polysynthesis Parameter. Oxford: Oxford University Press. Baker, Mark (2008a). The Syntax of Agreement and Concord. Cambridge: Cambridge University Press. Baker, Mark (2008b). The macroparameter in a microparametric world. In Theresa Biberauer (ed.), The Limits of Syntactic Variation. Amsterdam: Benjamins, 351–74. Barbi, Michele (ed.) (1932). La Vita nuova di Dante Alighieri. Florence: Bemporad. Barbieri, Alvaro, and Andreose, Alvise (eds) (1999). Marco Polo: Il Milione Veneto. Venice: Marsilio. Barbosa, Pilar (1995). Null subjects. Doctoral thesis, MIT. Barbosa, Pilar (2009). Two kinds of subject pro. Studia Linguistica 63: 2–58. Battisti, Carlo, and Alessio, Giovanni (eds) (1951). Dizionario etimologico italiano (5 vols). Florence: Barbèra. Battye, Adrian (1990). Quirky agreement in Genoese. Paper presented at the Crucial Languages Seminar, Geneva. Beaulieu, Louise, and Cichocki, Władisław (2009). Patrons sociolinguistiques chez trois générations de locuteurs acadiens. In Monique Dufresne, Fernande Dupuis, and Etleva Vocaj (eds), Historical Linguistics 2007: Selected Papers from the 18th International Conference on Historical Linguistics, Montréal, 6–11 August 2007. Amsterdam: Benjamins, 211–22. Belletti, Adriana (1990). Generalized Verb Movement. Turin: Rosenberg & Sellier. Belletti, Adriana (ed.) (1993). Syntactic Theory and the Dialects of Italy. Turin: Rosenberg & Sellier. Benincà, Paola (1988). L’ordine degli elementi della frase e le costruzioni marcate. In Renzi et al. (1988), i.115–92. Benincà, Paola (1993). Sintassi. In Alberto Sobrero (ed.), Introduzione all’italiano contemporaneo: le strutture. Bari: Laterza, 247–90. Benincà, Paola (2003). La frase relativa in fiorentino antico. Paper presented at the ‘V Incontro di dialettologia’, University of Bristol, Sept. 2003. Benincà, Paola (2006). A detailed map of the left periphery of Medieval Romance. In Raffaella Zanuttini, Hector Campos, Elena Herberger, and Paul Portner (eds), Crosslinguistic Research in Syntax and Semantics: Negation, Tense, and Clausal Architecture. Washington, DC: Georgetown University Press, 53–86. Benincà, Paola (2007). Clitici e ausiliari: gh ò, z é. In Bentley and Ledgeway (2007), 27–47. Benincà, Paola (2012a). Frasi relative e strutture copulari. In Vincenzo Orioles and Paolo Borghello (eds), Per Roberto Gusmani: studi in ricordo. Udine: Forum editrice, 251–67. Benincà, Paola (2012b). Lexical complementizers and headless relatives. In Laura Brugè, Anna Cardinaletti, Giuliana Giusti, Nicola Munaro, and Cecilia Poletto (eds), Functional Heads. New York: Oxford University Press, 29–41.

References

303

Benincà, Paola, and Cinque, Guglielmo (2010). La frase relativa. In Salvi and Renzi (2010), i.469–507. Benincà, Paola, Cinque, Guglielmo, De Mauro, Tullio, and Vincent, Nigel (eds) (1996). Italiano e dialetti nel tempo: saggi di grammatica per Giulio C. Lepschy. Rome: Bulzoni. Benincà, Paola, and Poletto, Cecilia (2005). On some descriptive generalizations in Romance. In Guglielmo Cinque and Richard Kayne (eds), The Oxford Handbook of Comparative Syntax. New York: Oxford University Press, 221–58. Benincà, Paola, and Tortora, Christina (2010). On clausal architecture: evidence from complement clitic placement in Romance. In Vicenç Torrens, Linda Escobar, Anna Gavarró, and Juncal Gutiérrez (eds), Movement and Clitics: Adult and Child Grammar. Newcastle: Cambridge Scholars, 219–37. Benincà, Paola, and Vanelli, Laura (1978). Il plurale friulano: contributo allo studio del plurale romanzo. Revue de linguistique romane 42: 167–8, 241–92. Bentley, Delia (2003). New linguistic research on Italo-Romance dialects. In Lepschy and Tosi (2003: 82–107). Bentley, Delia (2006). Split Intransitivity in Italian. Berlin: Mouton de Gruyter. Bentley, Delia (2010). Principles of subject markedness in Romance. Archivio glottologico italiano 95.2: 152–89. Bentley, Delia, and Eythórsson, Thórhallur (2001). Alternation according to person in ItaloRomance. In Laurel Brinton (ed.), Historical Linguistics 1999: Selected Papers from the 14th International Conference on Historical Linguistics (Vancouver, 9–13 August 1999). Amsterdam: Benjamins, 63–74. Bentley, Delia, and Ledgeway, Adam (eds) (2007). Sui dialetti italoromanzi: saggi in onore di Nigel B. Vincent (The Italianist 27, Special Supplement 1). Norfolk: Biddles. Benucci, Franco (1993). Temporal periphrasis and clitics in Central Romance. Catalan Working Papers in Linguistics 3.1: 51–83. Benveniste, Émile (1966). Problèmes de linguistique générale, vol. 1. Paris: Gallimard. Beretta, Claudio (1980). Contributo per una grammatica del milanese contemporaneo. Milan: Virgilio. Berizzi, Mariachiara (2001). The theory of relative clauses and the dialects of English. BA thesis, University of Padua. Berruto, Gaetano (1993). Varietà diamesiche, diastratiche, diafasiche. In Alberto Sobrero (ed.), Introduzione all’italiano contemporaneo: la variazione e gli usi. Rome: Laterza, 37–92. Bertocci, Davide, and Damonte, Federico (2007). Distribuzione e morfologia dei congiuntivi in alcune varietà salentine. In Federico Damonte and Jacopo Garzonio (eds), Studi sui dialetti della Puglia. Padua: Unipress, 3–28. Bertolucci Pizzorusso, Valeria (ed.) (1975). Il Milione di Marco Polo (versione toscana del Trecento). Milan: Adelphi. Berwick, Robert, and Chomsky, Noam (2011). The biolinguistic program: the current state of its evolution and development. In Anna Maria di Sciullo and Cedric Boeckx (eds), The Biolinguistic Enterprise: New Perspectives on the Evolution and Nature of the Human Language Faculty. New York: Oxford University Press, 19–41. Bianchi, Valentina (1999). Consequences of Antisymmetry: Headed Relative Clauses. Berlin: Mouton de Gruyter.

304

References

Biberauer, Theresa (2008). Introduction. In Theresa Biberauer (ed.), The Limits of Syntactic Variation. Amsterdam: Benjamins, 1–72. Biberauer, Theresa (2011). In defence of lexico-centric parametric variation: two 3rd factorconstrained case studies. Paper presented at the ‘Workshop on Formal Grammar and Syntactic Variation: Rethinking Parameters’, Madrid. Biberauer, Theresa, Holmberg, Anders, and Roberts, Ian (2007). Disharmonic word-order systems and the Final-over-Final-Constraint (FOFC). In Antonietta Bisetto and Francesco Barbieri (eds), Proceedings of XXXIII Incontro di Grammatica Generativa. Available on-line at: http://amsacta.cib.unibo.it/archive/00002397/01/PROCEEDINGS_IGG33.pdf Biberauer, Theresa, Holmberg, Anders, and Roberts, Ian (2008). Structure and linearization in disharmonic word orders. In Charles Chang and Hannah Haynie (eds), Proceedings of the 26th West Coast Conference on Formal Linguistics. Somerville, Mass.: Cascadilla Proceedings Project, 96–104. Biberauer, Theresa, Holmberg, Anders, and Roberts, Ian (2011). The Final Over Final Constraint in syntax and across domains. Paper presented at ‘Parallel Domains Workshop in Honour of Jean-Roger Vergnaud’, University of Southern California, Los Angeles. Biberauer, Theresa, Holmberg, Anders, Roberts, Ian, and Sheehan, Michelle (eds) (2010). Parametric Variation: Null Subjects in Minimalist Theory. Cambridge: Cambridge University Press. Biberauer, Theresa, and Roberts, Ian (forthcoming). Towards a parameter hierarchy for auxiliaries: diachronic considerations. In Anthony Kroch (ed.), Proceedings of DiGS XIII. Oxford: Oxford University Press. Biella, Angelo, Favaro Lanzetti, Virginia, Mondini, Luciana, and Scotti, Gianfranco (1992). Vocabolario italiano–lecchese, lecchese–italiano. Oggiono: Cattaneo. Blanchette, Frances (2012). Negative Concord in English, paper presented at the 79th Meeting of the Southeast Conference on Linguistics (SECOL79), April 2012, University of Kentucky, Lexington. Blanchette, Frances (2013). Negative Concord in English. Linguistic Variation 13.1: 1–47. Blasco Ferrer, Eduardo (1984). Storia linguistica della Sardegna. Tübingen: Niemeyer. Blasco Ferrer, Eduardo (1988). Le parlate dell’Alta Ogliastra: analisi dialettologica, saggio di storia linguistica e culturale. Cagliari: Edizioni della Torre. Blasco Ferrer, Eduardo (ed.) (2003). Crestomazia sarda dei primi secoli (2 vols). Nuoro: Ilisso. Bobaljik, Jonathan (2002). The Rich Agreement Hypothesis in review. In Yasuki Abe (ed.), Complex Predicates and Argument Structure. Nanzan University, Nagoya: research report for the Ministry of Education Grant-in-Aid for Scientific Research, 63–109. Bonet, Eulalia (1991). Morphology after syntax: pronominal clitics in Romance. Doctoral thesis, MIT. Borer, Hagit (1984). Parametric Syntax. Dordrecht: Foris. Borer, Hagit (1986). I-subjects. Linguistic Inquiry 17: 375–416. Borgogno, Giovanni Battista (1972). Note di linguistica ligure (in relazione con il dialetto di Perinaldo). Rivista ingauna e intemelia 27: 69–76. Börjars, Kersti, Vincent, Nigel, and Chapman, Carol (1997). Paradigms, periphrases and pronominal inflection: a feature-based approach. Yearbook of Morphology 1996: 155–80.

References

305

Bossong, Georg (1991). Differential object marking in Romance and beyond. In Dieter Wanner and Douglas Kibbee (eds), New Analyses in Romance Linguistics. Amsterdam: Benjamins, 143–70. Bottiglioni, Gino (1932–44). Atlante linguistico etnografico italiano della Corsica. Cagliari: University of Cagliari. Bourova, Viara, and Tasmowski, Liliane (2007). La préhistoire des futurs romans: ordre de constituants et sémantique. Cahiers Chronos 19: 25–41. Brambilla Ageno, Franca (1978). Aspetti della storia della lingua: la trasmissione dei moduli sintattici e le loro modificazioni attraverso il tempo. Studi di grammatica italiana 7: 353–73. Branca, Vittore (ed.) (1992). Giovanni Boccaccio, Il Decameron. Turin: Einaudi. Branciforti, Francesco (ed.) (1953). Regole, costituzioni, confessionali e rituali. Palermo: Centro di studi filologici e linguistici siciliani. Brandi, Luciana, and Cordin, Patrizia (1989). Two Italian dialects and the Null Subject Parameter. In Osvaldo Jaeggli and Ken Safir (eds), The Null Subject Parameter. Dordrecht: Kluwer, 111–42. Branigan, Phil (2011). In defense of macroparameters: Algonquian evidence. Paper presented at the ‘Workshop on Formal Grammar and Syntactic Variation: Rethinking Parameters’, Madrid. Branigan, Phil (2012). Macroparameter learnability: an Algonquian case study. MS, Memorial University of Newfoundland. Brasseur, Patrice (2001). Dictionnaire des régionalismes du français de Terre-Neuve. Tübingen: Niemeyer. Brero, Camillo, and Gandolfo, Renzo (1967). La letteratura in piemontese: dalle origini al Risorgimento. Turin: Casanova. Breu, Walter (ed.) (2011). L’influsso dell’italiano sul sistema del verbo delle lingue minoritarie (Resistenza e mutamento nella morfologia e nella sintassi). Bochum: Brockmeyer. Buono, Benedict (2011). ‘Sulle campagne d’Italia 1859–1866’, di Giovanni Rossetti: il percorso narrativo e linguistico di un soldato risorgimentale. Paper presented at the conference ‘Italian Dialects in Diachrony’, Leiden University, May 2011. Burzio, Luigi (1986). Italian Syntax. Dordrecht: Reidel. Butt, Miriam, and Lahiri, Aditi (2002). Historical stability vs. historical change. MS. http://ling. uni-konstanz.de/pages/home/butt/ Calabrese, Andrea (1993). The sentential complementation of Salentino: a study of a language without infinitival clauses. In Belletti (1993), 28–98. Calabrese, Andrea (1994). Syncretism phenomena in the clitic systems of Italian and Sardinian dialects and the notion or morphological change. In Jill Beckman (ed.), Proceedings of NELS 25.2. Amherst, Mass.: GLSA, University of Massachusetts, 151–74. Calabrese, Andrea (2008). On absolute and contextual syncretism: remarks on the structure of paradigms and on how to derive it. In Andrew Nevins and Asaf Bachrach (eds), The Bases of Inflectional Identity. Oxford: Oxford University Press, 156–205. Cardinaletti, Anna (1997). Subjects and clause structure. In Liliane Haegeman (ed.), The New Comparative Syntax. London: Longman, 33–63. Cardinaletti, Anna (2004). Toward a cartography of subject positions. In Rizzi (2004), 115–65.

306

References

Cardinaletti, Anna (2008). On different types of clitic clusters. In Cécile De Cat and Katherine Demuth (eds), The Bantu–Romance Connection. Amsterdam: Benjamins, 41–82. Cardinaletti, Anna, and Repetti, Lori (2008). The phonology and syntax of preverbal and postverbal subject clitics in northern Italian dialects. Linguistic Inquiry 39.4: 523–63. Cardinaletti, Anna, and Starke, Michal (1999). The typology of structural deficiency: a case study of the three classes of pronouns. In Hans van Riemsdijk (ed.), Clitics in the Languages of Europe. Berlin: de Gruyter, 145–233. Carstairs(-McCarthy), Andrew (1988). Some implications of phonologically conditioned suppletion. Yearbook of Morphology 1: 67–94. Carstairs(-McCarthy), Andrew (1990). Phonologically conditioned suppletion. In Wolfgang Dressler, Hans Luschützky, Oskar E. Pfeiffer, and John Rennison (eds), Contemporary Morphology. Berlin: Mouton de Gruyter, 17–23. Carstairs-McCarthy, Andrew (2010). The Evolution of Morphology. Oxford: Oxford University Press. Castelfranchi, Cristiano, and Parisi, Domenico (1976). Towards one si. Italian Linguistics 2: 83–122. Castellani, Arrigo (ed.) (1952). Nuovi testi fiorentini del Dugento. Florence: Sansoni. Castellani, Arrigo (ed.) (1982). La prosa italiana delle origini, vol. 1: Testi toscani di carattere pratico. Bologna: Pàtron. Castellani, Franco Elia (2002). Grammatica del dialetto di Cascinagrossa. Alessandria: Edizioni dell’Orso. Cecchetto, Carlo, and Oniga, Renato (2002). Consequences of the analysis of Latin infinitival clauses for the theory of case and control. Lingue e linguaggio 1: 151–89. Cennamo, Michela (1991). Se, suus, sibi nelle Inscriptiones Latinae Christianae Veteres ed i successivi sviluppi romanzi. Medioevo romanzo 16: 3–20. Cennamo, Michela (1993a). The Reanalysis of Reflexives: a Diachronic Perspective. Naples: Liguori. Cennamo, Michela (1993b). L’estensione del dominio referenziale del riflessivo in testi italiani antichi. Archivio glottologico italiano 78.1: 53–62. Cennamo, Michela (1995). Transitivity and VS order in Italian reflexives. In Yaron Matras and Hans-Jurgen Sasse (eds), Verb–Subject Order and Theticity in European Languages. Berlin: Akademie, 84–105. Cennamo, Michela (1997). Passive and impersonal constructions. In Maiden and Parry (1997), 145–61. Cennamo, Michela (1998). Transitivity in the Italian dialects: synchronic aspects and diachronic implications. In Hans Geisler and Daniel Jacob (eds), Transitivität und Diathese in romanischen Sprachen. Tübingen: Niemeyer, 73–87. Cennamo, Michela (2000). Costruzioni passive ed impersonali in veneziano e in napoletano antico. In Annick Englebert, Michel Perrard, Laurence Rosier, and Dan Van Raemdonck (eds), Actes du XXIIe Congrès international de linguistique et philologie romanes, vol. 2: Les nouvelles ambitions de la linguistique diachronique. Tübingen: Niemeyer, 91–103. Cennamo, Michela (2002). La selezione degli ausiliari perfettivi in napoletano antico: fenomeno sintattico o sintattico-semantico. Archivio glottologico italiano 87: 175–222. Cennamo, Michela (2010). Impersonali, verbi. In Raffaele Simone (ed.), Enciclopedia dell’italiano, vol. 1. Rome: Istituto dell’Enciclopedia italiana G. Treccani, 637–9.

References

307

Cennamo, Michela (2011a). Riflessivi, verbi. In Simone (2011), 320–22. Cennamo, Michela (2011b). Transitivi e intransitivi, verbi. In Simone (2011), 535–9. Cennamo, Michela, and Sorace, Antonella (2007). Unaccusativity at the syntax–lexicon interface: evidence from Paduan. In Raúl Aranovich (ed.), Split Auxiliary Systems: A Crosslinguistic Perspective. Amsterdam: Benjamins, 65–100. Ceruti, Antonio (ed.) (1878). Cronica degli imperadori. Archivio glottologico italiano 3: 177–243. Chafe, Wallace (1987). Cognitive constraints on information flow. In Russell Tomlin (ed.), Coherence and Grounding in Discourse. Amsterdam: Benjamins, 21–52. Chamberlain, Jeffrey T. (1986). Latin Antecedents of French Causative faire. New York: Lang. Chierchia, Gennaro (1998). Reference to kinds across languages. Natural Language Semantics: 339–405. Chomsky, Noam (1964). Current Issues in Linguistic Theory. The Hague: Mouton. Chomsky, Noam (1977). On wh-movement. In Peter W. Culicover, Thomas Wasow, and Adrian Akmajian (eds), Formal Syntax. New York: Academic Press, 71–132. Chomsky, Noam (1981a). Lectures on Government and Binding. Dordrecht: Foris. Chomsky, Noam (1981b). On the representation of form and function. Linguistic Review 1: 3–40. Chomsky, Noam (1986). Barriers. Cambridge, Mass.: MIT Press. Chomsky, Noam (1995). Categories and transformations. In Noam Chomsky, The Minimalist Program. Cambridge, Mass.: MIT Press. Chomsky, Noam (2000). Minimalist inquiries. In Roger Martin, David Michaels, and Juan Uriagereka (eds), Step by Step: Minimalist Syntax in Honor of Howard Lasnik. Cambridge, Mass.: MIT Press, 89–155. Chomsky, Noam (2001). Derivation by phase. In Michael Kenstowicz (ed.), Ken Hale: A Life in Language. Cambridge, Mass.: MIT Press, 1–53. Chomsky, Noam (2005). Three factors in language design. Linguistic Inquiry 36(1): 1–22. Chomsky, Noam (2007). Approaching UG from below. In Uli Sauerland and Hans-Martin Gärtner (eds), Interfaces + Recursion = Language? Chomsky’s Minimalism and the View from Syntax-Semantics. Berlin: Mouton, 1–29. Chomsky, Noam (2008). On phases. In Robert Freidin, Carlos Otero, and María Luisa Zubizarreta (eds), Foundational Issues in Linguistic Theory: Essays in Honor of Jean-Roger Vergnaud. Cambridge, Mass.: MIT Press, 133–66. Cinque, Guglielmo (1976). Mica. Annali della facoltà di lettere e filosofia dell’Università di Padova 1: 101–12. Repr. in G. Cinque, Teoria linguistica e sintassi italiana (Bologna: Il Mulino, 1991), 311–23. Cinque, Guglielmo (1978). La sintassi dei pronomi relativi ‘cui’ e ‘quale’ nell’italiano contemporaneo. Rivista di grammatica generativa 3: 31–126. Cinque, Guglielmo (1988a). On si constructions and the theory of arb. Linguistic Inquiry 19.4: 521–81. Cinque, Guglielmo (1988b). La costruzione relativa infinitiva. In Renzi et al. (1988), i.443–503. Cinque, Guglielmo (1994). On the evidence for partial N-movement in the Romance DP. In Guglielmo Cinque, Jan Koster, Jean-Yves Pollock, Luigi Rizzi, and Raffaella Zanuttini (eds), Paths Towards Universal Grammar. Washington, DC: Georgetown University Press, 85–110. Cinque, Guglielmo (1995). Ci si. In Guglielmo Cinque, Italian Syntax and Universal Grammar. Cambridge: Cambridge University Press, 193–98.

308

References

Cinque, Guglielmo (1999). Adverbs and Functional Heads: A Cross-Linguistic Perspective. Oxford: Oxford University Press. Cinque, Guglielmo (2006). Restructuring and Functional Heads: The Cartography of Syntactic Structures. New York: Oxford University Press. Cinque, Guglielmo (2007). La natura grammaticale del diminutivo e del vezzeggiativo. In Maschi et al. (2007), 229–36. Cinque, Guglielmo (2008a). Two types of nonrestrictive relatives. In Olivier Bonami and Patricia Cabredo Hofherr (eds), Empirical Issues in Syntax and Semantics 7. Paris: CNRS, 99–137. http://www.cssp.cnrs.fr/eiss7 Cinque, Guglielmo (2008b). More on the indefinite character of the head of restrictive relatives. In Paola Benincà, Federico Damonte, and Nicoletta Penello (eds), Selected Papers of the 34th Incontro di Grammatica Generativa, Unipress, Padova. Special issue of Rivista di grammatica generativa 33: 3–24. (http://lear.unive.it/handle/10278/2497) Cinque, Guglielmo (2010a). On a selective ‘violation’ of the Complex NP Constraint. In JanWouter Zwart and Mark de Vries (eds), Structure Preserved: Studies in Syntax for Jan Koster. Amsterdam: Benjamins, 81–90. Cinque, Guglielmo (2010b). Mapping spatial PPs: an introduction. In Cinque and Rizzi (2010), 3–25. Cinque, Guglielmo, and Kayne, Richard (eds) (2005). The Oxford Handbook of Comparative Syntax. Oxford: Oxford University Press. Cinque, Guglielmo, and Rizzi, Luigi (eds) (2010). Mapping Spatial PPs. Oxford: Oxford University Press. Clackson, James (2004). Latin. In Woodward (2004), 789–811. Clackson, James, and Horrocks, Geoffrey (2007). The Blackwell History of the Latin Language. Oxford: Blackwell. Cole, Peter, Gil, David, Hermon, Gabriella, and Tadmor, Uri (2001). The acquisition of in-situ WH-questions and WH-indefinites in Jakarta Indonesian. In Anna Do and Aimee Johansen (eds), Proceedings of the 25th Boston University Conference on Linguistic Development, vol. 1. Somerville, Mass.: Cascadilla Press, 169–79. Coleman, Robert (1971). The origin and development of Latin habeo + infinitive. Classical Quarterly 21: 215–32. Colombo, Michele (ed.) (2007). Benedetto Buommattei, Della lingua toscana. Florence: Presso L’Accademia. Coluccia, Rosario (1975). Tradizioni auliche e popolari nella poesia del Regno di Napoli in età angioina. Medioevo romanzo 2: 44–153. Comrie, Bernard (1999). Grammatical gender systems: a linguist’s assessment. Journal of Psycholinguistic Research 28.5: 457–66. Contini, Gianfranco (ed.) (1937). Cinque volgari di Bonvesin da la Riva. Modena: Società tipografica modenese, 27–40. Contini, Gianfranco (1960). Poeti del Duecento (2 vols). Milan: Ricciardi. Contini, Gianfranco (ed.) (1980). Dante Alighieri, Rime. Turin: Einaudi. Contini, Gianfranco (ed.) (1984). Dante Alighieri, Il Fiore e il Detto d’Amore attribuibili a Dante Alighieri. Milan: Mondadori.

References

309

Conzett, Philipp (2006). Gender assignment and the structure of the lexicon. Sprachtypologie und Universalienforschung 59.3: 223–40. Corbett, Greville G. (2013). Periphrasis and possible lexemes. In Marina Chumakina and Greville Corbett (eds), Periphrasis: The Role of Syntax and Morphology in Paradigms. (Proceedings of the British Academy 180). London and Oxford: British Academy and Oxford University Press, 169–89. Corbett, Greville, and Fraser, Norman (2000). Gender assignment: a typology and a model. In Gunter Senft (ed.), Systems of Nominal Classification. Cambridge: Cambridge University Press, 293–325. Cordin, Patrizia (1997). Tense, mood and aspect in the verb. In Maiden and Parry (1997), 87–98. Cornilescu, Alexandra (2000). The double subject construction in Romanian. In Alboiu and Motapanyane (2000), 83–133. Crocioni, Giovanni (1907). Il dialetto di Velletri e dei paesi finitimi. Studj romanzi 5: 27–88. Croft, William (1994). Typology and Universals. Cambridge: Cambridge University Press. Cruschina, Silvio (2012). Discourse-Related Features and Functional Projections. New York: Oxford University Press. Cuneo, Marco (1997). L’uso dell’infinito nei dialetti liguri: infinito con soggetto espresso e infinito flesso nel dialetto di Cicagna (GE). Rivista italiana di dialettologia 21: 99–132. Cyrino, Sonia (2010). On the loss of verb movement in Brazilian Portuguese. Paper presented at ‘DiGS XII’, Queens College, University of Cambridge. D’Achille, Paolo, and Thornton, Anna (2003). La flessione del nome dall’italiano antico all’italiano contemporaneo. In Nicoletta Maraschio and Teresa Poggi Salani (eds), Italia linguistica anno mille, Italia linguistica anno duemila: atti del XXXIV Congresso internazionale di studi della Società di linguistica italiana (SLI), Firenze, 19–21 ottobre 2000. Rome: Bulzoni, 211–30. D’Achille, Paolo, and Thornton, Anna (2008). I nomi femminili in -o. In Emanuela Cresti (ed.), Prospettive nello studio del lessico italiano. Florence: Firenze University Press, pt 2, 473–81. D’Alessandro, Roberta (2007). Impersonal si Constructions: Agreement and Interpretation. Berlin: Mouton de Gruyter. D’Alessandro, Roberta (2010). The verbal domain: TP-VP structure and auxiliaries. In D’Alessandro et al. (2010), 27–38. D’Alessandro, Roberta, and Ledgeway, Adam (2010). At the C–T boundary: investigating Abruzzese complementation. Lingua 120: 2040–60. D’Alessandro, Roberta, Ledgeway, Adam, and Roberts, Ian (eds) (2010). Syntactic Variation: The Dialects of Italy. Cambridge: Cambridge University Press. D’Alessandro, Roberta, and Roberts, Ian (2008). Movement and agreement in Italian past participles and defective phases. Linguistic Inquiry 39: 477–91. D’Alessandro, Roberta, and Roberts, Ian (2010). Past participle agreement in Abruzzese: split auxiliary selection. Natural Language & Linguistic Theory 28: 41–72. Da Tos, Martina (2011). The intramorphological meanings of thematic vowels in Italian verbs. Doctoral thesis, University of Padua. Damonte, Federico (2002). The syntax and semantics of Salentino ku. Paper presented at ‘XXVIII Incontro di Grammatica Generativa’, Lecce.

310

References

Damonte, Federico (2005). La diffusione della particella ‘mi’ in alcune varietà messinesi: problemi di metodo. In Gianna Marcato (ed.), Dialetti in città: atti del convegno, Sappada/ Plodn (Belluno). Padua: Unipress, 237–42. Damonte, Federico (2006). Differenze generazionali nell’uso del congiuntivo presente in salentino. In Gianna Marcato (ed.), Giovani, lingue e dialetti: atti del convegno Sappada/Plodn (Belluno), 29 giugno–3 luglio 2005. Padua: Unipress, 87–92. Damonte, Federico (2010). Matching moods: mood concord between CP and IP in Salentino and southern Calabrian subjunctive complements. In Paola Benincà and Nicola Munaro (eds), Mapping the Left Periphery. Oxford: Oxford University Press, 228–56. Dardano, Maurizio, and Frenguelli, Gianluca (eds) (2004). SintAnt: la sintassi dell’italiano antico. Atti del Convegno internazionale di studi (Università ‘Roma Tre’, 18–21 settembre 2002). Rome: Aracne. Davies, William, and Rosen, Carol (1988). Unions as multi-predicate clauses. Language 64: 52–88. De Blasi, Nicola (ed.) (1986). Libro de la Destructione de Troya: volgarizzamento napoletano trecentesco da Guido delle Colonne. Rome: Bulzoni. Delogu, Ignazio (ed.) (1997). Il Condaghe di San Pietro di Silki: testo logudorese inedito dei Secoli XI–XIII (CSPS). Sassari: Dessì. Dewaele, Jean-Marc, and Véronique, Daniel (2001). Gender assignment and gender agreement in advanced French interlanguage: a cross-sectional study. Bilingualism: Language and Cognition 4.3: 275–97. Dikken, Marcel den (2006). Relators and Linkers: The Syntax of Predication, Predicate Inversion, and Copulas. Cambridge, Mass.: MIT Press. Di Girolamo, Dino (1977). Gli Alfabetin e la lingua degli ebrei in Sicilia nel Medioevo. Materia giudaica 3: 50–55. Dionisotti, Carlo (ed.) (1966). Pietro Bembo, Prose e rime, 2nd edn. Turin: UTET. Dixon, Robert (1994). Ergativity. Cambridge: Cambridge University Press. D’Onghia, Luca (2003). Alcune osservazioni sul costrutto causativo nel pavano di Ruzante. Lingua e stile 38: 43–58. D’Ovidio, Francesco (1886). Ricerche sui pronomi personali e possessivi neolatini. Archivio glottologico italiano 9: 25–101. Dowty, David (1991). Thematic proto-roles and argument selection. Language 67: 547–619. Dresher, E. (1999). Charting the learning path: cues to parameter setting. Linguistic Inquiry 30: 27–68. Dressler, Wolfgang (1995). Interactions between iconicity and other semiotic parameters. In Raffaele Simone (ed.), Iconicity in Language. Amsterdam: Benjamins, 21–37. Dressler, Wolfgang, and Thornton, Anna (1996). Italian nominal inflection. Wiener Linguistische Gazette 55–57: 1–24. Dryer, Matthew (2011a). Order of object and verb. In Martin Haspelmath, Matthew Dryer, David Gil, and Bernard Comrie (eds), The World Atlas of Language Structures Online. Munich: Max Planck Digital Library, ch. 83A. Available online at: http://wals.info/feature/ 83A. Accessed 14 April 2012. Dryer, Matthew (2011b). Order of adposition and noun phrase. In Martin Haspelmath, Matthew Dryer, David Gil, and Bernard Comrie (eds) The World Atlas of Language

References

311

Structures Online. Munich: Max Planck Digital Library, ch. 85A. Available online at: http:// wals.info/feature/85A. Accessed 14 April 2012. Dubinsky, Stanley (1985). Union constructions in Japanese: a unified analysis of -sase and -rare. Doctoral thesis, Cornell University. Duraffour, Antonin (1932). Description morphologique avec notes syntaxiques du parler francoprovençal de Vaux (Ain) en 1919–1931. Grenoble: Institut phonétique de Grenoble. Durante, Marcello (1981). Dal latino all’italiano moderno. Bologna: Zanichelli. Egerland, Verner (2003). Impersonal pronouns in Scandinavian and Romance. Working Papers in Scandinavian Syntax 71: 75–102. Egerland, Verner (2005). Diachronic change and pronoun status: Italian dative ‘loro’. Linguistics 43: 1105–30. Emonds, Joseph (1978). The verbal complex V-V in French. Linguistic Inquiry 9: 151–75. Ewert, Alfred (1953). The French Language. London: Faber and Faber. Facchetti, Giovanni (1901). Il dialetto trevigliese. Treviglio: Messaggi. Favati, Guido (ed.) (1970). Il Novellino. Genoa: Bozzi. Fedele, Emily (2010). Verb movement and functional heads in Standard Italian and the dialects of Italy. MPhil. thesis, University of Cambridge. Fernández González, José Ramón (1985). Gramática histórica provenzal. Oviedo: Summa. Fernández-Ordóñez, Inés (2010). La grammaire dialectale de l’espagnol à travers le Corpus oral et sonore de l’espagnol rural (COSER, Corpus oral y sonoro del español rural). In Michèle Oliviéri (ed.), La syntaxe de corpus. Special issue of Corpus 9: 81–114. Finarelli, Serena (2002–3). Il vocalismo tonico agnonese tra metafonesi e DVP. BA thesis, Università di Roma ‘La Sapienza’. Fiorentino, Giuliana (1999). La relativa debole. Pavia: Angeli. Fløgstad, Guro (2012). Forking paths: the expansion of the preterite in Rioplatense Spanish. MS, University of Oslo. Förster, Wendelin (ed.) (1880–83). Antica parafrasi lombarda del ‘Neminem laedi nisi a se ipso’ di S. Giovanni Grisostomo. Archivio glottologico italiano 7: 1–20. Formentin, Vittorio (1996). Alcune considerazioni e un’ipotesi sull’articolo determinativo in area romanza. In Lucio Lugnani, Marco Santagata, and Alfredo Stussi (eds), Studi offerti a Luigi Blasucci dai colleghi e dagli allievi pisani. Lucca: Fazzi, 257–72. 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. Francez, Itamar (2007). Existental propositions. Doctoral thesis, Stanford University. Francez, Itamar (2009). Existentials, predication, and modification. Linguistics and Philosophy 32.1: 1–50. Frascarelli, Mara (ed.) (2006). Phases of Interpretation. Berlin: Mouton de Gruyter. Freeze, Ray (1992). Existentials and other locatives. Language 68: 553–95. Fusina, Jacques (1999). Parlons corse. Paris: L’Harmattan. Garnier, Christian (1898). Deux patois des Alpes-maritimes italiennes: grammaires et vocabulaires méthodiques des idiomes de Bordighera et de Realdo. Paris: Leroux.

312

References

Garzonio, Jacopo, and Poletto, Cecilia (2011). Avverbi e struttura frasale in alcune varietà delle Marche. Paper presented at ‘Giornata di studio sui dialetti delle Marche’, University of Padua. GDFC = Société du parler français au Canada (1930). Glossaire du parler français au Canada. Québec: Société du parler français au Canada. Gelderen, Elly van (2011). The Linguistic Cycle: Language Change and the Language Faculty. Oxford: Oxford University Press. Ghinassi, Ghino (2006). Studi sull’antico volgare di Mantova e sul ‘Cortegiano’. Florence: Olschki. Giacalone Ramat, Anna, and Sansò, Andrea. (2011). From passive to impersonal: a case study from Italian and its implications. In A. Malchukov and A. Siewiwerska (eds), Impersonal Constructions: A Cross-linguistic Perspective. Amsterdam: Benjamins, 189–228. Giamboni, Bono (1878–83). Il tesoro di Brunetto Latini volgarizzato da Bono Giamboni (4 vols). Bologna: Romagnoli. Giammarco, Ernesto (1968). Dizionario abruzzese e molisano, vol. 1. Rome: Ateneo. Gibson, Edward, and Fedorenko, Evelina (2010). Weak quantitative standards in linguistic research. Trends in Cognitive Sciences 14: 233–4. Giorgi, Alessandra (1983). Toward a theory of long distance anaphors: a GB approach. Linguistic Review 3: 307–61. Giorgi, Alessandra (2009). Toward a syntax of the subjunctive. In Quer (2009a: 1837–58). Giorgi, Alessandra (2010). About the Speaker: Towards a Syntax of Indexicality. Oxford: Oxford University Press. Giorgi, Alessandra, and Pianesi, Fabio (1997). Tense and Aspect: From Semantics to Morphosyntax. Oxford: Oxford University Press. Giorgi, Alessandra, and Pianesi, Fabio (2004). Complementizer deletion in Italian. In Rizzi (2004), 190–210. Giusti, Giuliana (2006). Parallels in clausal and nominal periphery. In Frascarelli (2006), 163–86. Giusti, Giuliana (in preparation). Nominal Expressions and Language Change: A Diachronic Study from Early Latin to Modern Romance. Oxford: Oxford University Press. Goodall, Grant (1991). Sec of IP and Spec of CP in Spanish Wh-questions. Paper presented at ‘21st Linguistic Symposium on Romance Languages’, University of California, Santa Barbara. Greenberg, Joseph (1963). Some universals of grammar with particular reference to the order of meaningful elements. In Joseph Greenberg, Universals of Language. Cambridge, Mass.: MIT Press, 58–90. Grevisse, Maurice, and Goosse, André (2011). Le bon usage: grammaire française, 15th edn. Paris: Duclot/Brussels: De Boeck. Guardiano, Cristina (2010). L’oggetto diretto preposizionale in siciliano: una breve rassegna e qualche domanda. In Jacopo Garzonio (ed.), Quaderni di lavoro ASIt 2010: studi sui dialetti della Sicilia. Padua: Unipress, 95–115. Guarnerio, Pier Enea (ed.) (1892–4). Gli statuti della repubblica sassarese: testo logudorese del secolo XIV, nuovamente edito d’in sul codice e annotato. Archivio glottologico italiano 13: 1–124.

References

313

Haegeman, Liliane (2010). The movement derivation of conditional clauses. Linguistic Inquiry 41: 595–621. Hall, Robert A. (1950). The reconstruction of Proto-Romance. Language 26: 6–27. Halle, Morris, and Marantz, Alec (1993). Distributed morphology and the pieces of inflection. In Kenneth Hale and S. Jay Keyser (eds), The View from Building 20. Cambridge, Mass.: MIT Press, 111–76. Harris, James (1994). The syntax–phonology mapping in Catalan and Spanish clitics. MIT Working Papers in Linguistics 21: 321–53. Harris, James, and Halle, Morris (2005). Unexpected plural inflections in Spanish: reduplication and metathesis. Linguistic Inquiry 36.2: 195–222. Harris, Martin (1982). The ‘past simple’ and the ‘present perfect’ in Romance. In Vincent and Harris (1982), 42–70. Harris, Martin, and Ramat, Paolo (eds) (1987). Historical Development of Auxiliaries. Berlin: Mouton de Gruyter. Harris, Martin, and Vincent, Nigel (1980). On zero relatives. Linguistic Inquiry 11: 805–7. Harris, Martin, and Vincent, Nigel (eds) (1988). The Romance Languages. London: Croom Helm. Hart, Victoria (2006). Subject positions in Salentino cu-clauses. MPhil. thesis, University of Cambridge. Haspelmath, Martin (2011). On S, A, P, T, and R as comparative concepts for alignment typology. Linguistic Typology 15.3: 535–689. Haverling, Gerd (2010). Sur l’expression du temps et de l’aspect grammatical en Latin tardif. De Lingua Latina 5: 1–23. Hermann, Tanja (2003). Relative clauses in dialects of English: a typological approach. Doctoral thesis, Freiburg i.Br., Albert-Ludwig University. Holmberg, Anders, and Roberts, Ian (2010). Introduction. In Biberauer et al. (2010: 1–57). Holmes, Virginia, and Segui, Juan (2004). Sublexical and lexical influences on gender assignment in French. Journal of Psycholinguistic Research 33.6: 425–57. Hout, Angeliek van (2004). Unaccusativity as telicity checking. In Artemis Alexiadou, Elena Anagnostopoulou, and Martin Everaert (eds), The Unaccusativity Puzzle: Explorations of the Syntax–Lexicon Interface. Oxford: Oxford University Press, 60–158. Huang, James (1984). On the distribution and reference of empty pronouns. Linguistic Inquiry 15: 531–74. Huang, James (2006). The macro-history of Chinese syntax and the theory of language change. Paper given at ‘Workshop on Chinese Linguistics’, University of Chicago, December. Huber, Joseph (1933). Altportugiesisches Elementarbuch. Heidelberg: Winter. Huber, Joseph (1958). Zur Verbalflexion der Mundart von Livigno. Vox Romanica 17: 82–128. Iliescu, Maria (1972). Le Frioulan à partir des dialectes parlés en Roumanie. The Hague: Mouton. Iliescu, Maria, and Mourin, Louis (1991). Typologie de la morphologie verbale romane, vol. 1: Vue synchronique. Innsbruck: Verlag des Instituts für Sprachwissenschaft der Universität Innsbruck. Ineichen, Gustav (1957). Die paduanische Mundart am Ende des 14. Jahrhunderts. Zeitschrift für romanische Philologie 73: 37–123.

314

References

Ineichen, Gustav (1966). El libro agregà de Serapiom (2 vols). Rome: Istituto per la Collaborazione culturale. Iordan, Iorgu, Guţu Romalo, Valeria, and Niculescu, Alexandru (1967). Structura morfologică a limbii române contemporane. Bucharest: Editura Ştiinţifică. Jaberg, Karl, and Jud, Jacob (1928–40). Sprach- und Sachatlas Italiens und der Südschweiz. Zofingen: Ringier. Jelinek, Eloise (1984). Empty categories, Case, and configurationality. Natural Language and Linguistic Theory 2: 39–76. Jeppesen Kragh, Kirsten (2010). Le remplacement de l’imparfait du subjonctif par le présent du subjonctif considéré dans une perspective de grammaticalisation. Copenhagen: Museum Tusculanum Press. Jezek, Elisabetta (2010). La struttura argomentale dei verbi. In Salvi and Renzi (2010), i.77–122. Jones, Michael (1988). Sardinian. In Harris and Vincent (1988), 314–50. Jones, Michael (1992). Infinitives with specified subjects in Sardinian. In Christiane Laeufer and Terrell Morgan (eds), Theoretical Analyses in Romance Linguistics. Amsterdam: Benjamins, 295–309. Jones, Michael (1993). Sardinian Syntax. London: Routledge. Joos, Martin (1957). Readings in Linguistics: The Development of Linguistics in America since 1925, 2nd edn. New York: American Council of Learned Societies. Kahane, Henry, and Kahane, Renée (1979, 1981, 1986). Graeca et Romanica Scripta Selecta (3 vols). Amsterdam: Hakkert. Kany, Charles (1951). American Spanish Syntax. Chicago: University of Chicago Press. Katz, Jerrold, and Postal, Paul (1964). An Integrated Theory of Linguistic Descriptions. Cambridge, Mass.: MIT Press. Kayne, Richard (1984). Connectedness and Binary Branching. Dordrecht: Foris. Kayne, Richard (1991). Romance clitics, verb movement and PRO. Linguistic Inquiry 22: 647–86. Kayne, Richard (1993). Toward a modular theory of auxiliary selection. Studia Linguistica 47: 3–31. Repr. in Kayne (2000), 107–30. Kayne, Richard (1994). The Antisymmetry of Syntax. Cambridge, Mass.: MIT Press. Kayne, Richard (2000). Parameters and Universals. Oxford: Oxford University Press. Kayne, Richard (2005). Remarks on comparative syntax, with special reference to English and French. In Cinque and Kayne (2005), 3–69. Kayne, Richard (2007). A short note on where vs. place. In Maschi et al. (2007), 245–57. Kayne, Richard (2009). Why is syntax asymmetric? Paper presented at ‘Workshop on Theoretical Approaches to Disharmonic Word Orders’, Newcastle upon Tyne. Kayne, Richard (2010a). Why isn’t this a complementizer? In Richard Kayne, Comparisons and Contrasts. Oxford: Oxford University Press, 190–227. Kayne, Richard (2010b). Toward a syntactic reinterpretation of Harris and Halle 2005. In Reineke Bok-Bennema, Brigitte Kampers-Manhe, and Bart Hollebrandse (eds), Romance Languages and Linguistic Theory 2008. Amsterdam: Benjamins, 145–70. Keller, Oskar (1928). La flexion du verbe dans le patois genevois. Geneva: Olschki. Koopman, Hilda (2010). Prepositions, postpositions, circumpositions and particles. In Cinque and Rizzi (2010), 26–73. Krämer, Martin (2009). The Phonology of Italian. Oxford: Oxford University Press.

References

315

Kroch, Anthony (1994). Morphosyntactic variation. In Katharine Beals, Jeanette Denton, and Robert Knippen (eds), Proceedings of the 30th Annual Meeting of the Chicago Linguistic Society (Parasession on Variation and Linguistic Theory), vol. 2. Chicago: CLS, 180–201. Kuroda, Sige-Yuki (1988). Whether we agree or not: a comparative syntax of English and Japanese. Lingvisticae Investigationes 12: 1–47. La Fauci, Nunzio (1984a). La formazione del siciliano nel Medioevo: uno sguardo oltre la storia della linguistica e la linguistica della storia. In Adriana Quattordio Moreschini (ed.), Tre millenni di storia linguistica della Sicilia. Pisa: Giardini, 105–38. La Fauci, Nunzio (1984b). Sulla natura assolutiva del controllore dell’accordo del participio passato in italiano. La Memoria: Annali della Facoltà di lettere e filosofia dell’Università di Palermo 3: 187–253. 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 Fabio Foresti, Elena Rizzi, and Paola Benedini (eds), L’italiano tra le lingue romanze: atti del XX Congresso internazionale di studi della SLI, Bologna, 25–27 settembre 1986 (SLI 27). Rome: Bulzoni, 213–42. La Fauci, Nunzio (1992). Capitoli di morfosintassi siciliana antica: tassonomia dei costrutti medi e ausiliari perfettivi. In Studi linguistici e filologici offerti a Girolamo Caracausi. Palermo: Centro di studi filologici e linguistici siciliani, 185–220. Repr. in Nunzio La Fauci, Forme romanze della funzione predicativa (Pisa: ETS, 2000), 41–73. La Fauci, Nunzio (1994). Objects and Subjects in the Formation of Romance Morphosyntax. Bloomington, Indiana: IULC. La Fauci, Nunzio (2004). Armonia differenziale dell’ausiliazione perfettiva nel volgare di Dante. In Dardano and Frenguelli (2004: 237–52). Repr. in Nunzio La Fauci, Relazionie differenze: questioni di linguistica razionale (Palermo: Sellerio, 2011), 153–67. La Fauci, Nunzio (2006). Dinamiche sistematiche: perifrasi perfettive e futuro sintetico, dal latino al romanzo. In Renato Oniga and Luigi Zennaro (eds), Atti della Giornata di linguistica latina, Venezia, 7 maggio 2004. Venice: Cafoscarina, 101–31. La Fauci, Nunzio, and Loporcaro, Michele (1993). Grammatical relations and syntactic levels in Bonorvese morphosyntax. In Belletti (1993), 155–203. La Fauci, Nunzio, and Loporcaro, Michele (1997). Outline of a theory of existentials on evidence from Romance. Studi italiani di linguistica teorica e applicata 26: 5–55. Larson, Pär (2010). Fonologia. In Salvi and Renzi (2010), ii.1515–46. Ledgeway, Adam (1998). Variation in the Romance infinitive: the case of the southern Calabrian inflected infinitive. Transactions of the Philological Society 96: 1–61. Ledgeway, Adam (2000). A Comparative Syntax of the Dialects of Southern Italy: A Minimalist Approach. Oxford: Blackwell. Ledgeway, Adam (2003). Linguistic theory and the mysteries of Italian dialects. In Lepschy and Tosi (2003), 108–40. Ledgeway, Adam (2004). Il sistema completivo dei dialetti meridionali: la doppia serie di complementatori. Rivista italiana di dialettologia 27: 89–147.

316

References

Ledgeway, Adam (2005). Moving through the left periphery: the dual complementiser system in the dialects of southern Italy. Transactions of the Philological Society 103: 336–96. Ledgeway, Adam (2006). The dual complementiser system in southern Italy: spirito greco, materia romanza? In Lepschy and Tosi (2006), 112–26. Ledgeway, Adam (2007). Diachrony and finiteness: subordination in the dialects of southern Italy. In Irina Nikolaeva (ed.), Finiteness: Theoretical and Empirical Foundations. Oxford: Oxford University Press, 335–65. Ledgeway, Adam (2009a). Grammatica diacronica del napoletano. Tübingen: Niemeyer. Ledgeway, Adam (2009b). Aspetti della sintassi della periferia sinistra del cosentino. In Diego Pescarini (ed.), Studi sui dialetti della Calabria. Padua: Unipress, 3–24. Ledgeway, Adam (2010). Syntactic variation and the dialects of Italy: an overview. In D’Alesssandro et al. (2010), 1–3. Ledgeway, Adam (2011). When data meet theory: the case of the Placiti cassinesi. In Sylvia Adamson and Wendy Bennett (eds), Linguistics and Philology in the TwentyFirst Century. Oxford: Blackwell. Special issue of the Transactions of the Philological Society 109: 213–19. Ledgeway, Adam (2012a). From Latin to Romance: Morphosyntactic Typology and Change. Oxford: Oxford University Press. Ledgeway, Adam (2012b). La sopravvivenza del sistema dei doppi complementatori nei dialetti meridionali. In Patrizia Del Puente (ed.), Atti del II Convegno Internazionale di Dialettologia: Progetto A.L.Ba. Rionero in Vulture: Calice, 151–76. Ledgeway, Adam (in press a). Reconstructing complementiser-drop in the dialects of the Salento: a syntactic or phonological phenomenon? In Theresa Biberauer and George Walkden (eds), Syntax Over Time: Lexical, Morphological, and Information-Structural Interactions. Oxford: Oxford University Press. Ledgeway, Adam (in press b). VP-related functional projections. In Giuseppe Longobardi (ed.), The Syntax of Italian. Cambridge: Cambridge University Press. Ledgeway, Adam (in press c). La morphologie flexionnelle de l’impératif des dialectes de l’Italie méridionale: la distribution de la métaphonie. In Jean Léo Léonard (ed.), Morphologie flexionnelle et dialectologie romane: typologie(s) et modélisation(s). Louvain: Peeters. Ledgeway, Adam, and Lombardi, Alessandra (2005). Verb movement, adverbs and clitic positions in Romance. Probus 17: 79–113. Legendre, Géraldine (2010). A formal typology of person-based auxiliary selection in ItaloRomance. In D’Alessandro et al. (2010), 186–200. Leone, Alfonso (1982). L’italiano regionale in Sicilia. Bologna: Il Mulino. Leone, Alfonso (1995). Profilo di sintassi siciliana. Palermo: Centro di studi filologici e linguistici siciliani. Lepschy, Giulio (1963). Morfologia veneziana. L’Italia dialettale 26: 129–144. Lepschy, Giulio (1986). Aspects of Italian constructions with si. The Italianist 6: 139–51. Lepschy, Giulio (1992). Proposte per l’accento secondario. The Italianist 12: 117–18. Lepschy, Giulio (1993). Altre note sull’accento secondario. The Italianist 13: 266–8; Lepschy, Anna Laura, and Lepschy, Giulio (1988). The Italian Language Today. London: Hutchinson.

References

317

Lepschy, Anna Laura, and Tosi, Arturo (eds) (2003). Multilingualism in Italy: Past and Present. Oxford: Legenda. Lepschy, Anna Laura, and Tosi, Arturo (eds) (2006). Rethinking Languages in Contact: The Case of Italian. Oxford: Legenda. Librandi, Rita (2006). Tre lettere di un mercante calabrese della fine del trecento. Contributi di filologia dell’Italia mediana 20: 25–50. Lichtenhahn, Anna (1951). La storia di ove, dove, onde, donde, di dove, da dove. Bern: Francke. Lightfoot, David (1979). Principles of Diachronic Syntax. Cambridge: Cambridge University Press. Lightfoot, David (1991). How to Set Parameters: Arguments from Language Change. Cambridge: Cambridge University Press. Lightfoot, David (1999). The Development of Language: Acquisition, Change and Evolution. Oxford: Blackwell. Limentani, Alberto (ed.) (1964). Tutte le opere di Giovanni Boccaccio, vol. 2. Milan: Mondadori. Lindsström, Anton (1907). Il vernacolo di Subiaco. Studj romanzi 5: 237–300. Locatelli, Libero (1970). Piccola grammatica del dialetto comasco: fonetica e morfologia. Como: Famiglia comasca. Lodge, Anthony (2004). A Sociolinguistic History of Parisian French. Cambridge: Cambridge University Press. Lois, Ximena (1989). Aspects de la syntaxe de l’espagnol et théorie de la grammaire. Thesis, Université de Paris VIII. Lombardi, Alessandra (1997). The grammar of complementation in the dialects of Calabria. Doctoral thesis, University of Manchester. Lombardi, Alessandra (1998). Calabria greca e Calabria latina da Rohlfs ai giorni nostri: la sintassi dei verbi modali-aspettuali. In Ramat and Roma (1998), 613–26. Longa, Glicerio (1912). Vocabolario bormino. Rome: Società filologica romana. Longhino, Paola (2002–3). L’ausiliazione perfettiva in padovano antico. BA thesis, Università di Udine. Loporcaro, Michele (1986). L’infinito coniugato nell’Italia centro-meridionale: ipotesi genetica e ricostruzione storica. L’Italia dialettale 49: 609–65. Loporcaro, Michele (1988). Grammatica storica del dialetto di Altamura. Pisa: Giardini. Loporcaro, Michele (1995). Un capitolo di morfologia storica italo-romanza: it. ant. ne ‘ci’ e forme meridionali congeneri. L’Italia dialettale 58: 1–48. Loporcaro, Michele (1997). Puglia and Salento. In Maiden and Parry (1997), 338–48. 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: 203–26. Loporcaro, Michele (2001). La selezione dell’ausiliare nei dialetti italiani: dati e teorie. In Federico Albano Leoni, Rosanna Sornicola, Eleonora Stenta Krosbakken, and Carolina Stromboli (eds), Dati empirici e teorie linguistiche: Atti del XXXIII Congresso della Società di Linguistica Italiana (Napoli, 28–30 ottobre 1999). Rome: Bulzoni, 455–76.

318

References

Loporcaro, Michele (2002). External and internal causation in morphological change: evidence from Italo-Romance dialects. In Sabrina Bendjaballah, Wolfgang Dressler, Oskar Pfeiffer, and Maria Voeikova (eds), Morphology 2000. Amsterdam: Benjamins, 227–40. Loporcaro, Michele (2007). On triple auxiliation in Romance. Linguistics 45: 173–222. Loporcaro, Michele (2008). Opposizioni di caso nel pronome personale: i dialetti del Mezzogiorno in prospettiva romanza. In Alessandro de Angelis (ed.), I dialetti meridionali tra arcaismo e interferenza. Palermo: Centro di studi filologici e linguistici siciliani, 207–35. Loporcaro, Michele (2011a). Two euroversals in a global perspective: auxiliation and alignment. In Peter Siemund (ed.), Linguistic Universals and Language Variation. Berlin: Mouton de Gruyter, 55–91. Loporcaro, Michele (2011b). Perfective auxiliaries over time: syntax and semantics. Paper given at 1st ‘Italian (Dialects) in Diachrony’ conference, Leiden University Centre for Linguistics, 19–21 May 2011. Luongo, Salvatore (1989). Ordini di la confessioni ‘Renovamini’. Palermo: Centro di studi filologici e linguistici siciliani. Maggini, Francesco (ed.) (1968). Brunetto Latini, La rettorica. Florence: Le Monnier. Magri, Riccardo (1995). Dialetto cremonese di città e dei paesi. Cremona: Turris. Maiden, Martin (1995). Storia linguistica dell’italiano. Bologna: Il Mulino. Maiden, Martin (1996). Ipotesi sulle origini del condizionale analitico come ‘futuro del passato’ in italiano. In Benincà et al. (1996), 149–73. Maiden, Martin (1996). On the Romance inflectional endings -i and -e. Romance Philology 50: 147–82. Maiden, Martin (2000). Il sistema desinenziale del sostantivo nell’italo-romanzo preletterario: ricostruzione parziale a partire dai dialetti moderni (il significato storico di plurali del tipo amici). In John Herman and Anna Marinetti (eds), La preistoria dell’italiano: atti della tavola rotonda di linguistica storica, Università Ca’ Foscari di Venezia, 11–13 giugno 1998. Tübingen: Niemeyer, 167–79. Maiden, Martin (2003). Verb augments and meaninglessness in Romance morphology. Studi di grammatica italiana 22: 1–61. Maiden, Martin (2004). When lexemes become allomorphs: on the genesis of suppletion. Folia Linguistica 38: 227–56. Maiden, Martin (2006). Accommodating synonymy: how some Italo-Romance verbs react to lexical and morphological borrowings. In Lepschy and Tosi (2006), 87–98. Maiden, Martin (2007). La linguistica romanza alla ricerca dell’arbitrario. In David Trotter (ed.), Actes du XXIVe CILPR. III. Tübingen: Niemeyer, 505–18. Maiden, Martin (2009). From pure phonology to pure morphology: the reshaping of the Romance verb. Recherches de linguistique de Vincennes 38: 45–82. Maiden, Martin (2010). Riflessioni comparative e storiche sulla sorte del congiuntivo presente nelle varietà italoromanze. In Giovanni Ruffino and Mari D’Agostino (eds), Storia della lingua italiana e dialettologia. Palermo: CFSLS, 129–49. Maiden, Martin (2011a). Morphological persistence. In Maiden, Smith, and Ledgeway (2011), 155–215. Maiden, Martin (2011b). Morphophonological innovation. In Maiden, Smith, and Ledgeway (2011), 216–67.

References

319

Maiden, Martin, and Parry, Mair (eds) (1997). The Dialects of Italy. London: Routledge. Maiden, Martin, Smith, John Charles, Goldbach, Maria, and Hinzelin, Marc-Oliver (eds) (2011). Morphological Autonomy: Perspectives from Romance Inflectional Morphology. Oxford: Oxford University Press. Maiden, Martin, Smith, John Charles, and Ledgeway, Adam (eds) (2011). The Cambridge History of the Romance Languages, vol. 1: Structures. Cambridge: Cambridge University Press. Malkiel, Yakov (1967). Multiple versus simple causation in linguistic change. In To Honor Roman Jakobson: Essays on the Occasion of his Seventieth Birthday (3 vols). The Hague: Mouton, ii.1228–46. Malkiel, Yakov (1969). The five sources of epenthetic /j/ in western Hispano-Romance: a study in multiple causation. Hispanic Review 37: 239–75. Mancarella, Giovan Battista (ed.) (1998). Salento: monografia regionale della ‘Carta dei dialetti italiani’. Lecce: Edizioni del Grifo. Manni, Paola (1979). Ricerche sui tratti fonetici e morfologici del fiorentino quattrocentesco. Studi di grammatica italiana 8: 115–71. Manni, Paola (2003). Storia della lingua italiana: il Trecento toscano. Bologna: Il Mulino. Manzini, Maria Rita, and Savoia, Leonardo (2004). Clitics: co-occurrence and mutual exclusion patterns. In Rizzi (2004), 211–50 Manzini, Maria Rita, and Savoia, Leonardo (2005). I dialetti italiani e romanci: morfosintassi generativa (3 vols). Alessandria: Edizioni dell’Orso. Manzini, Maria Rita, and Savoia, Leonardo (2007). A Unification of Morphology and Syntax: Investigations into Romance and Albanian Dialects. London: Routledge. Manzini, Maria Rita, and Savoia, Leonardo (2009). Morphology dissolves into syntax: infixation and doubling in Romance languages. Annali Online di Ferrara: Lettere 1: 1–28. Manzini, Rita, and Wexler, Kenneth (1987). Parameters, Binding Theory and learnability. Linguistic Inquiry 18: 413–44. Marazzini, Claudio, and Fornara, Simone (eds) (1999). Giovanni Francesco Fortunio, Regole grammaticali della volgar lingua. Pordenone: Accademia San Marco. Marchal, Harmony, Lemaire, Benoit, Bianco, Maryse, and Dessus, Philippe (2008). A MDLbased model of gender knowledge acquisition. In Alexander Clark and Kristina Toutanova (eds), Proceedings of the 12th Conference on Computational Natural Language Learning. Manchester: ACL, 73–80. Markun, Hans (1932). Vadere im Italienischen. Revue de linguistique romane 8: 281–354. Maschi, Roberta (2005). Processi analogici e naturalezza nella morfologia verbale romanza. Doctoral thesis, University of Padua. Maschi, Roberta, and Vanelli, Laura (2010). Morfologia verbale. In Salvi and Renzi (2010), ii.1431–91. Maschi, Roberta, Penello, Nicoletta, and Rizzolatti, Piera (eds) (2007). Miscellanea di studi linguistici offerti a Laura Vanelli da amici e allievi padovani. Udine: Forum. Mayerthaler, Willi (1977). Studien zur theoretischen und französischen Morphologie. Tübingen: Niemeyer.

320

References

Mayerthaler, Willi (1987). System-independent morphological naturalness. In Wolfgang Dressler, Willi Mayerthaler, Oswald Panagl, and Wolfgang Ullrich Wurzel (eds), Leitmotifs in Natural Morphology. Amsterdam: Benjamins, 25–58. Melander, Johan (1929). L’origine de l’italien me ne, me lo, te la, etc. Studia Neophilologica 2: 169–203. Melillo, Michele (1973). Le concordanze dei dialetti di Puglia. Bari: Università degli Studi di Bari. Menoni, Viviana (1982). Formazione e storia del gerundio composto nell’italiano antico. Studi di grammatica italiana 11: 5–88. Merci, Paolo (ed.) (1992). Il Condaghe di San Nicola di Trullas (CSNT). Sassari: Delfino. Meszler, Lenka, and Samu, Borbala (2007). ‘Deh, ragioniamo insieme un poco . . . ’: sulla generalizzazione della desinenza -iamo nell’indicativo presente in toscano antico. In Bentley and Ledgeway (2007), 165–83. Meul, Claire (2010). The intra-paradigmatic distribution of the infix -I/ESC- from Latin to Modern Romance: morphomic patterning and beyond. Morphology 20: 1–40. Meyer-Lübke, Wilhelm (1890). Grammaire des langues romanes, vol. 1: Phonétique. Paris: Welter. Meyer-Lübke, Wilhelm (1895). Grammaire des langues romanes, vol. 2: Morphologie. Paris: Welter. Meyer-Lübke, Wilhelm (1900). Grammaire des langues romanes, vol. 3: Syntaxe. Paris: Welter. Meyer-Lübke, Wilhelm ([1890–1906] 1974). Grammaire des langues romanes (4 vols). Geneva: Slatkine. Milsark, Gary (1977). Toward an explanation of certain peculiarities of the existential construction in English. Linguistic Analysis 3: 1–29. Minervini, Laura (in press). Gli altri siciliani: il sacrificio di Isacco in caratteri ebraici. In Studi in onore di Roberto Antonelli. Rome: Viella. Mithun, Marianne (1991). Active/agentive case marking and its motivations. Language 67: 510–46. Mocciaro, Antonia (1991). Italiano e siciliano nelle scritture dei semicolti. Palermo: Centro di studi filologici e linguistici siciliani. Monachesi, Paola (2005). The Verbal Complex in Romance: a Case Study in Grammatical Interfaces. Oxford: Oxford University Press. Monteil, Pierre (1963). La phrase relative en grec ancien: sa formation, son développement, sa structure des origines à la fin du Ve siècle A.C. Paris: Klincksieck. Moro, Andrea (1997). The Raising of Predicates: Predicative Noun Phrases and the Theory of Clause Structure. Cambridge: Cambridge University Press. Motolese, Matteo (ed.) (2004). Lodovico Castelvetro, Giunta fatta al ragionamento degli articoli et de’ verbi di messer Pietro Bembo. Rome: Antenore. Müller, Gary (2005). Pro-drop and impoverishment. In Patrick Brandt and Eric Fuss (eds), Form, Structure and Grammar: a Festschrift Presented to Günther Grewendorf on the Occasion of his 60th Birthday. Tübingen: Narr, 93–115. Munaro, Nicola (1999). Sintagmi interrogativi nei dialetti italiani settentrionali. Padua: Unipress.

References

321

Munaro, Nicola (2001). Free relatives as defective wh-elements. In Yves D’Hulst, Johan Rooryck, and Jan Schroten (eds), Romance Languages and Linguistic Theory 1999. Amsterdam: Benjamins, 281–306. Mussafia, Adolf (1868). Darstellung der altmailändischen Mundart nach Bonvesin’s Schriften. Kaiserliche Akademie der Wissenschaften in Wien: Philosophisch–Historische Klasse– Sitzungsberichte 59: 5–40. Napoli, Donna J. (1976). At least two si’s. Italian Linguistics 2: 123–47. Navone, Giulio (1922). Il dialetto di Paliano. Studi romanzi 17: 73–126. Neeleman, Ad, and Szendrői, Kriszta (2007). Radical pro-drop and the morphology of pronouns. Linguistic Inquiry 38: 671–714. Nencioni, Giovanni ([1953–4] 1989). Un caso di polimorfia della lingua letteraria dal secolo XIII al XVI. In Saggi di lingua antica e moderna. Turin: Rosenberg & Sellier, 11–188. Nesset, Tore (2006). Gender meets the usage-based model: four principles of rule interaction in gender assignment. Lingua 116: 1369–1493. Nichols, Johanna (1986). Head-marking and dependent-marking grammar. Language 62: 56–119. Nicoli, Franco (1983). Grammatica milanese. Busto Arsizio: Bramante. Niederländer, Johann Joseph (1900). Die Mundart von Namur. Zeitschrift für romanische Philologie 24: 1–32; 251–309. Noordhof, Harm (1937). La construction relative en italien. Doctoral thesis, The Hague: van Haeringen Nyrop, Kristian (1960). Grammaire historique de la langue française, vol. 2: Morphologie. Copenhagen: Gyldendal. Oliviéri, Michèle (2009). Typology or reconstruction: the benefits of dialectology for diachronic analysis. Paper presented at ‘Going Romance XXIII’, Nice, December. Orlandi, Giacomo (1980). Il dialetto di Roiate. Rome: Carocci. Ouali, Hamid (2008). On C-to-T -feature transfer: the nature of agreement and anti-agreement in Berber. In Roberta D’Alessandro, Gunnar Hrafnbjargarson, and Susan Fischer (eds). Agreement Restrictions. Berlin: Mouton, 159–80. Pace, Anna (1993–94). Ricerche di morfosintassi sui dialetti di Trebisacce e Castrovillari. BA thesis, University of Calabria. Paciaroni, Tania (2009). Sull’ausiliazione perfettiva in maceratese. In Bruno Moretti, Elena Pandolfi, and Matteo Casoni (eds), Linguisti in contatto: ricerche di linguistica italiana in Svizzera. Bellinzona: OLSI, 37–58. Pagani, Severino (1977). Come parla Meneghino: piccola grammatica del dialetto milanese. Milan: Virgilio. Pagano, Mario (2003). È davvero esistito il siciliano tre-quattrocentesco? Siculorum Gymnasium: rassegna della Facoltà di lettere e filosofia dell’Università di Catania 56: 161–77. Palmer, Frank (1961). The Latin Language. London: Allen & Unwin. Palumbo, Pietro (1954). Sposizione del Vangelo della Passione secondo Matteo (3 vols). Palermo: Centro di studi filologici e linguistici siciliani. Paoli, Sandra (2009). Gerarchie di caso e persona nei paradigmi dei pronomi atoni obliqui: il caso dei dialetti del Comelico. Rivista italiana di dialettologia 33: 59–84.

322

References

Papanti, Giovanni (1875). I parlari italiani in Certaldo alla festa del V centenario di messer Giovanni Boccacci: omaggio di Giovanni Papanti. Livorno: Vigo. Parodi, Ernesto Giacomo (1887). Illustrazioni linguistiche ai suddetti frammenti. Giornale storico della letteratura italiana 10: 178–96. Parodi, Ernesto Giacomo (1896). Studi liguri. Archivio glottologico italiano 14: 1–110. Parry, Mair (1991). Posizione dei clitici complemento nelle costruzioni verbali perifrastiche del piemontese. In Gianrenzo Clivio and Censin Pich (eds), Rëscontr antërnassional dë studi an sla lenga e la literatura piemontèisa VIII. Alba: Famija Albèisa, 247–59. Parry, Mair (1992). Innovazione e conservazione nell’entroterra savonese: appunti di morfosintassi valbormidese. In Lorenzo Massobrio and Giulia Petracco Sicardi (eds), L’anfizona ligure-padana. Alessandria: Edizioni dell’Orso, 49–72. Parry, Mair (1995). Some observations on the syntax of clitic pronouns in Piedmontese. In Martin Maiden and John Charles Smith (eds), Linguistic Theory and the Romance Languages. Amsterdam: Benjamins, 133–60. Parry, Mair (1996). La negazione italo-romanza: variazione tipologica e variazione strutturale. In Benincà et al. (1996), 225–57. Parry, Mair (1997). Preverbal negation and clitic ordering, with particular reference to a group of north-west Italian dialects. Zeitschrift für romanische Philologie 113.2: 243–70. Parry, Mair (1998a). The reinterpretation of the reflexive in Piedmontese: ‘impersonal’ se constructions. Transactions of the Philological Society 96.1: 63–116. Parry, Mair (1998b). La sintassi dei pronomi soggetto in piemontese. In Ramat and Roma (1998), 329–44. Parry, Mair (2000). Accordo e soggetti postverbali in piemontese. In Annick Englebert, Michel Pierrard, Laurence Rosier, and Dan Van Raemdonck (eds), Actes du XXIIe Congrès international de linguistique et philologie romanes, Brussels. Berlin: Mouton de Gruyter, 391–402. Parry, Mair (2005). Parluma ’D Còiri: sociolinguistica e grammatica del dialetto di Cairo Montenotte. Savona: Società savonese di storia patria, Editrice Liguria. Parry, Mair (2007a). ‘La frase relativa (con antecedente) negli antichi volgari dell’Italia nordoccidentale’, LabRomAn 1: 9–32 (http://www.maldura.unipd.it/ddlcs/laboratorio/numero1_ 2007.html) Parry, Mair (2007b). The interaction of semantics and syntax in the spread of relative che in the early vernaculars of Italy. In Bentley and Ledgeway (2007), 200–219. Pérez Saldanya, Manuel (1998). Del llatí al català: morfosintaxi verbal històrica. Valencia: Universitat de València. Perlmutter, David M. (1978). Impersonal passives and the unaccusative hypothesis. Proceedings of the 4th Annual Meeting of the Berkeley Linguistic Society, 157–89. Perlmutter, David M. (1989). Multiattachment and the Unaccusative Hypothesis: the perfect auxiliary in Italian. Probus 1: 63–119. Perna, Elena (2013). Sintassi delle relative in greco antico. Doctoral thesis, University of Padua. Pernicone, Vincenzo (1946). Franco Sacchetti, Il trecentonovelle. Florence: Sansoni. Péronnet, Louise (1990). Système des conjugaisons verbales dans le parler acadien du sud-est du Nouveau-Brunswick. Journal of the Atlantic Provinces Linguistic Association 12: 81–116. Pescarini, Diego (2010). Elsewhere in Romance: evidence from clitic clusters. Linguistic Inquiry 41.3: 427–44.

References

323

Pescia, Lorenza (2011). Sintassi e semantica delle forme participiali alternanti con particolare riguardo all’italo-romanzo. Doctoral thesis, University of Zürich. Pescia, Lorenza (2012). L’ausiliazione perfettiva nel dialetto galloitalico di Picerno (Basilicata). Poster presented at ‘Italian Dialect Meeting’, Leiden, May. Pesetsky, David (1987). Wh-in-situ: Movement and Unselective Binding. In Eric Reuland and Alice ter Meulen (eds), The Representation of Indefiniteness. Cambridge, Mass.: MIT Press. Petrocchi, Giorgio (ed.) (1966). Dante Alighieri, La Commedia secondo l’antica vulgata, vol. 2: Inferno. Milan: Mondadori. Pinkster, Harm (1987). The strategy and chronology of the development of future and perfect tense auxiliaries in Latin. In Harris and Ramat (1987), 193–223. Piras, Sara Silvia, and Dessì, Gisa (eds) (2003). Il Registro di S. Pietro di Sorres. Cagliari: Cuec. Pirrelli, Vito (2000). Paradigmi in morfologia: un approccio multidisciplinare alla flessione verbale dell’italiano. Pisa: Istituti editoriali e poligrafici internazionali. Pisano, Simone (2010). L’utilizzo dei tempi sovraccomposti in alcune varietà sarde moderne. Lingua e stile 45: 125–33. Pitré, Giuseppe (1875). Fiabe, novelle e racconti popolari siciliani, vol. 1. Catania: Clio. Pokorny, Julius (1930). Vergleichendes Wörterbuch der Indogermanischen Sprachen. Berlin: de Gruyter. 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 standard Italian. In Guglielmo Cinque and Giampaolo Salvi (eds), Current Studies in Italian Syntax: Essays Offered to Lorenzo Renzi. Amsterdam: Elsevier, 265–86. Poletto, Cecilia (2006). Parallel phases: a study on the high and low left periphery of Old Italian. In Frascarelli (2006: 261–94). 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, 499–517. Poletto, Cecilia, and Pollock, Jean-Yves (2009). Another look at wh-questions in Romance: the case of Mendrisiotto and its consequences for the analysis of French wh-in-situ and embedded interrogatives. In Leo Wetzels (ed.), Romance Languages and Linguistic Theory 2006. Amsterdam: Benjamins, 199–258. Politzer, Robert (1958). On the history of the third person ending in Italian. Italica 35: 192–7. Pollock, Jean-Yres (1989). Verb movement, Universal Grammar and the structure of IP. Linguistic Inquiry 20: 365–424. Pons, Cathy (1990). Language death among Waldensians of Valdese, North Carolina. Doctoral thesis, Indiana University. Pope, Mildred (1952). From Latin to Modern French, 2nd edn. Manchester University Press. Porro, Marzio (ed.) (1979). Domenico Cavalca, Dialogo de Sam Gregorio composito in vorgà. Florence: Accademia della Crusca. Porta, Giuseppe (ed.) (1979). Cronica di Anonimo romano. Milan: Adelphi. Porzio Gernia, Maria Luisa (1982). Il latino e le lingue indoeuropee dell’Italia antica. In Edoardo Vineis (ed.), Alle origini del Latino. Pisa: Giardini, 11–26.

324

References

Pountain, Christopher (1995). Infinitives with overt subjects: a pragmatic approach. In Tom Earle and Nigel Griffin (eds), Portuguese, Brazilian and African Studies: Studies Presented to Clive Willis on his Retirement. Warminster: Aris & Phillips, 11–25. Prader Schucany, Silvia (1970). Romanisch Bünden als selbständiges Sprachlandschaft. Bern: Francke. Price, Glanville (1971). The French Language: Present and Past. London: Arnold. Princivalli, Alessandra, and Ortalli, Gherardo (eds) (1993). Il Capitolare degli Ufficiali sopra Rialto: nei luoghi al centro del sistema economico veneziano (secoli XIII–XIV). Milan: La Storia. Quaglio, Antonio Enzo (ed.) (1967). Tutte le opere di Giovanni Boccaccio. Milan: Mondadori. Quer, Josep (ed.) (2009a). Twists of Mood: The Distribution and Interpretation of Indicative and Subjunctive. Special issue of Lingua 119.12. Quer, Josep (2009b). Twists of mood: the distribution and interpretation of indicative and subjunctive. In Quer (2009a: 1779–87). Ramat, Paolo, and Roma, Elisa (eds) (1998). Sintassi storica: atti del XXX Congresso internazionale della Società di linguistica italiana, Pavia, 26–28 settembre 1996. Rome: Bulzoni. Rasom, Sabrina (2006). La posizione del pronome clitico oggetto con le forme verbali ‘nonfinite’: differenze areali e generazionali nel fassano. In Gianna Marcato (ed.), Giovani, lingue e dialetti. Padua: Unipress, 109–114. Reinhart, Tanya (1998). Wh-in-situ in the framework of the Minimalist Program. Natural Language Semantics 6: 29–56. Remberger, Eva-Maria (2006). Hilfsverben: eine minimalistiche Analyse am Beispiel des Italienischen und Sardischen. Tübingen: Niemeyer. Renzi, Lorenzo, and Salvi, Giampaolo (2010). Italiano antico. LabRomAn 4.1: 35–57: http:// www.maldura.unipd.it/ddlcs/laboratorio/renzi-salvi-4-I-2011.pdf Renzi, Lorenzo, Salvi, Giampaolo, and Cardinaletti, Anna (eds) (1988). Grande grammatica italiana di consultazione, vol. 1: La frase: i sintagmi nominale e preposizionale. Bologna: Il Mulino. Renzi, Lorenzo, and Vanelli, Laura (1983). I pronomi soggetto in alcune varietà romanze. In Scritti linguistici in onore di Giovan Battista Pellegrini. Pisa: Pacini, 121–45. Retaro, Valentina (2005). La variabilità del dialetto in una comunità dell’area flegrea: il borgo di Cento Camerelle a Bacoli. BA thesis, University of Naples Federico II. Retaro, Valentina (2009). Dinamiche dialettali in area nord-vesuviana. Doctoral thesis, University of Turin. Ricca, Davide (1998). Una perifrasi continua-iterativa nei testi piemontesi dal Cinquecento all’Ottocento: tenere + participio passato. In Ramat and Roma (1998), 345–68. Rice, Curt (2006). Optimizing gender. Lingua 116: 1394–17. Richards, Marc (2004). Object Shift in North and West Germanic: optionality, scrambling and base-generated OV. Doctoral thesis, University of Cambridge. Rinaldi, Gaetana Maria (2005). Testi d’archivio del Trecento (2 vols). Palermo: Centro di studi filologici e linguistici siciliani. Rindler Schjerve, Rosita, and Kratschmer, Alexandra (1990). Zur Auxiliarselektion im periphrastischen Perfektparadigma der italienischen Pronominalverben. In Monica Berretta,

References

325

Piera Molinelli, and Ada Valentini (eds), Parallela 4: Morfologia/Morphologie. Atti del V Incontro italo-austriaco (Bergamo 2–4 ottobre 1989). Tübingen: Narr, 119–32. Rivero, María-Luisa (1978). Topicalization and WH-movement in Spanish. Linguistic Inquiry 9: 513–17. Rix, Helmut (2004). Etruscan. In Woodward (2004), 141–61. Rizzi, Luigi (1978). A restructuring rule in Italian syntax. In Samuel Jay Keyser (ed.), Recent Transformational Studies in European Languages. Cambridge, Mass.: MIT Press, 113–58. 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-Corvalán (eds), Studies in Romance Syntax. Dordrecht: Foris, 391–419. Rizzi, Luigi (1987). Three remarks on null subjects. Paper presented at ‘Workshop on Dialectology, GLOW X’, University of Venice. Rizzi, Luigi (1997). The fine structure of the left periphery. In Liliane Haegeman (ed.), Elements of Grammar. Dordrecht: Kluwer, 281–337. Rizzi, Luigi (ed.) (2004). The Structure of CP and IP. Oxford: Oxford University Press. Rizzi, Luigi, and Roberts, Ian (1989). Complex inversion in French. Probus 1: 1–30. Roberts, Ian (2007). Diachronic Syntax. Oxford: Oxford University Press. Roberts, Ian (2010a). Agreement and Head Movement: Clitics and Defective Goals. Cambridge, Mass.: MIT Press Roberts, Ian (2010b). A deletion analysis of null subjects. In Biberauer et al. (2010), 58–87. Roberts, Ian (2010c). Varieties of French and the Null-Subject Parameter. In Biberauer et al. (2010), 303–27. Roberts, Ian (2012). On the nature of syntactic parameters: a programme for research. In Juaita Avelar, Sonia Cyrino, Charlotte Galves, and Ruth Lopez (eds), Diacronic Syntax. Oxford: Oxford University Press, 319–34. Roberts, Ian, and Roussou, Anna (2003). Syntactic Change: A Minimalist Approach to Grammaticalization. Cambridge: Cambridge University Press. Robustelli, Cecilia (2000). Causativi in italiano antico e moderno. Modena: Il Fiorino. Rohlfs, Gerhard (1956–9). Vocabolario dei dialetti salentini (Terra d’Otranto). Munich: Verlag der Bayerischen Akademie der Wissenschaften. Repr. 1976 by Congedo, Lecce. Rohlfs, Gerhard (1966). Grammatica storica della lingua italiana e dei suoi dialetti, vol. 1: Fonetica. Turin: Einaudi. Rohlfs, Gerhard (1968). Grammatica storica della lingua italiana e dei suoi dialetti, vol. 2: Morfologia. Turin: Einaudi. Rohlfs, Gerhard (1969). Grammatica storica della lingua italiana e dei suoi dialetti, vol. 3: Sintassi e formazione delle parole. Turin: Einaudi. Rohlfs, Gerhard (1976). Vocabolario dei dialetti salentini (Terra d’Otranto) (3 vols). Lecce: Congedo. Rohlfs, Gerhard (1977). Nuovo dizionario dialettale della Calabria. Ravenna: Longo. Ronjat, Jules (1937). Grammaire istorique [sic] des parlers provençaux modernes, vol. 3: Morphologie et formation des mots. Montpellier: Société des langues romanes. Rosen, Carol ([1981] 1988). The Relational Structure of Reflexive Clauses. New York: Garland. Rosen, Carol (1982). The Unaccusative Hypothesis and the ‘inherent clitic’ phenomenon in Italian. Chicago Linguistic Society 18: 530–41.

326

References

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, 38–77. Rosen, Carol (1997). Auxiliation and serialization: on discerning the difference. In Alex Alsina, Joan Bresnan, and Peter Sells (eds), Complex Predicates. Stanford, Calif.: CSLI, 175–202. Rosén, Hannah (1999). Latine loqui: Trends and Directions in the Crystallisation of Classical Latin. Munich: Fink. Rowlett, Paul (2007). The Syntax of French. Cambridge: Cambridge University Press. Russell, Bertrand (1905). On denoting. Mind 14: 479–93. Sabatini, Francesco (1965a). Sull’origine dei plurali italiani: il tipo in -i. Studi linguistici italiani 5: 5–39. Sabatini, Francesco (1965b). Esigenze di realismo e dislocazione morfologica in testi preromanzi. In Studi in onore di Alfredo Schiaffini, Rivista di cultura classica e medioevale 7.2: 972–98. Sabatini, Francesco ([1983] 1996). Prospettive sul parlato nella storia linguistica italiana (con una lettura dell’Epistola napoletana del Boccaccio). In Vittorio Coletti, Rosario Coluccia, Paolo D’Achille, Nicola De Blasi, and Livio Petrucci (eds), Francesco Sabatini, Italia linguistica delle origini: saggi editi dal 1956 al 1996. Lecce: Argo, 425–66. Saito, Mamoru (2007). Notes on East Asian argument ellipsis. MS, Nanzan University and University of Connecticut. Salvi, Giampaolo (1988). La frase semplice. In Renzi et al. (1988), i.29–113. Salvi, Giampaolo (2008a). Imperfect systems and diachronic change. In Ulrich Detges and Richard Waltereit (eds), The Paradox of Grammatical Change: Perspectives from Romance. Amsterdam: Benjamins, 127–45. Salvi, Giampaolo (2008b). La formazione della costruzione impersonale in italiano. Revista de estudos linguísticos da Universidade do Porto 3.1: 13–37. Salvi, Giampaolo (2010a). La realizzazione sintattica della struttura argomentale. In Salvi and Renzi (2010), i.123–89. Salvi, Giampaolo (2010b). Az olasz személytelen szerkezet születése mint interferencia-jelenség. In Katalin É. Kiss and Attila Hegedűs (eds), Nyelvelmélet és kontaktológia. Piliscsaba: PPKE BTK Elmélet Nyelvészeti Tanszék–Magyar Nyelvészeti Tanszék, 99–115. Salvi, Giampaolo, and Renzi, Lorenzo (eds) (2010). Grammatica dell’italiano antico (2 vols). Bologna: Il Mulino. Salvi, Giampaolo, and Vanelli, Laura (2004). Nuova grammatica italiana. Bologna: Il Mulino. Salvioni, Carlo (ed.) (1886). La Passione e altre antiche scritture lombarde. Archivio glottologico italiano 9: 1–22. Salvioni, Carlo (1902). Dell’antico dialetto pavese. Pavia: Fusi. Salvioni, Carlo (1903). Del pronome enclitico oggetto suffisso ad altri elementi che non sieno la voce verbale, Rendiconti dell’Istituto lombardo (2nd series) 37: 1012–21. Repr. in Salvioni (2008), ii.96–105. Salvioni, Carlo (1919). Sul dialetto milanese arcaico. Rendiconti dell’Istituto lombardo di scienze e lettere (2nd series) 52: 517–40. Repr. in Salvioni (2008), iii.181–204. Salvioni, Carlo (1975). Fonetica e morfologia del dialetto milanese. L’Italia dialettale 38: 1–46.

References

327

Salvioni, Carlo (2008). Scritti linguistici, ed. Michele Loporcaro, Lorenza Pescia, Romano Broggini, and Paola Vecchio. 5 vols. Bellinzona: Edizioni dello Stato del Cantone Ticino. Sanga, Glauco (1979). Dialetto e folklore: ricerca a Cigole. Milan: Silvana. Sauerland, Uli (1999). Two structures for English restrictive relative clauses. In Mamoru Saito, Yasuaki Abe, Hiroshi Aoyagi, Masatake Arimoto, Keiko Murasugi, and Tatsuya Suzuki (eds), Proceedings of the Nanzan GLOW. Nagoya: Nanzan University, 351–66 (http://www2. sfs.nphil.uni-tuebingen.de/~uli/GlowNanzanPaper.pdf). Sauerland, Uli (2003). Unpronounced heads in relative clauses. In Kerstin Schwabe and Susanne Winkler (eds), The Interfaces: Deriving and Interpreting Omitted Structures. Amsterdam: Benjamins, 205–26. Săvescu, Oana (2007). Challenging the Person Case Constraint: evidence from Romanian. In José Camacho, Nydia Flores-Ferrán, Liliana Sánchez, Viviane Déprez, and María José Cabrera (eds), Romance Linguistics 2006. Amsterdam: Benjamins, 255–68. Savj-Lopez, Paolo (1900). Studi d’antico napoletano. Zeitschrift für romanische Philologie 24: 501–7. Schaden, Gerhard (2007). La sémantique du parfait: étude des ‘temps composés’ dans un choix de langues germaniques et romanes. Doctoral thesis, Université Paris 8. Schaden, Gerhard (2012). Modelling the ‘aoristic drift of the present perfect’ as inflation: an essay in historical pragmatics. International Review of Pragmatics 4: 261–92. Schiaffini, Alfredo (ed.) (1926/1954). Testi fiorentini del Dugento e dei primi del Trecento. Florence: Sansoni. Schifano, Norma (2011). Verb-movement in Italian, French and Spanish: a survey from written sources. MA thesis, Universities of Cambridge and Venice. Schuchardt, Hugo (1866–8). Der Vokalismus des Vulgärlatein. Leipzig: Teubner. Schwan, Dietrich, and Behrens, Edward (1932). Grammaire de l’ancien français: phonétique et morphologie. Leipzig: Reisland. Schwegler, Armin (1990). Analyticity and Syntheticity: A Diachronic Perspective with Special Reference to Romance Languages. Berlin: Mouton de Gruyter. Scorretti, Mauro (1994). Complementizer deletion. Doctoral thesis, University of Amsterdam. Selkirk, Elisabeth (1995). The prosodic structure of function words. In Jill Beckman, Laura Walsh Dickey, and Suzanne Urbanczyk (eds), Papers in Optimality Theory. Amherst, Mass.: GLSA, 439–70. Repr. in James Morgan and Katherine Demuth (eds), Signal to Syntax: Bootstrapping from Speech to Grammar in Early Acquisition (Mahwah, NJ: Erlbaum, 1996), 187–214. Sermoneta, Giuseppe (ed.) (1994). Alfabetin: traduzione giudeo-siciliana in carattere ebraici del servizio della Pentecoste. Palermo: Centro di studi filologici e linguistici siciliani. Sgrilli, Paola (1983). Il ‘libro di Sidrac’ salentino. Pisa: Pacini. Sheehan, Michelle (2013). Some implications of a Copy Theory of labeling. Syntax 16: 362–96. Shibatani, Masayosh, and Pardeshi, Prashant (2002). The causative continuum. In Masayosh Shibatani (ed.), The Grammar of Causation and Interpersonal Manipulation. Amsterdam: Benjamins, 85–126. Simone, Raffaele (ed.) (2011). Enciclopedia dell’italiano, vol. 2. Rome: Istituto dell’Enciclopedia italiana G. Treccani.

328

References

Skytte, Gunver (1983). La sintassi dell’infinito in italiano moderno (2 vols). Copenhagen: Études romanes de l’Université de Copenhague. Smith, John Charles (2011). Variable analyses of a verbal inflection in (mainly) Canadian French. In Maiden et al. (2011), 311–26. Sorace, Antonella (2000). Gradients in auxiliary selection with intransitive verbs. Language 76: 859–90. Sornicola, Rosanna (1992). ‘Col nostro simplice parlare et muliebre stilo’: ibridismi e registri linguistici nella Leggenda della Beata Eustochia da Messina. In Mario Fasino and Giovanni Ruffino (eds), Studi linguistici e filologici offerti a Girolamo Caracausi. Palermo: Centro di studi filologici e linguistici siciliani, 453–81. 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 (2000). Processi di convergenza nella formazione di un tipo sintattico: la genesi ibrida dell’oggetto preposizionale. In Actes du XXIIe Congrès international de linguistique et de philologie romanes (Bruxelles 23–29 Juillet 1998), vol. 2: Les nouvelles ambitions de la linguistique diachronique. Brussels: Niemeyer, 419–27. Sornicola, Rosanna (2003). Polimorfismo e instabilità strutturale: un esame della dittongazione spontanea dell’area flegrea in una prospettiva romanza. In Fernando Sánchez Miret (ed.), Actas del XXIII Congreso internacional de lingüística y filología románica (Salamanca, 24–30 settembre 2001). Tübingen: Niemeyer, 301–13. Sornicola, Rosanna (2011). Romance linguistics and historical linguistics: reflections on synchrony and diachrony. In Maiden et al. (2011), 1–49. Sornicola, Rosanna (2012). Bilinguismo e diglossia dei territori bizantini e longobardi del Mezzogiorno: le testimonianze dei documenti del IX e X secolo. Naples: Accademia Pontaniana. Sornicola, Rosanna (2013). Decomposizioni e ricomposizioni di sistemi: i pronomi personali delle lingue romanze tra paradigmatica e sintagmatica. In Emili Casanovo and Cesáreo Calvo Rigual (eds), Actas del XXVI Congres internacional de lingüística y filología romànicas (Valencia 6–11 Septiembre 2010). Berlin: Mouton de Gruyter, 1, 419–40. Sornicola, Rosanna, et al. (in preparation). Analisi linguistica del Codex Cavensis. Naples: Accademia di scienze, lettere e arti. Spencer, Andrew (2001). The paradigm-based model of morphosyntax. Transactions of the Philological Society 99: 279–313. Sportiche, Dominique (2008). Inward bound: splitting the wh-paradigm and French relative qui. MS, UCLA (http://ling.auf.net/lingBuzz/000623). Squartini, Mario (1998). Verbal Periphrases in Romance: Aspect, Actionality and Grammaticalization. Berlin: Mouton de Gruyter. Starke, Michal (2009). Nanosyntax: a short primer to a new approach to language. Nordlyd 36.1: 1–6. Stefanini, Ruggero (1983). Riflessivo, impersonale e passivo in italiano e fiorentino. Quaderni dell’atlante linguistico toscano 1: 103–14. Stussi, Alfredo (1965). Testi veneziani del Duecento e dei primi del Trecento. Pisa: Nistri-Lischi. Svenonius, Peter (2010). Spatial P in English. In Cinque and Rizzi (2010), 127–60.

References

329

Szabó, Zoltán (2012). The case for compositionality. In Markus Werning, Wolfram Hinzen, and Edouard Machery (eds), The Oxford Handbook of Compositionality. Oxford: Oxford University Press, 64–80. Szmrecsanyi, Benedikt (2012). Analyticity and syntheticity in the history of English. In Terttu Nevalainen and Elizabeth Closs Traugott (eds), The Oxford Handbook of the History of English. Oxford: Oxford University Press, 654–65. Taylor, Cameron (in preparation). Calabrian clausal syntax: the case of Nicoterese. Doctoral thesis, University of Cambridge. Tekavčić, Pavao (1980a). Grammatica storica della lingua italiana, vol. 1: Fonematica. Bologna: Il Mulino. Tekavčić, Pavao (1980b). Grammatica storica della lingua italiana, vol. 2: Morfosintassi. Bologna: Il Mulino. Thielmann, Phili (1885). Habere mit dem Infinitiv und die Entstehung des romanischen Futurums. Archiv für lateinische Lexikographie 2: 157–202. Thornton, Anna (1999). Diagrammaticità, uniformità di codifica e morfomicità. In Paola Benincà, Alberto Mioni, and Laura Vanelli (eds), Fonologia e morfologia dell’italiano e dei dialetti d’Italia. Rome: Bulzoni, 483–502. Thornton, Anna (2001). Some reflections on gender and inflectional class assignment in Italian. In Chris Schaner-Wolles, John Rennison, and Friedrich Neubarth (eds), Naturally! Linguistic Studies in Honour of Wolfgang Ulrich Dressler. Turin: Rosenberg & Sellier, 479–87. Thornton, Anna (2009). Overabundance in Italian verb morphology and its interactions with other non-canonical phenomena. Paper presented at ‘Irregularity in Morphology’, Bremen. Thornton, Anna (2011). Overabundance (multiple forms realizing the same cell): a noncanonical phenomenon in Italian verb morphology. In Maiden et al. (2011), 358–81. Tomioka, Satoshi (2003). The semantics of Japanese null pronouns and its cross-linguistic implications. In Kerstin Schwabe and Susanne Winkler (eds), The Interfaces: Deriving and Interpreting Omitted Structures. Amsterdam: Benjamins, 321–40. Toppino, Giuseppe (1926). Il dialetto di Castellinaldo: note di sintassi. L’Italia dialettale 2: 1–49. Tortora, Christina (2002). Romance enclisis, prepositions, and aspect. Natural Language & Linguistic Theory 20: 725–58. Tortora, Christina (2006). On morpho-syntactic variation in English dialects. Paper presented at the University of Padua, October–November. Tortora, Christina (2008). The comparative morpho-syntax of Appalachian English. Paper presented at the 3rd meeting of ‘European Dialect Syntax’, Venice, September. Tortora, Christina (2010). Domains of clitic placement in finite and non-finite clauses: evidence from a Piedmontese dialect. In D’Alessandro et al. (2010), 135–49. Tortora, Christina (in press). A Comparative Grammar of Borgomanerese. New York: Oxford University Press. Toscano, Reinat (1998). Gramàtica niçarda. Bélin-Béliet: Princi Negre. Toso, Fiorenzo (1997). Grammatica del genovese. Genoa: Le Mani. Trumper, John (2011). Lessici in contatto: l’elemento bizantino del veneziano-veneto-giuliano, ulteriori studi. In Antonio Mendicino, Nadia Prantera, and Cinzia Citraro (eds), Atti del Convegno ‘Parole: il lessico come strumento per organizzare e trasmettere gli etnosaperi’. Rende: Centro editoriale e librario, Università della Calabria, 791–855.

330

References

Trumper, John, and Vigolo, Maria Teresa (1995). Il Veneto centrale: problemi di classificazione dialettale e di fitonimia. Padua: CNR. Tuttle, Edward (1992). Del pronome d’oggetto suffisso al sintagma verbale: in calce ad una nota salvioniana del 1903. L’Italia dialettale 55: 13–63. Ugolini, Francesco A. (1982). Per la storia del dialetto di Roma nel Cinquecento. La ‘vecchia romanesca’, ne Le Stravaganze d’amore di Cristoforo Castelletti (1587). Contributi di dialettologia umbra 2.3: 73–199. Ulleland, Magnus (1961). Io canto come colui che . . . . Studia neophilologica 33: 329–34. Repr. in Paola Benincà and Lorenzo Renzi (eds) (2011), Magnus Ulleland. Studi di italiano antico. Padua: Unipress, 67–7. Ulleland, Magnus (1963). Francesca da Rimini e il costrutto Io canto come colui che . . . . Studia neophilologica 35: 290–94. Repr. in Paola Benincà and Lorenzo Renzi (eds) (2011), Magnus Ulleland. Studi di italiano antico. Padua: Unipress, 75–80. Väänänen, Veikko (1966). Le latin vulgaire des inscriptions pompéiennes. Berlin: Akademie. Väänänen, Veikko (1981). Introduction au latin vulgaire. Paris: Klincksieck. Vai, Massimo (1996). Per una storia della negazione in milanese in comparazione con altre varietà altoitaliane. ACME 49.1: 57–98. Valesio, Paolo (1968). The Romance synthetic future pattern and its first attestations. Lingua 20: 113–61. Valesio, Paolo (1969). The synthetic future again: phonology and morphosyntax. Lingua 24: 181–93. Van Valin, Robert, Jr (1990). Semantic parameters of split intransitivity. Language 66: 221–260. Van Valin, Robert, Jr (2005). Exploring the Syntax–Semantics Interface. Cambridge: Cambridge University Press. Van Valin, Robert, Jr, and LaPolla, Randy (1997). Syntax: Structure, Meaning and Function. Cambridge: Cambridge University Press. Vanelli, Laura (2007). Allomorfia e allotropia nella flessione verbale dell’italiano: le terminazioni di 3a persona plurale in italiano antico (e moderno). In Studi in onore di Pier Vincenzo Mengaldo per i suoi settant’anni. Florence: Galluzzo, ii.1797–1818. Vanelli, Laura (in press). Variazione desinenziale nella flessione verbale dell’italiano antico. Paper presented at ‘XXVIe Congrès International de Linguistique et de Philologie Romanes’ (Valencia, September 2010). Vangsnes, Øystein (2008). Decomposing manner how in colloquial Scandinavian. Studia Linguistica 62.1: 119–41. Vecchio, Paola (2006). L’ausiliazione perfettiva in napoletano: studio di sintassi diacronica. Revue de linguistique romane 70: 53–94. Vecchio, Paola (2010). The distribution of the complementizers /ka/ and /ku/ in the North Salentino dialect of Francavilla Fontana (Brindisi). In D’Alessandro et al. (2010), 312–22. Verrier, Anatole Joseph, and Onillon, René (1908). Glossaire des patois et parlers d’Anjou. Angers: Germain & Grassin. Viero, Franco (ed.) (1994). L’erbolato di Tubiolo: erbario rustico del ’600, 2nd edn. Bassano del Grappa: Ghedina & Tassotti.

References

331

Vignuzzi, Ugo (1992). Per la definizione della scripta romanesca ‘di tipo medio’ nel sec. XV: le due redazioni delle Visioni di S. Francesca Romana. Contributi di filologia del’Italia mediana 6: 49–130. Vincent, Nigel (1982). The development of the auxiliaries habere and esse in Romance. In Vincent and Harris (1982), 71–99. Vincent, Nigel (1996). L’infinito flesso in un testo napoletano del Trecento. In Benincà et al. (1996), 389–409. Vincent, Nigel (1997a). Synthetic and analytic structures. In Maiden and Parry (1997), 99–105. Vincent, Nigel (1997b). Complementation. In Maiden and Parry (1997), 171–8. Vincent, Nigel (1998). On the grammar of inflected non-finite forms (with special reference to Old Neapolitan). In Iørn Korzen and Michael Herslund (eds), Clause Combining and Text Structure. Frederiksberg: Samfundslitteratur, 135–58. Vincent, Nigel (2011). Non-finite forms, periphrases and autonomous morphology in Latin and Romance. In Maiden et al. (2011), 417–35. Vincent, Nigel (2013). Compositionality and change in conditionals and counterfactuals in Romance. In Silvio Cruschina, Martin Maiden, and John Charles Smith (eds), Boundaries of Pure Morphology. Oxford: Oxford University Press, 117–37. Vincent, Nigel (forthcoming). Compositionality and change. In Claire Bowern and Bethwyn Evans (eds), The Routledge Handbook of Historical Linguistics. London: Routledge. Vincent, Nigel, and Harris, Martin (eds) (1982). Studies in the Romance Verb. London: Croom Helm. Vincent, Nigel, Parry, Mair, and Hastings, Robert (2004). Il progetto SAVI: presentazione, procedure e problemi. In Dardano and Frenguelli (2004), 501–28. Virdis, Maurizio (1988). Sardische: Areallinguistik. In Günter Holtus, Michael Metzeltin, and Christian Schmitt (eds), Lexikon der romanistischen Linguistik, vol. 4: Italienisch, Korsisch, Sardisch. Tübingen: Niemeyer, 897–913. Vitale, Maurizio (1953). La lingua volgare della cancelleria visconteo-sforzesca nel Quattrocento. Varese: Istituto editoriale cisalpino. Wallace, Rex (2004). Sabellian. In Woodward (2004), 96–123. Wanner, Dieter (1977). On the order of clitics in Italian. Lingua 43: 101–28. Wanner, Dieter (1981). Surface complementizer deletion: Italian che ~ Ø. Journal of Italian Linguistics 6: 47–82. Wanner, Dieter (1983). Intorno alla sintassi pronominale alpina. In Renato Martinoni and Vittorio Raschèr (eds), Problemi linguistici nel mondo alpino. Napoli: Liguori, 169–85. Washington, Hannah (2008). Anomalies in Romance noun gender: the evolution of Latin nouns and cross-linguistic variation. Independent scholar thesis, Middlebury College, Vermont. Available at: http://middarchive.middlebury.edu/cdm/singleitem/collection/scholarship/id/ 241/rec/3 Watanabe, Akira (2004). The genesis of negative concord: syntax and morphology of negative doubling. Linguistic Inquiry 35: 559–612. Watumull, Jeffrey (2010). The emergence of Merge. MPhil. thesis, University of Cambridge. Wehr, Barbara (1995). SE-Diathese im Italienischen. Tübingen: Niemeyer. Wheeler, Max (1988). Occitan. In Harris and Vincent (1988), 246–78.

332

References

Wierzbicka, Anna (1981). Case marking and human nature. Australian Journal of Linguistics 1: 43–80. Wilhelm, Raymund (2006). Bonvesin da la Riva: la vita di Sant’Alessio. Tübingen: Niemeyer. Willis, R. Clive (1971). An Essential Course in Modern Portuguese. Walton-on-Thames: Nelson. Woodward, Roger (ed.) (2004). The Cambridge Encyclopedia of the World’s Ancient Languages. Cambridge: Cambridge University Press. Zagona, Karen (2002). The Syntax of Spanish. Cambridge: Cambridge University Press. Zahner, Gérard (1989). Il dialetto della Val San Giacomo (Valle Spluga). Milan: Università Cattolica. Zamboni, Alberto (1974). Veneto. Profilo dei dialetti italiani 5. Pisa: Pacini. Zanuttini, Raffaella (1997). Negation and Clausal Structure: A Comparative Study of Romance Languages. New York: Oxford University Press Ziccardi, Giovanni (1910). Il dialetto di Agnone: la fonetica e la flessione. Zeitschrift für romanische Philologie 34: 405–36.

Index Abeillé 11, 15 Abruzzese 30, 81 Acri 208 active – inactive alignment 51, 53 Adams 6, 12, 137, 187, 188 adverb positions 34 focused 36 higher adverb space (HAS) 34–5, 36, 37, 39, 45 lower adverb space (LAS) 34–5, 36, 37 Aebischer 202 Agnonese 49, 54, 63–8, 69, 70, 80 Agree 181, 182, 189, 192, 193, 198, 199 local 45 long-distance 44–5, 47 agreement (see also participial: agreement; verb: inflection) 196, 200 marker 196 syncretism 197 systems 186, 193 AgrSP 279 Aidone 33, 204 Airole 285 Alba 285 Alberti 122 Albin 221 Albogno 285 Alboiu 35 Aldon 284 Alexiadou 38, 178, 194 Alghero 2 Aliano 55 Alighieri 57–8, 61, 155, 209, 248, 297 Altamurano 4, 14, 55, 67 Amelia 82 Anagnostopoulou 38, 178, 194 analogical extension 126

analogical levelling 117, 123 analysis (analyticity) 188 typology 9–11 in verbal system 7–19, 137 anaphor binding 26–7 Anderson 128 Andreose 83 Annecy 124 Antronapiana 286 aoristic drift 12, 13 Apicius 10 apophony 123 Aquitanian 124 Aranovich 50, 56 Ardez 180 argument sharing 11 Arielli 38 Aronoff 140 Arsoli 82 ARTESIA 210–11 articles 186, 188 Aski 157 Aub-Büscher 124 augment 122, 128, 129, 130 Auvergnat 125 auxiliary (see also causative; periphrasis: modal; perfect; progressive) 7, 11, 12, 13, 15–16, 17, 34, 50 selection 20, 48–70, 179, 187 antipassive 66 person-driven 53, 64–7 reflexives 50–1, 57 et passim biargumental 51, 53 direct transitive 51–2, 57–8, 59, 62, 62, 64, 65, 68, 69 dyadic 59, 60, 61, 62, 64, 65, 67, 68, 69 indirect intransitive 52–3 indirect transitive 61, 62, 68

334

Index

auxiliary (cont.) indirect unergative 61, 62, 63, 64, 68, 69 monadic 60, 68 reciprocal 59 retroherent 51–2, 58, 59, 60, 62, 63, 64, 65, 66, 68, 69, 70 transitive 51, 52, 53, 61, 62, 64, 65, 66, 67, 68 triple 55–6, 58, 61–2, 67–8, 69 unaccusative 49, 51, 52, 57, 58, 59, 60, 61, 62, 64, 65, 66, 68 unergative 49, 50, 51, 52, 57, 60, 61, 62, 64, 65, 66, 67, 68 Avellinese 14 Baker 44, 156, 181, 185–6, 188–9, 279, 280 Balestrieri 15 Balme de Savoie 124 Bandello 15 Barbi 155 Barbieri 83 Barbosa 178, 194 Bardi 286 bare phrase structure 182 Barese 28 barrier(hood) 235–6, 237, 239 Bartoli 277 Basilicatese 56, 295 Battye 78 Beaulieu 125 Behrens 123, 126 Belletti 34, 35, 279 Belmonte Mezzagno 33 Bembo 60, 61 Benincà 21, 26, 36, 71, 96, 155, 156, 169, 170, 172, 174, 175, 186, 202, 219, 222, 228, 251, 257, 259, 260, 261, 262, 265, 277, 278, 279 Bentley 4, 7, 8, 20, 50, 54, 72, 74, 76, 94, 96, 106, 131 Benucci 219, 221 Benveniste 114, 137 Beretta 122 Berizzi 257, 263, 264

Berruto 2 Bertocci 29, 46 Bertolucci Pizzorusso 83 Berwick 181, 200 Bianchi 29, 274 Bianco 247 Biberauer 185, 190, 191, 200 Biella, A. 122, 220 Bienate 285 binary branching 182 Binding Theory Principle A 26 Principle C 274 Bisceglie 54 Blanchette 234 Blasco Ferrer 103, 107, 205, 215 Bobaljik 44 Boccaccio 57, 60, 61, 133, 139, 144, 155, 207 Börjars 16 Bolognese 284 old 207 Bonet 172 Bonorva 51 Bonvesin de La Riva 11, 13, 85, 288 Borer 178, 181, 182 Borer–Chomsky conjecture 181 Borgogno 158 Borgomanerese 78 Borgorose-Spedino 26 Bormio 121 borrowing 4 Bosa 169 Bossong 99 Bottiglioni 122 Bourova 137 Brach 225, 226 Brambilla Ageno 59, 61 Branciforti 98, 101, 102, 105, 108 Brandi 196 Brandner 297 Branigan 190 Brasseur 125 Brero 220 Bresciano 284, 285, 286

Index Breu 4 Brindisi 205 Brusson 284, 285 Bulgarian 188 Buommattei 61 Buono 12 Burzio 76, 194 Bussoleno 78, 79, 80 Butt 11, 15 Byzantium 244 C(omp) 45, 193, 199, 277 Cairese 2, 9, 10, 78, 219, 220, 222, 223, 224 Cairo Montenotte 2, 6 Calabrese, A. 2, 30, 31, 81, 167, 168 Calabrian 36, 56, 205, 206, 208, 209, 215, 295 early 210, 215 northern 208 southern 31, 38, 208 Calascibetta 204 Calizzano 285 Campanian 295, 296 Campidanese 96, 107, 168, 172 Campitello di Fassa 283 Capracotta 54 Cardinaletti 26, 156, 162, 164, 165, 167, 174, 176, 194 Carpaneto 285 Carrara 177, 178, 197, 198 Carstairs(-McCarthy) 128 cartography 15, 283 Casaccia 19 Cascinagrossa 120, 126 case (see also pronoun: tonic) 186, 187, 188, 202, 217 binary system 202–3 four-way systems 203 ternary 203, 205 Castelfranchi 76 Castellani, A. 83, 137, 140, 142–3, 144, 146, 149 Castellani, F. E. 120, 155 Castelletti 62 Castellinaldo 15 Castelpetroso 81

335

Castelvetro 60 Castrovillarese 55, 56, 67 Catalan 2, 27, 36, 56, 96, 145, 168, 203 Barceloní 172 old 56 Catanzarese 30 Catanzaro 205, 208 Catullus 206 causative 9–11, 16, 17, 21, 99, 111, 113 Cazet 225 Cecchetto 107 Cellini 291 Celtic 187, 188, 191, 196 Cennamo 20, 50, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 84, 89, 90, 93, 94, 157 Central Italian dialects 26, 56, 75, 76, 79, 92, 125, 179, 251, 264 Centranche 208 Cereda 12–13, 17 Ceresole 124 Ceruti 83 Cerva 208 Chamberlain 9 Chapman 16 Chierchia 183, 186 Chinese (see also Mandarin) 291 Chioggia 249, 280 Chomsky 44, 177, 178, 179, 180, 181, 182, 189, 196, 200, 235, 272 Cicagnese 97, 98, 103–4 Cicero 10 Cichocki 125 Cinque 21, 34, 35, 36, 75, 76, 94, 257, 259, 260, 262, 266, 270, 271, 272, 274, 275, 276, 280, 281, 282, 283, 289, 296 Città di Castello 82 Clackson 5, 187 clause union 235 Claut 283 clitic 7 pronouns (see also impersonal reflexives; reflexive) 4, 186, 187, 188, 277 object 218–40 allomorphy 163–7, 169, 170, 174, 175

336

Index

clitic (cont.) bi- vs monomorphemic 167–9, 170, 171, 173–4, 176 climbing 99, 179, 235 clusters 20, 155–76 mirror order 156–61, 169, 170, 176 separability 161–2 split sequence 156, 161–2, 175, 176, 229 doubling 156, 174–5, 176, 180, 196 enclisis with past participle in compound tenses 219–40 proclisis > enclisis 220–4, 234, 237–8 with restructuring infinitive 219, 221, 225 epenthesis 167 interpolation 36, 37–8, 212 locative 283 omission 180 placement 21 hierarchy of clitic positions 228–9, 230–2, 233, 236 proclisis > enclisis with infinitive 225–7 repetition 220–1, 222, 223, 224, 239 silent 172 spurious se 168, 169 suppletion 167–71, 175, 176 thematic vowel 167 subject 20–1, 77, 108, 177–201, 186, 191, 195–9, 233 doubling 195, 260 paradigmatic gaps 196 suppletion 178 syncretism 178, 197 vocalic 178 vs object 195–6, 199–200 tests for clitichood 293 Clitic Left Dislocation 277–8 Codex Cavensis 208 Cole 291 Coleman 137 Colledimacine 31, 54

Collodi 259 Colombo 61 Coluccia 208 Comacchio 285 Comano 285 Comasco 85–6, 177, 178, 197, 198 Comelico 180 complementation 179, 186, 187 complementizer 7, 115, 294 doubling / repetition 6 drop 27, 38–9, 46 dual complementizer system 31–3, 37, 39–44 apparent loss 40–4 Fin (head) 32, 38, 41, 44, 46, 47 Force (head) 32, 39, 41, 42, 44 non-finite 107 null 107 relexicalization 44 compositionality 9, 11–16 Comrie 241, 242 Comtois 125 conditional 3, 20, 30, 37, 137, 147, 148 stem 116–30 Connectionist Theory 244, 245 connectivity 184 conservation 202–5 contact (see also substratum) 21, 188, 190, 244 Contini 83, 155, 216 control obligatory 102, 106, 112, 115 non-obligatory 97, 103 Conzett 242, 245, 249 copular structure 257–8 inverse 273–4 locative 263 Corbett 54, 241 Cordin 26, 29, 196 Corio 284 Cornilescu 35 Corsican 122 Cortona 208 Corvara 294 Cosentino 13, 28, 30, 31, 35, 37, 38, 40–4, 80

Index Cosenza 205, 208 Cozzo 285 Cremona 122 Crocioni 121 Croft 184 Cruschina 36, 42, 114 Cuneo 98, 99, 103–4, 106 Cyrino 36 D(et) 178, 180, 181, 188, 193, 194, 195, 197, 198 D’Achille 48, 243, 248, 255 D’Alessandro 36, 38, 54, 72, 75, 76, 177, 180, 181, 199 D’Onghia 11 D’Ovidio 163–4, 209–10, 216 Da Tos 132 Dahl’s paradox 13 Damonte 29, 31, 38, 46 Danish 282 Dante (see Alighieri) Davies 49 De Blasi 29, 83 definiteness 194 Delogu 83 Delucchi 48 dependent-marking 113, 114 descriptive adequacy 185 Dessì 83 Dessus 247 Dewaele 241 Di Girolamo 212 diachrony 3–5 diagenerational variation 41 diagrammaticity 145, 146, 150, 151 dialects 2–3 diamesic variation 2, 5, 12, 137 diaphasic variation 2, 137, 204 diastratic variation 2, 5, 12, 137, 204, 216 diatopic variation 204 Dignano 285 Dikken, den 273 Dionisotti 60 diphthongization 129 in stressed open syllable 116–17

337

Distributed Morphology 182, 194, 229, 290 Dixon 72 Dizionario Treccani 136 Dowty 50 Dravidian 191 Dresher 190 Dressler 145, 242, 243 Dubinsky 53 Duraffour 124, 125 Durante 248 Dutch 191 Dryer 186 Egerland 94, 165 Emilian 248 Emonds 34 Engadine 19 English 7, 10, 15, 17, 35, 178, 179, 183, 191, 195, 234–5, 242, 257, 258, 263–4, 280, 282, 297, 298, 299 Appalachian 263 Greater Manchester 263 northern (England) 264 epenthesis 143, 286, 287 EPP 38, 183, 194, 196 Erstarrung 204, 217 Esperia 82 Etruscan 187, 188 etymology 21, 205, 206, 216, 217, 280, 298 Ewert 123 existentials 106, 265 explanatory adequacy 179, 201 explanatory power 185 extended accusative 217 Eythórsson 54 Facchetti 122 Fano 208 Faroese 282 Fassano 219, 224, 225–9, 231, 232 Fava, Guido 207 Favaro Lanzetti 122 Favati 155 feature (see also functional head / projection)

338

Index

feature (cont.) copying 198 -driven movement 290 economy 189, 190, 192 fission 198 harmony 234–5 neg 234–5 spreading 235, 236 retraction 236–8 transmission 45 Fedele 35 Fedorenko 69 Fernández González 242 Fernández-Ordóñez 28 Festus 206 Fiamenga 285 Final Over Final Constraint (FOFC) 200 Finarelli 64, 66 finiteness 8, 235–6, 238, 239 Fiorentino, G. 259 Fiume 284, 285 Flgstad 8 Florence 6 Florentine 62, 71, 77, 82, 83, 94, 127, 173, 196, 198, 209, 283, 297 old (early) 49, 57, 59–60, 62, 63, 70, 84, 87–89, 131, 137, 145, 148, 149, 150–1, 155–160, 175, 258, 259, 260, 261–2, 275 focus / focalization 104, 112, 114 contrastive 114 fronting 39, 42–3 Förster 83 Formentin 29, 48, 51, 60, 61–3, 68, 165 Fornara 60 Forni Avoltri 285 Fortunio 60 Francez 106 Francoprovençal 124, 125 Fraser 241 Freedman 257 French 12, 28, 35, 36, 68, 123, 124, 125, 126, 129, 183, 196, 197, 198, 199, 203, 243, 245, 247, 254, 258, 275, 276, 277, 284, 297, 299

Acadian 125 New Brunswick 125 Choletais 125 middle 247 Newfoundland 125 old 13, 57–61, 172, 202, 247 Parisian 125 Québécois 125 spoken 209, 277 Freeze 264 Frigento 55 Friulian 2, 78, 81–2, 122, 145, 280, 283, 292 functional head / projection (see also C; D; T) 181, 190, 194, 227, 228, 233–4, 235, 277, 279, 280, 282, 283, 287, 290 formal features 181–2, 183, 189, 194, 195, 198, 199, 280–1, 282 Functional Hierarchy Hypothesis 219, 227, 229, 231, 232–8, 240 Fusina 122 future 3, 4, 20, 137, 143 stem 116–30 Gainago 177, 178 Galician 96 Galliate 284 Gallo-Italic dialects 2–3, 20, 204 Gallo-Romance 116–30, 155, 158, 202, 203 Gallo Matese 54 Galtellì 169 Gandolfo 220 Garnier 120 Garzonio 36, 279 Gascon (see also Occitan) 202 Gelderen, van 277, 291 gemination (consonantal; see also raddoppiamento (fono)sintattico) 164–5 gender assignment 21, 241–56 compounds 248–9, 253 default 241, 241, 243, 245, 253, 255–6 incorrect 242 lexico-semantic 244–5, 247

Index loans 242, 253 morphological principles 242, 243–4, 245, 247, 255 phonological principles 241–2, 243 semantic principles 241, 242, 245, 247, 255 variation 246 default masculine 242 Latin neuter 243, 244, 249, 256 natural 243, 246 size 246, 249 Genoese 78, 121 old 158, 170 German 7, 191, 247, 291, 297 gerund 14, 15 inflected 8, 98 personal 108 Ghinassi 128 Giacalone Ramat 87 Giammarco 81 Gibson 69 Gigliese 81 Gil 291 Giorgi 26, 27, 30, 46 Giusti 188, 277 Godard 11, 15 Goodall 36 Goosse 124 Grado 284, 286 Græco-Latin loans 244 Grammatica dell’italiano antico 131 Gravina 55 grammaticalization 3, 11, 12, 14, 137, 188, 190, 291 cycle 277, 291 verbal periphrases 3, 7–19 Greek 1, 187, 188, 243, 244 Ancient 262 Greenberg 186 Grevisse 124 Gromo 286 Grosio 294 Guardiano 99 Guarniero 83

Gubbio 208 Guglionesi 32 Guittone 209 Guţu Romalo 205 Haegeman 296 Hall 187 Halle 171, 172, 182, 229, 230, 231 Harris, J. 167, 172, 229, 230, 231 Harris, M. 12, 263, 264 Hart 38 Haspelmath 72 Hastings 5, 69, 83 Haute-Marne 124 Haute-Saône 124 Haverling 13, 137 head-marking 108, 113, 114 Hermann 263, 264 Hermon 291 Holmberg 177, 180, 186, 189, 200 Holmes 244, 245 Horrocks Huang 183, 189, 193 Huber 56, 121 Ibero-Romance 2, 160, 168, 203–4, 210 Icelandic 282 iconicity 145, 150 Iliescu 122, 145 imperative 141, 208, 227, 229–32 imperfect indicative 3, 139, 147 impersonal reflexives 20, 71–95, 157, 162 indefinite subjects 80, 88, 94, 274 Indo-European 9, 188, 290 Ineichen 251 infinitive 3, 7, 117, 125 accusative and infinitive 107 avoidance 30 historic 96 inflected 4, 7, 8, 96, 109, 114, 183 personal 4, 8, 20, 96–115 clausal domain 99–100 distribution 97, 101–4, 106–7, 109–11 textual 108–9

339

340

Index

infinitive (cont.) negation 100 nominative subject 99, 104, 107, 108, 113 temporal reference 100–1 Infl (see also T) 34, 277 inflection (see also verb) parasitic gender 173 number (plural) 172, 175 inheritance (shared) 3, 187 innovation 3, 13, 137, 138, 139, 144, 202–5 Input Generalization 189, 190–1, 192 interfaces 200 interrogative 277 indirect 36, 46, 292 wh-interrogatives 279, 281–2, 289–90, 292 vs wh-relatives 292–5 ‘where’ 279–300 Iordan 205 irrealis modal marking 45–6 Isola 285 Isola Sant’Antonio 284, 285 isomorphism 139, 140–1, 144, 145, 150 Isone 283 Italian 2, 4, 7, 12, 13, 17, 18, 21, 25, 26, 27, 34, 35, 36, 39, 46, 50, 53, 56, 61, 62, 63, 64, 65, 66, 68, 69, 72, 73, 74, 92, 94, 95, 99, 103, 107, 116–18, 124, 125, 126, 127, 128, 131, 136, 137, 142, 145, 146, 147, 149, 155, 156, 157, 158, 159, 162, 163, 166, 171, 172, 173–5, 176, 178, 179, 180, 183, 194, 195, 198, 199, 221, 223, 225, 234, 235, 236, 237, 241, 242, 243, 244, 245, 246, 248, 249, 250, 251, 255, 256, 257–60, 262, 264–5, 275, 276, 278, 280, 284, 286, 292, 296, 297, 298, 299 colloquial 161–2, 258–60, 268, 270, 275, 276, 277, 278 old (early) (see also Florentine: old; Tuscan: old) 10, 21, 58, 125, 131–51, 155–60, 261–2, 277, 278, 296, 297, 298 regional 258 of Sicily 110–11

Renaissance 291, 295 Italo-Romance 2, 3, 6, 8, 10, 12, 13, 20, 21, 49, 63, 69, 116–30, 167, 168, 175, 177, 185, 186–8, 202 medieval (early) 169, 203, 258, 259 Ittiri 169 Japanese 191, 192–3, 194, 291 Jelinek 195 Jeppesen Kragh 28 Jespersen’s cycle 277 Jezek 58, 59 Jones, M. A. 10, 13, 18, 103, 106, 107, 168, 173 Joos 185 Judaeo-Italian 212 Judaeo-Spanish 229 Kahane, H. 244 Kahane, R. 244 Kany 96 Katz 279, 281–2, 289, 299 Kayne 35, 54, 55, 69, 156, 161, 164, 167–8, 171, 172, 176, 177, 181, 183–4, 185, 186, 192, 196, 200, 224, 227, 229, 230, 231, 232, 233, 258, 264, 275, 280, 296, 299, 300 Kayne’s Generalization 195–6 Keller 124 Koopman 280 Korean 191 Krämer 128 Kratschmer 58 Kroch 223 Kuroda 1988 La Fauci 51, 53, 56, 106, 109, 110, 137 La Thuile 124 Ladin 219 Lahiri 11, 15 Laino Borgo 33 language acquisition 190 LaPolla 99, 100, 113, 114 Larson 143 late insertion 182

Index Latin 1, 2, 3, 5–6, 8, 9–10, 12, 13, 14, 16, 20, 21, 95, 96, 107, 108, 109, 121, 123, 136, 137, 138, 139, 141, 142, 143, 144, 148, 164, 167, 168, 169, 187, 188, 200, 203, 206, 207, 208, 212, 213, 241, 243, 244, 248, 249, 256, 262, 283, 291 Classical 5–6, 187, 204, 206, 215 late 93–4, 95, 187, 216, 283 Pompei inscriptions 206 vulgar 6, 187, 188 Lazio 25 Leccese 53, 56, 63 Ledgeway 3, 5, 7, 10, 14, 19–20, 25, 26, 27, 29, 30, 31, 32, 35, 36, 37, 40, 42, 53, 71, 89, 96, 97, 98, 101, 106, 107, 108, 109, 110, 111, 113, 120, 129, 149, 177, 183, 187, 188, 196, 208, 227, 229, 257, 275 left-adjunction 161, 176 Legendre 54 Lemaire 247 Lengadocien 25 Leone 30, 109, 111, 113 Lepschy, A. L. xii–xiii, 76 Lepschy, G. xii–xiii, 72, 76, 126, 250 Leschaux 124 Lexicon (the) 182, 245, 255 Librandi 215 Lichtenhahn 283, 288, 291, 292, 294, 295, 297 light bi-clausality 233–4 Lightfoot 190 Ligurian 2, 78, 156, 257 northwestern 158 old 159, 169–70, 262, 276 Limentani 155 Limousin 125 Lindsström 121 Livigno 121, 285 Locatelli 121, 122 Lodge 125 Logudorese 20, 51–2, 71, 83, 168, 172, 173 old 71, 83, 90–1, 93, 283 Lois 35, 36 Lombard 2, 4, 14, 71, 83, 84, 122, 129, 209, 248, 293–4

341

Alpine dialects 283, 294 old 83, 85–7 Lombardi 20, 25, 30, 31, 35, 36, 37 Longa 121 Longhino 69 Loporcaro 4, 8, 12, 14, 16, 17, 20, 29, 48, 49, 50, 51, 53, 54, 55, 56, 57, 63, 69, 98, 106, 109, 167, 168, 203 Lorrain 125 Lucanian (see Basilicatese) Luongo 99, 108, 109 Luras 169 Macedonian 188 Maceratese 55, 56, 67 Maddalon 242 Maggi 15 Magri 122 Maiden xiii, 3, 20, 116, 122, 125, 130, 138, 148, 202, 241, 242, 246 Malesco 285, 286 Malkiel 204 Mancarella 29 Mandarin Chinese 179, 183 Manni 139, 140, 146, 148 Manzini 19, 26, 31, 54, 64, 65, 68, 76, 77, 78, 79, 80, 81, 82, 83, 92, 93, 158, 159, 168, 169, 172, 177, 178, 179, 180, 181, 183, 185, 186, 188, 194, 197, 198, 199, 228, 233 Manzoni 61 Marantz 171, 182 Maratea 26 Marazzini 60 Marchal 247 Marche, Le 25 Marchigiano 36, 82, 208 Markun 122 Marne 124 Marthod 124 Martial 206 Maruggio 39 Mascalucia 204 Maschi 131, 132

342

Index

Match 182, 192 Mathon 284, 285 Mayerthaler 139, 145 McArthur 29 Melander 164 Melicucco 32 Melillo 81 Menoni 59 Menton 26 Merci 83, 215 Merge 181–2, 200 Messinese 31, 32, 38 Meszler 138 metaphony 120, 129, 208 Meul 129 Meurthe-et-Moselle 124 Meuse 124 Meyer-Lübke 129, 141, 202, 206–7, 221 Milan 6 Milanese 2, 11, 13, 15, 80, 85–6, 128, 277 Milsark 106 Minervini 212 Minervino Murge 55 Minimalist Program 290 Mirror Principle 279 Mithun 50 Mocciaro 109, 113 Molfetta 55 Molisano 81 Monachesi 36 Mondini 122 Monno 293–4 Montagny 124 Montanaro 284, 285 Monteil 262 Monteparano 55 Monteroduni 33, 54 Morcone 55 Moro 106, 273 morphome 20, 130–1 N-pattern 130 morphological change phonological causation 125–6, 128–9 semantic causation 125

morphosyntactic change 4–5, 21, 278 Motapanyane 35 Motolese 60 Mourin 145 Müller 193–4 multi-attachment 50 Munaro 21, 279, 281, 293 Mussafia 128 Mussomelese 42–3 N-movement 277, 280 Namur 124 nanosyntax 282 Naples 6 Napoli, D. J. 76 Natural Morphology 139, 145 Navone 121 Neapolitan 2, 7, 8, 10, 14, 20, 28, 29, 30, 36, 42–3, 71, 79, 80, 82, 83, 129 Campi Flegrei 209 old (early) 47, 83, 89–90, 91, 92, 96, 98, 99, 120, 129, 149, 183 Neeleman 192 negation 179, 186, 277 clitic interpolation 212 mica in relative clauses 266 negative concord 234–5, 240 negative licensing 27 preverbal negators 231 Nencioni 137, 146, 150 Nesset 246 Nichols 108 Niçois 124 Nicoli 80, 122 Niculescu 205 Niederländer 124 Nievo 94 Nominal Mapping Parameter 186 non-agentivity 50 non-configurationality 182 Nònese 294 Noordhof 259, 275 Northern Italian dialects 20–1, 77, 79, 92, 125, 128, 158, 160, 177, 178, 179, 183, 188,

Index 191, 195, 202, 231, 250, 260, 275, 277, 280, 282, 283 northwestern 277 medieval (early) 259, 261 Norwegian 282 Novi Ligure 283 null argument 200 hierarchy 192–9 null subject parameter 177, 178–80, 197–9 radical/discourse pro-drop 192–3, 194, 195 Numeration 182 Nuorese 96, 107 Nye 279 Nyrop 123, 130 Occitan (see also Gascon) 26, 96, 103–4, 124, 125, 202, 227 old 202 southeastern 180 Oliva 29 Olivetta San Michele 158 Oliviéri 26 Oniga 107 Onillon 125 Optimality Theory 247 Orlandi 121 Ortalli 83 Oscan 188 Osco 284, 285 Ouali 44 overabundance 136–7 OVI 117, 133, 210 Pace 56 Paciaroni 51, 55 Padria 169 Paduan 2, 11, 14, 71, 77, 79–80, 81–2, 249, 251, 260 old 69, 92 Padula 54 Pagani 122 Pagano 210–11, 212 Palmanova 81–2

343

Palmer 187 Palumbo 98, 100, 101, 102, 103, 104, 105, 106, 108, 109 Paoli 180, 196, 227 Papanti 66 paradigmaticity 9, 16–19 paragoge 143, 209 parameter hierarchies 189–95 macroparameters 20–1, 181, 185–91 markedness 189–91 macro- vs microparameters 177, 180 macro- and microparametric variation 181–9 mesoparameters 191 microparameters 177–201 PF-realization 177, 181, 193 Pardeshi 9 Parisi 76 Parmigiano 284 Parodi 155, 164, 170 Parry xii–xiii, xxii, xxiv, 2, 3, 4, 5, 10, 13, 21, 25, 41, 69, 71, 78, 83, 96, 116, 155, 156, 158, 177, 202, 218–20, 221, 223, 231, 239, 257, 259, 261–2, 277, 279 participle 7 inflected 98 past 3, 12, 14, 15 agreement 51, 73, 179, 199 present 14 passato prossimo (see also perfect) 8, 12 passato remoto (see also perfect) 8, 9, 12, 13 passé surcomposé (see also perfect) 12–13, 17 passive 16, 20, 71–95 by-phrase 74, 87, 89 imperfective 8 periphrasis 16–17 with andare 18 with venire 18 synthetic 188 Pavese old 85–7 Pedretti 177 Peirce 145

344

Index

Pejo 284 Pelaez 29 Pentone 208 Pérez Saldanya 56 perfect (see also analysis; auxiliary selection) 3, 12–13, 16 aoristic drift 12, 13 current relevance 12 strong forms 146–7 structure of compound tenses 233–4 weak forms 139–40, 142–2, 147, 149 Perlmutter 49, 50, 51, 70, 231, 233 periphrasis (see also analysis; progressive) continuative 15 durative 15 iterative 15 modal 17 passive 16–17 verbal 3 Perna 257, 262 Pernicone 59, 60 Péronnet 125 Perrone 29 person endings (see also verb: inflection) 131–51 Pescarini 4, 20, 21, 71, 155, 167, 172, 229 Pescia 48, 64, 68 Pesetsky 279 Petrarch 139 Petrocchi 83, 155 PF 31, 171, 200, 290 Spell Out 193 phase edge 239–40, 296 Phase Impenetrability Condition (PIC) 181 Pianesi 27, 46 Picernese 49, 63–4, 68, 70 Piedmontese 2, 4, 14, 21, 78, 177, 218–40, 257 old 159, 169–70, 236, 237 Pietraporzio 284 Pinkster 137 Pinocchio 259 Piras 83 Pirrelli 131, 136 Pisano 13

Pitré 97, 110, 111, 112, 113, 114 Placiti Cassinesi 5 Plautus 188 plural formation in nouns 21, 202, 241–56 Poggio Imperiale 54 Pokorny 290 Poletto 13, 17, 21, 27, 36, 46, 172, 175, 178, 186, 233, 277, 279, 290, 293 Politzer 148 Pollock 34, 35, 279, 290 polymorphism 202, 204, 205 Pons 227 Pope 123, 124, 126, 129 Popoli 54 Porro 155 Porta 63 Portuguese 14, 56, 96, 145, 277 Brazilian 36, 180 old 56 Porzio Gernia 6 Poschiavino 284, 286 Postal 279, 281–2, 289, 299 Pountain 97 poverty of stimulus 179, 201 Prader Schucany 121 Pramollo 284 Predicate Union 49 Preposition PP internal structure 280 locative 280, 281, 287, 289–92 silent 258–9 stranding 258 prepositional accusative 99, 196 present indicative 3, 141, 142, 147 Price xiii, 123, 130 primary linguistic data (PLD) 179, 189, 190 Princivalli 83 pro-drop (see null argument) Procidano 140 progressive 4, 8, 13–14, 15, 16, 17 pronoun (see also clitic) 7, 20–1, 277 tonic

Index case distinctions 202 allomorphy 202, 203 oblique 21, 202–17 1sg/2sg 205–10 distribution according to clausal position 214–15 subject 113–14, 149, 180 expletive 84, 85, 86 omission 26 tonic > clitic 165 Proto(-Italo)-Romance 187 Pugliese northern 81 Putzu 71 Quaglio 155 Quer 26, 27 raddoppiamento (fono)sintattico (see also gemination) 39, 165 Ræto-Romance 3, 19, 180, 186, 202, 225, 277, 280, 294 Rak 29 Ranrupt 124 Rasom 225, 226, 227, 228 Realdo 121, 129 reanalysis 128–9, 143, 166 reflexive (see also auxiliary: selection; impersonal reflexive; passive) 168, 223–4, 226, 227, 228, 230–1, 232, 236–8, 240 impersonal/passive si/se 71–95, 157, 162, 240 agreement 73–4, 77–8, 87 animacy 80, 84, 87, 89 anticausative 77, 78, 79, 81 argument status 76 by-phrase 74, 87, 89, 91 generic vs indeterminate/existential interpretation 75–6, 82, 85, 86, 89, 94 passive vs impersonal 77–9, 84 reflexive ci 74–5, 76, 92–4 replacing 1pl 75, 94

345

tense/aspect contraints 81–2 word order 72–3, 77–9, 84, 86, 87, 89, 90 Reinhart 279 Relational Grammar 49, 52 relative clauses 21 compatibility with interrogative illocutionary force 268, 272 modals 267–8, 271–2 non-realis mood 266–7, 271 presuppositional negation 266, 268–9, 271 resumptive pronouns 270, 272 retention of internal Head 270–1, 272 stacking 269, 272 contact relative 263–4 headless 264–5 infinitival 265, 271, 276 introduced by che ‘that’ with gap 259 with resumptive clitic 258–9, 260, 261–2, 270, 272, 274 introduced by wh-phrase 258, 259 with copying clitic 259–60, 261–2, 275 kind-defining 257–8, 260, 261, 262–78 agreement with subject of copular sentence 275–6 matching derivation 274 post- and pre-copular position 273–4 post-copular position 275–6 raising derivation 274 non-restrictive 258, 260, 261, 262–3, 266–72, 274, 275 ordering 268–9, 272 restrictive 258, 260, 261, 262, 263, 264, 266–72, 274, 275 Remberger 17 Renzi 131, 155, 178 Repetti 167 restructuring 235 Retaro 209 rhizotony 128 Ricca 14, 71 Rice 246–7

346

Index

Richards 191 Rinaldi 97, 99, 100, 107, 108, 109 Rindler Schjerve 58 Ritmo Cassinese 207, 216 Rivolta D’Adda 285 Rix 188 Rizzi 39, 99, 161, 177, 178, 194, 195, 196 Roaf xii Roberts 4, 20–1, 31, 54, 177, 180, 181, 186, 187, 188, 189, 190, 191, 193, 194, 196, 199, 200 Robustelli 10, 11 Rocca Imperiale 168–9, 172 Roccasicura 54 Rochemolles 284, 285 Rohlfs 2, 14, 15, 25, 29, 30, 76, 81, 93, 141, 148, 165, 167, 205, 208, 209, 295 Romanesco old (early) 49, 61–3, 64, 65, 66, 68, 69, 70, 208 Romanian 10, 28, 35, 36, 37, 45, 46, 96, 144–5, 202, 205–7, 215, 233, 275, 297 Romansh 56, 128 Ronjat 124 root incorporation 156, 171–3, 176 Rosen 11, 49, 50, 51 Rosén 6 Rossi 155 Roussou 31, 188, 190 Rovani 94 Rovigo 249 Ruata 285 Ruda 284, 285 Rumilly 124 Russi 157 Russian 7 Ruvo di Puglia 33, 54 Ruzante 11, 14 Sabatini 202, 207 Sabellian 187, 188 Sacchetti 59–60 Saito 182, 193, 194

Salentino 2, 11, 29, 32, 38, 41, 46, 81, 205, 206, 208, 209, 215 old 210, 215 Salve 205 Salvi 74, 75, 76, 88, 94, 131, 155, 157, 172 Salvioni 15, 83, 84, 128, 221 Samolaco 283, 286 Samu 138 San Benedetto del Tronto 54 San Fratello 204 San Giorgio del Sannio 55 San Michele al Tagliamento 292–3 San Vigilio di Marebbe 285 Sanga 122 Sansò 87 Santo Stino di Livenza 294 Sardinian 2, 3, 6, 10, 13, 18, 53, 56, 63, 64, 66, 69, 71, 103, 106, 107, 159, 169, 170, 171, 172–3, 205–6, 296 northern 92 old (early) 6, 215 Sarroch 168 Sassello 285 Sassinoro 54 Satriano 208 Sauerland 274 Săvescu 233 Savigny 124 Savj-Lopez 98 Savoia 19, 31, 54, 64, 65, 68, 76, 77, 78, 79, 80, 81, 82, 83, 92, 93, 158, 159, 168, 169, 172, 177, 178, 179, 180, 181, 183, 185, 186, 188, 194, 197, 198, 199, 228, 233 Savoie 124 Savoyard 124 Scandinavian 282 Schaden 12 Schächter xii Schiaffini 83, 137, 146 Schifano 35 Schuchardt 216 Schwan 123, 126 Scotti 122

Index Scuol 19 Schwegler 7 Scorretti 27 Secinaro 54 secondary stress 126, 127, 129, 165, 166 Segui 244, 245 Selkirk 165 Semogo 283, 294 Sermoneta 212 Sgrilli 11, 215 Sheehan 192 Shibatani 9 Sicily 25, 32 Sicilian 2, 7, 8, 20, 21, 28, 30, 56, 80, 109–15, 202–17 old (early) 56, 97–109, 210–15 hagiographic texts 213 Sicilian School of Poetry 215 Sienese 2, 82 Simbario 208 Siniscola 169 Skytte 99, 107 Slavonic 188, 250 small clause 10 Smith, J.C. 3, 20, 116, 125 sociolinguistic variation 2, 137 Soglio 285 Sommaino 286 Sondrio 121 Sorace 50, 77 Sornicola 7, 21, 99, 104, 107, 202, 203, 208, 209, 210, 217 Southern Italian dialects 4, 19–20, 21, 25–47, 56, 63, 75, 76, 79, 92, 120, 168, 179, 188, 206, 208, 256, 264, 295 extreme south 30, 32, 33–4, 38, 45, 215 upper south 32, 33–4, 39, 45 Spanish 9, 13, 27, 35, 36, 53, 56, 63, 96, 145, 172, 174, 217, 219, 224, 227, 229–32, 244, 249 Colombian 96 central-northern peninsular varieties 28 dialects 229–32

347

old 203, 217 Castilian 56, 63 Rioplatense 8–9 South American 172 Spencer 12 Sperlinga 204 Sportiche 258 Squartini 13 Starke 165, 282 Stefanini 82 Stilo 208 Strong Minimalist Thesis (SMT) 200 Stussi 6, 83, 84 Subiaco 121–2 subject 113–14, 178, 183, 194 null (see also null argument: null subject parameter) 143, 194, 197 position 38, 46, 85, 107, 108, 113, 183, 194, 196 of predication 273 subjecthood 107–8, 113–14 subjunctive 103, 112, 125, 140–2, 143–4, 147, 148, 262, 263, 266–7 contrast with indicative 25–6 syntactic repercussions 26–7, 180 distribution in modern dialects of south 29–30 jussive 30 modal attraction 30 morphological marking 28–31, 37 loss 19–20, 25–47 imperfect subjunctive in French 28 optative 40–1 preservation in Salentino 46 replacement by indicative 25, 28, 30, 40–1 volitional 30 substratum 187 suppletion 121, 128, 130, 167–71 Svenonius 289 Swedish 282 synthesis (syntheticity; see also analysis) in verbal system 7–19, 187 Szabó 11–12

348

Index

Szendrő 192 Szmrecsanyi 7 T (‘tense’) 34, 36, 38, 39, 45, 178, 183, 194, 195, 196, 197, 198, 199, 235 Tadmor 291 Taggia 285 Taglio di Po 285 Tasmowski 137 Taylor 38 Tekavčić 32, 141, 148 telicity 50 tenere + past participle 4, 14–16 tense absolute / deictic 12 relative 12 Teolo 284 textual traditions 5–7 thematic vowel (TV; see also clitic: pronouns: object: thematic vowel) 131–2, 133, 143 Thielmann 137 Thornton 117, 136, 145, 151, 242, 243, 248, 255 Tobler 84 Todi 208 Tomioka 193 Tonezza 286 topic(alization) 39, 42–3, 202, 203 Clitic Left Dislocation 277–8 Hanging Topic 278 topic-worth 113 TopP 272 Toppin 15 Torinese 71, 78, 219, 220, 222, 223, 224, 239 Tortora 4, 21, 35, 36, 155, 218, 222, 227, 228, 233, 234, 235, 236, 240, 257, 263 Toscano 124 Toso 121 TP (‘tense phrase’) 279 Tramonti di Sotto 285 Travis 189 Trentino 55, 67, 80–1

Trumper 21, 241, 250 Tuscan 26, 75, 81, 94, 95, 108, 109, 117–18, 244, 254, 256 old 59, 88, 117–19, 129, 145, 146, 209, 241, 244, 248, 251, 252 southern 248 Tuttle 221 Ugolini 62 Ulleland 276 Umbrian 25, 75, 82, 188, 208 unaccuative hypothesis 49–50, 70 underspecification 183 uniform encoding 139, 145, 151 uniformitarianism 184 univerbation 164 Universal Grammar (UG) 177, 179, 180, 181, 182, 183, 184, 185, 189, 190, 281 (light) v 200 Väänänen 206 Vai 277 Valdese 227 Valesio 137 Valfura 280 van Hout 50 Van Valin 50, 99, 100, 107, 113, 114 Vanelli 8, 20, 74, 81, 131, 140, 141, 145, 151, 178, 202, 251 Vangsnes 282 Varro 206 Vastogirardi 32, 54 Vaux 124, 125 Vecchio 31, 51 Venetan 241–56 old 207 Venetian 2, 6, 71, 77, 83, 242, 244, 248, 254, 255, 284, 285, 286, 294 old (early) 79, 83, 84–5, 91, 93, 251, 254, 256 Venetic 187, 188 Veneto 21 northern 56 Venice 244

Index verb (see also auxiliary) allomorphy 117, 120, 124, 128, 130, 136, 137, 139, 140, 143, 149, 150 allotropy 136–137, 141, 146, 149, 150, 151 inflection (see also agreement) 20, 131–51 1sg 139–41 2sg 141–2 3sg 142–4 3pl 145–8, 229–32 imperative 208 light 7, 10, 11, 15–16 main 7, 15 movement 34–9, 46, 186, 239–40, 277 finite lexical verb 36 non-finite 222, 227, 228, 236 realis vs irrealis clauses 37–9, 44 polymorphy 137, 146 system 7, 16–17 V-to-T (V-to-Infl) 178, 180, 183, 277, 279, 292 vs XP movement 228–9 verb second (V2) 46, 277 Vernole 205 Véronique 241 Verrier 125 Viareggino 82 Vicenza 249 Vignuzzi 63 Vigolo 250 Viguzzolo 55 Vincent 1, 3, 5, 6, 7, 11, 12, 13, 16, 32, 59, 69, 71, 83, 96, 97, 98, 99, 183, 263, 264 Virdis 71, 205 Virgil 206 Vitale 128

Viticuso 54 Vosges 124 Wallace 188 Wallon 124 WALS (The World Atlas of Language Structures) 186 Wanner 27, 159, 221 Washington 249 Watanabe 193 Watumull 182 Wehr 85, 87, 88 Welsh 183 Wexler 26 wh-phrases (see interrogatives; relative clauses) Wheeler 103–4 Wierzbicka 114 Wilhelm 11 Willis 96 word order 187, 188, 189, 190, 191–2 head-finality/-initiality 191–2 OV > VO 187, 188 OV/VO and OP/PO correlations 186 SOV 188 SVO 186 VSO 183 Zagarolese 55, 56, 67, 69 Zagona 35, 36 Zahner 122 Zamboni 250 Zanuttini 231 Ziccardi 64, 65, 66 Zoagli 284–6

349

O X F O R D S T U D I E S I N D I A C H R O N I C A ND H I S T O R I C A L L I N G U I S T I CS general editors Adam Ledgeway and Ian Roberts, University of Cambridge advisory editors Cynthia Allen, Australian National University; Ricardo Bermúdez-Otero, University of Manchester; Theresa Biberauer, University of Cambridge; Charlotte Galves, University of Campinas; Geoff Horrocks, University of Cambridge; Paul Kiparsky, Stanford University; Anthony Kroch, University of Pennsylvania; David Lightfoot, Georgetown University; Giuseppe Longobardi, University of York; David Willis, University of Cambridge 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 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

in preparation Syntax over Time Lexical, Morphological, and Information-Structural Interactions Edited by Theresa Biberauer and George Walkden The History of Negation in Low German Anne Breitbarth Pragmatic Markers from Latin to the Romance Languages Edited by Chiara Ghezzi and Piera Molinelli Nominal Expressions and Language Change From Early Latin to Modern Romance Giuliana Giusti The Historical Dialectology of Arabic: Linguistic and Sociolinguistic Approaches Edited by Clive Holes A Study in Grammatical Change The Modern Greek Weak Subject Pronoun τος and its Implications for Language Change and Structure Brian D. Joseph Gender from Latin to Romance Michele Loporcaro Vowel Quantity from Latin to Romance Michele Loporcaro The Syntax and Semantics of Vedic Particles John J. Lowe Syntactic Change and Stability Joel Wallenberg The History of Negation in the Languages of Europe and the Mediterranean Volume II: Patterns and Processes Edited by David Willis, Christopher Lucas, and Anne Breitbarth