Variation and Change in Gallo-Romance Grammar (Oxford Studies in Diachronic and Historical Linguistics) 9780198840176, 0198840179
This volume offers a wide-range of case studies on variation and change in the sub-family of the Romance languages that
136 26 10MB
English Pages 496 [485] Year 2020
Cover
Variation and Change in Gallo-Romance Grammar
Copyright
Dedication
Contents
Series preface
Abbreviations and symbols for this volume
The contributors
1: Introduction
1.1 Background to the volume
1.2 Smith’s contribution to linguistics and the chapters in the volume
PART I: SENTENCE STRUCTURE
2: Old Gallo-Romance, periodization, and the left periphery
2.1 Background
2.1.1 Linguistic change in old French and old Occitan
2.1.2 Classification and the North / South split
2.1.3 Periodization
2.1.4 Previous research on old Occitan
2.2 Linguistic change in French
2.2.1 Background
2.2.2 V2 and inversion
2.2.3 Change in the left periphery
2.2.4 Change in the syntax of SI
2.2.5 Change in the subject system
2.3 Occitan and early Gallo-Romance syntax
2.3.1 Common syntactic properties
2.3.2 Occitan and the late medieval split
2.3.2.1 Verb placement
2.3.2.2 Discourse-pragmatics and the prefield
2.3.2.3 Verb-third
2.3.2.4 Null arguments
2.4 Conclusion
2.4.1 Empirical summary
2.4.2 Consequences
Primary sources
3: Resumptive structures in a Gallo-Romance perspective
3.1 Introduction
3.2 Theoretical and methodical backdrop
3.2.1 Sources
3.2.2 Verb Second
3.2.3 The lexeme SI
3.2.4 Specialized and generalized resumption
3.3 Resumption in old French
3.3.1 After fronted adverbial constituents
3.3.2 After fronted arguments
3.4 Resumption in old Occitan
3.5 Discourse function of the fronted argument
3.5.1 Old Occitan
3.6 A structural analysis of old French
3.6.1 Contexts where resumption is ruled out
3.6.1.1 Stylistic fronting
3.6.1.2 Wh-words
3.6.1.3 Fronted VP elements
3.6.2 Maximal or minimal category?
3.6.3 Specialized or generalized resumptive?
3.6.4 Embedded contexts
3.6.4.1 Embedded [SUBJECT + SI]
3.6.4.2 Doubly embedded clauses
3.6.4.3 Left dislocation
3.6.5 Summing up
3.6.6 Old Occitan
3.7 Conclusion
Texts cited
Old French
Old Occitan
4: Variation in the Gallo-Romance left periphery: V2, complementizers, and the Gascon enunciative system
4.1 Introduction
4.2. Strong / weak C dimension
4.2.1 Medieval Gallo-Romance: strong C
4.2.2 Modern Gallo-Romance: weak C
4.3 The Gascon enunciative system
4.3.1 Introduction
4.3.2 Strong C: Merge vs Move
4.3.3 Sentential negation
4.3.4 Root clauses
4.3.5 Embedded clauses
4.4 Summary and conclusion
5: Dialectological evidence for a predicate focus analysis of Gascon que
5.1 Introduction
5.2 The controversial nature and origin of Gascon ‘que’
5.2.1 The geographic extension of Gascon ‘que’
5.2.2 Interpretations of the functional value of Gascon ‘que’
5.3 Gascon ‘que’ and focalization
5.4 The dialect of Les Esseintes
5.4.1 Introduction
5.4.2 The ‘enunciative’ es ke in the dialect of Les Esseintes
5.4.3 Focus marking and focus marking suspension
5.4.4 A contact-induced phenomenon?
5.5 Conclusion
6: Postverbal negators in Gallo-Romance: The view from Old Occitan
6.1 Introduction
6.2 Pas
6.2.1 Introduction
6.2.2 Pas through the centuries
6.2.3 Pas in negative polar questions
6.3 Ges
6.3.1 Introduction
6.3.2 Semantics and pragmatics of ges
6.3.3 Syntax of ges
6.3.4 Discussion
6.4 Conclusions
7: The loss of clitic climbing in French: A Gallo-Romance perspective
7.1 Introduction
7.1.1 Overview
7.1.2 Clitic climbing in French
7.1.3 Clitic climbing in (varieties of) Occitan
7.1.4 Clitic climbing in other Romance varieties
7.1.5 The interest of clitic climbing
7.2 Corpus and methodology
7.3 Results and discussion
7.3.1 Frequency of clitic climbing
7.3.2 The process of change
7.3.2.1 Matrix verbs
7.3.2.2 Complexity of the environment
7.3.3 Discussion
7.4 Conclusion
PART II: THE VERB COMPLEX
8: Motivating the North–South continuum: Evidence from the perfects of Gallo-Romance
8.1 Introduction
8.2 The North–South Continuum
8.2.1 Zamboni
8.2.2 Ledgeway
8.3 Discussion of Ledgeway’s characterization
8.4 Motivating the North–South Continuum: a stratified model of the development of the perfect in Europe
8.4.1 Why should the Carolingian realm be the locale of theinnovative expansion of the HAVE / BE dichotomy?
8.4.2 HAVE perfects in Carolingian Latin
8.4.3 Deponents and perfects in Carolingian Latin
8.5 Evidence from southern Gallo-Romance varieties
8.5.1 Introduction
8.5.2 Occitan
8.5.3 Catalan
8.6 Evidence from northern oïl varieties
8.6.1 Introduction
8.6.2 Northeastern oïl varieties
8.6.3 Northwestern oïl varieties
8.6.4 Comparison of northeastern and northwestern oïl varieties
8.6.4.1 Introduction
8.6.4.2 Exclusive use of HAVE auxiliary, especially in thenortheastern oïl varieties
8.6.4.3 Non-agreement of past participles
8.6.4.4 Persistence of the preterite especially in the northwestern oïl varieties
8.7 Conclusions
9: Active-middle alignment and the aoristic drift: The North–South divide in the Romània on evidence from northern Gallo-Romance
9.1 Introduction
9.2 The North–South divide in the Romània
9.3 The case of northern Gallo-Romance
9.4 Making sense of northern Gallo-Romance
9.5 The North–South classification reconsidered
9.6 Conclusion
10: A comparative analysis of French auxiliation, with new evidence from Montréal
10.1 Introduction
10.2 Theoretical approaches to split intransitivity
10.2.1 The interface of syntax and semantics: the Auxiliary Selection Hierarchy
10.2.2 Syntactic approach to split intransitivity
10.2.2.1 French as an outlier among Romance auxiliation systems?
10.2.2.2 Argument structure and unaccusativity
10.2.3 Issues with theoretical approaches and the problem of ‘free’ variation
10.3 Recorded variation in French auxiliary selection
10.3.1 Variationist sociolinguistic studies on North American French auxiliary alternation
10.3.2 Factors of influence
10.3.2.1 Sociolinguistic factors
10.3.2.2 Linguistic factors
10.4 Evidence from new Montréal French data
10.4.1 Sociolinguistic corpus of Rea (2016)
10.4.2 The crowdsourcing platform Les français de nos régions (2017)
10.5 Discussion
11: Présent inclusif and passé composé à valeur de présent accompli in modern French and Occitan
11.1 Introduction
11.2 Theoretical considerations
11.3 Présent inclusif
11.4 Passé composé à valeur de présent accompli
11.5 Passé composé with future projection
11.6 Conclusions
12: Future temporal reference in French and Gascon: Aller / anar + infinitive periphrasis and structural transfer in the bilingual grammar
12.1 Introduction
12.2 Research context: French and Gascon in Béarn
12.3 Theoretical framework: the bilingual grammar
12.4 Linguistic background: future temporal reference
12.4.1 Historical evolution of future temporal reference variants
12.4.2 Circumscription of the variable context
12.4.3 Previous studies of future temporal reference
12.5 Methodology
12.5.1 Data collection
12.5.2 Coding the dependent variable
12.5.3 Independent variables
12.5.3.1 Sentential polarity
12.5.3.2 Temporal distance
12.5.3.3 Imminence
12.5.3.4 Verb stem regularity
12.5.3.5 Certainty
12.5.3.6 Adverbial specification
12.5.3.7 Grammatical person
12.5.3.8 Subject type
12.5.3.9 Contingency
12.5.3.10 Presence of ‘quand’
12.5.3.11 Extralinguistic factors
12.5.4 Statistical analysis
12.6 Results
12.6.1 Frequency distributions
12.6.2 Modelling future temporal reference in French
12.6.3 Modelling future temporal reference in Gascon
12.7 Discussion: structural transfer in the bilingual grammar
12.8 Conclusion
13: Mainland and insular Norman: Pronoun sharing and pronoun sparing
13.1 Introduction
13.2 Methodology
13.3 The pronouns of Norman
13.4 Results
13.4.1 Pronoun sharing with French
13.4.1.1 Conjunctive pronouns
13.4.1.1.1 Conjunctive third-person singular feminine subject pronouns
13.4.1.1.2 Conjunctive first person plural subject pronouns
13.4.1.1.3 Conjunctive third person plural subject pronouns
13.4.1.1.4 Conjunctive impersonal pronouns
13.4.1.2 Disjunctive pronouns
13.4.1.3 Reflexive pronouns
13.4.2 Pronoun sharing with English
13.4.2.1 Conjunctive pronouns
13.4.2.1.1 Gender-marking
13.4.2.1.2 Pronoun calquing
13.4.2.2 Pronominal reflexive verbs
13.5 Conclusion
PART III: WORD STRUCTURE
14: On the origins of French and Occitan
14.1 Introduction
14.1.1 The problem
14.1.2 Identifying a vernacular language
14.1.3 The nature of the evidence
14.1.4 The books we have for early texts
14.2 Comparing the vernacular forms
14.2.1 Writing the vernacular alongside Latin
14.2.2 Definite articles in the earliest texts
14.2.3 Unstressed pronouns in the earliest texts
14.2.4 The preposition ab in the earliest texts
14.2.5 Third person plural endings from unstressed Latin vowels
14.2.6 Some inconsistent usages
14.2.7 Possible inferences
14.3 Writing the vernacular
14.3.1 The Augsburg Passion
14.3.2 The circumstances of writing
14.4 Conclusion
List of primary sources
15: Appositive compounds in dialectal and sociolinguistic varieties of French
15.1 Introduction
15.2 Appositive compounds
15.2.1 Introduction
15.2.2 Main characteristics
15.2.3 Appositive compounds in French
15.3 Appositive compounds in Gallo-Romance dialects
15.3.1 Introduction
15.3.2 Sources and general trends
15.3.3 Appositive compounds: hybrids
15.3.4 Appositive compounds: additional compounds
15.3.5 Appositive compounds: type-of compounds
15.3.6 Appositive compounds: the role of metaphors
15.3.7 The sequence of the component nouns
15.3.8 Preliminary conclusions
15.4 Le poilu
15.4.1 Introduction
15.4.2 Appositive compounds in poilu
15.4.3 Summary of findings
15.5 Discussion of the findings and conclusions
16: Complex versus compound prepositions: Evidence from Gallo-Romance
16.1 Introduction
16.2 Prepositions: some preliminary considerations
16.3 Origins
16.4 Diachronic stages
16.5 The typological context
16.6 Modelling P and PP
16.7 Prefixes and PPs
16.8 Conclusions
17: Syncretism and metamorphomes in northern Occitan (Lemosin) varieties
17.1 Introduction
17.2 Syncretism patterns in the north Lemosin variety of Gartempe (Creuse)
17.3 The origin of the syncretism patterns
17.3.1 Introduction
17.3.2 Person syncretism: 1PL=3PL
17.3.3 Person syncretism: 2SG=2PL
17.3.4 Person syncretism: 1SG=3SG
17.3.5 TAM syncretism: imperfect indicative and conditional forms
17.3.6 TAM syncretism: present indicative and present subjunctive forms
17.3.7 Summary
17.4 The relationship between patterns of syncretism and metamorphomes
17.4.1 Introduction
17.4.2 Syncretism patterns as templates
17.4.3 The source of syncretism patterns
17.4.4 Interaction between syncretism patterns and metamorphomes
17.5 Conclusions
18: The verbs ‘rain’ and ‘snow’ in Gallo-Romance, and other morphological mismatches in diachrony
18.1 The morphology of ‘rain’ and ‘snow’ in Romance
18.2 How and why ‘snow’ follows ‘rain’ in Gallo-Romance
18.3 Other Gallo-Romance cases of lexical contamination manifested in desinential morphology
18.3.1 Osoir, falloir, and estovoir
18.3.2 The type il trouverra
18.3.3 The type amerrir, alunir
18.3.4 First conjugation reflexes of third conjugation MITTERE and third conjugation reflexes of first conjugation *boˈtare
18.4 The role of phonological resemblance
18.5 The Gallo-Romance verbs ‘snow’ and ‘rain’ and their incompatibility with ‘constructivist’ models of morphology
References and bibliographical abbreviations
Index
Recommend Papers
![Micro-change and Macro-change in Diachronic Syntax (Oxford Studies in Diachronic and Historical Linguistics) [1 ed.]
9780198747840, 0198747845](https://ebin.pub/img/200x200/micro-change-and-macro-change-in-diachronic-syntax-oxford-studies-in-diachronic-and-historical-linguistics-1nbsped-9780198747840-0198747845.jpg)
![Parameter Theory and Linguistic Change (Oxford Studies in Diachronic and Historical Linguistics) [Illustrated]
9780199659203, 0199659206](https://ebin.pub/img/200x200/parameter-theory-and-linguistic-change-oxford-studies-in-diachronic-and-historical-linguistics-illustrated-9780199659203-0199659206.jpg)
![Word Order Change (Oxford Studies in Diachronic and Historical Linguistics) [Illustrated]
9780198747307, 0198747306](https://ebin.pub/img/200x200/word-order-change-oxford-studies-in-diachronic-and-historical-linguistics-illustrated-9780198747307-0198747306.jpg)
![Syllable and Segment in Latin (Oxford Studies in Diachronic and Historical Linguistics) [Illustrated]
9780199660186, 0199660182](https://ebin.pub/img/200x200/syllable-and-segment-in-latin-oxford-studies-in-diachronic-and-historical-linguistics-illustrated-9780199660186-0199660182.jpg)
- Author / Uploaded
- Sam Wolfe (editor)
- Martin Maiden (editor)
File loading please wait...
Citation preview
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
Variation and Change in Gallo-Romance Grammar
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
OXFORD STUDIES IN DIACHRONIC AND HISTORICAL LINGUISTICS General Editors Adam Ledgeway and Ian Roberts, University of Cambridge Advisory Editors Cynthia L. 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; George Walkden, University of Konstanz; David Willis, University of Cambridge 33 Indefinites between Latin and Romance Chiara Gianollo 34 Verb Second in Medieval Romance Sam Wolfe 35 Referential Null Subjects in Early English Kristian A. Rusten 36 Word Order and Parameter Change in Romanian A Comparative Romance Perspective Alexandru Nicolae 37 Cycles in Language Change Edited by Miriam Bouzouita, Anne Breitbarth, Lieven Danckaert, and Elisabeth Witzenhausen 38 Palatal Sound Change in the Romance Languages Diachronic and Synchronic Perspectives André Zampaulo 39 Dative External Possessors in Early English Cynthia L. Allen 40 The History of Negation in the Languages of Europe and the Mediterranean Volume II: Patterns and Processes Anne Breitbarth, Christopher Lucas, and David Willis 41 Variation and Change in Gallo-Romance Grammar Edited by Sam Wolfe and Martin Maiden For a complete list of titles published and in preparation for the series, see pp. 460–464
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
Variation and Change in Gallo-Romance Grammar Edited by SAM WOLFE AND MARTIN MAIDEN
1
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
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 Oxford University Press 2020 © the chapters their several authors 2020 The moral rights of the authors have been asserted First Edition published in 2020 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: 2019949628 ISBN 978–0–19–884017–6 Printed and bound in Great Britain by Clays Ltd, Elcograf S.p.A. 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.
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
In honour of John Charles Smith in the year of his seventieth birthday
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
Contents Series preface Abbreviations and symbols for this volume The contributors
1. Introduction Sam Wolfe and Martin Maiden
ix xi xvii
1
PART I. SENTENCE STRUCTURE 2. Old Gallo-Romance, periodization, and the left periphery Sam Wolfe 3. Resumptive structures in a Gallo-Romance perspective Christine Meklenborg
9 41
4. Variation in the Gallo-Romance left periphery: V2, complementizers, and the Gascon enunciative system Adam Ledgeway
71
5. Dialectological evidence for a predicate focus analysis of Gascon que Franck Floricic
100
6. Postverbal negators in Gallo-Romance: The view from Old Occitan Sandra Paoli and Xavier Bach
117
7. The loss of clitic climbing in French: A Gallo-Romance perspective Zack Bekowies and Mairi McLaughlin
138
PART II. THE VERB COMPLEX 8. Motivating the North–South continuum: Evidence from the perfects of Gallo-Romance Bridget Drinka
161
9. Active-middle alignment and the aoristic drift: The North–South divide in the Romània on evidence from northern Gallo-Romance Delia Bentley
191
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
viii
10. A comparative analysis of French auxiliation, with new evidence from Montréal Béatrice Rea
213
11. Présent inclusif and passé composé à valeur de présent accompli in modern French and Occitan Ingmar Söhrman
241
12. Future temporal reference in French and Gascon: Aller / anar + infinitive periphrasis and structural transfer in the bilingual grammar Damien Mooney 13. Mainland and insular Norman: Pronoun sharing and pronoun sparing Mari C. Jones
258
279
PART III. WORD STRUCTURE 14. On the origins of French and Occitan Clive R. Sneddon
307
15. Appositive compounds in dialectal and sociolinguistic varieties of French Brigitte L. M. Bauer
326
16. Complex versus compound prepositions: Evidence from Gallo-Romance Nigel Vincent
347
17. Syncretism and metamorphomes in northern Occitan (Lemosin) varieties Louise Esher
364
18. The verbs ‘rain’ and ‘snow’ in Gallo-Romance, and other morphological mismatches in diachrony Martin Maiden
385
References and bibliographical abbreviations Index
403 451
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
Series preface Modern diachronic linguistics has important contacts with other subdisciplines, notably first-language acquisition, learnability theory, computational linguistics, sociolinguistics, and the traditional philological study of texts. It is now recognized in the wider field that diachronic linguistics can make a novel contribution to linguistic theory, to historical linguistics, and arguably to cognitive science more widely. This series provides a forum for work in both diachronic and historical linguistics, including work on change in grammar, sound, and meaning within and across languages; synchronic studies of languages in the past; and descriptive histories of one or more languages. It is intended to reflect and encourage the links between these subjects and fields such as those mentioned above. The goal of the series is to publish high-quality monographs and collections of papers in diachronic linguistics generally, i.e. studies focusing on change in linguistic structure, and/or change in grammars, which are also intended to make a contribution to linguistic theory, by developing and adopting a current theoretical model, by raising wider questions concerning the nature of language change or by developing theoretical connections with other areas of linguistics and cognitive science as listed above. There is no bias towards a particular language or language family, or towards a particular theoretical framework; work in all theoretical frameworks, and work based on the descriptive tradition of language typology, as well as quantitatively based work using theoretical ideas, also feature in the series. Adam Ledgeway and Ian Roberts University of Cambridge
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
Abbreviations and symbols for this volume * ** (?)? ?
% > = Ø 1/2/3 I / II / III/ IV acc.mrk AcI AD Adj(P) Adv(P) Agr ALEN ALG ALLOC an. Arg. ASH Ba. BFM c. Cat. ch. conjug.
unattested form or usage ungrammatical form or usage (very) dubious form or usage substandard / non-standard form or usage form or usage which is not universally accepted as grammatical becomes, yields cliticized to null argument (subject or complement) first / second / third person first / second/ third / fourth conjugation ablative accusative accusative marker accusative and infinitive construction active (voice) Anno Domini adjective (phrase) adnominal adverbal adverb(ial) (phrase) assertive focus agreement Atlas Linguistique et Ethnographique Normand Atlas linguistique de la Gascogne Atlas linguistique et ethnographique du Languedoc Occidental animate Argot Auxiliary Selection Hierarchy auxiliary Dialect of Bagnes Base de Français Médiéval circa Catalan chapter clitic complementizer (position) conditional conjunction conjugation
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
xii
CP DP Du. E
e.g. Eng. EPP f. Fl. foc Foc(P) Fr. Frp. fut G GaR. Ger. Gsc. H H~E HT(LD) i.e. IbR. IE imp In inan.
copula complementizer phrase dative declarative determiner phrase Dutch (i) east(ern) (ii) event time (iii) Romance outcome of ‘be’ exempli gratia (for example) enunciative English Extended Projection Principle exclamative expletive feminine (i) folio (ii) and following page; finite Flemish focus (feature) focus (phrase) futurate present French Francoprovençal (Gallo-Romance dialects spoken in central eastern France, western Switzerland, and northwestern Italy) future temporal reference future Guernesiais Gallo-Romance genitive gender German Gascon Romance outcome of ‘have’ habitual aspect ~ hanging topic (left-dislocation) id est (that is) Ibero-Romance Indo-European inflected future imperative insular Norman inanimate indicative
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
indef inf int IP IPF ipfv IO It. J K(P) Lat. LD LG Lgd. Lim. lit. LMI loc LP m Med MHG Mid MN mod Mod ms. n N neg neut nom NP NPI num NW O Occ. ocl OE OFOH OHG
indefinite infinitive interrogative inflexion phrase imperfect imperfective (aspect) indirect object Italian Jèrriais case (phrase) Latin left-dislocation Dialect of La Gleize Lengadocien (Occitan dialects of Languedoc, southern France) Limousin (Occitan dialect spoken in Départements of Limousin, Charente, and Dordogne in southwestern France) literally Linguistic Market Index locative left periphery masculine medieval Middle High German Middle mainland Norman modal(ity) modern manuscript neuter (i) north(ern) (ii) noun negator neuter nominative noun phrase negative polarity item number Northwestern (i) old (ii) object Occitan object clitic Old English One Feature One Head Old High German
xiii
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
xiv
op opt OVS P₁ part pass PC per perf pf pfv PI Pic. pl plpfv PN Poi. poss P(P) prep prog prs prt Prv. pst Pt. ptcp q Q(P) quot R ref. refl reg. rel Ro. S
S SA sbjv ScS(P)
operator optative object-verb-subject order today past tense partitive passive Compound Past person perfect periphrastic future perfective (aspect) Inclusive Present Picard (dialects spoken in Picardy and Pas-de-Calais, northern France) plural pluperfect person and number Poilu possessive preposition(al phrase) preposition(al) progressive present tense preterite Provençal Occitan (Occitan dialects spoken in Provence, southeastern France) past Portuguese participle question particle/marker quantifier (phrase) quotative reference time referential reflexive (i) regional (ii) regular relative Romanian (i) south(ern) (ii) speech time (iii) subject Sercquiais intransitive (Actor/Agent) subject of an unergative clause subjunctive Scene-Setter (Phrase)
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
SF sg Skt SO/U SOV Sp. Srs. suf SV SVO t T TAM top Top(P) U UBH USH v V V2 Vgl. VOS VP VPA VS VSO W Wal. X XP
Stylistic Fronting singular Sanskrit intransitive (Undergoer) subject of an unaccusative clause subject-object-verb order Spanish Surselvan suffix subject-verb order subject-verb-object order trace (= underlying / intermediate position) of moved element Tense (position) tense, aspect, and mood topic (feature) topic (phrase) undergoer Universal Base Hypothesis Universal Spine Hypothesis verso (i) vowel; (ii) verb; verb second (syntax) Vegliote verb-object-subject order verb phrase Perfect with the value of fulfilled present action verb-subject order verb-subject-object order west(ern) Wallon (French dialect of Wallonia, southern Belgium) unspecified head element unspecified phrasal category
xv
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
The contributors Xavier Bach is Junior Research Fellow in Linguistics at Trinity College, Oxford. He completed his DPhil at the University of Oxford with a thesis on the typology of the diachronic development of inflexional classes. He works on Gallo-Romance, especially on Occitan varieties (mesoclisis, morphosemantics of the augmentative, number), as well as on morphological typology. Brigitte L. M. Bauer is Professor then Research Fellow at the University of Texas at Austin and Research Fellow at the Max Planck Institute for Psycholinguistics, Nijmegen. She specializes in historical comparative linguistics, with a specific focus on major grammatical changes in Indo-European languages (word order, transitivity, word formation) and processes of grammatical evolution in the shift from Latin to French/Romance. Her latest book, Nominal Apposition in Indo-European. Its forms and functions, and its evolution in LatinRomance (De Gruyter, 2017), demonstrates the prominence of nominal apposition in both syntax and morphology (compounding) and traces its long-term shift from asyndetic grammatical equivalence to hypotaxis. Zack Bekowies is a PhD candidate in Romance Languages and Literatures at the University of California, Berkeley, with specializations in Italian, Romanian, and French. His interests include comparative Romance linguistics, cognitive linguistics, semantics, and pragmatics, and his dissertation project explores the application of linguistic tools and theory as explanatory models for broader systems of discourse, focusing on authoritarian-regime propaganda. Delia Bentley is Professor of Romance Linguistics in the School of Arts, Languages and Cultures of the University of Manchester. Her main research interest is in dialect morphosyntax and how it interfaces with lexical semantics and discourse. She has published extensively on alignment, split intransitivity, existentials, locatives, presentationals, and subjecthood in Romance. She is the author of Split intransitivity in Italian (Mouton, 2006) and a co-author of Existentials and Locatives in the Romance Dialects of Italy (Oxford University Press, 2015). The latter is the principal output of the research project entitled Existential Constructions: an investigation into the Italo-Romance dialects (2010–2014), funded by the Arts and Humanities Research Council, on which Delia was the Principal Investigator. Bridget Drinka is Professor of Linguistics at the University of Texas at San Antonio and specializes in the sociolinguistic motivations for language change, the role of contact in linguistic innovation, and the importance of geographical contiguity in the diffusion of changes across the Indo-European languages. Her book, Language Contact in Europe: the periphrastic perfect through history (Cambridge University Press, 2017), explores the complex development of a grammatical category as it spread across the map of Europe and won the Linguistic Society of America’s Leonard Bloomfield Book Award for 2019.
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
xviii
A Fulbright Senior Lecturer at Moscow State University in 1998 and visiting professor in Germany, Italy, and Japan, she served as President of the International Society for Historical Linguistics (2015–17), and was responsible for organizing the International Conference on Historical Linguistics in 2017. She currently serves as Associate Editor of Folia Linguistica Historica. Louise Esher is a CNRS researcher based in the unit CLLE-ERSS (UMR 5263, Université Toulouse Jean Jaurès), with particular interests in autonomous morphology, historical and comparative linguistics, inflexion, and varieties of Occitan. She studied Romance linguistics at Oxford with J. C. Smith and Martin Maiden, and previously held a Junior Research Fellowship in Modern Languages at St John’s College, Oxford. Franck Floricic is Associate Professor in Italian linguistics in the department of Italian and Romanian of Paris 3—Sorbonne Nouvelle. His works focus mainly on the morphosyntax of Italian (including issues related to negation, modality, clitics, and indefinites). He also devoted various works to the morphosyntax of other Romance varieties, especially Sardinian and Occitan. Mari C. Jones is Professor of French Linguistics and Language Change at the University of Cambridge and Fellow in Modern and Medieval Languages at Peterhouse. Her extensive publications on language endangerment and revitalization include Variation and Change in Mainland and Insular Norman (Brill, 2015); The Guernsey Norman French Translations of Thomas Martin (Peeters, 2008); Exploring Language Change (Routledge, 2013); Jersey Norman French (Blackwell, 2001); Language Obsolescence and Revitalization (Clarendon Press, 1998). She is Fellow of the Learned Society of Wales. Adam Ledgeway is Professor of Italian and Romance Linguistics at the University of Cambridge, a Professorial Fellow of Downing College, and a Fellow of the British Academy. His research interests include the comparative history and morphosyntax of the Romance languages, Italian dialectology, Latin, Italo-Greek, syntactic theory, and linguistic change. Martin Maiden is Professor of the Romance Languages, and a Fellow of Trinity College, Oxford. He became a Fellow of the British Academy in 2003 and a Member of Academia Europaea in 2018, and is Director of the Oxford Research Centre for Romance Linguistics. His recent publications include The Cambridge History of the Romance Languages (co-edited with Adam Ledgeway and John Charles Smith; Cambridge University Press, 2011/2013), Morphological Autonomy (co-edited with John Charles Smith, Maria Goldbach, and Marc-Olivier Hinzelin, Oxford University Press, 2011), The Boundaries of Pure Morphology (co-edited with Silvio Cruschina and John Charles Smith, Oxford University Press, 2013), The Oxford Guide to the Romance Languages (co-edited with Adam Ledgeway, Oxford University Press, 2016), and The Romance Verb: morphomic structure and diachrony (Oxford University Press, 2018). Mairi McLaughlin is Associate Professor of French and an Affiliated Member of the Departments of Linguistics and Italian Studies at the University of California, Berkeley. She specializes in French and Romance Linguistics, and also in Translation Studies. Her first book, Syntactic Borrowing in Contemporary French: a linguistic analysis of news translation, was published by Legenda in 2011. Her monograph La presse française historique: histoire d’un genre et l’histoire de la langue will be published by Garnier.
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
xix
Christine Meklenborg is Professor of French Linguistics at the University of Oslo, Norway. She specializes in Medieval French but in recent years has also been working on medieval Germanic languages, especially Old Swedish. From 2014 to 2019 she was Principal Investigator of the research project Traces of History. Damien Mooney is Senior Lecturer in French Linguistics at the University of Bristol. His research focuses on contact-induced transfer in bilingual speech, language death theory, and the role of language and dialect contact in the loss or retention of pronunciation and grammatical features in regional varieties of French and the regional languages of France. He is the author of Southern Regional French: a linguistic analysis of language and dialect contact (Legenda, 2016). Sandra Paoli is Associate Professor in Linguistics (Romance languages) at the University of Oxford. Her main interests lie in comparative Romance linguistics, especially within grammaticalization and the interface between grammar and pragmatics. Béatrice Rea is a DPhil candidate at the Faculty of Linguistics, Philology & Phonetics of Oxford University. Her doctoral research, for which she has conducted extensive fieldwork in Montréal (Canada) creating a new corpus of sociolinguistic interviews, focuses on variation and change in the morphosyntax of Laurentian French. She also teaches the history and structure of French, as well as French grammar and translation at Oxford. Clive R. Sneddon has worked as a lexicographer in Mediaeval French for the Clarendon Press, and as the French Linguistics specialist at St Andrews University. He is now working there on the Old French Bible Project as an Honorary Research Fellow in the School of History. His research interests involve both medieval translation studies and the origins of the Romance languages. Ingmar Söhrman is Professor emeritus of Romance Languages at the University of Gothenburg. He obtained his doctorate from the University of Uppsala and has held positions at the Universities of Alcalá and Umeå. His main linguistic interests are semantics, syntax, history of the language, minority languages, Medieval history, Romance and Slavic languages and cultures, and cultural contacts. Nigel Vincent is Professor Emeritus of General & Romance Linguistics at the University of Manchester, following his retirement from the Mont Follick Chair in Comparative Philology in 2011. His research focuses on the modelling of morphosyntactic change, with special reference to Latin, Italian, and the dialects of Italy. He is a Fellow of the British Academy and a Member of Academia Europaea. Sam Wolfe is Associate Professor in French Linguistics at the University of Oxford and a Fellow of St Catherine’s College. His research interests include French and comparative Romance linguistics, formal syntactic theory and historical linguistics. His monograph Verb Second in Medieval Romance appeared with Oxford University Press in 2019.
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
1 Introduction Sam Wolfe and Martin Maiden
1.1 Background to the volume Variation and change are inextricably linked. This simple yet powerful maxim has been a central intuition in J. C. Smith’s research and it is his example that has inspired this volume. This book, which we dedicate to JC, focuses on Gallo-Romance data to show that formal theories and functionalist theories of synchronic language variation and usage can both fruitfully inform our understanding of language change. Conversely, theories of diachronic change nearly always bring ‘explanation’ within our grasp when they are informed by theories of variation, themselves developed primarily from synchronic sources.¹ The domain of diachronic grammatical change is hardly an unploughed furrow for French. Studies on the medieval language abound (e.g. Foulet 1919; Herman 1954; Moignet 1973; Skårup 1975; Vance 1997; Zink 1997; Buridant 2000) as do histories of the language (e.g. Nyrop 1904–30; Brunot 1933; Picoche and Marchello-Nizia 1989; Marchello-Nizia 1979, 1995; Marchello-Nizia, Combettes, Prévost, and Scheer 2019). However, in contrast to the Italian peninsula, where the development of standard Italian is regularly tracked alongside other ItaloRomance varieties (e.g. Rohlfs 1968, 1969; Maiden 1995; Benincà, Ledgeway, and Vincent 2014), the development of grammatical phenomena in French is rarely considered alongside the other Gallo-Romance languages.² This neglect is regrettable, because as JC himself has remarked in various works (cf. Smith 2002: 437–40; 2016: 317–18; 2019), French, specifically, and Gallo-Romance varieties, in general, often constitute Romance typological outliers in grammatical terms. Indeed, a central concern of many of the volume’s chapters (cf. §1.2) involves tracking the precise mechanisms through which the Gallo-Romance languages
¹ The symbiotic nature of language variation and change are most closely associated with variationist sociolinguistics (Labov 1966, 1972, 1994, 2001, 2010). However, formal generative approaches to morphosyntax have led to synchronic theories of language universals and variation informing diachronic theory and vice versa (see in particular Roberts 2007, 2019 and Ringe and Eska 2013). ² Notable exceptions to this are to be found in Jensen’s (1990) comparative treatment of historical Gallo-Romance syntax, Hinzelin’s (2007) comparative study object pronouns, and a number of recent generative studies comparing old French with Occitan (e.g. Vance, Donaldson, and Steiner 2009; Wolfe 2018b).
Variation and Change in Gallo-Romance Grammar. First edition. Sam Wolfe and Martin Maiden (eds). This chapter © Sam Wolfe and Martin Maiden 2020. First published in 2020 by Oxford University Press.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
2
acquired a distinctive typological footprint. Central to this issue is the question of how typologically similar contemporary and historical varieties of French are to other Gallo-Romance languages and whether the Gallo-Romance languages belong to the ‘northern Romance’ morphosyntactic prototype proposed by Ledgeway (2012: 252). To summarize, two key questions to be explored in the volume are the precise degree of continuity and variation between the GalloRomance varieties discussed and their level of similarity to, or distinctness from, Romance varieties more generally. The guiding principle behind the volume is that only by considering the standard and non-standard Gallo-Romance languages together can progress be made in answering these foundational issues. France historically has been home to enormous linguistic diversity,³ with issues of language contact and dialect-mixing at the heart of historical accounts of the Gallo-Romance languages’ history (Lodge 1993, 2004; Posner 1996; AyresBennett 2004) and, to a lesser extent, the linguistic situation in France today (Hornsby 2006; Beeching, Amstrong, and Gadet 2009; Jones 2011; Mooney 2016; Gadet and Hambye 2018). Nevertheless, variationist sociolinguistic concerns have never been central to the discipline of linguistics in France, as discussed in detail in Jones (2011: 505–6). There are however signs of a renewed interest in applying sociolinguistic theory to the linguistic situation in France, reflected in both a number of recent works seeking to re-evaluate the status of regional variation in France (Pooley 2004, 2006; Hornsby 2006, 2013) and studies adopting an explicitly variationist methodology to understand variation and change in progress (i.e. Ashby 1981; Coveney 2010; Mooney 2018). Indeed, both these developments are parallel to an increasing body of theoretically rich formal grammatical research on both French and other Gallo-Romance varieties.⁴ Although an apparent tension between sociolinguistic and formal grammatical insights into a range of phenomena is either tacit or explicit in the literature, our intention in assembling the volume is to reflect a fundamental tenet of J. C. Smith’s research, namely that seemingly divergent theoretical approaches have easily overlooked complementary value. We hope that this book, inspired as it is by JC’s intellectual example, will serve as a model and springboard for new research, which seeks to explain the historical development of French and other Gallo-Romance varieties in the wider comparative context.
³ One cannot but mention here the pioneering dialectological work of Gilliéron and Edmont (1902–10) on the ALF. For recent discussion of their legacy, see Manzano (2019). ⁴ On syntax see in particular Kayne (2000, 2005), Hulk and Pollock (2001), Lahousse (2003), Poletto and Pollock (2004), Oliviéri (2004), Rowlett (2007), De Cat (2009), Kristol (2009), Roberts (2010), Hinzelin and Kaiser (2012) and Kayne and Pollock (2012). On morphology see Maiden (2005, 2009, 2018), Smith (2011, 2013), and Esher (2012, 2015a, 2016, 2017a, 2017b).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
3
1.2 Smith’s contribution to linguistics and the chapters in the volume The chapters that follow have been assembled in honour of J. C. Smith’s contributions to French, Romance, and general linguistics. While the present volume reflects Smith’s central research focus on French and Gallo-Romance (see in particular Smith 1989a, 1989b, 1995c, 2002, 2005, 2007, 2011, 2016), he has also presented and written widely on a range of phenomena across the full range of Romance languages, notably Italo-Romance (Smith 1991, 2019; Cappellaro, Maiden, and Smith 2012) and Ibero-Romance (Smith 1993, 2018), Romanian, as well as non-Romance languages (Donohue and Smith 1998; Smith 2008). Indeed, any appreciation of Smith’s contribution to the field must reflect the sheer versatility of his contributions to linguistics. JC’s research has spanned work on Latin (1989c, 1995b, 2004, Ashdowne and Smith 2007), comparative studies of Romance languages (Smith 1989c, 2001, 1992, 1995a, 1999b, 2006, 2013a, 2019; Maiden, Smith, and Ledgeway 2011, 2013) as well as studies of particular Romance varieties. Crucially, while showing admirable rigour in how the linguistic data are handled in his work, almost all his contributions to the field are informed by, or lead to the refinement of, a body of formal linguistic theory. In this domain, Smith’s legacy to the field has been one of profound intellectual breadth, spanning Government and Binding Theory (Smith 1991), Functional Grammar (Smith 2001), morphological theory in general and morphomic theory in particular (Smith 2011, 2013a; Maiden, Smith, Goldbach, and Hinzelin 2011; Cruschina, Maiden, and Smith 2013), discourse-pragmatics (Smith 1989a, 1995b, 2001) and sociolinguistics (Smith 1995a, 2011, forthcoming). Smith’s work has had a particular impact in the field of historical linguistics. His distinction in this area led to his appointment as Secretary to the International Society for Historical Linguistics and his publications in this area have shown a particular focus on the development of theories of markedness (Smith 1989c, 1997, 1999b, 2006), reanalysis (Smith 1989c, 2011, 2013a), the Actuation Problem (Smith 1989c), the role of semantics and pragmatics in conditioning change (Smith 1999a, 1999b, 2007) and large-scale questions of periodization and classification (Smith 2002, 2011, 2013a, 2016, 2019; Smith and Sneddon 2002). In this volume we have sought to pursue the scholarly values characteristic of Smith’s work: a belief that careful empirical work can fruitfully be combined with formal theoretical approaches, the overriding conviction that seemingly conflicting theoretical ‘schools’ can in fact offer complementary insights, and, first and foremost, a challenge to the dogma, too often present in Gallo-Romance linguistics historically, that only standard languages have anything of value to offer the linguist. The first chapter of Part I takes as its point of departure Smith’s (2002) work on the issue of language classification and periodization in middle French.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
4
Wolfe’s chapter assesses evidence for syntactic change in medieval French and Occitan. He uses data from left-peripheral word order properties, null arguments, and the syntax-pragmatics mapping, to evaluate the usefulness of conventional periodization labels such as ‘new’ or ‘old’. Meklenborg’s chapter is also a comparative study of the syntax of the left periphery in old French and Occitan. Exploring the data on the now hotly debated issue of the resumptive particle (see Marchello-Nizia 1985; Fleischman 1991; Ledgeway 2007) she concludes that French and Occitan both feature multiple sub-types of with distinct distributional properties. The following two chapters, although informed by historical concerns, focus principally on the left periphery of contemporary Gallo-Romance varieties. Ledgeway offers a formal analysis of V2 ‘relics’ in French and Gascon. Building on Smith’s (2016: 310) discussion of French inversion, he offers a novel pragmatico-syntactic account of both the French contexts where V-to-C movement still obtains alongside an analysis of the Gascon left periphery, which he argues can still be conceived of as V2 at a certain level of abstraction. Also looking at Gascon left-peripheral particles, Floricic offers a focused analysis of the enunciative particle es kə in the dialect of Les Esseintes. He argues that a particular particle is a fossilized instantiation of a cleft construction, which may shed new light on the genesis of Gascon enunciative particles more generally. Although the diachronic trajectory of French negation has accrued a vast literature (see Hansen 2013 for recent review), Occitan negation has received much less attention. Paoli’s and Bach’s analysis of the development of the postverbal negators pas and ges, is therefore a welcome contribution on this topic which offers an account of syntactic and discourse-pragmatic factors which conditioned the rise and fall (in the case of ges) of the two negators. Bekowies and McLaughlin deal with a distinct syntactic phenomenon, namely clitic-climbing, which is still present in modern Occitan varieties but absent in modern French. Based on a new corpus study of hagiographical texts, they argue that clitic-climbing was retained longer in Occitan-speaking areas than was the case elsewhere in France. Part II deals with a range of grammatical phenomena within the verbal domain, a topic which has been of central interest to Smith throughout his career (Smith 1991a, 1991b, 1992a, 1993a, 1993b, 1995a, 1995b, 1996, 1999b). The first four chapters are concerned with Gallo-Romance past constructions. Drinka engages with Smith’s (2016) revisions to Ledgeway’s (2012) proposed North–South continuum on the basis of Francoprovençal, Occitan, and Catalan data and argues that the key factor differentiating northern varieties from their southern counterparts in terms of split intransitivity and their use of the perfect is whether or not they fell within the Carolingian realm. Similar phenomena are discussed by Bentley who also adopts a pan-Romance perspective. Her analysis links developments in the subject clitic system of Gallo-Romance to its distinct profile in terms of alignment and the / split. Rea’s chapter is concerned with considering this precise split between auxiliary avoir ‘have’ and être ‘be’ in
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
5
Montréal French. She demonstrates that the use of avoir with unaccusative predicates has declined in recent years and argues for the importance of a variationist sociolinguistic analysis to understand the relevant facts. Söhrman also deals with the periphrastic past, but his chapter is principally devoted to the temporal-aspectual readings which it allows. He demonstrates that both the present and passé composé can be used with a projection towards the present time and future. The chapters by Mooney and Jones conclude Part II, and both have questions regarding the effect of language contact upon the grammar. Mooney discusses competing strategies to encode future temporal reference by comparing the grammars of Gascon-French bilinguals and monolingual French speakers. He finds that temporal distance is the defining factor constraining use across all groups, but that there are important differences between the different grammars analysed which challenge certain existing theories of bilingualism. Jones presents a detailed discussion of the pronominal system of mainland and Channel Islands Norman. She argues that the grammars of both Norman varieties are diverging in the pronominal system and that French is exerting both ‘overt’ and ‘covert’ influence on the pronoun system of the languages examined. The focus of Part III is word structure. Sneddon opens this part of the book with a discussion of the origins of Gallo-Romance, through a philological study of word structure in the earliest available manuscripts. He shows that multilingualism (cf. also the chapters by Mooney and Jones) is key to understanding the early medieval textual records and provides a systematic account of, principally morphological, points of innovation and conservatism. Bauer’s chapter looks at an aspect of word structure which is widespread in Gallo-Romance, namely appositive compounding, e.g. homme-grenouille ‘man-frog’ > ‘diver’. Wide-ranging empirical analysis of data from French, Picard, Walloon, Gascon, Occitan, Limousin, and le poilu lead to the conclusion that appositive compounds are a widespread feature of Gallo-Romance grammar. Compounding is also a central concern for Vincent, in his chapter on compound and complex prepositions in Gallo-Romance. However, in contrast to the relatively uniform picture painted by Bauer for nominal compounding, Vincent finds that the prepositional domain yields evidence of extensive microvariation. This degree of relative heterogeneity, he argues, provides further evidence for the theoretical claim (cf. Déchaine 2005) that prepositions constitute a (borderline-)lexical category. The final two chapters are focused on morphological structure in the verbal domain. Esher shows that person-syncretism is a systematic feature of Lemosin varieties, with syncretism particularly widespread in the variety of Gartempte, on which she focusses. She argues that the data available are a classic case-study where paradigm cells are reassigned from one metamorphomic template to another. In his study of the verbs ‘rain’ and ‘snow’, Maiden shows that the former has ‘contaminated’ the latter diachronically in terms of its morphological structure.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
6
The Gallo-Romance data for these forms are shown to have wide-ranging significance for morphological theory in arguing for theoretical approaches which put word-forms at the centre of morphological analysis. The contributions to this volume have certain characteristics in common. They have a thorough empirical base in philological, corpus-based, or spoken data. These data cover both standard and non-standard Gallo-Romance varieties, and formal grammatical or sociolinguistic theory is engaged with critically in all the chapters. Furthermore, they offer a synthesis of big-picture typological analysis alongside fine-grained explanation of variation and change at the micro-level. We believe that the chapters assembled here meet J. C. Smith’s very high standard of methodological, empirical, and theoretical originality and rigour, which he has always combined in equal measure in his research and in his teaching.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
PART I
SENTENCE STRUCTURE
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
2 Old Gallo-Romance, periodization, and the left periphery Sam Wolfe
2.1 Background 2.1.1 Linguistic change in old French and old Occitan There is a long tradition of descriptive-philological studies on both medieval French and Occitan, which has often led to the identification of common morphosyntactic properties between the surviving texts from medieval France (Tobler 1875; Thurneysen 1892; Foulet 1919; Brunel 1935; Anglade 1930; Priestley 1955; Hamlin, Ricketts, and Hathaway 1967; Moignet 1973; Skårup 1975; Ménard 1988; Jensen 1990, 1994; Romieu and Bianchi 1999, 2006; Buridant 2000). Largely due to its more extensive descriptive tradition, formal syntactic studies of medieval French proliferated from the mid-1980s onwards, with important ramifications within Government and Binding Theory (Chomsky 1981, 1982, 1986) and early Minimalism (Chomsky 1995) for theories of Verb Second (V2) and null subjects, to name just two phenomena.¹ In the past two decades, the syntax of early French has again seen a resurgent research interest, which has not only led to the formulation of a number of significant empirical generalizations about the data from the medieval period, but has also placed historical French data at the forefront of novel theoretical developments in formal syntactic theory (Labelle and Hirschbühler 2005, 2017, 2018; Rouveret 2004; Mathieu 2006, 2009, 2012; Labelle 2007, 2016; Larrivée 2011; Salvesen 2011, 2013, 2019; Hansen 2013; Salvesen and Walkden 2017; Simonenko and Hirschbühler 2012; Wolfe 2016a, 2018a, 2018b: chaps 4, 7; Simonenko, Crabbé, and Prévost 2018). However, while French received considerable attention in theoretically informed work in the 1980s and 1990s, its Gallo-Romance sister language, Occitan, has only accrued a small literature on its clitic pronominal
¹ On V2 see in particular Vance (1995, 1997: chaps 2–4), Roberts (1993: chap. 2), Hulk and Van Kemenade (1995), Platzack (1995), Lemieux and Dupuis (1995), and Cardinaletti and Roberts (2002). See Vanelli, Renzi, and Benincà (1986: 58–62), Adams (1987a, 1987b, 1987c), Vance (1987, 1993, 1997: chap. 4), Hirschbühler (1990) and Roberts (1993: chap. 2) on null subjects.
Variation and Change in Gallo-Romance Grammar. First edition. Sam Wolfe and Martin Maiden (eds). This chapter © Sam Wolfe 2020. First published in 2020 by Oxford University Press.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
10
system (Hinzelin 2004, 2007), resumptive particles (Donaldson 2015; Meklenborg this volume), V2 syntax (Vance, Donaldson, and Steiner 2009; Donaldson 2016; Wolfe 2017, 2018b: chap. 4), and null argument properties (Wolfe 2018c: 327–30). Given the rich picture of regional variation which characterized much of France until the early twentieth century and still characterizes certain Gallo-Romance varieties,² can similar claims be made about the earliest Gallo-Romance textual records? Are the medieval textual records characterized by homogeneity or is syntactic variation at the macro-, meso-, or micro-level (cf. Roberts 2019: chap. 1 on these labels) already discernible at an early stage in the languages’ history?
2.1.2 Classification and the North / South split The issue of classification has been prominent in nearly all manuals of Romance linguistics published within the past century (cf. Wartburg 1950; Posner 1996: chap. 2; Ledgeway 2012; Bossong 2016; Heinz, Filipponio, and Hinzelin forthcoming), yet it is an understatement to say that many issues remain unresolved. Two issues are particularly relevant to the present chapter. First, the exact status of Occitan varieties within a wider Romance typology still eludes a commonly accepted classification. While many of the authors in this volume and elsewhere include Occitan within Gallo-Romance (Wheeler 1988; Oliviéri and Sauzet 2016) this stance is controversial, with Bec (1967) famously arguing for an Occitano-Romance subgroup, including both Occitan and Catalan on the basis of shared linguistic features, which are principally morpho-phonological. A second relevant issue concerns the formulation of a proposed North–South axis defining the Romània (Zamboni 2000; Ledgeway 2012: 314; Drinka 2017: 166–7: §7.9). Specifically, Ledgeway (2012: 314) suggests that six major morphosyntactic isoglosses separate northern Romance varieties (langue d’oïl, langue d’oc, RaetoRomance, and northern Italian dialects) from their southern counterparts. Ledgeway (2012: 314) lists the following properties as central to this division: (i) prolonged retention/early loss of V2 syntax; (ii) marking of A/S (subject clitics, generalized preverbal position) vs marking of O (prepositional accusative, object clitic doubling); (iii) prolonged retention/early loss of binary (or ternary) case system; (iv) HABERE/ESSE auxiliary alternation vs generalized auxiliary (either HABERE or ESSE depending on variety and/or syntactic context); (v) retention vs loss of participial agreement; (vi) loss vs retention (and reinforcement) of preterite
² See Bec (1967, 1973) and Mooney (Chapter 12, this volume) on Occitan-speaking France, Kristol (2016: §20.1) on Francoprovençal-speaking areas and in particular the discussion of regional French in Jones (2011), and Hornsby (2006).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-, ,
11
Features (i) and (ii) are directly relevant to the word order properties of the medieval languages. We can add an additional feature to Ledgeway’s typology, drawing on insights in Wolfe (forthcoming: § 5), namely the widespread availability of argument-fronting operations reaching the left periphery via movement in southern Romance (1) (cf. for overview Cruschina 2012: chap. 3), in contrast to contemporary French and Occitan in particular where alternative strategies such as left-dislocation and clefting (2,3) are strongly preferred to give discourseprominence to a constituent (Sauzet 1989; Belletti 2005; De Cat 2009; Faure and Oliviéri 2013): (1)
a. Dada l ana tottu culthos pannos a samunare give. her. have.3 all these clothes to wash. ‘They have given her all these clothes to wash’ (Sardinian, Remberger 2010: 560) b. Un libbru ci a book him. ‘I gave him a book’
detti give.1. (Sicilian, Cruschina 2012: 54)
c. Albă e doar pe margini white be.3. only on margins ‘[The dress] is white only on the margins’ (Romanian, Zafiu 2013a: 570) d. Con NADIE compartió María su secreto with nobody share.3. María her secret ‘María shared her secret with nobody’ (Spanish, Zubizarreta 1998: 103) (2)
(3)
a. Lo libre de Joan the book of Joan ‘Joan’s book, I read it’
l’ it.
b. Le livre de Jean the book of Jean ‘Jean’s book, I read it’
je I
ai have.1 l’ it.
legit read. (Occitan, Sauzet 1989: 237)
ai have.1.
lu read. (French)
a. Son las mandarinas qu’ ai balhadas a la be.1 the tangerines that have.1. give. to the Maria Maria ‘I gave the tangerines to Maria (Occitan, Frota and Prieto 2015: 209) b. Ce sont les mandarines it be.3 the tangerines à Marie to Marie ‘I gave the tangerines to Marie’
que that
j’ I
ai have.1.
données give.
(French)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
12
The question then arises as to whether the two features identified by Ledgeway and the additional related property introduced above have their roots in the early textual records, or, as suggested by Smith (2019), the North—South split gets into trouble when faced with the empirical data from Gallo-Romance.
2.1.3 Periodization Within the domain of historical Romance syntax, period divisions are frequent, with a general consensus that syntactic criteria distinguish an entity called ‘old French’ from one called ‘middle French’.³ Treatments of the periodization debate, such as Smith (2002) for French specifically, and Wright (2013) for Romance, highlight numerous issues with these labels including the arbitrary nature of the nomenclature and the fuzzy and frequently unfalsifiable internal and external criteria used to justify them.⁴ Pertinent here is the discussion in Smith (2002: 434f.), who notes that the decline in V2 and increase in SVO orders in the history of French is viewed as an intrinsic characteristic of the ‘middle’ period by a number of scholars (Ayres-Bennett 1996: 98; Price 1971: 261; Marchello-Nizia 1980: 329–33, 1999: 48f.; Prévost 2002: 14–16; Muller 2005: 10f., 2009: §5; Steiner 2014: 128–43). Intriguingly, Smith (2002: 424) also notes that the term ‘middle Occitan’ or ‘moyen occitan’ is a ‘rarity’. While it is undoubtedly true that the form is not used as frequently as in the French linguistic literature, it does appear in a number of handbook and reference treatments such as Bec (1967: 90), which Smith notes, as well as Smith and Bergin (1984: 2, 6) and Perugi (2003: 246). Given that the ‘old’ vs ‘middle’ distinction is axiomatic in historical French syntax and frequently supplemented with an ‘early old’ and ‘later old’ subdivision (Roberts 1993: 135f.; de Bakker 1997: 35; Labelle 2007; Wolfe 2016b: 479–83; Mathieu 2012: 333–5; Zimmermann 2014: 9, 17f., 81, 108; Carlier et al. forthcoming) a major consideration in the empirical discussion that follows is whether these labels correspond to a linguistic reality in syntactic terms. That is to say, semi-independently of the value they may or may not have elsewhere, are there broad periods in the history of French which can be usefully labelled for a syntactician? Furthermore, to what do we attribute the near-total absence (Smith 2002: 424) or, as I have suggested, paucity in the use of the term ‘middle Occitan’? Is its absence or rarity due to major structural continuity in the medieval texts or a matter or mere convention?
³ See for example this division in Roberts (1993: 142–86), Vance (1995; 1997: chap. 6), Lemieux and Dupuis (1995), Bakker (1997: chap. 4), Sprouse and Vance (1999: 266), and Muller (2009). ⁴ Cf. Wright’s observation (Wright 2013: 121) that ‘old’ is frequently an inappropriate term for early Romance forms on the basis that they were ‘in context new and even revolutionary’.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-, ,
13
2.1.4 Previous research on old Occitan It is absolutely clear that medieval Occitan syntax does not enjoy the same depth of scholarship as French. Nevertheless, the body of formally oriented syntactic scholarship on early Occitan texts is growing and therefore this area no longer constitutes a total lacuna. In contrast to some or all of the medieval French texts (cf. discussion in §7.2), which are often described as instantiating a strict form of V2 syntax (Vanelli, Renzi, and Benincà 1986: 53; Benincà 1995: 329; Vance, Donaldson, and Steiner 2009; Sitaridou 2012: 578; Wolfe 2018b: 86), the literature on old Occitan often stresses its ‘relaxed V2’ status. This is the case in Donaldson’s analysis (Donaldson 2015, 2016) of subject-left dislocation in old Occitan, where he explicitly adopts Benincà’s analysis (Benincà 2004, 2006, 2013) of medieval Italo-Romance V2, involving verb-movement to a low Focus-head within the left periphery and corresponding merger of a moved constituent within the Focus field. Building in many ways on intuitions in Vanelli, Renzi, and Benincà (1986), Wolfe (2017, 2018b, 2018c) also adopts a version of this relaxed V2 analysis, but with Fin, the lowest head in Rizzi’s left-peripheral schema (Rizzi 1997, 2001a, 2004, 2013, 2015), as the locus of V2. In empirical terms, data from the late fourteenth-century prose text Douceline are also used to argue that Occitan patterns distinctly from French and northern Italian dialects in a number of significant ways. These include, widespread attestation of V1 orders featuring a form of null topic, the attestation of - orders which are restricted in certain medieval Romance varieties, the licensing of verb-fourth or greater orders (henceforth V>4), less significant matrix / embedded asymmetries concerning word order and symmetrical licensing of null subjects in contrast to the asymmetric distribution in thirteenth-century French (Adams 1987a, 1987b, 1987c). Although the details of the analysis need not concern us here, Meklenborg (Chapter 3, this volume) also notices distinctions between French and Occitan regarding the behaviour of the particle (on which see §7.2.5). Despite the upsurge in research on Occitan, there are numerous factors that call for a reappraisal. First, despite the acknowledgement since at least Vanelli, Renzi, and Benincà (1986: 49f.) that there is internal intertextual variation within medieval Occitan varieties, some of the studies above draw only on a small number of texts. Second, there is a clear need to better understand patterns of diachronic progression within the medieval period and whether points of variation can in any way be given useful diachronic labels. Third, as will be set out in detail in §7.2, the core syntactic properties of French and their diachronic progression are now better understood than before. Comparison between these closely related sets of varieties is therefore potentially enlightening.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
14
2.2 Linguistic change in French 2.2.1 Background In this section I highlight the major characteristics of early French word order, paying particular attention to properties which show unambiguous evidence for diachronic variation. All the properties surveyed here have in common that they are linked to the V2 syntax of early French, but this should not be taken to mean that diachronic variation is not found within other important domains.⁵
2.2.2 V2 and inversion Although the hypothesis is controversial (Kaiser 2002b; Rinke and Meisel 2009), there is increasing consensus that French until at least the fifteenth century (Muller 2005, 2009; Vance 1995, 1997: chap. 6) was a V2 language (Vance 1987, 1988, 1995, 1997; Adams 1987a, 1987b, 1987c; Roberts 1993: chap. 2; Cardinaletti and Roberts 2002: §1.2; Labelle and Hirschbühler 2005, 2017, 2018; Mathieu 2006, 2009, 2012; Labelle 2007, 2016; Salvesen 2011, 2013, 2019; Vance, Donaldson, and Steiner 2009; Wolfe 2016b, 2017, 2018a, 2018b: chap. 4). This is not to be understood as a linear order constraint but rather a structural requirement that the finite verb raise into the left periphery of the clause and that an additional constituent be merged in the left periphery, in a parallel fashion to contemporary French where verb movement targets the T-layer (Pollock 1989; Belletti 1990; Schifano 2018) and a subject constituent must be merged in SpecTP (Roberts 2010). This is schematized in a simplified format in (4): (4)
a. [CP . . . XPV2 [C VFinite] [TP . . . [vP . . . .]]] b. [CP . . . [TP XPSubject [T VFinite] [vP . . . .]]]
Evidence for such an analysis of the early French textual sources is abundant within the textual records and rests upon a number of core empirical observations. First sits the very basic observation that nearly all scholars of early French syntax identify second-position as the preferred position for the finite verb (Thurneysen 1892; Wartburg 1958: 103; Price 1971: §11.5; Moignet 1973: 287; Skårup 1975: 290; Harris 1978: 18–22, 1984: 189–98; Benincà 1983b: 5; Adams 1987c: 2f.; Fleischman 1991: 267–70; Roberts 1993: 88; Bakker 1997: chap. 1; Vance 1997: 37–9; Salvi 2000b: 665; Labelle 2007: 290; Vance, Donaldson, and Steiner 2009: 302–7; Hansch 2014: 81–118; Steiner 2014: 10–23). Wolfe ⁵ See for example Zaring (1998, 2010, 2011) on the progressive loss of OV and Labelle and Hirschbühler (2005) alongside Rouveret (2004) on important changes in the clitic system.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-, ,
15
(2018b: 66) reinforces Vance’s (1997) early claims about the thirteenth-century prose text La Queste, in showing that 75.16% of matrix clauses in his 1000-clause sample are descriptively V2. Ledgeway’s analysis (Ledgeway 2019) of the early thirteenth-century Histoire ancienne jusqu’à César (1208–1213) reaches a similar conclusion, namely that of the 540 matrix clauses examined, 79.4% are also descriptively V2. Dealing with a smaller sample size assembled for this chapter, of 200 clauses from the Grandes chroniques de France, which dates from approximately fifty years later than La Queste, the finding is still similar with 81% (162/200) of matrix clauses displaying linear V2. As we shall see below there is more diachronic and intertextual variation in the verb-placement data than is conventionally assumed, but the basic observation holds that there is no published study of early French, to the best of my knowledge, which suggests that the unmarked position for the finite verb is anything but second. Second, the preverbal field in early French is noted by a wide range of scholars to differ radically from the modern language. Put simply, in contrast to the prefield in modern French, which typically hosts the subject, earlier stages of French manifest a prefield which can be lexicalized by a wide-range of constituents belonging to various grammatical categories with varying discoursepragmatic values (Foulet 1919: §§306–32; Wartburg 1958: 103; Moignet 1973: 367; Skårup 1975: 9-69; Vanelli, Renzi, and Benincà 1986: 58; Roberts 1993: 85–7; Hulk and van Kemenade 1995: 235f.; Donaldson 2012: 1025; Mathieu 2012: 327; Wolfe 2018b: 68–72, forthcoming):⁶ (5)
(6)
(7)
Ce oïrent el pales that hear.3. in=the palace ‘Many in the palace heard this’
maint many
De sa parole ne fut mie of his words be.3. ‘He was not hasty in response to these words’
(Charrette 80) hastifs hasty (Roland 140)
Longuement parlerent ensemble entre le preudome et long. speak.3. together among the man and Lancelot Lancelot ‘The man and Lancelot spoke together for a long time’ (Queste 178, 26–7)
Third, clear evidence that finite verb movement targets the clausal left periphery and not a position within the clause-medial T-layer comes from well-known Germanic-inversion structures (Adams 1987c: 9; Dupuis 1989: 119f.; Roberts 1993: 56; Vance 1995: 174–6; Salvesen and Bech 2014: 213f.; Wolfe 2018b: 72–4), ⁶ The primary textual sources used in this study are listed in an appendix at the end of this chapter.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
16
where the subject appears ‘sandwiched’ between the left-peripheral finite verb and verbal complements demarcating the left-edge of the v-VP-complex: (8)
a. Son cors ne poï his body can.1. ‘I cannot see his body’
je I
veoir see. (Queste 122)
b. Et quant il vinrent la, si s’en estoient and when they come.3. there =.=be.3 ja li Grieu fui already the Greeks flee. ‘And when they arrived there, the Greeks had already fled’ (Clari 67) Fourth, early French varieties show major word order asymmetries between matrix and embedded clauses, which in the paradigm case fall out naturally from the observation in the classic Germanic V2 literature that the raised verb and the complementizer are in complementary distribution (Den Besten 1983). Thus V2 and the related properties of Germanic-inversion and CP-particles such as (cf. §7.2.5, this volume) are typically restricted in the embedded domain (Vanelli, Renzi, and Benincà 1986: 59; Adams 1987c: 5; Roberts 1993: 142; Roberts 2007: 61–3; Vance 1997: 133; Labelle 2016; Wolfe 2018b: §4.3), where, aside from certain exceptions, SV(O) typically obtains: (9)
a. De cez paroles que vos avez ci dit of these words that you have.2. here say. ‘Of these words which you have said here’ (Roland 145) b. Et quant il vit qu’i ne vendroient pas and when he saw that=they come.3. ‘And when he saw that they were not coming . . . ’ (Queste 190b, 24–5)
All these core indicators serve the important purpose of highlighting that V2 and the related properties are one of the major distinguishing features of early French (cf. also Smith 2016: 310). We shall now see, however, that there is notable evidence for diachronic variation in a number of the V2 correlates.
2.2.3 Change in the left periphery Although certain core properties of the V2 syntax are uniform throughout the medieval period, in keeping with variation and change observable in other uncontroversial V2 systems,⁷ there is detailed empirical evidence that the structure of the ⁷ See for example Axel (2007) for a historical perspective on German, and Walkden (2017) for a review of developments in contemporary Germanic varieties.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-, ,
17
left periphery is not uniform throughout the early history of French, which I take to run from the ninth to the fifteenth centuries. Looking at the earliest French texts up to approximately 1200, the preverbal field appears non-specialized in terms of discourse-pragmatics and can clearly host constituents which are both unambiguously discourse- and . In this regard, consider the wide body of scholarship on ‘early old’ French, an entity often interpreted as corresponding to verse texts from the eleventh and twelfth centuries, which highlight the possibility for an initial focal element to satisfy V2 (Labelle 2007: 302–5; Mathieu 2012: 341; Donaldson 2012: 1040–4; Wolfe 2016b: 480; Wolfe 2018a: 349f.; Labelle and Hirschbühler 2018).⁸ In less formal terms, Marchello-Nizia (1995: 95–100) notes that preverbal objects in early old French verse are more frequently informationally that fronted objects in old French prose after 1200, a finding confirmed with certain caveats in Labelle and Hirschbühler (2018: 276f.) and Steiner (2014: 205–26). I offer relevant examples of initial focal constituents as follows: (10) a. Buona pulcella fut good girl be.3. ‘Eulalia was a good girl’
Eulalia Eulalia (Eulalia 1)
b. Un en i out ki sempres vint one . . have.3. who always come.3 avant forward ‘There is one who always comes forth’ (Alexis 228) c. En ceste tere ad asez in this land have.3. enough ‘He has stayed long enough in this land’
osteiet stay. (Roland 35)
d. Un faldestoet out suz l’umbre d’un pin a folding-chair have.3. under the-shade of-a pine-tree ‘There is a folding chair under the shade of a pine tree’ (Roland 407) Alongside these qualitative observations, we can add results of a quantitative nature, in addition to Marchello-Nizia’s. Labelle and Hirschbühler (2018) is by far the most thorough study and makes use of a large-scale diachronic corpus. They note a strong tendency for objects, predicates, and quantifiers not to be placed in initial-position when focal after approximately 1200, but highlight that the change is not categorical, with constituents coded as focal within their corpus still present in initial-position after this point (2018: 277). As small-scale ⁸ Note that the relatively widespread attestation of focal constituents—often verbal complements— across the earliest old French and Romance texts (Wolfe 2016b: 466–72) may go some way towards accounting for Dardel’s otherwise somewhat curious proposal (Dardel 1996) that ‘Proto-Romance’ went through an OVS stage.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
18
corroboration of their findings, I tagged 100 main clause-initial arguments for their discourse-pragmatic status from La Queste, the text typically used as indicative of early thirteenth-century French prose, the Grandes chroniques de France, a text dating from fifty years later which certain scholars cite as the transition period into ‘middle’ French, and Enguerrand de Monstrelet’s Chronique, which dates from 1441–1444, a period which Smith (2002: 427) notes is indisputably classed as ‘middle’ by those using the terms. Given the small size of the corpus, these findings should be taken with some caution, but nevertheless are broadly in line with the findings of previous studies, in showing only c.10% of initial constituents which are discourse- (Table 2.1). Table 2.1. Discourse-pragmatic values of preverbal constituents in French New Graal 1225 GChron 1275 Monstre 1442
10 10 11
10% 10% 11%
Accessible 17 26 7
17% 26% 7%
Active 11 28 46
11% 28% 46%
Pro 56 22 32
56% 22% 32%
Q 6 14 4
6% 14% 4%
Total 100 100 100
We therefore see a major diachronic division between French texts from before c.1200 and those dating from afterwards. This concerns whether there is widespread licensing of left-peripheral new information focus. Importantly, this is not the only area of the left periphery where there is solid evidence for diachronic change. Since the seminal work of Benincà (1995) which applied emerging insights on the cartography of the left periphery to comparative medieval Romance data, the concept of a ‘relaxed’ V2 system has had a firm foothold in studies of medieval Italo-Romance varieties in particular. In the early Italo-Romance texts, as already touched upon above, the finite verb is assumed to move to a very low leftperipheral position, either Fin⁰ or Focus⁰, with the result that a full range of hierarchically ordered left-peripheral projections are free to host constituents (Benincà 2004: 275; Lombardi 2007: 124, 137f.; Poletto 2006: 263f., 2014: 16; Ledgeway 2007: 124–42, 2008: 439; Salvi 2012: 105; Wolfe 2015d: 16). Descriptively, this yields a grammatical system where orders where the verb is third, fourth, or greater are relatively numerous: Caim del so lavor ofria (11) a. Or now Cain of-the his work offer.3. ‘Now Cain offered from his toil . . . ’ (old Piedmontese, Sermoni 139, 19) b. Et dall’ altra parte Aiaces era . . . and on-the other side Ajax be.3. ‘On the other hand Ajax was . . . ’ (old Italian, Rettorica 94, Poletto 2014: 16)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-, ,
19
c. Unu iornu, volendu li monachi edificari one day want. the monks build. una chella, in menzo chillu locu sì era a cell in middle that place be.3. ‘One day, as the monks were want to build a cell, in the middle of that place there was . . . ’ (old Sicilian, DSG 92, Wolfe 2015d: 16) This stands in stark contrast, however, to what is frequently termed ‘old French’. Thus in the highly influential studies of Roberts (1993) and Vance (1997), verbfourth or greater is not discussed at all and V3 is treated as an exception to be explained. It is telling in this respect that a number of scholars have thus sought to produce exhaustive lists of V3-triggers for their datasets in old French (Foulet 1919: 245, 252; Skårup 1975: 435–59; Jensen 1990: 539f.; Roberts 1993: 144; Vance 1995: 183, 1997: 61–5; Wolfe 2016b: 475f.), and, furthermore, that the more formal literature frequently postulates a less elaborate left-peripheral structure for old French V2 than is the case elsewhere in medieval Romance (Skårup 1975: 179; Ferraresi and Goldbach 2002; Labelle and Hirschbühler 2005; Labelle 2007: 302; Simonenko and Hirschbühler 2012). In support of these general conclusions, it is correct that V>4 orders are almost entirely absent in Robert de Clari’s Conquête de Constantinople (c.1205), La Queste (c.1225) and the Grandes chroniques de France (c.1275). Likewise, in keeping with the consensus view in the literature cited immediately above, V3 orders in these texts are qualitatively restricted, involving an initial clause, dislocated element or speaker-oriented or scene-setting adverbial:⁹ quant li message a l’empereur vinrent (12) a. Et and when the messengers to the-emperor come.3. as Franchois, il leur dissent to-the French they them. say.3. ‘And when the messengers of the emperor came to the French, they said to them . . . ’ (Clari 17) b. Quant retornez fu il vit when return. be.3. he see.3. ‘When he had returned, he saw . . . ’ message respondirent (13) a. Adont li then the messengers respond.3. ‘Then the messengers responded . . . ’ b. Et neporec Nostre Sires and nevetheless our Lord ‘And nevertheless our Lord had . . . ’ ⁹ I treat -clauses separately below in §2.4.
(GChron 10, 35)
(Clari 41)
avoit . . . have.3. (La Queste 166, 26)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
20
If, however, we intend to equate ‘old French’ also with the language prior to 1200, there is every reason to think that the descriptive generalizations above and their theoretical ramifications do not hold. Witness the following examples where multiple constituents co-occur before the finite verb which do not belong to the restricted class outlined above for later texts. As highlighted in various works, these bear a striking resemblance to the V3 and V>4 clauses found in old ItaloRomance, evidencing an articulated C-space (Labelle 2007: 303; Donaldson 2012: 1038; Mathieu 2012: 339–41; Wolfe 2016b: 468). empereres Carles (14) a. Li the emperor Charles En cest païs nos in this land us. ‘The Emperor Charles of destroy us’
de France of France est be.3. sweet France
dulce// sweet venuz cunfundre come. destroy. has come to this country to (Roland 16–17)
b. É la grace Deu la dame visitá and the grace God the woman visit.3. ‘And the woman was visited by the grace of God’
(QLR 7)
c. d’ist di in avant, in quant Deus savir from-this day in forward in sofar-as God knowledge et podir me dunat, si salvarai . . . and power me give.3 support.1. ‘From this day forward, insofar as God gives me knowledge and power, I will support . . . ’ (Strasbourg Oaths) Crucially therefore we have an established case of diachronic progression in both these domains: preverbal new information focus is licensed in the earliest French texts, whereas it is far more restricted in the old French prose from the thirteenth century. Relatedly, V>3 orders are highly restricted in thirteenth-century prose, but not in the texts dating from prior to 1200. Both these points caution strongly against generalizing findings on thirteenth century texts to the whole period covered by the label ‘old French’.
2.2.4 Change in the syntax of We have already seen that a number of core word order phenomena show more diachronic variation than is conventionally assumed within old French texts. Another area where this holds true is the syntax of the particle . This particle, derived from Latin ‘like this, thus’, has accrued a vast literature but
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-, ,
21
enjoys no widely accepted analysis.¹⁰ Previous treatments have classed as a simple adverbial (Foulet 1928: 301–3; Skårup 1975: 238f.; Jensen 1990: 472f.), a Topic-continuity marker (Diez 1882: 2060; Fleischman 1991; Reenen and Schøsler 1992: 102; Benincà 1995: 333; Buridant 2000: 508) or a phrasal- or head-category expletive linked to old French’s V2 property (Lemieux and Dupuis 1995: 95; Ferraresi and Goldbach 2002: 18–23; Rouveret 2004: 193–5; Ledgeway 2008: 452–65; Salvesen 2013: 143; Wolfe 2016b: 469f., 474). For an evaluation of these claims, see Wolfe (2018a: 337–41). Note that the simple adverbial analysis fails to capture the sheer ubiquity of across the early French textual records, while the Topic-continuity analysis captures a portion of the data but is directly incompatible with a large number of cases, particularly in later texts. Although the phrasal V2-expletive analysis has a number of merits, in particular the very basic observation that essentially only occurs in environments where V2 is known to be licensed (Vance 1995; Ledgeway 2008; Salvesen 2013; Wolfe 2016b), the facts still remain to be addressed that does have clear discourse-pragmatic correlates in certain texts that need to be captured by any formal analysis. The core claim in Wolfe (2018a) is that the conflicting observations regarding , most of which have both empirical and theoretical merits, can be reconciled under an account where follows an ‘upwards’ grammaticalization pathway in the sense of Roberts and Roussou (2002b). Under such an account, undergoes fundamental syntactic change throughout the course of the twelfth century in particular and performs distinct functions in different texts. First, in the Roland (c.1100) typically occurs as the first constituent of a V2 clause alongside a null subject (180/201 occurrences). This is what we would expect in light of Fleischman’s (1991) analysis, where satisfies V2 as a form of ‘default topic’ in initial position. Her analysis also predicts the low occurrence of overt subjects with , since as a marker of ‘samesubject reference’ in her terms, an overt subject expression alongside should be rendered unnecessary on discourse-pragmatic grounds. Structurally, we observe at this stage that can co-occur with focal elements, such as an informationally direct object and manner adverbials, which suggests it cannot occupy a position higher in the left periphery than SpecFinP. I indicate that can optionally co-occur with frame-setters, topics and foci during this stage in (16). en Imphe (15) a. Reis Vivien si succuras King Vivien help.2. in Imphe ‘You will help King Vivien there in Imphe’
(Roland 3995–3996)
b. Cunquerrantment si finereit li bers conquer. end.3. the nobleman ‘The nobleman would end his life as a conquering hero’ (Roland 2867) ¹⁰ See Marchello-Nizia (1985), Fleischman (1991), Van Reenen and Schøsler (1992, 2000), Ledgeway (2008), and Wolfe (2018a) for review of the relevant literature.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
22
(16) [FrameP (Frame-Setter)[ForceP [TopicP (Topic) [FocusP (Focus) [FinP TopicCont [Fin VFin][TP . . . ]]]] While the analysis of Fleischman (1991) and Van Reenen and Schøsler (1992; 2000) holds up well against the data from Roland, it struggles with a wider sample of data from twelfth-century texts. Looking to Charrette (c.1180) and Thèbes (c.1150) we observe a concomitant decline in -initial clauses and a rise in clauses where is preceded by a clause. While focus-initial clauses with are a rarity in these texts and others from the mid-twelfth century onwards (Marchello-Nizia 1985: 158; Wolfe 2018a: 353; Meklenborg, Chapter 3, this volume), examples where is preceded by a topic or frame-setter become increasingly widespread: si a (17) Pharamont Pharamond have.3 ‘Pharamond said to the king’
dit say.
au to-the
roi king (Thèbes 230, 7091)
mialz li pot, (18) Mes, quant respondre but when respond. better him. can.3 Si li a dit him. have.3. say. ‘But once he could best respond to him, he said . . . ’ (Charrette 27d, 94–5) Given that constituents occupying the Frame-Topic layer, in contrast to those in Focus, are taken by Benincà (2004) and Benincà and Poletto (2004) to be base-generated in their left-peripheral field rather than moved there, Wolfe (2018a: 353) formulates the hypothesis that in certain texts from the mid-twelfth century onwards satisfies V2 when no other V2-satisfier can reach the left periphery through movement.¹¹ (19) [FrameP (Frame-Setter)[ForceP [TopicP (Topic) [FocusP [FinP V2Exp [Fin VFin][TP . . . ]]]] So far we have seen that prose texts from the very end of the twelfth century onwards show a number of syntactic characteristics that separate them from earlier texts in both prose and verse. This is also the case with . In Clari and La Queste is preceded by a clause in a large number of cases (32.25% and 59% of cases in the 2018 corpus respectively). Moreover, direct-object fronting with is entirely absent in both texts, which is telling in formal terms as object-fronting is taken to target positions within the Topic-Focus layer in the cartographic ¹¹ See Holmberg (2015, 2019) for elucidation of the claim that only moved items can satisfy V2.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-, ,
23
literature (Rizzi 1997: 286f.; Benincà 2001: 39–50; Benincà and Poletto 2004: §3–4). Finally, while does still co-occur with adverbials in Clari, these are of a homogeneous frame-setting class in that they anchor the clause in terms of locative or temporal coordinates (Poletto 2000: 100; Benincà and Poletto 2004: 66; Öhl 2010: 62). departirent (20) a. puis si se then . leave.3. ‘Then they departed’ b. Adont si atorna then return.3. ‘Then the king returned’
li the
(Clari 4, 2) rois king (Clari 19, 20)
In Wolfe (2018a) it is proposed that these data lend themselves to an analysis whereby, in certain thirteenth-century prose texts, can only be preceded by constituents lexicalizing the Frame-field of the left periphery, such as clauses and frame-setting adverbials. This follows from ’s upwards grammaticalization (Roberts and Roussou 2002b) within the C-domain from a FinP to a ForceP V2-expletive, with concomitant loss of its original pragmatico-semantic featural content encoding thematic continuity.¹² (21) [FrameP (Frame-Setter)[ForceP V2Exp [Force VFin] [TopicP [FocusP [FinP [TP . . . ]]]] To conclude, we have identified another area where data often treated as heterogeneous are nothing of the sort and show clear evidence of an upwards grammaticalization pathway for the particle . Hence, once again we are led to identify clear diachronic change in a period conventionally periodized as a single entity.
2.2.5 Change in the subject system In contrast to the other phenomena surveyed in this section, the core observation that the null subject system of early French changes fundamentally is well established. The syntactic system instantiated in the earliest French texts is that of a full null subject language in the sense of Rizzi (1986b). That is to say, null subjects are licensed in a wide range of matrix and embedded clauses (Adams 1987a; Dupuis 1988; Vance 1988; Roberts 1993: 136–47), and the presence of a null subject in a matrix clause can yield a surface verb-initial order
¹² See in particular Van Gelderen (2009) on the role of featural-loss in grammaticalization.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
24
(Hirschbühler and Junker 1988; Roberts 1993: 94–103; Labelle and Hirschbühler 2005: 62; Simonenko and Hirschbühler 2012: 30; Zimmermann 2014: 36; Wolfe 2016b: 470f.): Rollant einz qu’il fust mort (22) a. Se veïssum if see.1.. Roland before that-he be.3.. dead ‘If we saw Roland before he was dead . . . ’ (Roland 1804) b. Getet le a terre throw.3.= it. to ground ‘He throws it to the ground . . . ’
(Roland 464)
This system referred to by Roberts (1993: 135) as ‘Conservative Old French’ changes fundamentally from approximately 1180 onwards. It is in prose texts from this date onwards that we observe the progressive attestation of the wellknown asymmetric null subject system where null subjects show a far wider distribution in matrix clauses than in embedded clauses (1987c: 3; Adams 1987b; Roberts 1993: 139; Vance 1997: chap. 5; Poletto 2013: 160). This coincides with a rapid fall in the proportion of V1 licensed (Skårup 1975: 298; MarchelloNizia 1980: 331; Vance 1997: 32; Rouveret 2004: 193–5). This latter point has been demonstrated quantitatively by Simonenko and Hirschbühler (2012) who show through corpus analysis that V1-clauses start to decline after 1160 and are almost entirely absent after 1190. Thus we can conclude that yet another major distinction between the texts prior to approximately 1190 and the majority of those composed after concerns the status of their null argument system.
2.3 Occitan and early Gallo-Romance syntax 2.3.1 Common syntactic properties I here outline various respects in which the earliest Occitan texts available pattern in a homogeneous fashion with those available from other Romance languages (Wolfe 2016b: 466–71) and, in particular, French prior to approximately 1200 (henceforth ‘early old French’). Considering first the preverbal field, in a parallel fashion to early old French, we find clear evidence that a range of focal (23) and thematic (24) elements can satisfy the V2 constraint operative in old Occitan (Donaldson 2015; Donaldson 2016; Wolfe 2017, 2018c; Vance, Donaldson, and Steiner 2009). The following examples are taken from twelfth-century texts, which are not numerous for Occitan:
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-, ,
25
(23) a. Arnals Dutranz e sei fraire donero tot . . . Arnal Dutranz and his brothers give.3. all ‘Arnal Dutranz and his brothers gave all . . . ’ (Chartes 20, 1120) b. De Guimerra porta semblan of Chimera bear.3. resemblance ‘She bears resemblance to Chimera’ (La Puta, 1130–1150) c. Perte avas loss have.2. ‘You have lost . . . ’
fete make. (Aigar et Maurin 48, 1175–1200)
fo traita de la carta . . . (24) a. Aquesta carta this charter be.3. linked of the chart . . . ‘This charter was linked to the one that . . . .’ (Chartes 32, 1140) b. Tot aizo voz do all this you. give.1. ‘I give you all this for my soul . . . ’
per for
m’arma . . . my-soul (Chartes 60, 1150)
We saw that in the case of French, the ability to license preverbal new information focus also correlates with V>4 orders which arise typically when a focal and thematic element coincide (cf. Benincà 2004, 2006 in particular). Such orders are also licensed in twelfth-century Occitan prose and verse: sobre tot aizo per amor de Deu e de sant (25) a. E and above all this for love of God and of saint Antoní donam . . . Anthony give.1. ‘And on top of all this, for the love of God and for Saint Anthony, we donate . . . ’ (Chartes 41, 1143) b. [Co]ntra [l’i]vern que s’e[n]anasa ab cossi[ri]er against the-winter that .=advance.3 with perplexities que m as[sa]lh, m’es belh [q]ue . . . that me.cl assail.3 me.=be.3. good that ‘With winter advancing and perplexities assailing me it is good that . . . ’ (L’ivern, 1130–1150) Furthermore and in keeping with these other findings, the earliest Occitan texts correspond to Roberts’ (1993: 135) ‘Conservative Old French’ null argument system.¹³
¹³ See also Wolfe (2016b: 470f.) for discussion of these V1 structures across early medieval Romance texts.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
26
That is to say that in early poetry and the twelfth-century Chartes, we find both verb-initial matrix clauses and embedded null subjects: tot lo quart del blat que avia (26) a. Laiset leave.3. all the section of-the grain that have.3. a Solpiac at Solpiac ‘[She] left all the portion of the land that she had at Solpiac’ (Chartes 92, 1160) b. Fraignent los astes mantenent li donzel break.3. the pikes now the girls ‘The girls now break the pikes’ (Aigar et Maurin 1377, 1175–1200) lo matin, quant uira lo ior clar (27) a. E and the morning when see.3. the day clear ‘And tomorrow when [he] sees the light of day’ (Aigar et Maurin 424, 1175–1200) b. E cel qui plus l’am acuillir, Quan venra and he who most it-love.3 amass. when come.3. al derrier badaill to-the last breath ‘And he who most likes to amass it, when [he] comes to his final breath . . . ’ (Emperaire 1130–1150) The final area of the grammar that we examined above as showing evidence for diachronic change in early French was the particle . Crucially, this only occurs in certain Occitan texts and is not attested in any full-length texts from the eleventh or twelfth centuries. I therefore save discussion of this part of the syntactic system until §7.4. We have seen from this brief qualitative survey that even from the sporadic evidence available, early Occitan texts pattern similarly to early French texts in their V2 syntax. Thus, as a notion taken in isolation, the concept of an ‘early old GalloRomance’ V2 syntax is not an implausible one, if used in a similar sense to Benincà’s (2004: 245) ‘abstract “Medieval Romance” syntax’ which is based on syntactic points of commonality. As we begin to examine textual evidence from the thirteenth century onwards, however, this notion becomes increasingly unsustainable.
2.3.2 Occitan and the late medieval split I now compare and contrast the findings presented above for change in early French against the available data available for Occitan after approximately 1200.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-, ,
27
This period is important as earlier work on thirteenth- and fourteenth-century Occitan has suggested that it diverges syntactically in certain respects from French of the same period (Vance, Donaldson, and Steiner 2009; Donaldson 2016; Wolfe 2017, 2018c). While the analysis of a number of texts will show this is true to an extent, I argue that there is greater continuity in the later medieval Gallo-Romance textual record than has previously been claimed (for example by Wolfe 2018c).
2.3.2.1 Verb placement Although Vance, Donaldson, and Steiner (2009) and Donaldson’s analysis (Donaldson 2015, 2016) of old Occitan as a ‘relaxed’ V2 language is based on a relatively large corpus of texts, Wolfe (2016b, 2017, 2018c) has used the late thirteenth-century text Douceline as exemplary of the later old Occitan textual record. This text has various merits, including the fact that its region of composition is known as Provence and that its religious narrative style is comparable to many of the other Gallo-Romance texts of the period on which major studies of word order are based. Significantly, Douceline diverges markedly from these very same old French prose texts which, as we saw above, instantiate a strict V2 syntax (§7.2.2), in that V1, V3 and V>4 orders are widely attested (Table 2.2). Table 2.2. Verb placement in Douceline Verb Position
N
%
V1 V2 V3 V4 V5 V6 Total
47 328 185 50 8 4 622
7.56% 52.73% 29.74% 8.04% 1.29% 0.64% 100%
This table is reproduced from Wolfe (2018b: 69)
Therefore, alongside verb-initial structures such as (28), found elsewhere in the Occitan textual record (Vanelli, Renzi, and Benincà 1986: 53; Romieu and Bianchi 1999; Kunert 2003: 196f.; Sitaridou 2005: 366–9), which are traditionally analysed as featuring a null subject or topic in the prefield (Benincà 2004: 290; Poletto 2014: 21–3; Salvi 2012: 106f.; Wolfe 2015e), we also find cases where a frame-setter, thematic expression, and focus co-occur, yielding orders where the finite verb is fourth or greater in the linear ordering: e queria luechs solitaris (28) Amava love.3. and want.3. places solitary ‘She loved and wanted places where she could be alone’
(Douceline 107)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
28
adoncs illi, ab amars critz, (29) E and thus she with bitter cries ‘And she thus cried in a bitter tone . . . ’
dizia . . . say.3. (Douceline 136)
As already noted, V1 and V>3 orders are often considered typical of the relaxed V2 status of old Occitan and the data from Douceline are therefore unsurprising. However, to test how typical such orders are in comparison with other texts, I undertook a small pilot study on verb placement in Damiette (c.1250) and Jerusalem (c.1473), looking at 200 matrix clauses in each text. Although the sample size is not directly comparable with that for Douceline above, the results are nevertheless notable in that V1 and V>4 orders are attested but to a lesser extent than in Douceline. To ensure better comparability, the results from the pilot are presented alongside the first 200 clauses of the analysis for Douceline in Table 2.3. Table 2.3. Verb placement in medieval Occitan V1 Damiette 1250 Douceline 1300 Jerusalem 1473
5 17 3
3% 9% 2%
V2 167 111 128
84% 56% 64%
V3 25 50 65
13% 25% 33%
V4 3 18 4
2% 9% 2%
V5 0 3 0
0% 2% 0%
V6 0 1 0
0% 1% 0%
Total 200 200 200
Looking at Damiette, we might hypothesize, pending a qualitative analysis below, that it constitutes a form of ‘halfway-house’ between Douceline, on the one hand, and thirteenth-century French prose on the other, in licensing V1 and V>4 orders which are generally absent from French prose texts of the period, but to a lesser extent than is found in Douceline and medieval Italo-Romance varieties. Our later text, Jerusalem is firmly within the period of composition that would be classified as ‘middle’ in the French tradition (Marchello-Nizia 1980: chap. 1; Smith 2002). It is established that the strict constraints on V>3 and V>4 orders operative in French for much of the thirteenth century are not applicable to middle French texts (Roberts 1993; Vance 1995), but it is telling in comparison that V1 orders are found to a limited extent in Jerusalem when they are absent from the French of the same period (Simonenko and Hirschbühler 2012). Pending more detailed analysis below, we can therefore surmise that Douceline may not be entirely typical of the early Occitan textual record, with wide-ranging implications as to how we formally analyse and periodize early Gallo-Romance.
2.3.2.2 Discourse-pragmatics and the prefield I now consider the prefield in these three texts in detail. There are two major hypotheses which we could entertain on the basis of previous research. First, the texts could show a retention of the common early old Gallo-Romance system
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-, ,
29
outlined in §7.2 and §7.3, whereby focal constituents as well as the universally attested thematic and frame-setting constituents occupy the prefield. This would be consistent with a number of earlier claims in the literature about thirteenthand fourteenth-century Occitan (Donaldson 2015, 2016; Wolfe 2018c) and would mean that these Occitan texts have a similar grammar to that reported for early Italo-Romance varieties (Benincà 1983a, 1995; Lombardi and Middleton 2004; Ledgeway 2007; Cruschina 2011; Wolfe 2015d). A second alternative hypothesis would be that the Occitan texts examined no longer license left-peripheral information focus. This would be consistent with the observations made in §7.1.2 that contemporary Occitan varieties do not license focus-fronting operations and would also reflect recent findings on medieval Venetan varieties which lose the property of focus-fronting in certain dialect areas in the fourteenth century (Wolfe 2018b: 43–5). The findings are more consistent with the first hypothesis. Indeed, while unsurprisingly the prefield can be lexicalized by thematic constituents in all three texts examined, focal constituents can also satisfy V2 in all three of the later texts, in contrast to the generally observed pattern for French texts of the same period (cf. §7.2.3). (30) a. Cascun jorn, aquist verge each day this virgin ‘This virgin went every day . . . ’
annet . . . go.3.
b. Aquil de la ciutat viron thos of the town see.3. ‘Those of the town saw the bodies . . . ’
(Douceline 44) los the
c. Aquelas cozian e manjavon those cook.3. and eat.3. ‘They cooked those and ate them’
cors . . . bodies (Damiette 537, 550–551) las them
de la ciutat de (31) a. Uns homs fon a man be.3. of the town of ‘There was a man from the town of Digne . . . ’
(Jerusalem 32) Dinha . . . . Digne
b. Mas mosenher Savarix de Malleo era vengutz but monsenher Savarix de Malleo be.3. come. ‘But M. Savarix de Malleo had come . . . ’ (Damiette 535, 495–496) c. Mal cosselh donet bad advice give.3. ‘Pilate gave bad advice’
Pilat . . . Pilate (Jerusalem 41)
In order to support these qualitative observations, a parallel analysis was undertaken to complement that presented for later medieval French texts in
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
30
Table 2.4. Discourse-pragmatic values of preverbal constituents New Damiette 1250 26 Douceline 1300 16 Jerusalem 1473 20
26% 16% 20%
Accessible 23 8 16
23% 8% 16%
Active 26 29 36
26% 29% 36%
Pro 15 32 20
15% 32% 20%
Q 10 15 8
Total 10% 15% 8%
100 100 100
Table 2.1. Tellingly, in all the texts in the 100-clause sample were tagged as having a greater proportion of discourse- constituents in initial position that in French texts of similar dating (Table 2.4). As noted above for the parallel French analysis, the corpus analysis is small and should be treated as a springboard for future research. Nevertheless, the findings offer pause for thought on the grounds that (i) a greater incidence of initial foci has been reported for early Occitan by a number of scholars (Kunert 2003: 200; Donaldson 2015: 164f., 2016; Wolfe 2018c: §2.2.2) and (ii) the corpus corroborates qualitative findings outlined above. We can therefore reach the tentative conclusion that, based on the evidence available, there is greater continuity in the syntaxdiscourse mapping for Occitan from the eleventh to fifteenth centuries than is the case for French, where major changes are attested.
2.3.2.3 Verb-third Orders where two or more constituents precede the finite verb are a topic of major interest within the literature on medieval Romance. For scholars such as Benincà (2004, 2006), Poletto (2006, 2014: chap. 1), and Salvi (2000b, 2012), V3 and V>4 orders have been viewed as an inevitable result of the coalescent properties of a V2 syntax and an articulated left-peripheral structure. We have already seen that early old French and early old Occitan are clear instantiations of such a system with a range of V>3 structures attested evidencing an articulated left periphery. Here, we set out to establish whether these V>3 structures show continuity across the medieval textual records as has been suggested for southern Italo-Romance by Wolfe (2016b: §3.1; 2018b: 50–6) or whether there is major discontinuity as is found for French (§7.2.3).¹⁴ If we first consider Douceline, the patterns attested appear identical to those found in the earliest Occitan texts surveyed in §7.3.1. As Tables 2.2 and 2.3 indicate, V3 accounts for a substantial proportion of the text and V4 is also found to a degree comparable with the more ‘relaxed’ V2 systems elsewhere in Romance. While certain V3 cases involve an initial frame-setting clause or adverbial, there are others which show that constituents lexicalising the ¹⁴ This is also implicit for old Italian in the case of Poletto (2005; 2006; 2014), where no diachronic variation, aside from the eventual loss of V2, is discussed for the medieval period.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-, ,
31
Topic-Focus layer can also co-occur, as in the earliest French texts (Labelle 2007: 303; Donaldson 2012: 1038; Mathieu 2012: 339–41): adoncs el conoc (32) E and thus he know.3. ‘And he thus knew the shame’
la the
vergonha . . . shame
Senhor lur (33) Illi per amor del she for love of-the Lord them. ‘For the love of the lord, she washed their feet’
(Douceline 46) lavava los pes wash.3. the feet (Douceline 45)
These qualitative observations on this text from Provence from the very end of the thirteenth century also extend to the troubadour bibliographies (thirteenth and fourteenth centuries) and Jerusalem (c.1473). Alongside the frame-setter-initial examples such as (34), cases such as (35) are also attested where we observe a thematic and focal constituent co-occurring: ja per la cieutat non pogratz anar (34) a. E and already through the city can.2. go. ‘And now you could not go through the city . . . ’ (Jerusalem 41) b. Cant auziro aquest cossel, mot foron dolens when hear.3. this advice much be.3. sad ‘When they heard this news, they were very sad . . . ’ (Jerusalem 35) (35) a. En
Savaric per far a luy honor Savaric to make.inf to him honour li mostret . . . him.= show.3. ‘Savaric, in order to honour him, revealed . . . ’
(Vidas 432, 3)
b. Pilat, can o auzi, mot ac gran fereza Pilate when it.= here.3. much have.3. great ferocity ‘Pilate, when he heard this, was very angry’ (Jerusalem 34) The particular case of Damiette (c.1250) is somewhat ambiguous. Alongside V3 cases which are clearly not of the + type (36), there remain examples which could either be V3 or V4 depending on how one subdivides the relevant constituents (37): cant li Roman gitavan, li Sarrazin issiron . . . (36) E and when the Romans fire.3. the Sarracens come-out.3. ‘And when the Romans fired, the Saracens came out . . . ’ (Damiette 527)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
32
(37) e .j. jorn, cant li Turc assalhiron a las lizas, and one day when the Turks assail.3. to the fields li Sarrazin issiron . . . the Sarracens come-out.3. ‘And one day when the Turks besieged the fields of battle, the Saracens came out..’ (Damiette 526) As such we have a situation where certain texts show evidence of a rich leftperipheral structure and a single text where the evidence is ambiguous. Our handsearch of the entirety of Damiette, Douceline, the Vidas, and Jerusalem suggests, however, that with the exception of Douceline V>4 orders and V3 orders triggered by co-occurring and elements are rare. Note in this regard the pilot findings in Table 2.2, where V4 is found in all three texts but is noticeably more frequent in Douceline. The overall importance of the findings in this section is that the textual evidence from the thirteenth to fifteenth centuries available for Occitan is neither directly comparable to that found in the earliest Gallo-Romance texts nor to that found in thirteenth-century French onwards. In fact, the data are qualitatively similar to the early old Gallo-Romance texts, but quantitatively more restricted. In this regard Occitan post-1200 can be conceived as a form of ‘halfway house’ between the original grammar found in the earliest texts and the innovative grammar of later old French.
2.3.2.4 Null arguments We observed above for French that the null argument system changes quite drastically from 1180 to 1200, with verb-initial matrix clauses disappearing almost completely during this period and embedded null subjects also declining in all but a small number of environments (see Roberts 1993, Vance 1997, and references in §7.2.5). It is worth noting in this regard the apparently split in the literature concerning the correct classification of the null argument properties of old Occitan; Vanelli, Renzi and Benincà (1986: 53, 58f.) classify ‘Provençal’ varieties as showing asymmetric null subjects and restricted attestation of V1, while others such as Sitaridou (2005: 369) and Wolfe (2015b: 12) have argued for relatively widespread V1 and a symmetrical null subject system (Wolfe 2018b: 84), Sitaridou on the basis of the thirteenth- and fourteenth-century Chartes, and Wolfe on the basis of Douceline. The main findings for Douceline stand in relatively stark contrast to French. As Table 2.2 shows, V1 cases make up over 7% of matrix clauses. This figure is broadly comparable to findings for a number of Venetan and Sicilian texts discussed in Wolfe (2018b: chaps. 3, 4), where such clauses across Romance are analysed following Benincà (2004: 289) and others as featuring a null topic. See in this regard (28) as well as (38):
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-, , (38) Fugia totz plazers temporals avoid.3. all pleasures temporal ‘She avoided all temporal pleasures’
33
(Douceline 270)
Furthermore, null subjects, unlike French of the same period, show a symmetrical distribution and are readily licensed in a range of embedded environments, so thus pattern with early old French and Occitan in this regard (Dupuis 1989 and references in §7.2.5): (39) a. E adoncs li sancta maire, per consell del an thus the saintly mother by advice of-the vole que elegissan mejana want.3. that choose.3.. moderate ‘So the Holy Mother, following the advice of the Holy them to choose [a life of ] moderation and poverty’
sant paire, saintly father paupertat poverty Father, wanted (Douceline 82)
b. . . . e ell . . . am gran dolor de cor dels mals and she at great pain of heart of-the bad que fach avia that do. have.3. ‘And she . . . had great sorrow in her heart for the bad that she had done’ (Douceline 261) Moving on to consider our other two core texts, we see the pattern in Douceline replicated. That is to say, as well as appearing in the complements of so-called ‘bridge’ verbs (40a)(41a) (Vikner 1995: 65–86) where null subjects even in post-1200 French texts are attested (Vance 1988; Roberts 1993: 183–6), we find null subjects in embedded clauses where they would not be typical of French after 1200 (40b,c)(41b,c): disseron que petit avian de gens a cavalh (40) a. e and say.3. that few have.3. of people at horse ‘And they said that [they] had few cavalry left’ (Damiette 532) b. e de gens menudas si perderon tans que and of people humble . lose.3. such that non say nombre know.1 number ‘And they lost so many humble people, that [I] cannot say how many’ (Damiette 533) c. E cant li Sarrazin viron que ren and when the Saracens see.3. that nothing no y podian far . can.3. do. ‘And when the Sarcens saw that [they] could do nothing . . . ’ (Damiette 536)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
34
que en totas guizas volia que (41) a. Pilat dis Pilate say.3. that in all ways want.3. that ‘And Pilate said that in all respects [he] wanted that . . . ’ (Jerusalem 35) b. E manda Pilat que parlar volia amb elh and send.3. Pilate that speak. want.3. with him ‘And Pilate send orders that [he] wanted to speak with him’ (Jerusalem 36) c. Pilat, cant auzic Pilate when hear.3. ‘Pilate, when [he] heard this’
ayso . . . . this (Jerusalem 39)
Although I do not discount the possibility that analysis of further texts could lead to support for Vanelli, Renzi, and Benincà’s (1986) classification of old Occitan as an asymmetric null subject system, my finding is that it shows a similar symmetrical system to the earliest French and Occitan texts, with V1 also likewise attested (§7.4.1).
2.3.2.5 Particle In our discussion of old French above, we saw that this particle, frequently described in the literature as a semi-homogeneous entity (Marchello-Nizia 1985; Fleischman 1991; Ferraresi and Goldbach 2002) does not warrant such an analysis.¹⁵ In particular we observed, alongside a host of other changes, that the constituents preceding become increasingly restricted diachronically along a cline which we can schematize as follows:¹⁶ (42) XPFocal(Argument/Non-Argument) > XPTopic(Argument/Non-Argument) > XPFrame-Setter(Non-Argument) Thus, it was observed that fronted focal elements such as informationally objects and VP-adverbials are only found in the very earliest French texts (Marchello-Nizia 1985: 158), thematic constituents are found in a variety of texts up to (and in some cases after) 1200, and in certain thirteenth-century prose texts only clauses or certain classes of high adverbials can precede . How then does early Occitan compare to French in this regard? The first central observation to make is that in contrast to the ubiquity of in the early French textual record (Foulet 1928: 300; Marchello-Nizia 1985: 2; Fleischman 1991: 261; Buridant 2000: §408), Occitan is relatively rare. ¹⁵ But see Reenen and Schøsler (1992, 1993, 2000) for discussion of diachronic, regional and intertextual variation. ¹⁶ There is likely further diachronic variation within the topical/thematic XP category, discussed in detail by Meklenborg (Chapter 3, this volume).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-, ,
35
Meklenborg (this volume) refers to the ‘scarcity’ of the -construction in old Occitan and this observation is also echoed in Wolfe (2018c: §2.2.4). Thus, of the core text used in this chapter, only the Vidas and Razos make extensive use of . In this text, as observed by Donaldson (2015), Meklenborg (this volume) and Wolfe (2018c), can readily be preceded by subjects encoding -information and can also be preceded by multiple constituents within an articulated leftperipheral space.¹⁷ qe Gauselm Faiditz (43) Una sason si fo a time be.3. that Gauselm Faidit s’ entendet lonjamen .= desire.3. long. ‘There was a time when Gauselm Faidit long desired . . . ’ (Razo 167, 40, Donaldson 2015: 176) en aquesta [a]legressa, lo marques de Monferrat (44) a. E and in this happiness the marquis of Monferrat si se croset .= cross.3. ‘In the midst of this happiness, the Marquis of Moferrat crossed himself ’ (Vidas 167, 33) b. Peire Vidal per la mort del bon comte Raiom Peire Vidal on the death of-the good Count Raimon de Toloza si se marri molt of Tolouse .= sadden.3. much ‘Peire Vidal was greatly saddened on account of the death of the good Count Raimon of Toulouse’ (Razo 364, 16, Donaldson 2015: 174) Commenting on the discourse-pragmatic value of the initial constituent, Donaldson (2015: 173) remarks ‘[t]he fact that nearly all the vidas and a number of the razos introduce the name of the troubadour in question through Sn-si-V [Sn = ‘nominal subject’-SW] suggests that a major function of this structure is to present a new discourse referent’. Although as we saw above French does license foci in initial position of clauses at an early stage, this generalization about thirteenth and fourteenth-century Occitan stands in stark contrast to the later old French prose where the reverse observation holds, namely that passages of text never begin with (Marchello-Nizia 1985: 25; Ménard 1988: 328; Lemieux and Dupuis 1995: 96; Vance 1997: 54; Reenen and Schøsler 2000: 86;
¹⁷ Crucially Donaldson (2015) considers a new-referent to still be a form of initial-topic, an analysis I do not adopt here or in Wolfe (2018a) on which the comparison with French is based.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
36
Buridant 2000: 508; Bonnard and Régnier 2008: 209). Furthermore, although there is not yet a quantitative study of the frequency of clauses such as (40) where stands in third-or-greater position of the clause, it is important to note in comparison with French that -third orders decline with frequency diachronically and represent a pattern which is more typical of ninth-twelfth century texts (Wolfe 2018a: 348, 355). The overall significance of these data should be considered cautiously in light of their restricted attestation. Nevertheless, it is potentially telling that certain thirteenth and fourteenth century Occitan texts pattern with the earliest old French texts but not those from 1200 onwards in the syntax of .
2.4 Conclusion 2.4.1 Empirical summary i. From the earliest textual attestations until the end of the fifteenth-century, when our corpus analysis ends, both French and Occitan show the hallmarks of a V2 syntax. ii. There is robust evidence that the possibility for French to license CP-new information focus declines during the medieval period, whereas in Occitan initial foci appear to be licensed throughout the period examined. iii. T + F orders and V>4 are licensed in the earliest texts in both French and Occitan. They disappear completely in the majority of French prose after 1200, are still widely attested in the Provençal text Douceline (c.1300), and are attested but sporadically in other thirteenth and fourteenth century Occitan texts. iv. Verb-first matrix clauses are widely attested in early old French and Occitan. This system is maintained in Douceline. They are still attested but decline in frequency in French prose from 1160–1190 and in the majority of later medieval Occitan texts. In French texts after 1190 verbinitial matrix declaratives are almost entirely absent. v. In the earliest French and Occitan texts there is a symmetrical system of null subject licensing in matrix and embedded clauses. While this system appears to be maintained in Occitan throughout the medieval period, the asymmetric system characterizes French after approximately 1200. vi. In the earliest French texts the particle can be preceded by focal and topical arguments and non-arguments. While French sees the progressive loss of topical arguments and then arguments in general to precede , the small number of thirteenth and fourteenth-century Occitan texts featuring the particle show a system cognate with the original French one, where topical and focal verbal arguments can precede the particle.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-, ,
37
2.4.2 Consequences We began by noting that there is no established account of how syntactically similar French and Occitan were during the medieval period; the data and analysis offered so far go some way towards addressing this issue. The broad generalization on the basis of the textual evidence is that both French and Occitan are strikingly similar in many aspects of their clausal syntax in the earliest available texts, but that French diverges from this early norm in a number of respects as the Middle Ages progress. In formal terms, we can sketch the main features of the shared early syntactic system as follows. Finite verb movement targets a low left-peripheral head, Fin⁰, in line with common assumptions about the nature of the V2 constraint outlined in §7.2.3. There are varying strategies available to satisfy the phrasal merger-triggering Edge Feature which forms the other half of the V2 constraint (Roberts 2004; Holmberg 2015; Wolfe and Woods forthcoming): an overt thematic or focal constituent can undergo fronting into the left periphery, the particle can be merged in a low position within the hierarchy, likely SpecFinP (Wolfe 2018a: §5) or a null topic can satisfy V2, which following Poletto (2014: chap. 1) I take to target a specifier position within the Topiclayer. Crucially under such a system frame-setters, topics, foci and the particle can co-occur in certain structurally constrained contexts, due to the articulated nature of the left-peripheral structure above the finite verb: (45) [FrameP (Frame-Setter)[ForceP [TopicP ((Null) Topic) [FocusP (Focus) [FinP XPV2/ [Fin VFin][TP . . . ]]]] A reanalysis of this system clearly occurs in the history of French. As we have seen, from 1180 to 1200 depending on the phenomenon in question, CP-new information focus is no longer licensed, the possibility for and to cooccur ceases, the syntax of changes so that it is increasingly incompatible with, first, focal and, later, thematic arguments and the null argument system also alters so that verb-initial matrix clauses and embedded null subjects are no longer licensed. These empirical generalizations hold regardless of the formal analysis adopted, but in Wolfe (2016b: 472–8; 2018b: chap. 7) it is argued that these changes can be understood as the result of an ‘upstairs’ shift in the locus of V2 from Fin to Force, such that the left-peripheral structure to the left of the finite verb and the resulting possible combinations of constituents are heavily restricted:¹⁸
¹⁸ For the sake of clarity I do not include lower copies of moved constituents in the schema here and elsewhere in the chapter.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
38
(46) [FrameP (Frame-Setter)[ForceP XPV2/ [Force VFin][TopicP ((Null) Topic) [FocusP (Focus) [FinP [TP . . . ]]]] With these observations in mind, we can turn to the question of periodization, explored in detail for French in Smith (2002) and outlined in §7.1.3. From the narrow perspective of the phenomena discussed in this chapter, is there any value in the traditional labels ascribed to various ‘periods’ in the history of French and Occitan? Looking first to French, the answer should be a cautious positive one. While the textual record is far from homogeneous, it is nevertheless the case that a substantial number of typologically significant syntactic changes are all instantiated in the textual record in a relatively short period of twenty to forty years (~1160–1200). For a historical syntactician’s purposes, having a shorthand to differentiate between the texts from the ninth to late twelfth centuries and those written after is, therefore, of use, suggesting that the ‘early old’ and ‘later old’ labels should not be shunned entirely, if used with appropriate caution. It is less clear that these labels are of use for Occitan, which is unsurprising given that, independent of common external factors, there is no reason a priori why Occitan should undergo major syntactic change in the same time period as French. Furthermore, we have seen that the shared syntactic similarity of the French and Occitan textual records from the first vernacular attestations through to approximately 1200 is striking and could motivate the use of the term ‘early old Gallo-Romance’ in certain domains.¹⁹ Space constraints have precluded a thorough discussion of the syntactic status of ‘middle French’ here. Suffice to say, however, that a large body of scholarship has suggested that the left periphery (Adams 1988; Vance 1995; Lemieux and Dupuis 1995), null argument properties (Hirschbühler and Junker 1988; Hirschbühler 1995), and the overt subject system (Wolfe ms) do change markedly from the late thirteenth century onwards. Although it is not wise to take a stance on periodizing the syntax of a language on the basis of a single text, there is striking continuity between our latest Occitan text, Jerusalem, and those from earlier in Occitan’s textual history, which may in some respects account for the rarity of the term ‘moyen occitan’ observed by Smith (2002: 424). Relatedly, however, Smith (2002: 437–40; 2019), following insights in Lass (2000) and Coseriu (1988), argues for a ‘typological rather than temporal’ (Smith 2002: 441) definition of labels such as ‘middle’ based on internal features of the languages concerned, in particular the extent to which they may diverge from a morphosyntactic prototype of their various families. Consider in this regard the outline schema of the various features discussed in this chapter in a comparative Romance perspective, given in Table 2.5. ¹⁹ Consider the comparative study of Jensen (1990) in this respect, alongside Benincà’s (2004: 245) discussion of a shared ‘abstract “Medieval Romance” syntax’.
Table 2.5. Syntactic typology—medieval and modern Late Latin
Old Sardinian
Old Sicilian, Neapolitan, Florentine
VerbMovement
V/v (Ledgeway 2012)
Fin (Ledgeway 2017c)
Fin (Wolfe 2015f )
EPP-Effects (EF)
V/v (Ledgeway 2012)
V/v (Ledgeway 2012)
Access to Full CP
þ (Salvi 2000b; Spevak 2010; Devine and Stephens 2006)
New Information Focus Verb-Initial Orders
Early Old GalloRomance
Modern Occitan
Modern French
T (Low) T (Ledgeway and Lombardi 2005; Schifano 2015)
T
T (High) (Pollock 1989; Schifano 2018)
Force (Wolfe 2018b: chap. 3)
T
T
T
T (Pollock 1989; Rowlett 2007)
þ
þ
þ
þ
CP
vP/-
vP/-
CP
vP/-
vP/-
(vP)/-
(+)
(+)
þ
+/
Later Old Later Old Occitan French (Post-1200) (post-1200)
Later Old Northern ItaloRomance
Modern Southern and Central ItaloRomance
Fin Fin (Ledgeway 2007; Poletto 2014; Wolfe 2015d)
Fin
Force
Force (Wolfe 2018b: chap. 3)
V/v (Wolfe 2015f )
Fin Fin (Ledgeway 2007; Poletto 2014; Wolfe 2015d)
Fin
Force
þ (Clackson and Horrocks 2007; Ledgeway 2017c)
þ (Lombardi 2007; Wolfe 2015f )
þ þ (Ledgeway 2007; Wolfe 2015d; Poletto 2014)
þ
CP
CP
CP
CP
CP
þ
þ
þ
þ
þ
Modern Northern ItaloRomance
Null Subjects Symmetrical Symmetrical Symmetrical Symmetrical
Symmetrical Symmetrical Asymmetric Asymmetric Symmetrical
þ/
SI
N/A
N/A
N/A
Topic/Focus/ Frame
Topic/ Topic/ (Topic)/ Focus/Frame Focus/Frame Frame
(Topic)/ Frame
N/A
N/A
N/A
N/A
Argument Fronting
þ
þ
þ
þ
þ
þ
þ
þ
þ
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
Classical Latin
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
40
Two related observations are worth making in respect of the table, which is intended to stimulate further research. First, we see that French from 1200 until 1475, where the analysis in this chapter ends, is always more innovative than Occitan, which patterns with the conservative syntactic system of southern ItaloRomance. Second, in the spirit of Ledgeway (2012), we note that if one looks at the encoding of focus, argument-fronting, and the null subject system of Occitan and French today, the same observation essentially holds in modified form, namely that Occitan is syntactically closer to southern Italo-Romance than French is, though the Gallo-Romance varieties form a homogeneous grouping in many respects. Thus, French corresponds to the ‘northern’ Romance prototype more than Occitan. This comment is not divorced from the observations throughout this chapter. Rather, as we have seen, in the early textual records for both Occitan and French there is evidence for both continuity and variation in syntactic terms.
Primary sources Boutière, J. (1964). Biographies des troubadours: textes provençaux des XIIIe et XIVe siècles. Paris: Nizet. Brossmer, A. (1902). ‘Aigar et Maurin, Bruchstücke einer Chanson de geste nach der einzigen Handschrift in Gent neu herausgegeben’. Romanische Forschungen 14: 1–102. Brunel, C. (1926). Les plus anciennes chartes en langue provençale: recueil de piéces originales antérieures au XIIIe siècle. Paris: Picard. Chabaneau, C. (1888–9). ‘La Prise de Jérusalem’. Revue des langues romanes 32: 581–608, 33: 31–46, 600–9. Clari, R de. (1924). Conquête de Constantinople. Paris: Champion. Curtius, E. (1911). Quatre livres des rois. Dresden : Niemeyer. Delfuoco, S. and Bernadi, P. (2005). Sermoni subalpini. Turin: Centro Studi Piemontesi. Gout, R. (1927). La Vie de sainte Douceline. Paris: Ars et Fides. Guillot-Barbance, C., Rainsford, T., and Lavrentiev, A. Séquence de sainte Eulalie. http://txm.bfm-corpus.org (16 April 2018). Marchello-Nizia, C. (2013). Queste del saint graal. http://txm.bfm-corpus.org (12 January 2016). Meyer, P. (1877). La Prise de Damiette en 1219. Paris: F. Vieweg. Moignet, G. (1972). Chanson de Roland. Paris: Bordas. Raynaud de Lage, G. (1966). Roman de Thèbes. Paris: Champion. Taylor, M. (2000). ‘The Cansos of the Troubadour Marcabru. Critical texts and a commentary’. Romania 118: 336–74. Viard, J. (1920). Grandes chroniques de France. Paris: Renouard.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
3 Resumptive structures in a Gallo-Romance perspective Christine Meklenborg
3.1 Introduction Resumption is a strategy in which an element is resumed by another element in the clause. In this chapter, I will use the term resumption for instances where a resumptive particle occurs in second position in a Verb Second (V2) language, yielding a surface V3 structure (1). (1)
[Quant Lancelot l’ entend]1, si2 [en=est]3 tot esbahi when Lancelot him hears of.it=is all surprised ‘When Lancelot hears this he is very surprised’ (old French, MortArtu III-24)
This structure is well known from the old and modern Germanic languages (with the exception of English) (for an overview, see Meklenborg 2020), but is also attested in some old Romance languages such as old French (see among numerous others Skårup 1975; van Reenen and Schøsler 2000; Ferraresi and Goldbach 2003; Wolfe 2018; Salvesen 2013, Meklenborg 2020), old Occitan (Donaldson 2015; Wolfe 2015c, 2018b), old Sicilian (Wolfe 2015c, 2018b), old Neapolitan (Ledgeway 2008), old Florentine (Poletto 2014), and old Venetian (Wolfe 2018b). In is not attested in old Sardinian and old Spanish (Wolfe 2015c, 2018b), so it is not a panRomance phenomenon. In the modern Germanic languages, there are essentially two types of resumption: adverbial and pronominal. Adverbial resumption resumes an initial adverbial element, while pronominal resumption resumes an initial argument. In (2) the adverbial is followed by the semantically bleached resumptive så (litt. ‘so’), while in (3), the initial DP is followed by the weak personal pronoun han ‘he’. Here, the DP corresponds to the thematic subject of the clause. In both cases, the two initial elements are co-referential.
Variation and Change in Gallo-Romance Grammar. First edition. Sam Wolfe and Martin Maiden (eds). This chapter © Christine Meklenborg 2020. First published in 2020 by Oxford University Press.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
42
(2)
[ADV I dag ]i såi ble det kjent at Axel Lund Svindal legger in day SÅ became it known that Axel Lund Svindal lays opp etter VM. up after WC ‘Today it was made known that Axel Lund Svindal will retire after the World Championships.’ (Norwegian)
(3)
[DP θ Axel Lund Svindal]i, hani legger opp etter VM. Axel Lund Svindal HE lays up after WC ‘Axel Lund Svindal will retire after the World Championships.’ (Norwegian)
In Fenno-Swedish, the variety of Swedish spoken in Finland, both structures may be generated by using så, which is linked to adverbial resumption in the other Scandinavian languages (Holmberg 1986, 2020; Meklenborg 2020). (4)
Ja, och [DP θ julgran]i såi har vi nog alla yes and christmas.spruce have we MOD all tider nästan. times almost ‘Yes, and we almost always had a Christmas tree.’ (Fenno-Swedish; Harling-Kranck 1998)
Thus, there are three different ways of generating a resumptive structure (5). (5)
a. thematic DP + pronoun b. adverb + adverbial resumptive c. thematic DP + adverbial resumptive
In old French and old Occitan, all these options are present. In these two languages, (5a, c) are more common than (5b). In this chapter, I will examine resumptive structures involving in both old French and old Occitan. The quantitative weight will be on old French, while old Occitan will be used for comparison. I will argue that is not a homogenous structure, contrary to the underlying assumptions made by previous attempts to analyse it (see among others Fleischman 1991; Vance 1997; Ferraresi and Goldbach 2003; van Reenen and Schøsler 2000; Salvesen 2013). I will argue that it may be both a head and a phrase, and that its antecedent may have reached its surface position by movement or by base generation. Section 3.2 provides a theoretical and methodical backdrop to this chapter. Section 3.3 presents the data from old French including a qualitative study of two thirteenth-century texts. In §3.4, the data from old Occitan are presented, and in §3.5, I compare the information-structural properties of the particle in the two
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
43
languages. In §3.6, I provide an analysis of in old French, while comparing the major findings to old Occitan. Section 3.7 sums up the chapter.¹
3.2 Theoretical and methodical backdrop 3.2.1 Sources In order to retrieve data for old French, I used data from the Base de français médiéval (BFM) in addition to data retrieved manually from La Mort Artu. I also formed a database of -constructions from two thirteenth century texts: La Queste del Saint Graal and Robert de Clari’s La Conquête de Constantinople that serves as the base of a quantitative study. For old Occitan, I used the Canso d’Antioca (Sweetenham and Paterson 2003), a chanson de geste from the twelfth century. Further, I searched for occurrences of in Lo Roman de Flamenca, edited and published Michael McGuire and Olga Scrivner at Indiana University.² Finally, I use data from the so-called Vidas, the biographies of the troubadours (Boutière 1964). I do not present original statistical material from this texts, but build on work done by Donaldson (2015).
3.2.2 Verb Second Both old French and old Occitan are V2 languages, in which the finite verb regularly occupies second position of the declarative main clause (for old French, see among numerous others Thurneysen 1892; Foulet 1930; Skårup 1975; Vanelli et al. 1985; Adams 1989; Roberts 1993; Vance 1997; Rouveret 2004; Labelle 2007; Ledgeway 2012:65; Salvesen 2013; Klævik-Pettersen 2018; Wolfe 2015c, 2018b; for old Occitan, see Jensen 1994; Paden 1998; Donaldson 2015; Wolfe 2015c, 2018b).³ When a non-subject constituent moves to initial position of the clause, the subject (if expressed) occurs postverbally. In (6a) the adverb or ‘now’ occurs preverbally, while the subject, Carles, is postverbal. In (6b), the temporal adverb lonc temps is in clause-initial position, while the subject lor amors is postverbal. ¹ I would like to thank the audience at DiGS 20 in York 2018 and the participants at the seminar on the languages of the Veneto in Oxford in April 2019, where these ideas were presented. Bryan Donaldsson has been a great support and has helped me access the Occitan material. Furthermore, I would like to thank Ans van Kemenade for fruitful discussions about the issues discussed here. A big thank you goes to Silvia Rossi for an enthusiastic exchange of ideas. Finally, I am greatly indebted to Sam Wolfe and Martin Maiden for insightful comments, stimulating discussions, and their meticulous editorial work. All mistakes are my own. ² http://cl.indiana.edu/mpmcguir/flamenca01.php ³ Note that not all authors use the term V2, but merely describe the word order.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
44
(6)
a. Or ad Carles grant ire. now has Charles big anger ‘Now, Charles is very angry.’ (old French, Roland, p. 212, v. 2944) b. Lonc temps duret lor amors long time lasted their love ‘Their love lasted a long time’ (old Occitan, BdV 7)
While thirteenth-century French observed a strict V2 word order, old Occitan has more V2 violations and is labelled a relaxed V2 language (Wolfe 2018b: 77). The definition of V2 that I will use is linked to the obligatory movement of the finite verb to a head in the left periphery (Benincà 2006; Ledgeway 2012; Salvesen 2013; Holmberg 2015; Wolfe 2015c, 2018b). This head is endowed with an EPP feature, which has to be satisfied, either by the movement of the subject into preverbal position, or by any other element, leaving the subject in a postverbal position. The head associated with the EPP feature functions as a ‘bottleneck’ (Haegeman 1996; Roberts 2004) in that no element may move into the left periphery without moving through this position. In other words, any element occurring to the left of the V2 structure must have been merged there (Holmberg 2015). Further, I will assume that there is a visibility requirement associated with V2, namely that the preverbal element must contain overt phonological material. The CP is not understood here as a single position to the left of TP, but rather as a fixed field of designated projections forming a hierarchy. The idea of a split CP was originally introduced by Rizzi (1997), who suggested that the CP must be understood as a field of positions between ForceP and Fin(iteness)P, where elements such as topics, foci and wh-words may move. Rizzi’s model is presented in (7), where the asterisk marks the possibility of having recursive topics. (7)
Force . . . Top* . . . Foc . . . Top* . . . Fin . . .
Rizzi’s model has since been elaborated, but the basic insight has been retained. Benincà and Poletto (2004) have suggested that the left periphery has the following make-up (8), where Topic and Focus must be understood as fields, comprising several subcategories. (8)
FrameP . . . ForceP . . . Topic field . . . Focus field . . . FinP . . .
Examining topics in German and Italian, Frascarelli and Hinterhölzl (2007) have suggested that the left periphery has at least three different topic positions, depending on the information-structural value of the topic (9). The topic positions are divided by a position for focus, in a model that resembles Rizzi’s original proposal.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
- (9)
45
ForceP . . . Shift Topic . . . Contrastive Topic . . . Focus . . . Familiar Topic . . . FinP . . .
Wolfe has exploited the split CP in his seminal work on word order in old Romance (see among several other chapters Wolfe 2014, 2015a, 2015c, 2015g, 2018b). His suggestion is that a language that observes a strict V2 word order has the finite verb under Forceo, while a more relaxed language has the verb under Fino. When the finite verb sits under Forceo, the only positions available for a preV2 element is in Frame, which consists of Hanging Topics and Scene-Setters (Benincà and Poletto 2004). When the verb sits in Fino, all the other positions of the left periphery are accessible to preverbal elements, which may explain word orders where the finite verb occurs in not only second, but also in third, fourth, or even fifth position. In Wolfe’s framework, early old French and old Occitan are Fin-V2 languages, while old French (thirteenth century) is a Force-V2 language (Wolfe 2015c, 2019).
3.2.3 The lexeme S is the most common resumptive element in old French and old Occitan. It is derived from the Latin adverb ‘thus, so’. In the old French texts, it has several homographs: it may be an adverb, the possessive determiner in the third person or an adverbial conjunction. It is very common to find main clauses describing consecutive events introduced by . These clauses often have a null subject (Foulet 1930; Fleischman 1991). The structure in (10) is typical: the initial adverbial clause is followed by two main clauses introduced by . The subject of the fronted embedded clause, Lanceloz, is co-referent with the null subject of the two following clauses. Wolfe (2018a) has shown that clauses with an initial si are more common in the oldest texts, and that their frequency decreases over time. la dedenz], si ferma bien l’ (10) [Quant Lanceloz vint when Lancelot came there inside closed well the uis aprés soi, einsi com aventure estoit, si deschauça lock behind him such that adventure was took.off.shoes maintenant et despoilla et se coucha avuec la roïne now and undressed and lay with the queen ‘When Lancelot came inside, by chance he closed the lock behind him and then he took off his shoes and his clothes and lay down with the queen.’ (old French, MortArtu X-28) As for old Occitan, may be the reflexive pronoun, the stressed object pronoun or an adverb (Anglade 1921; Donaldson 2015). In some texts the conditional adverb se ‘if ’ may also be written .
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
46
Anglade (1921: 367) notes that while the coordinating conjunction is et, there are ‘a few examples’ (quelques exemples) of in this function. This function is not attested in the Antiocha and hardly found in the Flamenca. There are, however, several in the Vidas. I will restrict the discussion of to cases where it occupies the second position between a fronted XP and the finite verb.
3.2.4 Specialized and generalized resumption One can propose a distinction between specialized and generalized resumption (Meklenborg 2020). A specialized resumptive is a resumptive element that is not completely semantically bleached, and where there must be semantic mapping between the antecedent and the resumptive. An example of this may be found in Dutch, which makes a distinction between the temporal resumptives dan and toen ‘then’, where the former is reserved future contexts and the latter is used for events in the past. With a fronted adverbial temporal clause denoting a future event, the resumptive toen is ruled out (11). terugkomt naar Griekenland], dan / *toen moet. (11) [Wanneer je when you back.come to Greece DAN / TOEN must je ons bezoeken you us visit ‘When you come back to Greece, you must visit us.’ (Dutch; Meklenborg 2020) A generalized resumptive, on the other hand, is completely semantically bleached and may be used regardless of the semantic properties of its antecedent. This is the case for the Scandinavian resumptive . du kommer tilbake til Hellas], så må du besøke oss. (12) [Når when you come back to Greece must you visit us ‘When you come back to Greece, you must visit us.’ (Norwegian; Meklenborg 2020) In this chapter, I will argue that in old French may be both a generalized and a specialized resumptive, both a head and a phrase, something which represents a completely new take on the long-standing -discussion.
3.3 Resumption in old French In old French, resumptive structures with are extremely common. However, resumptive disappears along with V2 in the middle French period. Resumptive
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
47
is found in two contexts: After initial adverbial elements (§3.3.1) and after initial arguments (§3.3.2).
3.3.1 After fronted adverbial constituents The most common context for resumptive is after a fronted adverbial clause. This clause is typically temporal (13a) or conditional (13b). [quant il entrerent en la grant sale de leenz], si (13) a. et and when they entered in the big hall of inside encontrerent Morgue la desloial met Morgan the disloyal ‘and when they entered the great hall inside, they met Morgan, the disloyal’ (old French, MortArtu IV-6) b. [Se je en deusse orendroit morir] si le=feïsse je por la if I of-it should now die it=make I for the volenté mon seignor acomplir. will my lord finish ‘If I should die now because of this, I will do so in order to fulfil the wish of my lord.’ (old French, qgraal_cm, col. 161b, l. 12) It is also possible to find after initial PPs (14). d’ or, (14) [Seur chele porte], si avoit un pumel on this door had a pommel of gold ‘On this door there was a golden door handle’ (old French, Clari, p.87) often occurs after initial adverbs, both temporal (15a) and locative (15b). qu’ il s’ (15) a. [Dont] si kemanda li empereres a se gent then asked the emperor to his people that they refl armaissent et qu’ i li=aidaissent armed and that they him=helped ‘Then the emperor told his people to arm themselves and to help him.’ (old French, Clari, p.18) b. [Illueques] si vi un vaslet qui estoit freres a le there saw a valet who was brother to the femme l’ empereur d’ Alemaingne wife the emperor of Alemania ‘There they met a valet who was the brother of the wife of the emperor of Alemania.’ (old French, Clari, p.16)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
48
3.3.2 After fronted arguments S may occur after a fronted argument, normally the subject. This subject may have both an animate (16a) or an inanimate reference (16b), and it may be both nominal (16a–b) and pronominal (16c). (16) a. et [li rois Artus] si savoit bien tant de letres que il pooit and the king Arthur knew well much of letters that it could bien .i. escrit entendre well un text understand ‘and King Arthur knew enought about letters that he could read a text. (old French, MortArtu IV-11) b. Et [cil Tanebors] si estoit .i. chastiax molt forz et mot and this Taneborg was a castle much strong and much bien seant a l’ entree de Norgales. well situated on the entrance of North Wales ‘and this Taneborg was a very strong castle, well situated at the entrance to North Wales.’ (old French, MortArtu II-4) c. Et [cele] si li=demanda qui and this him=asked who ‘And she asked him who he was.’
il he
estoit. was (old French, MortArtu II-7)
While only occurs after an initial subject in the thirteenth century, there are some examples of after other types of arguments in the earliest texts. Wolfe (2018a) explores in seven texts spanning from the eleventh to the thirteenth centuries and only finds seven examples of Non-Subject DP + constructions. S often occurs after a fronted free relative clause or after a DP followed by a relative clause. (17) a. Ains s’= en= ala qui aler rather = of.it= went who go vaut, et [qui vaut] si remest will and who will remained ‘Then the ones who wanted to leave went, remained.’
s’= =
en= of.it=
and whoever wanted to (old French, Clari, p.80)
b. [Cil qui n’ a esté compainz de la queste del saint the.one who has been party of the quest of=the holy Graal] si se= departe de ci, car il n’ est pas droit Grail = leave of here for it is right qu’ il i= remaigne that he here= stay ‘Anyone who has not taken part in the quest for the holy Grail should leave, for it is not right that he stays here.’ (old French, qgraal_cm, col. 223d, l. 31)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
49
c. [La damoisele a qui tu as parlé] si est li anemis li the maiden to whom you has talked is the enemy the mestres d’ enfer cil qui a pooir sor toz autres. master of hell he who has power over all others ‘The maiden you talked to is the enemy of the master of hell, the one who has more power than all others’ (old French, qgraal_cm, col. 186d, l. 12) In order to get an impression of the frequency of the different constructions, I extracted all instances of resumptive in two thirteenth-century prose texts, both accessed through BFM: La Queste del Saint Graal (henceforth La Queste) and Robert de Clari’s La Conquête de Constantinople (henceforth Clari), with results shown in Table 3.1. Table 3.1. Different kinds of antecedents La Queste Antecedent
no
CP PP AdvP DP other Grand Total No. words
Clari %
no.
%
701 3 17 13 –
95.5% 0.4% 2.3% 1.7% –
286 25 86 18 1
68.8% 6.0% 20.7% 4.3% 0.2%
734 107,677
99.9% 0.7%
416 34,013
100.0% 1.2%
We find that the resumptive structure is much more common in Clari than in La Queste. In the former, there are 1158 instances of the -construction in 34,013 words, which gives the ratio of 1.2%. In La Queste, there are only 734 occurrences of resumptive in 107,677 words, which gives us a ratio of 0.7%. The difference is statistically significant.⁴ The frequency is not the only striking difference between the two texts. We see that while fronted adverbial clauses account for 95.5% of all the -structures in La Queste, the amount is only 68.8% in Clari.⁵ While fronted adverbs followed by are hardly attested in La Queste, this structure accounts for more than 20% in Clari.⁶ Further, while fronted arguments followed by are almost nonexistent in La Queste (1.7% of all -structures), they are well attested in Clari (4.5%).⁷
⁴ Likelihood Ratio = 92.497 (p=0). ⁶ Likelihood Ratio = 107.916 (p=0).
⁵ Likelihood Ratio = 152.679 (p=0). ⁷ Likelihood Ratio = 6.294 (p=0.01212).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
50
When we look at the semantics of the antecedent (Table 3.2), we see that adverbs and adverbial clauses are predominantly temporal. However, after fronted PPs, locative antecedents are most common in both texts.⁸ Table 3.2. The semantic value of the adverbial antecedent CP La Queste temporal 687 causal 4 conditional 9 comparative 1 locative manner instrumental Total
Clari
98.0% 285 0.6% 1 1.3% 0.1%
PP
ADV
La Queste Clari
La Queste
99.7% 1 0.33% 0.3%
7 2
Clari
29.2% 13 8.3%
76.5% 78 89.7%
2 0.66% 14
58.3%
23.5%
1
4.2%
4
1 3 4 1
1.1% 3.4% 4.6% 1.1%
701 100.0% 286 100.0% 3 0.99% 24 100.0% 17 100.0% 87 99.9%
The -clauses introduced by a DP may be subdivided into three different cases: simple DPs, [DP + relative clause], and free relatives (Table 3.3). Table 3.3. Different kinds of DP antecedents La Queste antecedent
no.
Clari %
no.
%
simple DP DP + relative free relative
9 2 2
69.2% 15.4% 15.4%
8 5 5
44.4% 27.8% 27.8%
Grand Total
13
100.0%
18
100.0%
Information-structurally, we find that when the antecedent is a simple DP, it conveys or information, normally information that has been introduced in the preceding clause. With the structure [DP + relative clause] we find an equal share of - and - information. When the antecedent is a free relative, it inherently conveys - information that has an unknown referent. Table 3.4 shows these results. These data tell us that there are two triggers for the use of resumptive after a fronted argument: is inserted after a topic shift or after a finite antecedent. We see that in certain cases may be inserted to mark that the subject of the clause is ⁸ This is a rather puzzling result, given that Wolfe (2018a) did not find any locative PPs + in Clari. While this chapter builds on all instances of resumptive in the two texts in question; Wolfe uses a sample of approximately 300 examples from these two texts. His examples include all kinds of -clauses, also the instances where is in clause-initial position. Thus, our data sets differ.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
51
Table 3.4. The discourse value of the nominal antecedent -
-
antecedent
no.
%
no.
%
simple DP DP + relative free relative
0 4 7
0.0% 57.1% 100.0%
17 3 0
100.0% 42.9% 0.0%
different from the previous one, but that it is a - topic. In other cases, occurs after an initial constituent that contains a relative clause, and where the information-structural value may differ. This is very different from the claim made by Fleischman, who says that ‘si ( . . . ) never follows a major break or turning point in the thematic continuity of the discourse’ (Fleischman 1991: 270). We have seen that resumptive may follow fronted adverbial clauses, fronted PPs, fronted adverbs, and fronted arguments. In the majority of cases, the antecedent contains a finite verb—either because it is an adverbial clause or because it is a DP that either contains a relative or that is itself a free relative clause. Kiparsky (1995) has argued that fronted clauses in older stages of Indo-European were not integrated but, rather, paratactic to the main clause. It seems that this observation is valid also for old French. In a cartographic framework these clauses could be associated with Frame, in other words the field preceding ForceP.
3.4 Resumption in old Occitan In old Occitan, resumption is much less common than in old French, and there is widespread intertextual variation. In the Canso d’Antioca, there is only one example of , while examples in the Vidas are numerous. I have not found any examples at all in the Flamenca. All the data below must be read with this in mind. While we have seen that occurs in a vast variety of constructions in old French, in old Occitan, the construction is limited to fronted adverbial clauses and fronted arguments. After a fronted adverbial clause, it is possible to use the resumptive . [quant lo vescons s’ en aperceup], si s’= estranjet (18) E and when the viscount of.it discovered si = alienated de lui e la moillier fetz serar e gardar of him and the wife made bind and watch ‘And when the viscount realised this, he distanced himself from him and locked up his wife and put her under surveillance.’ (old Occitan, BdV)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
52
In the Vidas, almost every text starts with the sequence [ + ] (19). In his study, Donaldson found that 98 of 105 of the Vidas started with the sequence [ + ] + (Donaldson 2015: 172). de Ventador] si fo de Lemoisin (19) [Bernartz Bernhard of Ventador was of Limousin ‘Bernhard of Ventador came from Limousin’ (old Occitan, BdV)
3.5 Discourse function of the fronted argument As we have seen, may follow an initial DP argument in both old French and old Occitan (at least in the Vidas). This option seems to be restricted to subject-initial clauses. There is, however, an important difference between the two languages when it comes to the information-structural value of the initial element.
3.5.1 Old Occitan In old Occitan, one discourse function of the [ + ]-structure is either to mark a topic shift or to introduce a new referent (Donaldson 2015). Both options can be illustrated from the Vida telling the story of Giraut de Borneil. The example in (20) is the opening line of the entire Vida, and Giraut de Borneil is a completely new referent. This - element is followed by (20). de (20) [Girautz de Borneill] si fo Giraut de Borneil was of ‘Giraut de Borneil was from Limousin’
Limozi Limousin (old Occitan, GdB)
The same happens in one of the Razos (‘chapters’) of this story. Giraut has been mentioned before, but not in this particular chapter. (21) [Girautz de Borneil] si Giraut de Borneil qu avia nom N’ who had name Lady ‘Girautz de Borneil loved a Alamanda d’Estanc.’
amava una dompna de Gascoina loved a lady of Gascony Alamanda d’Estanc. Alamanda d’Estanc lady from Gascogne by the name of Lady (old Occitan, GdB, Razo 242,69)
We do not find a similar use of in old French (Fleischman 1991). In a text such as the Chanson de Roland (late eleventh century), 140 of 291 laisses start with a subject, and 113 of these have the structure nominal subject + finite verb. There is
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
53
no example of [ + ] at the beginning of laisses. In the younger Lapidaire (from around 1200), which has a similar structure to the Vidas in that each paragraph treats a new stone (like each Vida presents the life of a new troubadour), the structure [ + ] is not attested. All the chapters follow the same pattern: the name of the stone is introduced first, then followed by a description of its qualities. is never used in second position, but may head the following clause, after the stone has been introduced (22a). In some cases, the writer needs an additional syllable in the opening verse. In those cases, we find ço (22b), and in one case ce. Thus, it seems that in this position is ruled out. (22) a. Alabaustre est une pere, si est de mult tendre manere. alabaster is a stone is of much tender way ‘Alabaster is a stone which is very soft.’ (old French, Lapidfp v.137–138) b. Hyenia ço est une pere ki est trovee en tel manere, hyena that is a stone that is found in such way e dedenz les oilz la trovum d’une beste, hyenie ad num. and inside the eyes it= find of a beast hyena has name ‘Hyena is a stone that is found inside the eyes of a beast that is called hyena.’ (old French, Lapidfp v.1171–1174) Going back to old Occitan and the story about Giraut de Borneil, we find a different use of the structure [ + ]. In the paragraph following the example in (21), we read the story of how Lady Alamanda gives Guiraut a glove, which he loses. Consequently, the lady lets him know that she will no longer stay acquainted with him. Giraut becomes very sad and seeks out one of Lady Alamanda’s maids (23). (23)
Qant Girautz ausi la novella [o]caison e·l comjat qe la domna li=dava, mout fo dolens e trist[tz], e ven[c]=s’=en ad une donzella q’ ell’ avia nom Alamanda, si com la domna. La doncella si era mout savia e cortesa, e sabia trobar ben et entendre. E Girautz si · l=dis so qe la domna li=avia dit . . . ‘When Giraut heard the new accusation and that the lady no longer wanted to see him, he became very sad, and he went to a maiden who was called Alamanda, just like the lady. The maiden was very wise and courteous and knew how to sing and [create songs?]⁹ And Giraut told her what the lady had said to him’ (old Occitan, GdB, Razo 242,69)
⁹ The meaning of entendre in this clause is unclear. Boutière suggests that it might mean ‘create’.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
54
In this paragraph, Giraut is the subject of the first clause, both the embedded clause and the matrix clause that follows. In the second clause, there is a topic shift from Giraut to la doncella. ‘the maiden’, who has been introduced in the preceding clause. This shift is marked by the insertion of after the initial DP. In the third clause, the topic shifts again, and Giraut is once more the subject of the clause. Here too, the shift is marked by the insertion of . We find the exact same pattern in old French. Examining all the [ + ]-structures in La Queste and Clari, we find that in all the cases where the subject is a nominal DP, it is - and marks a topic shift. ai devisees avoies tu en (24) Aprés ces .ii. vertuz que je t’= after these two virtues that I you= have told had you in you pain. Souffrance si est semblable a esmeraude que toi souffrance pain is comparable to emerald that toz jorz est vert. all days is green ‘After these two virtues that I have told you about, you had pain. Pain is comparable to the emerald that is always green.’ (old French, qgraal_cm, col. 189c, l. 18) It must however be noted that is never obligatory, either in old Occitan or in old French. In a later Razo, we find the following sequence (25). (25)
Girautz de Borneil si era partitz del bon rei Anfons de Castella e si · l=avia dat lo reis un mout ric palafre ferran [ . . . ] ; e venia=s’=en en Gasconia, e passava per la terra del rei de Navarra. E·l reis Gascony o=saub qe Girautz era cossi ric[s] e qe passava per la soa terra . . . ‘Giraut de Boreil had left the good king Alfons of Castilla and the king had given him a very good grey palfrey, and he entered into Gascony and rode through the land belonging to the king of Navarra. The king knew that Giraut was rich and that he travelled through his lands.’ (old Occitan, GdB, Razo 242,46)
Again, the new referent Girautz de Borneil is followed by . The king of Navarre is introduced towards the end of this clause and immediately resumed as the subject of the next. This resembles the context where we found in (24), but in this clause, the subject directly precedes the finite verb. In the following example from old French (26), we see that follows the fronted adverbial clause, and that the subject is the same in the subordinate and in the main clause. In the next main clause, there is a topic shift, and Boort is now the subject. This is, however, not marked by .
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
- (26)
55
Et tantost come cil de l’ agait la=choisirent, si la=mostrerent a Boort et a Hestor. Et Boorz dist maintenant a ses compaignons . . . ‘And immediately as the guard saw her [= the banner] they showed it to Boort and Hestor. And Boort then told his companions . . . ’ (old French, MortArtu XIII-16)
In other words, we see that in old Occitan may be used to mark a new topic, an option which is not possible in old French. Both languages, however, use to mark topic shift.
3.6 A structural analysis of old French So far, we have examined the structures where we find resumption from a descriptive and an information-structural angle. A different question is where it occurs in the clausal hierarchy. Different proposals have been put forward in the literature: Fino (Ferraresi and Goldbach 2003), SpecFinP (Salvesen 2013, and Wolfe 2018a for early old French), SpecFocP (Benincà 2006; Donaldson 2012), and SpecForceP (Wolfe 2018a for thirteenth-century French). A first observation is that does not occur in clause-medial or final position (Herman 1954), and it must thus be analysed as a CP-element. Based on data from Ledgeway (2017a), Wolfe (2018a) shows that could occur clause-medially in late Latin. By early old French, he assumes that is base generated in the High Adverb field and obligatorily moved to SpecFinP. In the next stage it has been basegenerated there, before the locus of merge moves to SpecForceP. I will not subdivide old French into different periods, and the analysis will lean heavily on data from thirteenth-century prose. The analysis I will pursue is that may be both a head and a phrase, depending on the environments in which it occurs. In §3.6.1, I will discuss the environments where is disallowed as this will give us an indication of its position in a cartographic framework. In §3.6.2, I will discuss whether is a maximal or a minimal category, and in §3.6.3, whether it is a generalized or a specialized resumptive.
3.6.1 Contexts where resumption is ruled out The first piece of the puzzle stems from the fact that there are certain structures where is ruled out. It is beneficial to examine these structures in order to gain information about the structural position of the resumptive particle.
3.6.1.1 Stylistic fronting Numerous scholars assume that old French has some sort of Stylistic Fronting (SF) (Maling 1990), defined by the fronting of what appears to be a head in the
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
56
absence of an overt subject (Mathieu 2006; Salvesen 2011). This is typically an infinitive or a participle. In the BFM we find a total of 194 occurrences of infinitives and participles in absolute clause-initial position but no instances where the participle or infinitive is followed by .¹⁰ les chevaliers (27) Desarmer fet disarm make the knights ‘[She] had the knights disarmed.’
(old French, CharetteKu, v 2533)
There is no consensus in the literature with respect to the position of SF elements, but they are generally assumed to occupy a low position in the CP field (Salvesen 2009, 2011), or a high position in the TP field (Mathieu 2006; Klævik-Pettersen 2018)—or both depending on the exact structure (Labelle and Hirschbühler 2017). Regardless of the precise analysis of SF-moved elements, must minimally be in SpecFinP or higher. This way, we may rule out Fino (Ferraresi and Goldbach 2003) as a possible position for .
3.6.1.2 Wh-words Resumption is never found after a wh-word in old French (28). Searching all kinds of wh-words + yields no results in the entire BFM corpus.¹¹ This tendency is attested cross-linguistically and seems as such to be a feature of resumption in all V2 languages (for Swedish, see Holmberg forthcoming). as tu (28) Pur quei plures, qu’ for what cries what have you ‘Why do you cry, what has happened?’
eü ? had (old French, brut2 v 14078)
Within a cartographic framework, wh-words are typically assumed to occupy two different positions within the CP field: one high and one low (Rizzi 2001a). The high field typically hosts why, while the lower hosts interrogative operators. Leftdislocated elements occur between the two. In old French, however, left dislocation precedes porquoi ‘why’. Compare the two clauses below. n’ (29) a. Et je por coi et je for what ‘And why don’t I go there?’
i there
partirai go (old French; Salvesen 2009: 218)
¹⁰ The following search string was used in these queries: [pos="PONfrt"][pos="VERppe|VERinf "] [pos="VERcjg"] and [pos="PONfrt"][pos="VERppe|VERinf "] [word="si"]. ¹¹ The following search string was used in this query: [pos="PROint|DETint|ADVint|ADVing"] [word="si"].
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
- b. Pourquoi l’ homme serait-il why the man is=he ‘Why then is man virtuous.’
57
alors vertueux then virtuous (modern French; Salvesen 2009: 75)
In (29a), je ‘I’, corresponding to the thematic subject precedes por coi ‘why’.¹² In (29b), we have a so-called complex inversion. This complex inversion contains a left-dislocated DP immediately following pourquoi, whereas its co-referential pronoun is expressed postverbally. This is a structural difference between old and modern French. In the early languages, there is no left-dislocation intervening between the wh-word and the finite verb. In other words, we may assume that wh-words in old French target the same WhP, located to the right of the Left Dislocation Phrase (Salvesen 2009). (30)
old French: Left dislocation . . . Wh-word modern French: Wh-word . . . Left Dislocation . . . Wh-word
If we further base this on the insight from central cartographic work (Rizzi 1997, 2001a; Benincà and Poletto 2004), we assume that Wh-words move to FocusP, while leftdislocated elements belong to the Topic field. A consequence is that in the case of modern French, we must assume an Int(errogation) Phrase directly below ForceP (Rizzi 2001a). In old French, wh-words uniformly move to a position below the Topic field. The implication this has for the analysis of is that it must occur in a position that is structurally higher than Focuso. Recall the two different hierarchies in (8) and (9), repeated here as (31) and (32) (from Benincà and Poletto 2004; Frascarelli and Hinterhölzl 2007). (31)
FrameP . . . ForceP . . . Topic field . . . Focus field . . . FinP
(32)
ForceP . . . Shift Topic . . . Contrastive Topic . . . Focus . . . Familiar Topic . . . FinP
We see that, if we assume that is a head, it must be the head of a topic phrase. This is the same conclusion we drew with respect to information structure in §3.5.
3.6.1.3 Fronted VP elements In the BFM, the bare quantifier mult is fronted in a total of 208 cases, but there is no instance of mult followed by .¹³,¹⁴,¹⁵ Mult may be a fronted adverb (33a), or it may be extracted from a constituent, such as the predicative adjective in (33b). ¹² From a theoretical standpoint this is technically not the subject, but a left-dislocated element that is co-referential to a null element in the clause proper. ¹³ The following search string was used in this query: [pos="PONfrt"][word="Molt|Mult|Mout"] [pos="VERcjg"]. ¹⁴ The following search string was used in this query: [pos="PONfrt"][word="Molt|Mult|Mout"] [word="si"]. ¹⁵ I am grateful to Silvia Rossi for this observation.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
58
(33) a. Mout ont parlé li baron tuit del noir much have talked the noblemen all of=the black chevalier cele nuit knight this night ‘This night, the noblemen talked a lot about the black night.’ (old French, CligesKu v. 4705–4706) b. Mout sont andui li much sont both the ‘Both noblemen were tall’
vasal noblemen
large tall (old French, CligesKu v. 4033)
Adverbs from the Lower Adverb Space (Cinque 1999) cannot be followed by . The adverb bien ‘well’ is fronted in 193 cases (34), but there are no instances of bien followed by .¹⁶ (34) Bien aïde a Tristran son mestre well help to Tristan her master ‘[She] offers good assistance to Tristan, her master.’ (old French, Béroul v. 1263) As for arguments, Wolfe (2018a) observes that it is only in the earliest texts that non-subjects may be followed by . In other words: no element that has been first merged in the VP may be moved to a clause-initial position followed by .¹⁷
3.6.2 Maximal or minimal category? It is common to view in old French as a maximal category (Adams 1987; Salvi 2004; Rouveret 2004; Benincà 2006; Donaldson 2012; Wolfe 2018a). Only Ferraresi and Goldbach (2003) have suggested in old French is a head. However, in his analysis of in old Neapolitan, Ledgeway (2008) suggests that the particle is a minimal category, and a similar analysis has been put forward for the Scandinavian generalized resumptive (Egerland and Falk 2010; Eide 2011; Holmberg 2020; Meklenborg 2020). Resumptive is never obligatory in old French. It is therefore interesting to see what happens when is not present in the structure. What elements may follow the antecedent of the -structure other than ? What we find, is that when is not present, nothing normally intervenes between the finite verb and an argument, a PP, or a temporal or locative adverb. However, when the antecedent contains a finite verb (e.g. adverbial clauses ¹⁶ The following search string was used in this query: [pos="PONfrt"][word="Bien"] [pos="VERcjg"]. ¹⁷ Note that if we adapt a Remnant movement-analysis for the Stylistically Fronted structures above, as suggested by Salvesen (2011), the ban on after these structures is naturally accounted for.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
59
(35a, b) or arguments that either are free relatives (35c) or that contain a relative clause), it may be followed by an argument (35a, c) or an adverb (35b), which in turn is followed by the finite verb of the main clause (Vance et al. 2010; Donaldson 2012, 2019). baron], la dame lor= dist (35) a. [Quant venu furent li when came were the noblemen the lady them= said sa raisson her reason ‘When the noblemen had arrived, the lady told them her plans.’ (old French, belinc, p.154, v.5039) b. [Quant ot Rollant qu’ il ert en la rereguarde], when heard Roland that he was in the rearguard ireement parlat a sun parastre angrily talked to his godfather ‘When Roland heard that he should be in the rearguard, he spoke angrily to his godfather’ (old French, Roland v.761) c. [Qui chu vaslet porroit avoir], fist li marchis, il porroit who this valet could have made the marshal he could bien aler en le tere de Constantinoble et prendre well go in the land of Constantinople and take viandes et autres coses, car li vaslés en est drois oirs. meat and other things for the valet of.it is right heir ‘Whoever could have this valet, said the marshal, he could go to the land of Constantinople and take meat and other things, for he is its rightful heir.’ (old French, Clari, p. 16) In some cases, the antecedent may immediately precede the finite verb (36), an option that is less common than the one where an element occurs between the clause and the verb (Donaldson 2012, 2019). il, [se il i= fu et gié l’= i= vi], ne (36) Dame, fet lady made he if he there= was and I him= there= saw le= conui ge pas; et [se il i= fust], je croi que il him= knew I and if he there =was I think that he eüst veincu le tornoiement. had won the tournament ‘My lady, he said, if he was there and I saw him, I did not recognise him. And if he were there, I think he would have won the tournament.’ (old French, MortArtu II-16) Thus, we can make a distinction between two different groups of resumptive structures: in the first group, the antecedent may be followed by an ordinary V2
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
60
structure, and may be exchanged for any other XP that precedes the finite verb. In the second group, the antecedents may either be followed by or the finite verb. This is a fundamental difference. (37) a. Group 1: fronted clause + / subject / argument / adverbial / finite verb b. Group 2: fronted subject / PP / adverb + / finite verb In all the literature on , has been taken to be a rather unitary phenomenon. The patterns in (37) are a strong indication that this is not the case. The only logical conclusion to be drawn from the data is that must be a maximal category when the initial element is a clause (or contains a clause), while this is less clear in group 2. In fact, if is also a maximal category in group 2, we are faced with the problem of explaining why is the only maximal category permitted in this position. Further, we are left with the rather problematic observation that old French permits a lot more V3 structures than previously assumed. The data forces us to conclude that in group 2 is a minimal category. The idea is summed up in (38). (38) a. Group 1: = a maximal category b. Group 2: = a minimal category This implies that in group 1, it is that fulfils the V2 requirement and occupies the specifier position of the left peripheral head associated with V2. In group 2, it is its antecedent and that satisfies the EPP, which we will see in §6.4.3. Additional evidence for the existence of two different positions is found in certain V3- structures. In (39a) the fronted adverbial clause is followed by the adverb lors ‘than’, which in turn is followed by . [quant li rois voit mon seignor Gauvain son nevo (39) Et and when the king sees my lord Gauvain his nephew revenu], [lors] si li=demande: . . . returned then him=asks ‘And when the king sees that my lord Gauvain, his nephew is back, he asks him . . . ’ (old French, MortArtu II-2) In (39), it is lors that satisfies the V2 requirement. I will come back to the details of the analysis under §6.4.3.
3.6.3 Specialized or generalized resumptive? In the introduction, we sketched a distinction between specialized and generalized resumptives. While special resumptives need to match the semantics of their
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
61
antecedents, generalized resumptives are completely semantically bleached. In old French we see that the two different kinds of pattern differently. Even though the fronted adverbial clause retains a certain independence with respect to the main clause, there is a link between the semantics of the initial adverbial clause and the following V2 structure introduced by . As we saw in §3.2.3, may be an adverb that assumes a coordinating function (Raynaud de Lage 1983: 88) expressing temporal succession (Ménard 1994: 190). With a fronted adverbial clause which predominantly is temporal (see §3.3.2), is the natural choice, as it matches the semantics of the fronted clause. This is exactly what we have seen is the characteristics of a specialized resumptive—there is semantic mapping between the initial element and the resumptive. The quasiabsence of the adverbial conjunction in old Occitan may thus be a key to understanding why SI-resumption is so rare. However, in the case of fronted adverbs, PPs, and subjects, the semantic matching is less obvious. In the case of [ + ], there is no matching whatsoever. In the case of adverbs, we see that first and foremost follow temporal antecedents, something that points towards semantic mapping. However, when the antecedent is a PP, there does not seem to be any mapping at all as the fronted PPs commonly are locative. We may thus conclude with the following (40). (40) a. Group 1: = specialized resumptive b. Group 2: = generalized resumptive
3.6.4 Embedded contexts So far, we have concluded that may be divided into two different groups and that it must occupy some position higher than Focus in the left periphery. But where? Embedded clauses may give us an indication.
3.6.4.1 Embedded [ + ] In his study, Wolfe (2018b) does not discuss any instances of as the initial element in embedded contexts in old French. However, while is absent in the case of fronted adverbials, there are cases of with fronted subjects. The following examples are all of the [ + ] construction in embedded clauses. In (41a), the matrix predicate is savoir ‘know’. In (41c), it can be interpreted both as embedded under savoir and estre voirs ‘be true’. In the system of Hooper and Thompson (1973), estre voirs is a Class A predicate, while savoir is a Class B predicate. Both classes are known to permit embedded root clauses, including in old French (Salvesen and Walkden 2017). The existence of in these cases is thus compatible with an analysis that states that is a left-peripheral element.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
62
(41) a. Dame, fet il, or sachiez veraiement que [cil lady made he now know truly that the.one as armes vermeilles qui porta la manche on+the weapons red who carried the sleeve sor son heaume a cest tornoiement qui a esté a on his helmet on this tournament that has been at Wincestre] si fu Lanceloz del Lac Wincester was Lancelot of.the Lake ‘Lady, he said, now you should know truly that the one with the red armour and who carried the sleeve on his helmet at the tournament at Wincester was Lancelot of the Lake.’ (old French, MortArtu II-17) b. Voirs est, fet soi Morgain, je ne sai se vos le= true is made Morgane I know if you it savez encore, que [Lanceloz del Lac] si aime la roïne Guinievre, know already that Lancelot of-the lake loves the queen Guinievre vostre feme, des icelui premier jor que il reçut l’ your wife from this first day that he received the ordre de chevalerie. order of chivalry ‘It is true, said Morgan, I don’t know if you already know this, that Lancelot of the Lake has loved the queen Guinievre, your wife, from the first day he received knighthood. (old French, MortArtu IV-15) It is puzzling that the [ + ] construction is possible in embedded clauses while the [/ + ] construction is not. We have already shown that these three antecedents pattern together, and we would expect that they behaved similarly way in embedded contexts. The reason for this might be that they target different projections and that only the projection hosting the [ + ] construction is the only one available in embedded contexts. This is a question I leave to further research.
3.6.4.2 Doubly embedded clauses As for fronted clauses (adverbial clauses and DP + relatives), we have concluded that they occupy a clause-peripheral position, at the very least a position that is to the left of the V2 schema (Donaldson 2012, 2019). This independence was already witnessed in §3.2, where we saw that the information-structural value was very different when the fronted DP was a simple noun or pronoun and when it contained a relative clause. In other words: when the antecedent is clause-internal, it conveys information and marks a topic shift, while there are no restrictions on the information value of the antecedent when it is clause-external.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
63
In the model proposed by Benincà and Poletto (2004), a non-integrated clause would occupy a position in the Frame field, to the left of FocusP. However, these clauses may be embedded under certain verbs (Salvesen and Walkden 2017). In these cases, there is often complementizer doubling (Rouveret 2004; Paoli 2006; Ledgeway 2012: 169–76; Salvesen 2014; Salvesen and Walkden 2017): il disent que s’ il les= en= pooit (42) Et and they said that if he them= of.it= could delivrer, qu’ il le= feroient empereur rescue that they him= make emperor ‘They said that if he could free them of this danger, they would make him emperor.’ (old French, Clari p. 61) This might point in the direction that fronted clauses occupy a position to the right of ForceP, or that there is a complementizer phrase higher than ForceP. There is, however, a third solution: if we accept Kiparsky’s proposal that fronted subordinate clauses are not incorporated into the main clause (Kiparsky 1995), we see that what is embedded under dire in (42) is not the structure in (43), but rather that in (44). The difference is that in (43), the fronted adverbial clause occurs as the initial element of the embedded clause, whereas in (44), there are two independent CPs. (43)
[CP [CP s’il les en pooit delivrer] [CP il le feroient empereur]]
(44)
[CP s’il les en pooit delivrer] [CP il le feroient empereur]
In this way, the paratactical nature is observed in (46). We assume that both clauses are headed by a C head, and that it is these C heads that host the complementizers. This also solves the problem pointed out by Ferraresi and Goldbach (2003), who use the complementary distribution of and the complementizer as an argument for being a head. We are thus compelled to assume that fronted clauses are paratactic and that they occupy a position in the Frame field. Examples such as the one in (45) strongly indicates that DPs that contain a relative clause occupy the Hanging Topic position, while adverbial clauses are Scene-Setters (46). [mes sires Gauvains, qui bien conut Boort], [quant il (45) Et and my lord Gaucains who well knew Boort when he le= vit a terre], si dist au roi Artur . . . him= sees on earth says to.the king Arthur ‘And when my lord Gauvain, who well knew Boort, sees him on the ground, he says to king Arthur . . . ’ (old French, MortArtu I-21)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
64 (46)
[HTP mes sires Gauvains, qui bien conut Boort] [ScSP quant il le=vit a terre]
In the example in (45), is merged in the specifier position of the head associated with V2.
3.6.4.3 Left dislocation We conclude that in Group 1, the antecedent is clause-external and that is a maximal category that is merged in the specifier of the V2 head, as such checking the EPP-feature. In Group 2, the situation is different. We have seen that cannot be exchanged for a different XP, and we thus assume that is a head, possibly a clitic. The structure strongly resembles Germanic Left Dislocation where the leftdislocated DP is always adjacent to the resumptive pronoun (47). ingenting. (47) [Per], han sa Per said nothing ‘Per didn’t say anything.’
(Norwegian)
In the Germanic languages, this structure is often analysed as an instance of movement, where the DP moves to a Topic position, while the resumptive pronoun sits in the specifier of the V2 head (for two different derivations of this principle, see Grohmann 2000; Grewendorf 2002). This presence of the pronoun is necessary as this position must contain overt phonetic material. The derivation is illustrated in Figure 3.1. LDP Per
LD’ LDo Per
FinP han
Fino
Fino
IP
sa
…
Figure 3.1. The derivation of Germanic Left Dislocation Following Grohmann (2000)
When Holmberg (forthcoming) analyses the Swedish -structure, he suggests that the mechanism is almost the same as in the case of Left Dislocation, with the difference that rather than having a pronoun in the specifier of the V2 head, is the lexicalization of the head, to which the antecedent moves. The EPP feature of the V2-head is checked as the antecedent moves through it.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
65
The specifier does not contain overt phonetic material, but the closest head, the one containing , is expressed, something that satisfies the visibility requirement of V2. If we adopt this analysis for old French, the derivation of a simple structure such as the one in (48) would be as in Figure 3.2. For now, the projection is labelled ZP, while the V2 projection is labelled YP: ZP lors
Z’ Zo si
YP Yo
lors Yo
IP
dit
…
Figure 3.2. Derivation of Group 2 structures with
(48) Lors si dit . . . then says ‘Then he says . . . ’ The implication of this analysis is that there is one more difference between group 1 and group 2 regarding the derivation of the structure (49). (49) a. Group 1: = the antecedent has been merged in a clause-external position b. Group 2: = the antecedent has been moved to its surface position from a clause-internal position In clause (39), repeated as (50), we see how the two mechanisms interact. One element is merged in the left periphery, the other is moved to its surface position. S is the spell-out to the head to which the operator moves, and the bottleneck assures that there may not be two instances of in the same clause. The derivation is shown in Figure 3.3. [quant li rois voit mon seignor Gauvain son nevo (50) Et and when the king sees my lord Gauvain his nephew revenu], [lors] si li= demande: . . . returned then him= asks ‘And when the king sees that my lord Gauvain, his nephew is back, he asks him . . . ’ (old French, MortArtu II-2)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
66
ScSP ScS’
CP quant li rois voit mon seignor Sco Gauvain son nevo revenu
ForceP Forceo Forceo
TopP Top’
lors Topo si
FinP lors
Fin’ Fino
IP
li demande
Figure 3.3. Merged clause-external constituent followed by moved adverbial +
We have seen that a fronted adverbial clause must be merged in the Scene Setting Projection (ScSP), and that all instances of must occur above the Focus field. This leads us to posit that Group 2 resumption involves a topic head located between Focus and Force. Consequently, the verb must target Fino, which I associate with the V2 head. In the tree in Figure 3.3, I have thus exchanged ZP for TopP and YP for FinP. This is not in line with the analysis proposed by Wolfe (2015c, 2018b), who suggests that the finite verb targets Forceo in thirteenth century French. However, given the possibility of structures such as the one in (39), the analysis in Figure 3.3 imposes itself. This approach implies that the Group 1 antecedent is either a Hanging Topic or a Scene Setter, while the Group 2 antecedent is a Left-Dislocated Topic. By following a very different path, we thus end up with the same conclusion as the one reached by previous researchers, namely that the DP preceding has been left-dislocated (Marchello-Nizia 1985; Vance 1997; Donaldson 2015).
3.6.5 Summing up In the previous section, we have seen that falls into two different categories, labelled Group 1 and Group 2. The groups’ characteristics are summarized in Table 3.5.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
67
Table 3.5. The distribution of the different types of antecedent
Group 1 Group 2
resumptive
type
origin
type
category
adverbial clauses (DP +) relative clauses DP, PP, AdvP,
merged
specialized
maximal
moved
generalized
minimal
3.6.6 Old Occitan Can the same model be transferred to old Occitan? It is clear that fronted clausal elements may be clause-external in old Occitan. In the 714 verses of the Canso d’Antioca (Table 3.6), there are a total of eighteen fronted adverbial clauses: ten of these are temporal clauses introduced by cant ‘when’; eight are conditional clauses introduced by se/si ‘if ’. The majority of these are followed by an XP (either the subject or any other nonverbal constituent). Only one clause is introduced by (51).¹⁸ Table 3.6. Fronted adverbial clauses in the Canso d’Antioca cant + V +S + XP + total
se/si
3 3 3 1
2 1 5 –
10
8
dux pren sas armas e vai las revestir], (51) [Cant lo when the duke takes his arms and goes them put.on si fa tota la trera deotz sos pes fremir makes all the earth underneath his feet tremble ‘When the duke takes up his arms and goes to put them on, he makes the whole earth shake beneath his feet’ (old Occitan, Antioca 65–66) There are several structures where a fronted adverbial clause is immediately followed by the subject (52a) or an adverb (52b). The initial adverb of the main clause does not have to reflect the semantics of the fronted adverbial clause. This strongly suggests that belongs to Group 1 in these cases.
¹⁸ All translations from the Canso d’Antioca by Sweetenham and Paterson (2003).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
68
(52) a. [Cant elh esperonero a milhers e a sans] lo when they spur to millions and to hundreds the vasal qe· il encontro abb aqels martal[h]ans es fis de knight that= they meet by blade deadly is sure to mortz recebre death receive ‘When they spur forward in hundreds and thousands, any warrior they come across with their deadly cutting-blades is certain to suffer death’ (old Occitan, Antioca 574f.) b. Reis, [se· l laissa[s] de te aisir ni propïar] king if it let of you approach not approach ja ers d’ aqui enant no ·t poiria aiudar already thing from here forward you can help ‘King, if you let this man approach or come near you, certainly nothing hence-forward will be able to help you.’ (old Occitan, Antioca 93f.) Interestingly, we find structures where an adverbial clause precedes an argument, which in turn is followed by (53). We saw the same structure in old French in (39), only that the fronted clause was followed by an adverb and not the leftdislocated subject. [qan la ciutatz ne fon presa e tuich[h] li baron s= (53) E and when the city was taken and all the barons en= torneren], [Girautz de Borneil] si s’= en= anet al of.it= return Giraut de Borneil = of.it= went to.the bon prince d’ Antiocha good prince of Antioca ‘And when the city had been taken and all the barons had returned, Guiraut de Borneil went to the good prince of Antioche’ (old Occitan, GdB 1) The clause in (53) does suggest that the structure we proposed for old French may also be valid for old Occitan. However, given the scarcity of the -construction and the relaxed V2 status of the language, it is hard to provide a detailed analysis. Further research should take into consideration a larger set of -structures from old Occitan and try to draw out a more detailed map of the structure than what has previously been done. This chapter is only a suggestion of the direction which such a project might take.
3.7 Conclusion We have compared -resumption in old French and old Occitan. S is abundant in old French, both as a resumptive particle and as an adverbial connector.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
69
The latter use is hardly attested in old Occitan, and the use of as a resumptive element is limited, with great intertextual variation. While we find after fronted DPs, AdvPs, PPs, and adverbial clauses in old French, the use of is much more limited in old Occitan, and we have only found it after fronted clauses and DP subjects. The formal analysis of the -structure seems to be similar in the two languages, but they differ with respect to information structure. While the [ + ]-structure is restricted to - subjects that represent a topic shift in old French, old Occitan also uses [ + ] for - subjects, in addition to the usage we find in old French. In the case of old French, we have seen that the uses of fall into two groups, labelled Group 1 and Group 2. In the former, is a maximal category and a specialized resumptive, while in the latter it is a minimal category and a generalized resumptive. The antecedents of Group 1 have been first-merged in a clauseexternal position, while the antecedents of Group 2 have reached their surface position by movement through SpecFinP. We may thus conclude that the -structure is more common in old French than in old Occitan, but that its syntactic properties are probably similar. They do, however, diverge when it comes to information structure.
Texts cited Old French Note: All texts with the exception of La Mort du roi Arthur have been accessed through the BFM. belinc béroul brut2 CharetteKu Clari CligesKu qgraal_cm Lapidfp MortArtu Roland
Williams, G. P. (1929). Bel Inconnu. Paris: Champion. Defourques, L. M. and Muret, E. (1947). Tristan. Paris : Champion. Arnold, I. D. O. (1938–40). Brut. London: Anglo-Norman Text Society. Kunstmann, P. (2009). Chevalier de la Charrette ou Lancelot. Ottowa: Laboratoire de Français Ancien. Lauer, P. (1924). Conquête de Constantinople. Paris: Champion. Kunstmann, P. (2009). Cligés. Ottowa: Laboratoire de Français Ancien. Marchello-Nizia, C. (2013). Queste del saint Graal. Lyon: Equipe BFM. Studer, P. and Evans, J. (1924). Lapidaire en prose. Paris: Champion. Hult, D. F. (2009). La Mort du roi Arthur. Paris: Livre de Poche. Moignet, G. (1972). La Chanson de Roland. Paris: Bordas.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
70
Old Occitan Antioca BdV GdB
Sweetenham, C. and Paterson, L. M. (2003). The Canso a Antioca. An Occitan Epic Chronicle of the First Crusade. Aldershot: Ashgate. ‘Bertrand de Ventadorn’. (1964). In Boutière, J. (ed.) Biographies des troubadours: Textes provençaux des XIIIe et XIVe siècles. Paris: Nizet, 20–31. ‘Guiraut de Borneill’. (1964). In Boutière, J. (ed.) Biographies des troubadours Textes provençaux des XIIIe et XIVe siècles. Paris: Nizet, 37–58.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
4 Variation in the Gallo-Romance left periphery V2, complementizers, and the Gascon enunciative system Adam Ledgeway
4.1 Introduction In his overview of langue d’oïl varieties, Smith (2016: 310) perceptively notes that ‘[k]ey to a diachronic understanding of two major aspects of contemporary French syntax—inversion and the use of subject clitics—is the role played by the “verb-second” (V2) constraint at earlier stages of the language.’ This V2 constraint consists in an operation which moves the finite verb to the vacant C(omplementizer) position, with concomitant fronting of a pragmatically-salient focus / theme constituent to an operator position to its immediate left (for a recent overview and references, see Wolfe 2019). This regularly results in so-called verb-subject inversion whenever constituent-fronting does not target the subject (1a), and also explains the emergence of subject clitics from weakened forms of erstwhile tonic subject pronouns: the latter, despite the positive setting of the null subject parameter, exceptionally surface in the operator position as a last resort mechanism to satisfy the second-position requirement of the verb in thetic clauses and utterances consisting of a simple verb. Witness the use of overt expletives (1b).¹ fu (1) a. [CP [la] there be...3 ‘he drowned there’
[TP
il he
fu
noiés la]] drown... (§608,2)
b. [CP [il] lor avint [TP lor avint une grans mescheance]] it them.=happen...3 a great misfortune ‘a great misfortune befell them’ (§608, 2) (old French (13th c.): Histoire ancienne jusqu’à César, Eneas (https: //tvof.ac.uk)) ¹ In what follows free translations are only provided where the intended meaning cannot otherwise be inferred from the glosses.
Variation and Change in Gallo-Romance Grammar. First edition. Sam Wolfe and Martin Maiden (eds). This chapter © Adam Ledgeway 2020. First published in 2020 by Oxford University Press.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
72
Thus, although the syntax of modern French is no longer V2, the persistence, at least in higher registers, of inversion in certain non-veridical polarity contexts (2a) and the rise of subject clitics, with concomitant reversal in the null subject parameter such that today all finite clauses must contain an overt pronominal / lexical preverbal subject (2b), are manifestly the residue and reinterpretation of an original V2 rule. (2)
a. [CP [Que] lui as[TP tu as dit que?]] what him.=have...2 you said b. [TP Elle / Marie / **pro she Marie pro
mange.] eat...3 (modern French)
The diachronic path and facts sketched here for French, although broadly true of other Gallo-Romance varieties including Occitan, northern Italian dialects, Friulian, and Francoprovençal (see §2.2), do not, however, hold for (central) Gascon varieties. Although medieval Occitan varieties, including Gascon (cf. Joseph 1992: 486–90), have been argued to be V2 (Vance, Donaldson, and Steiner 2009; Donaldson 2015, 2016; Wolfe 2018a, 2018c; Meklenborg, Chapter 3, this volume), regularly licensing inversion structures such as (3a) as in old French, medieval Gascon differs from the latter in freely licensing apparent V1 structures (see 3b) concealing a phonologically null pronominal (shift/default) topic in the initial operator position (cf. Wolfe 2018c: §2.2.5). [TP nos totz três pausam aqueste carta (3) a. [CP [aqueste carta] pausam this charter place...1 we all three sober l’ autar de Santa-Fee]] on the altar of holy-faith ‘we all three place this charter on the altar of Holy Faith’ (XXVI, 42) b. [CP [proTop] Judya [TP la cort judya a Morlaas que [ . . . ]]] judge...3 the court at Morlaàs that ‘The court in Morlaàs judged that . . . ’ (XVIII, 30) (old Gascon (11th c.): Fors de Béarn) Given the availability of V1 structures such as (3b) in apparent violation of the superficial PF-driven V2 requirement otherwise active in old French, the last resort mechanism of pronominal insertion never arose in Gascon which remains a null subject language displaying neither subject clitics (4b) nor French-style inversion (4a).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
- (4)
a. E tu and you
73
[CP [que] [TP l’ as (**tu) dit?]] what him.=have...2 you say. (modern Gascon; Palay 1927: 9)
b. Tu / pro que you pro que ‘You are eating.’
minyos. eat...2 (modern Gascon; Bonaparte 1878: 3)
This suggests that changes in medieval V2 syntax were not uniform across Gallo-Romance, but involve at least two possible outcomes as exemplified by the contrasting behaviour of modern French and Gascon in (2) and (4). These differences in the availability of inversion (viz. V-to-C movement) and subject clitics, the latter an instantiation of a broader structural requirement (viz. EPP feature on T) that all finite clauses realize a dedicated preverbal subject position, are argued below to follow from changes in the V2 constraint singled out by Smith. In particular, I will demonstrate that whereas modern French and other Gallo-Romance varieties exhibit, at best, residual effects of an original V2 syntax (§2.2), modern (central) Gascon can be argued, at a certain level of abstraction, to continue a fully active V2 syntax (§3). In the passage from medieval to modern Gascon, this V2 syntax has undergone a radical change in its formal realization such that it is no longer satisfied through the Move option raising the finite verb to the C position, but through the Merge option directly lexicalizing the latter position with a so-called ‘enunciative’ particle such as que in (4b).
4.2. Strong / weak C dimension 4.2.1 Medieval Gallo-Romance: strong C I adopt here the traditional intuition that clauses are nominal, as evidenced by the fact that in embedded contexts they are headed in Romance by complementizers which continue original D elements (cf. Latin relativizer / interrogative paradigms in - > [k]-, > > si/se ‘if ’ from proto-IE *[so] ‘this/that’). Indeed, according to Manzini and Savoia (2003, 2011) C(omplementizer) is merely a descriptive label for a particular set of occurrences of the nominal D(eterminer) that binds a propositional variable with sentential content restricted by the (embedded) sentence (cf. also Poletto and Sanfelici 2018). On this view, we expect parallels in the distribution and development of articles and complementizers.² One such case is the strong/weak D parameter for nominals of Guardiano and Longobardi (2005), which I assume can be extended to the clausal domain ² On the parallels of (in)definiteness and (ir)realis marking in the Romance nominal and clausal domains, see Ledgeway (2015).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
74
through a parallel strong / weak dimension of parametric variation for the C head which, if strong, must be associated with (a) V(-feature) overtly in the syntax. From this perspective, most modern Romance varieties qualify as weak C languages, inasmuch as there is no systematic association in the syntax between V and [+declarative] root C, as witness the ungrammaticality of auxiliary-subject inversion in modern French declarative contexts: (5)
[TP Elle avait trop bu.]] (Fr.) [CP (**Avait-) have...3 she had...3 too.much drink.
As observed in (1) and (3), by contrast, the medieval Romance languages, as well as some modern Ladin varieties (Salvi 2000a; Kaiser 2002a; Poletto 2002; Benincà 2013; Casalicchio and Cognola 2015), are arguably strong C languages, in that root C (as well as some embedded cases in so-called ‘bridge’ contexts) is characterized by a V2 constraint which imposes generalized V-to-C movement on the finite verb and fronting of a (covert) constituent to the operator position. This latter operation can be viewed as a generalized EPP effect (cf. Holmberg 2015) if we assume that when C is strong (i.e. bears an uninterpretable V-feature) it also comes with a corresponding uninterpretable edge-feature satisfied by constituent-fronting.
4.2.2 Modern Gallo-Romance: weak C Above we have seen (cf. 5) that modern Gallo-Romance varieties should be considered weak C languages, as further highlighted by the fact that the EPP is checked on T rather than C in these varieties which obligatorily project and lexicalize a dedicated preverbal subject position (cf. 2b). Nonetheless, C may still probe V (and hence license V-to-C movement) under particular marked conditions, as variously reflected in (simple / complex) verb-subject inversion, enclisis of object clitics, and the complementary distribution of subjunctive verb forms and complementizers (cf. Ronjat 1913: §142; Poletto and Tortora 2016: 779–81). Following Rizzi and Roberts (1989) and Rizzi (1990b), this more constrained type of V-to-C movement can be considered a synchronic residue of generalized V2 movement from the medieval period—as indirectly supported by its greater productivity in French, for instance, in higher and hence more archaicizing registers— which is today licensed only in a restricted number of non-veridical polarity contexts tied to specific types of illocutionary force, including values variously labelled in the traditional literature as interrogative, optative, (ex)hortative or jussive, hypothetical, concessive, disjunctive, exclamative, imperatival, and quotative. Thus, while generalized V2 movement, triggered by a semantically uninterpretable V-feature in declarative contexts, has been systematically lost in weak C varieties, V-to-C movement is exceptionally retained just in those contexts
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
75
where movement plays a role in interpretation (and hence associated with a semantically interpretable V-feature) licensing the observed non-veridical polarity values (cf. Munaro 2004; Manzini and Savoia 2005: 398f.). Nonetheless, the distribution of such semantically driven V-to-C movement is not uniform across Romance displaying different and often unpredictable degrees of productivity and attrition. Conflating some of the traditional labels above, I distinguish here between interrogative, optative (subsuming (ex)hortative, jussive, hypothetical, concessive), exclamative, quotative, and imperatival illocutionary forces (for an overview, see Cruschina and Ledgeway 2016: 568–71; Giurgea and Remberger 2016), the precise distribution of which can be tentatively modelled, at least for Romance, in terms of the microparametric choices presented in the hierarchy in (6). (6)
(a) Is C strong? Yes: MedR. No (b) Extended to all [–declarative] force types? Yes: GaR. No (c) Restricted to optative, exclamative, quotative, imperative? Yes: It., Ro. No (d) Restricted to exclamative, quotative, imperative? Yes: IbR. No (e) Restricted to quotative/imperative? Yes: SIDs
…
The positive setting for option (6a) isolates languages with a strong C, hence endowed with a semantically uninterpretable V-feature which indiscriminately probes all finite verbs, a situation which we have seen obtains in medieval Romance V2 varieties with generalized V-to-C movement (cf. examples 1 and 3). The negative setting, by contrast, broadly identifies most modern Romance varieties where the relevant parametric setting for C is weak, hence variably endowed with a semantically interpretable V-feature such that V-to-C movement represents a marked option limited to one or more [–declarative] clause types. Among this group we can identify through the positive specification for option (6b) those more liberal Gallo-Romance varieties such as French and especially north(east)ern Italian dialects (cf. Poletto 2000: chaps 3 and 5; Manzini and Savoia 2005, I: 384–7; Munaro 2010; Benincà, Parry, and Pescarini 2016: 200) which continue to license V-to-C movement across all marked clause types:
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
76
(7)
a. Vient-il? come...3=he b. Puisse-t-elle may...3=she c. Est-elle be...3=she d. «Oui», yes
réussir! succeed.
jolie! pretty
répondit-elle reply...3=she
e. Demande-le-lui! ask..2=it=him.
(French)
However, some of these cases of V-to-C movement are not particularly productive even in higher registers (arguably lexicalized in many cases, cf. Fr. Vive / **Meure le roi! ‘live/die...3 the king!’), and are often subject to additional restrictions related to verb class, mood, and grammatical person (cf. also Biberauer and Roberts 2012, 2017). For example, V-to-C movement in French interrogatives (but not generally in northern Italian dialects) is more readily licensed by functional rather than lexical predicates (cf. je suis ‘I be...1’ (< être) or ‘I follow.. .1’ (< suivre) ) suis-je? ‘be/**follow...1=I?’) and by 2nd/3rd-person rather that 1st-person subjects (cf. prends-tu/prend-il? ‘take...3/2=you/he?’ vs **prends-je? ‘take...1=I?’). Similarly, the distribution of V-to-C movement in French optatives is increasingly limited to a handful of principally functional predicates (viz. être ‘be’, avoir ‘have’, devoir ‘must’, pouvoir ‘can’, vouloir ‘want’, and venir ‘come’), and occurs above all in the third person, e.g. M’eût-il encouragé . . . ‘me=have...3=he encouraged (= Had he encouraged me)’, Voulût-il le faire . . . ‘want...3=he it=do. (= Even if he wanted to do so)’, Vienne le printemps et tout semblera plus souriant ‘come...3 the spring and everything seem..3 more jolly’. We witness in such behaviours often well-advanced and ongoing morphosyntactic and lexical restrictions on a once fully productive movement operation which in lower registers is now predominantly replaced, with the exception of quotatives and positive true imperatives, by the Merge option with lexicalization of the C-head with a relevant modal particle and concomitant impossibility of inversion and enclisis of object clitics.³
³ The increasing diachronic restriction of inversion to functional predicates in French reflects a progressive loss of V-(to-T-)to-C movement (manifested in the growing infelicity of inversion with lexical predicates), with verb movement now increasingly limited to T-to-C movement (hence the greater acceptability of inversion with functional predicates). Significantly, this also explains the observed greater propensity of inversion with verbs in the subjunctive, including lexical predicates since, as shown in Ledgeway (2009), Ledgeway and Lombardi (2014) and Schifano (2018), Romance irrealis verb forms typically raise to the highest available position within the T-domain. By the same token, the near total loss of inversion in southern Italian dialects (with the exception of quotatives and positive imperatives) discussed in the text can be explained by the independent observation that finite
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
- (8)
Est-ce qu’ [TP il mange?]] he=eat...3 ‘Is he eating?’ b. [CP Que [TP cela ne se répète that. that self=repeat...3 ‘May that not happen again!’
77
a. [CP
c. [CP
S’ [TP il m’ eût / avait if he me=have../.3 ‘If (only) he had encouraged me . . . ’
d. [CP
((Qu’est-)ce) qu’ (what)that. ‘How pretty she is!’
[TP
elle she
pas!]]
encouragé . . . ]] encourage.
est be...3
jolie !]] pretty (French)
This weakening of the Move option is even more evident in those varieties singled out by the positive specifications of options (6c) and (6d) such as Italian / Romanian, and Ibero-Romance, respectively, which have both lost semanticallydriven V-to-C movement with polar interrogatives,⁴ but continue to display it with (some types of) exclamatives and imperatives, though differing with respect to the availability of such movement in optatives, e.g. Ro. Arză-l focul!⁵ ‘burn. .3=him fire.’ vs Sp.¡**(Que) le queme el fuego! ‘that him= burn.. .3 the fire’. In these varieties, too, non-declarative illocutionary force is in many cases more readily licensed through the Merge option with lexicalization of C by various C-heads (cf. Ledgeway 2012: 175f.; Corr 2017) such as Sardinian interrogative a, Portuguese / Spanish optative oxalá / ojalá. Finally, option (7e) identifies those varieties such as southern Italian dialects in which V-to-C movement shows the most restrictive distribution, having all but disappeared from the grammar with the exception of quotatives and positive true imperatives,⁶ where V-to-C verbs typically raise to a very low position within the sentential core (the lower adverb space in Ledgeway and Lombardi 2005, 2014; Ledgeway 2009, 2012; Schifano 2015, 2018) and hence are not available for T-to-C movement (cf. discussion of inversion in the history of English in Biberauer and Roberts 2012, 2017). ⁴ In the highest literary and archaicizing styles V-to-C movement is still occasionally found in polar interrogatives involving marked irrealis modal interpretations (Poletto 2000: 156; Giurgea and Remberger 2016: §53.3.1.2). ⁵ Cf. also the archaic morphology in the subjunctive form arză (cf. modRo. ardă). ⁶ Although the licensing of V-to-C movement in Romance appears to follow the implicational scale interrogative > optative > exclamative > quotative/imperative formalized in (6), there is little, if any, evidence for a particular ordering of the latter two clause types. However, there are some Romance varieties which, though robustly preserving V-to-C movement in imperatives, are reported to show a more relaxed behaviour with quotatives suggesting an order quotative > imperative. Relevant cases include Brazilian Portuguese where, unlike European Portuguese, quotative inversion proves optional (Kato and Martins 2016; Lobo and Martins 2017: 36), and, within Gallo-Romance, the Occitan dialect of Périgord (Miremont 1976: 88).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
78
movement proves most resilient across Romance,⁷ e.g. Cal. Chi vò scattà! ‘that. want...3 explode. (= May he keel over!)’ vs «Mannamillu!», dicìa Cicciu ‘send..2=me=it!, say....3 Ciccio’.
4.3 The Gascon enunciative system 4.3.1 Introduction In an area south of the Garonne running east to west through the departments of southern central-western Ariège, southwestern Haute-Garonne, Hautes-Pyrénées, and Pyrénées-Atlantiques and stretching north through the Landes and the western and central areas of Gers with occasional incursions into southwestern Gironde and southwestern Lot-et-Garot,⁸ as well as to a lesser extent in the Val D’Aran in northern Spain (Pusch 2000b: 627; Marcus 2010: 36; pace Rohlfs 1970a: 206), the relevant Gascon varieties display a system of so-called enunciative particles marking different types of illocutionary force in both main and embedded clauses.⁹ By way of illustration, consider the French-Gascon root contrasts in Table 4.1.¹⁰ Apart from positive imperatives which display V-to-C movement in both varieties,¹¹ there are striking differences between French and Gascon in the formal licensing of different clause types and, especially, in the licensing of declaratives, the least marked of sentential types. As the unmarked clause type, declaratives in
⁷ For Romance quotatives see Ambar (1992), Vanelli (1995), Maldonado (1999), Suñer (2000), Bonami and Godard (2008), Matos (2013), Pană Dindelegan (2013: 121), Lobo and Martins (2017: 36), and for imperatives Rivero (1994), Graffi (1996), Zanuttini (1997). ⁸ Cf. Ronjat (1913: 80), Rohlfs (1970: 206), Darrigrand (1974: 32), Marcus (2010: 35f.), Floricic (2012: 3). See also, for instance, ALF maps 10A, 23, 24, 27, 28, 32, 34, 84, 91, 92, 136, 462. ⁹ All Gascon examples either come from my own personal examination of a number of Gascon texts (plays, prose) and grammars or are those reported in the vast literature on the topic. In what follows all examples will simply be referred to as Gascon without any further differentiation, unless finer diatopic distinctions prove relevant. ¹⁰ Cf. Lespy (1858: 223–5, 258f.), Ronjat (1913: 80–5, 142f., 231), Bouzet (1963: 26f.; 1975: 66–8), Rohlfs (1970: 205–11), Darrigrand and Grosclaude (1971: 3), Darrigrand (1974: 32, 56, 84), Joseph (1992: 481–3), Pusch (2000a: 189), Morin (2005: 60f.; 2006: 3f.; 2008: 138–40), Fossat (2006: 161f.), Karenova (2006: 4; 2008: 48f.), Marcus (2010: 31f.), Joly (2013: 247f.), Puyau (2013: 135f.), Giurgea and Remberger (2016: 863f., 872). ¹¹ The robustness of V-to-C movement with positive imperatives is argued in Ledgeway (forthcoming) to follow from the widespread idea that imperatival clauses do not project the full array of functional projections associated with the T-domain, as reflected cross-linguistically in the absence of any inflexional marking or, at the very least, very minimal inflexional marking on second-person imperatives (Bybee 1985: 173; Floricic 2008: 10; Ledgeway 2014a). Theoretically, the observed inflexional impoverishment of the imperative can be interpreted in terms of the mechanisms of feature transmission and inheritance (Chomsky 2007, 2008): whereas phi-features that originate on C are usually ‘transferred’ down to T in root declaratives, in the absence of T in imperatives these same features fail to be passed down, such that the imperatival verb is forced to raise to C to license its inflexional features.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
79
Table 4.1. Gascon enunciative C-system French Move D –
Gascon Merge
Move
Merge
–
–
Que cantas. que sing...2 E cantas? e sing...2
I
Chantes-tu? sing.. Est-ce que tu .2=you chantes? you sing.. .2 O Puisse-t-il chanter! Qu’il chante! may...3=he that. he sing. sing. ..3 E Est-il bête! ((Qu’est-)ce) qu’il be...3=he est bête! silly (what)that. he be...3 silly Q « . . . », répondit la fille – « . . . », reply.. .3 the girl I
Porte-le-moi! – bring..2=it=me
–
–
E cantèsse! e sing...3
–
B’ei plan pèc! be be...3 very silly
–
« . . . », e responou la maynade « . . . », e reply.. .3 the girl –
Pòrta-m-ac! bring. .2=me=it
modern French are not formally marked in the C-system, with subject and finite verb occurring within the sentential core as the default option, namely [CP [IP Tu chantes]]. By contrast, Gascon displays differential marking of declaratives through lexicalization of the C-head with the erstwhile finite complementizer que ‘that’. In particular, it should be noted that que is not optional, but obligatorily introduces all finite verbs in all person, temporal, aspectual, and modal specifications such that its omission in root declaratives, except when negated (see §3.1.1), invariably results in ungrammaticality.¹² Although cross-linguistically explicit typing of declarative force represents a very rare option (cf. Lyons 1968: 307; Bybee 1985: 147; Cinque 1999: 130; Franco 2013), recourse to que in this context is unsurprising since it represents the default, unmarked complementizer in (Gallo-)Romance.
¹² Cf. De Grateloup ([1734]1887: 16), Lespy (1858: 223), Cénac-Moncaut (1863: 131), Bonaparte (1878: 1), Ronjat (1913: 80), Camproux (1958: 388-90), Bouzet (1963: 26; 1975: 66), Rohlfs (1970a: 205), Darrigrand and Grosclaude (1971: 3), Darrigrand (1974: 32), Pusch (1999: 113; 2000a: 192; 2000b: 627), Marcus (2010: 31), Joly (2013: 248), Puyau (2013: 136), Suïls Subirà and Ribes (2015: 550).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
80
In non-declarative root clauses, Gascon also lexicalizes the C-head with one of two other clause-typing particles: be and e.¹³ While the former, the typical marker of total exclamatives (Fossat 2006), represents the grammaticalization of the adverb > be(n) ‘well, indeed’,¹⁴ the latter is most commonly, though not uncontroversially (cf. Pusch 2001), claimed to continue the coordinator ‘and’ (Lafont 1964; Rohlfs 1970a: 210 n.375), a conclusion not immediately reconcilable with its synchronic uses in typing polar interrogatives, optatives, and quotatives.¹⁵ Thus, here I adopt the view that, at least synchronically, e constitutes a reduced form of the irrealis/interrogative complementizer se / si ‘if ’, a conclusion readily compatible with the observed non-veridical root uses of e and, as we shall see in §4.3.3, its distribution in non-asserted embedded contexts such as subjunctive and ‘central’ adverbial clauses. Furthermore, in some dialects, and especially those from the southeast of the region (e.g. Couserans; cf. points 695, 607 (Haute Pyrénées) and 782 (Ariège) of ALF map 575), interrogative e is often replaced and / or alternates with the form se / si,¹⁶ e.g. (s)e benguerats? ‘(s)e come..2 (= will you come)?’.¹⁷ Similarly, in optatives e is also reported to alternate with se/ si (Marcus 2010: 53f., 132; Joly 213: 249), exactly along the lines of the optative use of Fr. si ‘if ’ witnessed in (9c). Quotative e, presumably an evidential marker (cf. Pusch 2002: 114; cf. Cinque 1999: 85f.), is also widely reported (cf. Bouzet 1975: 67; Marcus 2010: 53f., 130f.; Puyau 2013: 115) to alternate with forms variously spelt as se / si / ce / ci / çò / ça, e.g. > e / se / si / ce digou lou pay ‘ e / se / si / ce said the father’ (Joly 2013: 247), forms which can
¹³ Although explicitly excluded by some studies (e.g. Pusch 2002: 112), other studies (Rohlfs 1970a: 209; Bouzet 1975: 68; Karenova 2006: 3; Marcus 2010: 32, 50–2; Joly 2013: 247; Puyau 2013: 128, 130) report an additional enunciative particle ja(ya) / je(ye) (< ‘now; already’) which, possibly in accordance with diatopic variation, is used to reinforce an assertion (i.a; cf. modal uses of IbR. preverbal ya / já / ja) or mark an exclamative (i.b): (i)
a. Ya la bés. ya her=see...2 ‘You can indeed/certainly see her.’
(Rohlfs 1970a: 209)
b. Lous chins já sou au lheyt! the kids já be...3 to.the bed ‘What, the kids are in bed!’
(Karenova 2006: 3)
¹⁴ The manner adverb ‘well’ in Gascon is pla(n) < () ‘full’. Note furthermore that grammaticalized forms of functioning as left-peripheral or high sentential modal / polarity particles / adverbs are not uncommon in Romance (cf. Belletti 1994; Hernanz 2007, 2010; Cognola and Schifano 2018). ¹⁵ Both be and e, as well as que, also exhibit related extensions of these basic functions, as detailed in the literature. Unless immediately relevant, we will limit our attention in what follows to these fundamental values and functions of the three particles. ¹⁶ See Ronjat (1913: 226), Field (1985), Pusch (2000a: 198), Fossat (2006: 161), Marcus (2010: 52f., 127, 130f.), Puyau (2013: 49, 65), Rigau and Suïls (2010: 154f.), Joly (2013: 247, 249), Suïls Subirà and Ribes (2015: 553f.). ¹⁷ Note that before vowels non-veridical e is regularly elided (Lespy 1858: 258; Bouzet 1963: 28; 1975: 67; Rohlfs 1970a: 210 n.376; Marcus 2010: §2.4), e.g. Etz anatas tà la hèsta? lit. ‘be...2 go. . to the party?’ (Darrigrand 1974: 115).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
81
all be straightforwardly derived from an original ‘if’ given the notable diatopic variation in the realization of final unstressed vowels across the region (viz. [se / si / sә / sɔ / sa]). In these non-veridical contexts, Gascon therefore differs from (formal registers of ) French in that the relevant illocutionary force cannot be licensed on the C-head, even in quotatives, through the Move option (viz. V-to-C movement), but, only through the Merge option. Superficially, we might therefore be tempted to interpret the Gascon facts on par with the Merge option observed in interrogatives, optatives, and exclamatives in less formal registers of French. If this were the case, then the Gascon data would not prove particularly spectacular. However, I am claiming here that, unlike modern French, modern Gascon is a V2 language such that the distribution and licensing of be and e (se / si), as we shall see in detail below, is quite different from that of Fr. est-ce que, que, and ((qu’est-)ce)que. Just consider, for example, that whereas the latter are followed by a pronominal/lexical subject, this is excluded with be and e (se / si), which must immediately precede the finite verb and any complement clitics (cf. 11). Furthermore, interrogative e (se / si) is incompatible with wh-interrogatives and is found in both root and embedded polar interrogatives, whereas Fr. interrogative est-ce que is compatible with wh-interrogatives but is excluded from embedded contexts.
4.3.2 Strong C: Merge vs Move The traditional interpretation noted above (cf. examples 1 and 3) of satisfying the V2 requirement on strong C in terms of V-to-C movement represents just one of two possible licensing mechanisms made available by the grammar: alongside the more marked Move option, the system also makes available the less costly Merge option whereby the ‘strong’ uninterpretable V-feature requirement on C can be satisfied by direct lexical insertion of a suitable head (cf. Roberts’ (2004) claims about PF-realization of C-Fin in V2 contexts).¹⁸ Ledgeway (2008) shows that in some medieval Romance varieties this latter option is realized by sì / si (< ‘thus’) insertion, as illustrated by the old Neapolitan near minimal pair in (9a–b) exemplifying the competing Move and Merge options, respectively:
¹⁸ Cf. the parallel use of expletive articles in the nominal domain in conjunction with proper names as an alternative to N-to-D raising to satisfy the strong D requirement (Ledgeway 2015: §2), e.g. Gsc. [DP la [NP Yane]] ‘the Jeanne’ vs Fr. [DP Jeanne [NP Jeanne]] ‘Jeanne’. Note that, although basegeneration (viz. first Merge) is generally agreed to be prove less costly than movement, this does not imply that Gascon-type V2 systems are necessarily more frequent cross-linguistically than classic V2 systems involving verb raising – or, for that matter, that expletive articles should prove more frequent than N-to-D raising, inasmuch as the two strategies never actually compete. The base-generation Merge option will only arise in those languages whose lexicon includes a relevant functional item (e.g. C-particle, expletive article) and then only in those particular derivations whose lexical array happens to contain the relevant lexical item.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
82
(9)
a. [CP [sì fuorti cuolpi] li donava [TP li donava sì fuorti cuolpi]] such strong blows him.=give...3 (66.12) b. [CP[spissi cuolpi mortali] sì [TP le dava many blows mortal sì him.=give...3 spissi cuolpi mortali]] ‘He struck him with such strong blows / many mortal blows’ (133.66) (old Neapolitan: 14th c. Libro de la destructione de Troya)
Thus, a positive specification for (6a) above actually leads to the two further parametric choices under (10b).
(a) Is C strong?
(10)
Yes No (= marked V-to-C mvt) (b) Satisfied through Merge? … Yes: sì
No (⇒ generalized V-to-C mvt)
Given the proposed parallel between clausal and nominal structures according to which C is nothing more than a descriptive label for a subset of occurrences of D, I assume that, when strong, C must be associated with an N- / V-feature overtly in the syntax, a requirement which, in accordance with the variation formalized in (10b), can be met by either the Merge or Move options. If saturated by a V-feature this gives rise to the Move option which we have seen uniformly characterizes medieval Romance (cf. 9a), including old Gascon (cf. 3a–b), whereas the Merge option obtains whenever strong C is satisfied by an N-feature, witness its lexicalization by sì (cf. 9b). Following initial ideas proposed in Ledgeway (2012: 167f.; 2015: §3.2), I argue that this same Merge option identified for medieval Romance sì also characterizes modern (central) Gascon which must also be considered a strong C (and hence V2) language since, although it does not display the Move option (namely, generalized V-to-C movement), it does obligatorily lexicalize [+declarative] root (and embedded) C with the so-called enunciative particle que ‘that’ alongside non-veridical values on C through the other enunciative particles e (se/si) and be (cf. Table 4.1).¹⁹ Significantly, just as in medieval Romance, the strong specification of modern Gascon C predicts that the EPP edge-feature also appears on C, rather than on T. As (11) shows, this prediction is indeed borne out for Gascon where preverbal subjects are always left-peripheral
¹⁹ Benincà (2017), by contrast, sees the functions of Gascon que as a residue of a former V2 constraint.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
83
and can never occur between que and the finite verb (Pusch 1999: 144; Morin 2006: 25) since T lacks the relevant EPP feature. (11) [CP [Lou roument] que [TP (**lou roument) madura.]] the wheat que the wheat ripen...3 ‘The wheat is ripening.’ (Bouzet 1963: 26) The evidence from examples such as (11) leads us to conclude that modern Gascon, unlike modern French, is not an EPP-language in that T fails to project a preverbal subject position. Rather, subjects, just like all other constituents, are restricted to occurring in their base position within the sentential core, unless they receive particular pragmatic salience, in which case they are fronted to the left periphery. This explains why, in contrast to all other Gallo-Romance varieties, Gascon is reported to license VSO (Bouzet 1963: 35f.), witness examples such as (12a–b) where the postverbal lexical/pronominal subject, which immediately follows the lexical verb (raised to the T-domain) and immediately precedes the direct object, must occupy SpecvP. This conclusion is all the more compelling in (12a) where the v-VP-internal position of the subject is highlighted by its occurrence to the immediate right of the participle, and not just the finite verb. (12) a. Qu’ que
a have...3
b. Que haram nous que do..1 we ‘We’ll do what’s best.’
hèyt do. ço that
tou your qui which
pay father
ua a
bestiessa. stupidity
coumbienga. suit...3
The absence of a T-related preverbal subject position in Gascon is entirely consistent with the proposed V2 nature of the language, as independently maintained for other V2 languages such as late Latin (Ledgeway 2017: 186f.), medieval Romance (Ledgeway 2007: §2.2.6; 2008: 452f.), and Germanic (cf. Haider 1993; Roberts and Roussou 2002: 145; Biberauer 2003, 2004; Biberauer and Roberts 2005). Whereas in non-V2 languages like modern French and Italian (cf. Cardinaletti 1997, 2004) the dedicated preverbal subject position licenses, although not exclusively, both thematic and rhematic subjects, in a V2 language like Gascon these same pragmatic functions are typically licensed by fronting of the subject to a specifier position within the C-space. It follows that there would be very little motivation, empirical or theoretical, for a T-related preverbal subject position in a V2 language like Gascon, hence the absence of the order que+Subject+VFinite. From a diachronic perspective, the relevant parametric change in the history of Gascon involves therefore, not a loss of V2 (viz. a change from strong to weak C, with concomitant change in the EPP) as happened in other Gallo-Romance varieties, but a change in the formal realization and satisfaction of strong
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
84
C manifested in a shift from the Move option in favour of the less costly Merge option formalized under (10b). Indeed, although it has been claimed that the current system of enunciative particles and, in particular, que is the result of a relatively recent grammaticalization process not robustly attested, according to different sources, until between the sixteenth and eighteenth centuries (Ronjat 1937; Lafont 1967: 352; Hetzron 1977; Bouzet 1933; Darrigrand 1974: 32; Wüest 1985: 287; Pusch 2002), others have convincingly argued that the enunciative system represents the outcome of an original Basque sub-/adstrate influence. As such, it goes back at least as far as the medieval period (Bouzet 1932; 1933; 1951; Grosclaude 1986; Haase 1994; Marcus 2010: chap. 38; cf. also Bourciez 1946: 384; Ravier and Curesente 2005: 254; Lafitte 2019: 4–7), with attestations from as early as the twelfth century (13), although not systematically represented in early texts where it was frequently suppressed under prescriptive pressures (Joseph 1992; cf. also Bouzet 1933: 33f.; Grosclaude 1979: 7; 1986: 7). deant que i auie en pengs MDCC solis. (13) E and before que there= have...3 in mortgage 1700 coins ‘And before there was a mortgage payment of 1700 sous.’ (1179–92, Laloubère) On this view, the exceptional retention of V2 in Gascon finds a straightforward contact-induced explanation: following centuries of close contact with Basque, which is independently known to present similar typing and polarity particles (cf. affirmative / emphatic declarative ba(i), polar interrogative al, dubitative / evaluative ote, and evidential / reportative omen / bide),²⁰ the medieval (Gallo-)Romance V2 constraint was reinforced, but at the same time also aligned with the Basque model leading to a shift from the Move to the Merge options in satisfaction of the strong C setting.²¹
4.3.3 Sentential negation Further proof of the role of Basque in the emergence of the Gascon enunciative system comes from the observation that declarative que and the non-veridical ²⁰ See, among others, de Rijk (1969), Ortiz de Urbina (1989), Laka (1990), Elordieta (2001), Haddican (2004). ²¹ Note that this contact-induced explanation relates to the parametric shift from Move to Merge in the satisfaction of the V2 constraint, but has nothing to say about the individual grammaticalization processes underlying the individual enunciative particles. Indeed, the analysis developed in Floricic (2012) and this volume that declarative que represents the reanalysis of an original cleft es que . . . ‘it.is that . . . ’ as a predicate focus construction (cf. also Pusch 1999), as witnessed by its continuation to the present day in some Gascon varieties (cf. also Rohlfs 1970a: 206 n.366), provides a highly plausible account.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
85
C-particles generally prove incompatible with sentential negation,²² witness the absence of que in (13): (13) Ne
’ve =you
parli speak...1
pas,
Sénher. sir
(Darrigrand 1974: 41)
This incompatibility finds a striking, and I would argue not coincidental, parallel in Basque where the (emphatic) declarative particle ba(i) also proves incompatible with negator ez (Laka 1990: 103–6).²³ In her classic analysis of such facts, Laka (1990) proposes that the Basque affirmative and negative particles ba(i) and ez spell out opposite values of a left-peripheral polarity projection (ΣP), hence their complementary distribution. It is tempting therefore to see in the complementary distribution of Gascon que and (>) nou(n)/non/ne a contact-induced replication of a Basque model, such that two variants of the unmarked declarative C-particle must be recognised: que [+affirmative] and nou [–affirmative]. Indeed, many traditional descriptions and analyses of Gascon enunciatives (e.g. Bouzet 1932, 1933, 1975: 66; Wüest 1985; Marcus 2010: 32f., 54f.) explicitly include nou among the members of the system. Assuming this to be correct, I take que and nou to be the affirmative and non-affirmative lexicalizations of the default spell-out of the declarative C-head (cf. Suïls Subirà and Ribes 2015: 550), a conclusion which immediately explains their incompatibility in (13) in line with an original Basque pattern. This conclusion also has significant repercussions for our understanding of Gascon negation. Unlike other Romance reflexes of the preverbal negator , which are standardly taken to lexicalize a head within the T-domain, I am claiming that that Gascon nou is a C-head. This difference explains why, in contrast to neighbouring Occitan and (spoken) langue d’oïl varieties where negation has now generally reached Stage III of Jespersen’s Cycle following loss of preverbal (e.g. Fr. ne . . . pas > pas; cf. Smith 2016: 309f.), Gascon apparently continues Stage II in the cycle with obligatory retention of nou in conjunction with pas (Darrigrand 1974: 4; Puyau 2013: 33; Sauzet and Oliviéri 2016: 346f.). In the former case loss of represents a reversal in the relationship between the head (e.g. ne) and specifier (e.g. pas) of a T-related NegP (Poletto 2016: 836f.), whereas in Gascon nou is not part of a NegP in conjunction with a postverbal pas. Rather, it is a non-affirmative marker of declarative sentence mood, an obligatory element of the C-system not readily susceptible to attrition which therefore falls outside of
²² Ronjat (1913: 80), Bouzet (1963: 26), Pusch (1999: 113), Morin (2006: 12, 45f.), Karenova (2008: 48), Marcus (2010: 43, §5.2.7), Floricic (2012: 10f.), Puyau (2013: 32), Suïls Subirà and Ribes (2015: 550). ²³ Floricic (2012; Chapter 5, this volume) argues that incompatibility of que with negation represents a residual effect of the former predicate focus construction from which it originates, inasmuch as focus marking of predication frequently clashes with negation cross-linguistically.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
86
Jespersen’s Cycle. Consequently, on the surface Gascon negation appears to be at stage II (14a), but underlying it is at stage III (14b) just like its Occitan neighbours and spoken French (14c): (14) a. b. [CP c. [CP
Ne Ne
[TP [TP
je I
souy souy suis be...1
pas pas pas
bàscou. bàscou.]] basque.]] (Fr.) Basque (Puyau 2013: 33)
Another interesting consequence of this analysis is the observation that in certain marked contexts que and nou can co-occur (cf. Bouzet 1963: 26; Field 1985: 83; Rohlfs 1970: 208; Fossat 2006: 161; González i Planas 2009: 90 fn.8; Marcus 2010: 43; Floricic 2012: 5; Puyau 2013: 33): bouy d’ (15) Que ne want...1 of ‘But I don’t want any of that wine!’
aquét(h) that
bî! wine (Puyau 2013: 33)
As the English translation of (15) indicates, whenever que co-occurs with nou, and only in that order, the negator receives a strong presuppositional interpretation serving to deny a previous assertion, e.g. as a response to ‘Here you are, have some of this wine’. Crucially, as (15) illustrates, in these cases postverbal pas is not realized (Bouzet 1963: 26; Puyau 2013: 32), a fact which I take to indicate that nou in such examples is not the non-affirmative declarative C-head (that position is already filled by que), but realizes SpecNegP, the position otherwise lexicalized by pas.²⁴ From SpecNegP nou then cliticizes to the finite verb raised to the highest position of the T-domain (cf. Cinque 1999: 124), transparently giving rise to the order que+nou+V and the observed emphatic declarative interpretation.²⁵
4.3.4 Root clauses Following Cinque’s (1999: 84–6) claims about the fine structure of the sentential core, I assume that the highest portion of the T-domain includes projections specialized for speech act mood (declarative, interrogative, optative, exclamative, imperative) and evidential mood (quotative) which, for expositional convenience, ²⁴ This also explains the observed presuppositional reading of nou in such contexts if, with Cinque (1999: 121, 220 fn.38), we take SpecNegP2 to be the position in which presuppositional negators are licensed (cf. also Garzonio and Poletto 2010). ²⁵ In some dialects the co-occurrence of que and nou does not knock out pas (Field 1985; Pusch 2000a; Marcus 2010: 43), suggesting that in these varieties nou can lexicalize the head of NegP, and not its specifier.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
87
I conflate here into a single projection MoodP.²⁶ As part of a clause-typing operation, this projection must, in turn, enter into a checking relation with the Force-Fin system where the sentence mood of the T-domain can be formally licensed in a local relation with the C-domain.²⁷ With these assumptions in mind, I propose that non-veridical polar (viz. [–declarative]) clause types involve the generation of a relevant null operator in SpecMoodP which raises to SpecCP where it enters into a Spec-Head relation with Cº. This feature-checking operation is spelt out at PF in the lexicalization of the C-head with one of the various enunciative particles in accordance with the particular interpretable features of the modal operator, as sketched in (16a–b). (16) a. [CP-Force-Fin [Spec Opi] [C’ Ci (= (s) e, si, be, ço/ça) [MoodP [Spec Opi] . . . ]]] b. [ForceP Forcei *Top [FinP [Spec Opi] [Fin’ Fini (= (s)e, si, be, ço / ça) [MoodP [Spec Opi] . . . ]]]] On this view, the overt lexicalization of the C-head, a morphophonological reflex of the Spec-Head relation, can be understood to spell out and make visible the content of the modal illocutionary force associated with the null operator raised to SpecCP. In accordance with Giorgi and Pianesi’s (1996) Feature Scattering Hypothesis, in the absence of left-peripheral topic constituents the C-head instantiates a syncretic realization of the Force-Fin system, viz. CFin-Force (cf. 16a), whereas in the presence of left-peripheral topics the features of the Force-Fin system are scattered such that they head their own projections (cf. 16b). In both cases the modal force of the clause is invariably checked against Fin, a head traditionally associated with licensing modal properties of the clause, as well as against Force, the locus of sentential force in Rizzi (1997; cf. also Munaro 2010), with which Fin is either bundled into a single head or by which Fin is bound within the C-domain (cf. fn.28). The result is invariably a V2 structure involving the obligatory lexicalization of the C-head and movement of a null modal operator to its associated Specifier position. Turning now to declarative root clauses, I assume these do not involve a modal operator, in that declarative is the unmarked sentence mood which obtains whenever SpecCP is not targeted by a specific modal operator, yielding a default
²⁶ Although not considered here, MoodP can also be assumed to subsume Cinque’s (1999: 84f.) evaluative mood, thereby also accounting for ya / ye / ja / je mentioned in fn.13 above. ²⁷ Though not explored further here, the C-domain can be further assumed to interface with a higher speech act domain (cf. Speas and Tenny 2003) where the basic sentence moods are formally licensed in a layered Speech Act Phrase that configurationally encodes the P(ragmatic)-roles speaker, hearer, and utterance content (cf. also Giorgi’s (2010) claims about the encoding of speaker in C). Such an account would offer an elegant explanation, for example, of the restriction of interrogative que to just the speaker in some Piedmontese Occitan varieties analysed by Benincà (2014, 2017).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
88
value spelt out on the C-head with the lexicalization of the erstwhile unmarked complementizer que ‘that’ or its non-affirmative variant nou: (17) [CP-Force-Fin [Spec____] [C0 C (= que/nou) [MoodP . . . ]]] Significantly, the differing structural representations of [–declarative] and [+declarative] clauses in (16)–(17) make some non-trivial predictions about the nature and distribution of constituents that can surface in the left periphery, replicating surface V2 effects similar to those observed in Germanic and medieval Romance. In particular, the derivation of [–declarative] clauses in (16) leads us to expect a ‘bottleneck effect’ (Haegeman 1996; Roberts 2004; Cardinaletti 2010), since raising of the relevant null modal operator to SpecCP(Force-)Fin precludes under Rizzi’s (1990a) Relativized Minimality any further movement through that same position, thereby limiting movement to the left periphery to one constituent. It follows that in [–declarative] clauses the only overt constituents that can occur before the C-particles are those which are base-generated in the left-periphery, namely frame and theme elements (henceforth underlined; cf. Anagnostopoulou 1997; Wiltschko 1997; Frey 2004), but not those which are moved to the focus field (henceforth in small caps; cf. Poletto 2002). Direct evidence for this bottleneck effect comes from examples like those in (18)–(20). (18) a. E cerques? search...2 b. Q (**e) boulét(s) when want...2 c. Maria e Maria
parla speal...3
(Lespy 1858: 258) biéne? come.
gascon? Gascon
(Puyau 2013: 65) (Morin 2006: 26)
donc lo vin! (19) a. B’ aimas love...2 so the wine (Darrigrand and Grosclaude 1971: 3) b. Q (**b’) ei cadut sus l’ ostau! which fate be...3 fall. on the house ‘What a fatality fell upon the house!’ (Darrigrand 1974: 119) c. Maria be canta Maria sing...3 ‘Doesn’t Maria sing well!’
plan! well (Morin 2006: 27)
podossi aver un setmanèr Occitan! (20) a. E can...1 have. a weekly Occitan ‘If only I could have an Occitan weekly newspaper!’ (Darrigrand 1974: 238) b. Lo Diable se t’ en carrege! the devil you=therefrom=remove...3 ‘May the devil carry you away!’ (Darrigrand 1974: 168)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
89
The (a)-examples in (18)–(20) demonstrate the raising of a null interrogative, exclamative, and optative modal operator to SpecCP, the content of which is spelt out on the C-head through lexicalization of the appropriate C-particle. By contrast, the (b)-examples in (18)–(19) involve raising of an overt wh-operator to SpecCP, in which case the relevant C-particle is now excluded since its presence would involve a checking operation with a null modal operator whose presence would inhibit wh-movement. In short, sentential mood is spelt out just once in the C-system, either on the C-head when SpecCP hosts a null modal operator or directly through an overt wh-operator raised to SpecCP,²⁸ correctly predicting the restriction of the C-particles e/be to total interrogatives/exclamatives (Bouzet 1963: 27; 1975: 67; Rohlfs 1970: 210; Darrigrand 1974: 84; Pusch 2001: 385; Marcus 2010: 38, 48; Rigau and Suïls 2010: 154f.; Joly 213: 249) and their complementary distribution with overt wh-phrases (Karenova 2006: 4; Marcus 2010: §5.2.8; Rigau and Suïls 2010: 155; Puyau 2013: 65). It follows that the only elements which can precede C-particles are those which avoid the bottleneck effect through base-generation in the left periphery such as the topical subjects in the (c)-examples in (18)–(20).²⁹ A similar explanation applies to apparent counterexamples such as those involving the overt wh-operator ‘why’ (21a–b) which has been argued by Rizzi (2001a) to be first-merged above the Focus space (in SpecIntP), as well as examples of D-linked wh-phrases (Pesetsky 1987, 2000) such as (22a–b) where the lexical restriction licenses a referential reading reflected in their basegeneration within the theme space (Grohmann 1998, 2003; Rizzi 2001b). devath (21) a. Perqué lo bohon e viu why the mole live...3 under b. Enta qué e bos aquére for what want...2 that (22) a. A quina pena to which punishment condamnar? condemn.
e
terra? earth (Darrigrand 1974: 90) moundéde? money (Puyau 2013: 81)
’u cau him=be.necessary...3 (Darrigrand 1974: 90)
²⁸ In polar root (and embedded) interrogatives in the south-eastern Gascon varieties studied by Rigau and Suïls (2010), the CFin-particle e can also be doubled by a higher C-particle se ‘if’ in Intº whenever the left periphery hosts an intervening topic, e.g. (Sabes) se Joan e poirà vier? ‘(Do you know) se Joan e will come?’ (Rigau and Suïls 2010: 160). As predicted, this same higher interrogative C-particle se, unlike the lower C-particle e, can also precede wh-phrases in partial root (and embedded) questions: (Demana-les) se ’ vien aqueras gojatas? ‘(Ask them) se from where do those girls come?’ (Rigau and Suïls 2010: 155). ²⁹ We can include here also the fronted quote in quotative inversion if we assume that the quote is base-generated in the Topic Field of the left periphery co-indexed with a null quotative operator raised from MoodP to SpecCP which binds the null complement of the parenthetical verb (for various ideas along these lines, see discussion in Matos 2013).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
90
b. Des(em)puch since
quoan(t) how.much
de of
téms time
e
demourat(s) aci? live...2 here (Puyau 2013: 67)
By contrast, in declarative clauses the analysis outlined in (17) predicts that no such bottleneck effect should ever arise, since SpecCP remains available for focus-movement. Consequently, in declaratives we find immediately before que (and nou) all types of elements, irrespective of their grammatical function, that are argued to target the Focus field under movement, including bare quantifiers and indefinites (23a–b), information and mirative foci (23c–e), and contrastive foci (23f–h):³⁰ hè grand oumprère. (23) a. que-m all que=self=do...3 great shade ‘I’m eclipsed by everything.’ (Lespy 1876: 97) b. Q’ que truca à la porte. someone que knock...3 at the door (J. Hustach, de Camelat 1933: 110) c. U ` que l’ a hissat. a snake que him=have...3 bite. (Darrigrand 1974: 177) d. Q qu’ èm! quits. que be...1 (de Camelat n.d.: 42) e. ` que l’ aura empipautit! in two hours que it= have..3 soil. ‘in two hours he’ll have soiled it all!’ (Darrigrand 1974: 103) f. P que cau! pay. taxes que be.necessary...3 ‘Taxes have to be paid!’ (Puyau 2013: 85) g. Y ´ que-b bieni cerca. I also que=you. come...1 search. (de Camelat n.d.: 14) h. M ’ ´ que sabou tourneya lou petit more well than nobody que knew...3 turn. the small counde en prose. tale in prose ‘Better than anybody else he could turn short stories into prose.’ (M. de Camelat 1933: 87) Naturally, in declaratives all sorts of theme constituent can also occur before que (and nou), including null (24a) and overt (24b) referential subjects, complements ³⁰ Examples (23a–c, g) in which que is immediately preceded by an informationally or contrastively focused subject invalidate the claims of González i Planas (2009: 87–9) that preverbal subjects in Gascon can only be (aboutness) topics, with informationally focused subjects obligatorily occurring in postverbal position and contrastively focused subjects in a fronted cleft structure.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
91
(24c–d), contrastive subjects (24e), phrasal (24f) and clausal (24g) adverbials, and scene-setters (24h). Furthermore, these can combine among themselves (25a–b) and, in turn, with focalized constituents in the order Theme-Focus (25c–e). tot lo (24) a. Que’m mingi que=me eat...1 all the b. Lous due amics, eths, que the two friends they que
pan bread (Darrigrand 1974: 25) plouraben cry...3 (J. Hustach, de Camelat 1933: 110)
c. Ço que vedi que’m platz. that which see...1 que=me please...3 (Darrigrand 1974: 72) d. Aus mainatages que los va parlar Maria to.the children que them.=go...3 speak. Maria (Morin 2008: 141) e. La Yane que-m bôu e you que the Jeanne que=me want...3 and I que la bouy her=want...1 (M. de Camelat 1916: 2) f. Labéts que m’ en as mandade tau marcat then que me=thence=have...2 send. to.the market toute soule. all alone (Palay 1927: 3) g. Se demoras tròp au só que’t vas if stay...2 too.much to.the sun que=you go...2 usclar. burn. (Darrigrand 1974: 102) h. Un dia en se passejant tot plentiu au ras d’ un a day in self= stroll. all plaintive at.the edge of a broishagar, que trobè lo bohon. undergrowth que find...3 the mole ‘One day walking mournfully through the undergrowth he came across the mole.’ (Darrigrand 1974: 82) (25) a. A la fin de at the end of hèits make..
la partida los jogaors que’s son the match the players que=self be...3 ahupar. boo. (Darriagrdan 1974: 112)
b. Joan lo Pèc, bèth còp on sa mair èra tau marcat, Jean the silly once where his mother be...3 to.the market que ’s cope ua bèra tranche de jambon que self=cut...3 a nice slice of ham (Darrigrand 1974: 100)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
92
c. Aquéste, ’ que là-m pagara! this someone que it=me= pay..3 d. L’ artiste que the artist already que cau be.necessary...3 e. Per u bèt by a nice ère be...3
sabè knew...3
(de Camelat n.d.: 9) béde see.
coume as
(De Caillabère 1965: 13)
dilus d’ estiéu, ´ qu’ Monday of summer a friend my que partit d’ Arguilès left from Argelès (Y. Bourdéte, de Camelat 1933: 58)
4.3.5 Embedded clauses The facts outlined above for Gascon root clauses also carry over, albeit with some minor modifications, to embedded clauses, where C must also be considered strong yielding a symmetrical V2 syntax. Given that quotative (s)e / si / ço / ça and exclamative be are independently excluded from embedded contexts (on the absence of total embedded exclamatives, see Bosque 2017: 45), the number of C-particles in embedded contexts is reduced to two: declarative que / nou and non-veridical e. The former introduces both propositional complements (generally requiring an indicative verb) and those adverbial clauses (e.g. causal, result, non-restrictive relatives) which, since the seminal work of Hooper and Thompson (1973), have been described in the literature as assertive, whereas the latter introduces irrealis complements (typically containing a subjunctive verb) and adverbial clauses (e.g. temporal, purpose, conditional, restrictive relatives) whose propositional content is presupposed (cf. Pusch 2000a: 198; 2000b: 628; 2002: 111; 2007: 99f.; Marcus 2010: 68). While assertive complement and adverbial clauses are known to display main clause phenomena, non-assertive clauses do not, a distinction which Haegeman derives from their respective ‘peripheral’ and ‘central’ integration within the matrix domain:³¹ in contrast to ‘central’ clauses which modify the matrix predicate and are structurally integrated into the matrix speech act by merging with the matrix clause before the completion of TP, ‘peripheral’ clauses display independent illocutionary force merging with the matrix clause after the completion of CP. Following a long tradition, Haegeman further argues that, in terms of their internal syntax, ‘central’ clauses are derived by movement of a TP-internal null temporal/modal operator to the C-domain, unlike ‘peripheral’ clauses whose derivation does not involve any such movement. Extending the
³¹ See Haegeman (2003, 2007, 2009, 2010a,b, 2012a,b, 2013, 2014), Haegeman and Ürögdi (2010), Danckaert and Haegeman (2012).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
93
analysis of [–declarative] root clauses above (cf. 16a–b), I therefore assume a derivation of embedded ‘central’ clauses along the lines of (26a–b). (26) a. [CP-Force-Fin [Spec Opi] [C’ Ci (= , , , . . . ) [MoodP [Spec Opi] . . . ]]] b. [ForceP [Force’ Forcei (= , , , . . . ) [TopP *Top [FinP [Spec Opi] [Fin’ Fini (= e) [MoodP [Spec Opi] . . . ]]]]]] In (26) a null temporal/modal operator base-generated in MoodP is raised to SpecCP where it enters into a Spec-Head relation with Cº, a checking operation spelt out at PF in the lexicalization of the C-head as , , , , etc. in accordance with the temporal/modal features of the raised null operator. Relevant Gascon examples from Bouzet (1963: 27) are given in (27a–b) where, in the absence of left-peripheral topicalized constituents, the C-head involves a syncretic instantiation of the Force-Fin system (cf. 26a). (27) a. Coan when
tribalhaba work...3
aquiu there
Yanto Jeannot
b. Si hasè l’ amic tout ço qui ditz if do...3 the friend all that that say...3 When left-peripheral topics are present (28a–b), the features of the Force-Fin system are necessarily scattered each heading their own projection (cf. 26b), with Forceº lexicalized by the relevant temporal/modal adverbial (, , , ) and Finº by the non-veridical C-particle e. I assume that the adverbial is first licensed in C-Finº as a reflex of the Spec-Head checking relation with the null operator, before raising to Forceº where it is spelt out at PF, with Finº lexicalized by the non-veridical C-particle e in accordance with our previous observation that sentential mood is spelt out at least once in the CFin-system, either on C-Finº when SpecCP hosts a null modal operator or directly through an overt wh-operator raised to SpecCP.³²
³² As also noted above (fn.17) in relation to root clauses, e is elided before vowels, though underlying still present. (i) a. Que l’
u de nous e se-n ane.
b. Que l’ u de nous Ø ane t’ aciu. que the one of us e self=thence= go...3 to
there (Bouzet, 1963: 28)
Our corpus also includes a fair number of ‘central’ clauses in which e is replaced by que as in (ii): (ii)
Que que
bouleri, petit Jesus, que la may qu’ y biencousse tabéy want..1 small Jesus that the mother que =there come...3 also (P-D. Lafore, de Camelat 1933: 49)
Although others have noted this tendency (Pusch 1999: 115; 2000a: 193f., 203f. n.13; 2002: 107; Marcus 2010: §5.2.5), albeit without any explanation, a closer examination of the relevant examples reveals that
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
94
(28) a. Coan when
Yantot Jeannot
e e
tribalhaba work...3
aquiu there
b. Si l’amic e hasè tout ço qui ditz if the friend e do...3 all that that say...3 This analysis can be straightforwardly extended to restrictive relative clauses which are also assumed to involve movement of a null wh-operator, as well as subjunctive complement clauses where the licensing of subjunctive mood— presumably coinciding with the marked value of Cinque’s (1999: 88) Moodirrealis conflated here with MoodP—has been argued by Manzini (1996) to involve the presence of a relevant operator in SpecCP. Once again, in the absence of leftperipheral topics the Force-Fin system is realized syncretically, with Cº spelt out at PF as que / qui ‘that’ once checked against the null operator in SpecCP (29a–b).³³ However, when topics are hosted in the left-periphery, Forceº and Finº are scattered and spelt out at PF as que / qui and e, respectively (30a–b). Following examples from Bouzet (1963: 27): ço qui-t (29) a. Entenes understand...2 that that=you b. Que cau que que be.necessary...3 that l’ u de nous-autis the one of we-others
ditz tou may? say...3 your mother se-n ane self=thence=go...3
ço qui tou (30) a. Entenes understand...2 that that your b. Que cau que que be.necessary...3 that se-n ane. self=thence= go...3
may e-t ditz? mother e=you say...3 l’ u de nous-autis e the one of we-others e
Consequently, this analysis provides a principled explanation for the traditional observation that the enunciative particle e surfaces in a subset of adverbial clauses and subjunctive complements whenever the subordinator and the verb are not
the overwhelming majority involve the non-veridical C-particle before a vowel-initial verb or proclitic adjoined to the verb as in (ii). This suggests the existence of a morphophonological rule before vowels yielding either elision (e > Ø) or reinforcement (e > que + e > qu’) of the non-veridical particle. ³³ Cf. variation in the realization of the non-veridical C-particle se/si in §4.3.1. At a certain level of abstraction, lexicalization of syncretic Force-Fin by que / qui can be transparently analysed as an amalgam of Force qu- ‘that’ and non-veridical Fin e, namely qu-+e > que/-i, which are visible in the scattered realizations (cf. 30a–b).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
95
strictly adjacent.³⁴ In our analysis, e surfaces precisely in these ‘central’ embedded clause types, and only these, since their derivation alone involves movement of a null operator from TP to CP, the content of which is spelt out by the non-veridical C-particle e lexicalized in C-Fin. Turning now to ‘peripheral’ clauses, given their resemblance to root clauses I extend to these the analysis of root declaratives in (16b), according to which they do not involve raising of a temporal/modal operator: (31) a. [CP-Force-Fin [Spec____] [C’ C (= que/nou) [MoodP . . . ]]] b. [ForceP [Force’ Force (= que) [Top/FocP *Top/Foc [FinP [Spec____] [Fin’ Fin (= que/nou) [MoodP . . . ]]]]]] When the embedded left periphery does not host topics or foci (cf. 31a), Force-Fin are bundled together in a syncretic head which, in the absence of an operator, receives a default interpretation spelt out as que, a lexicalization of the C-head which felicitously coincides with both that of the declarative complementizer/ subordinator (viz. Force) and the declarative enunciative particle (viz. Fin):³⁵ i pois pas anar pr’amor que soi (32) N’ there= can...1 go. because that be...1 carcat de coientas. charged of commitments ‘I can’t go because I’m busy.’ (Darrigrand 1974: 237) When, however, the embedded Topic-Focus spaces are activated (cf. 31b), Force and Fin head independent projections both spelt out as que: (33) permou because
que lous cepàyres que soun matiès that the mushroom.foragers que be...3 early.risers (O. Coustet, de Camelat 1933: 117)
It is important to note that examples such as (33) should not be confused with cases of Romance recomplementation in which, under specific structural
³⁴ Cf. Lespy (1858: 259), Ronjat (1913: 84), Bouzet (1963: 27; 1975: 67), Lafont (1964: 41), Rohlfs (1970a: 210f.), Darrigrand and Grosclaude (1971: 3), Darrigrand (1974: 84, 239), Pusch (2000a: 192f.; 2000b: 628; 2001: 385–91; 2007: 99f.), Morin (2005: 65; 2006: 37f.), Karenova (2006: 3f.; 2008: 50), Marcus (2010: 39–42, 68; §5.2.5), Floricic (2012: 5f.), Joly (2013: 249, 253), Suïls Subirà and Ribes (2015: 552). ³⁵ In negative clauses (cf. i) where the form of Force (que) and Fin (nou/non/ne) do not coincide, the head cannot be lexicalized syncretically and Force and Fin head their own projections: (i) Que soi que be...1
segur sure
que that
ne
son be...3
pas
maduras ripe
las arhagas. the strawberries (Morin 2006: 47)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
96
conditions, complementizers can simultaneously lexicalize more than one head of the C-domain.³⁶ First, Gascon structures such as (33) are obligatory, not optional as with Romance recomplementation, whenever the Topic-Focus spaces are activated. Second, the lower occurrence of que in examples such as (33) is not a complementizer but, rather, a C-particle marking the sentence mood of the clause as part of a system of enunciative particles not found outside of Gascon. Indeed, the particle status of the lower que is also shown by examples (34a–b). (34) a. Que m’ a que me=have...3 maridabes marry...2
dit say.
Yantot Jeannot
de of
que-t that=you
(Bouzet 1963: 35)
b. Per’mor que m’ avisi de que despuish hèra, un because that me=realize...3 of that since much a fenomèn centralista e centralizador que torna com phenomen centralist and centralizing que return...3 as um simptòma. a symptom (S. Javaloyès, https://ljgascon.wordpress.com/2015/04/23/lenga-de-doman/) On a par with many Ibero-Romance varieties, Gascon displays the phenomenon of dequeísmo (Bouzet 1963: 58f.; 1975: 78f.; Joly 2013: 256f.; Puyau 2013: 124f.), whereby the embedded finite complementizer que can be pleonastically preceded by the preposition de ‘of’ (34a). However, as (34b) illustrates, while the higher occurrence of que can be preceded by de in accordance with its complementizer status, the lower occurrence, the so-called enunciative particle, cannot. Another piece of evidence that points in the same direction is the behaviour of negation. As already noted in §4.3.3, unlike the complementizer que (cf. (i) in fn. 35), the root C-particle que is generally incompatible with negator nou since both represent different (viz. ±affirmative) instantiations of the same [+declarative] C-head. Therefore as predicted, in embedded contexts the lower que cannot occur with negation (** . . . que+XP+que+nou), since it is not a complementizer and must be substituted by nou: aryént qui nou s’ emplegue pas (35) permou que l’ because that the money that self=employ...3 tau bé de touts, aus noustes oelhs nou pot so well of all to.the our eyes can...3 pas da mey de balou give. more of value (S. Palay, de Camelat 1933: 74)
³⁶ For an overview and relevant bibliography, see Ledgeway (2012: 165–8) and Villa-García (2015).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
97
Significantly, the differing structural representations of [–declarative] and [+declarative] clauses in (26) and (31) once again make some non-trivial predictions about the nature and distribution of constituents that can be fronted to the left periphery.³⁷ In accordance with the bottleneck effect introduced in §4.3.2, the only constituents that can occur before the non-veridical e particle are those first-merged above the Focus Field, witness the left-peripheral topicalizations in (36): lous lous arrepics (36) a. coum las campanes e dingouleyaben as the bells e rang...3 the their chimes (de Camelat 1933: 67) b. Si lou maridatje e-s’ if the marriage e=self ‘If the marriage is called off’
coupe cut...3 (Palay 1927 : 3)
c. Qu’ aténdi que lou dinna e sie près. que wait...1 that the dinner e be...3 ready (Joly 2013: 249) By contrast, in ‘peripheral’ clauses whose derivation does not involve the movement of a null operator, both foci (37a) and theme elements (37b) may legitimately (co-)occur (37c–d) before que/nou. que que pouderém sénse (37) a. que sémble [ . . . ] que seem...3 that all that can..1 without pénes debisa parié difficulties speak. equally ‘it seems [ . . . ] that we can all speak equally without problems’ (de Camelat 1933: 107) b. Sapiat d’abord que l’ oustau de Casadbath qu’ know..2 first that the house of Casadbath que ey ue de las mey bielhes cases be...3 one of the more old houses (F. Mascaraux, de Camelat 1933: 94) c. dab l’ ahide que lhèu ’ que l’ embitère with the hope that perhaps someone que him=invite..3 (Bouzet 1963: 63)
³⁷ As the discussion below shows, there is no evidence to support the claim of González i Planas (2009) that in embedded contexts preverbal subjects preceding que and e can only be aboutness and familiar, topics, respectively, nor for the concomitant implication that focused subjects are invariably excluded before que.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
98
d. Que crési qu’ aquère neyt las aurelhes qu’ous que believe...1 that that night the shepherds que=them deoun turla must...3 drink. (F. Mascaraux, de Camelat 1933: 101)
Observe, finally, how given our proposed analysis of the Gascon embedded enunciative system in terms of a strong C requirement, we are led to conclude that Gascon is characterized by a symmetrical V2 syntax. ‘Central’ clauses (cf. examples 27–30, 36) formally qualify as examples of embedded V2 structures, since they invariably involve both lexicalization of the (syncretic) C(Force-)Fin head and raising of a null modal/temporal operator to its associated specifier in satisfaction of an edge feature (viz. full V2). ‘Peripheral’ clauses, by contrast, only present the first, and arguably most crucial, of these two ingredients of the V2 constraint (viz. partial V2) since, while (syncretic) C(Force-)Fin is invariably lexicalized, operator movement to SpecCP(Force-)Fin in satisfaction of an edge feature is conditioned by the particular information structure of the utterance (cf. examples 37a,c).³⁸ There thus emerges a major difference in the distribution of embedded V2 in medieval Romance (old Sardinian excepted) and modern Gascon: while in the former embedded V2 is restricted to ‘peripheral’ clauses (so-called bridge contexts) but excluded from ‘central’ clauses, in the latter embedded V2 is systematic in both ‘peripheral’ clauses (partial / full V2) and ‘central’ clauses (full V2).
4.4 Summary and conclusion Exploiting parallels between nominal and clausal structures, I have argued that the strong / weak D dimension of parametric variation for nominals can be extended to clauses, such that V2 syntax can be reinterpreted as the reflex of a strong C setting. On this view, we observe in the history of most Gallo-Romance varieties a parametric shift from strong to weak C manifested in the loss of generalized Vto-C movement and the concomitant reassignment of the EPP edge-feature from CP to TP, as witnessed in the emergence of a dedicated preverbal subject position and reversal in the null subject parameter. Thus, while generalized V2 movement triggered by a semantically uninterpretable V-feature in declarative contexts has been systematically lost, V-to-C movement manifested in verb-subject inversion and pronominal enclisis is residually retained just in those contexts where movement plays a role in interpretation (and is hence associated with a semantically interpretable V-feature) licensing a series of non-veridical polarity values. ³⁸ Cf. the analysis of embedded VSO in old Sardinian in Lombardi (2007) and Wolfe (2012, 2015a, 2015b, 2018c: §§3.1–2).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
99
Within this scenario, I have shown that Gascon represents a major exception having uniquely retained its medieval V2 syntax and, indeed, further extended it to embedded contexts. In particular, in the passage from medieval to modern Gascon, the grammar has witnessed a radical change in the formal realization of the strong C head requirement (while the accompanying EPP edge-feature remains unchanged) such that strong C is no longer satisfied through the Move option raising the finite verb to the C position, but through the Merge option directly lexicalizing the latter position with a so-called ‘enunciative’ particle (cf. V2 satisfaction in Brythonic Celtic argued by Roberts 2004, as well as the Late Merge Principle developed in van Gelderen 2004). This development represents the result of intensive contact with Basque, a language independently known to present similar preverbal particles, highlighting how the medieval Gallo-Romance V2 constraint was exceptionally reinforced in this area, but at the same time aligned with a Basque model triggering a shift from the Move to the Merge options in satisfaction of strong C and the emergence of an elaborate system of C-particles.³⁹
³⁹ In contrast to Gascon, (some of) the Basque preverbal particles are argued to be phrasal and hence target a C-related operator position within a particular type of partial V2 syntax (see Haddican and Eldorieta 2013).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
5 Dialectological evidence for a predicate focus analysis of Gascon que Franck Floricic
Et il est certain que la répartition géographique de l’emploi du que déclaratif en proposition principale affirmative, jou que suy bienut = « je suis venu », est bien près d’être linguistiquement aussi caractéristique que tel ou tel trait phonétique ou morphologique propre aux parlers gascons. (Millardet 1923: 453)¹
5.1 Introduction Gascon is an Occitan variety spoken in the southwestern part of the Occitan domain (Map 5.1).² Many of the most original features of Gascon are shared with at least some Ibero-Romance varieties (cf. Gavel 1920: 303): debuccalization of word-initial [f ], vowel prosthesis before the rhotic [r], [e] prosthesis in [s] + stop sequences, merger of Latin [b] and [w], evolution of medial [ll] towards [r], velarization of final [l], the use of nat (< , see Bourciez 1910: 266)) and digỹŋ (< ()) as negative indefinites, etc. (cf. Bourciez 1892).³ The question I will address here, however, regards the so-called ‘que déclaratif’ or ‘que énonciatif ’, whose distribution is illustrated in (1) with examples from Fleischer (1913: 121) and Ronjat (1913: 80): ¹ ‘And it is certain that the geographical distribution of the use of the declarative que in the main affirmative clause, jou que suy bienut = ‘I have come’, is very close to being linguistically as characteristic as any phonetic or morphological feature specific to Gascon varieties.’ ² My warmest thanks go to Denis Creissels, Jean-Louis Fossat, Bernd Heine, Jean Lafitte, Martin Maiden, Jean-Louis Massourre, Lucia Molinu, Paolo Ramat, Xavier Ravier, and Sam Wolfe for their comments and criticisms. The ideas developed in this chapter were first presented in Toulouse at the seminar Linguistique et dialectologie occitane et romane, in 2013, in Split at the Societas Linguistica Europaea, 46th Annual Meeting, in 2013 and in Wrocław (Languages in Contact 2018). Needless to say, I am solely responsible for any shortcomings in the final version of this chapter. ³ The main phonetic features of Gascon are summarized by Baldinger (1958: 245f.), who crucially relies on Luchaire (1877: 23). For a more extensive presentation of typical phonetic and morphosyntactic features of Gascon, see Ravier (1991), Haase (1997), etc. Variation and Change in Gallo-Romance Grammar. First edition. Sam Wolfe and Martin Maiden (eds). This chapter © Franck Floricic 2020. First published in 2020 by Oxford University Press.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
Médoc
Bordeaux Entre-deux-Mers Gironde Buch
Bazadais
nd
e
Born
Gra
nd
eL
a
Albiet Landes
Marensin
Mont-deMarsan
Armagnac
Lomagne
Chalosse
Bayonne
Toulouse Vic-Bilh
Béarn
Pau Tarbes Bigorre
Comminges
Barége Lavedan
Couserans
Aran
Map 5.1. The Gascon linguistic domain described by Lafitte (2005: 377) Adapted from Lartigue (1998: 23)
(1)
a. Era ceremounia que the.. ceremony. ‘The ceremony is beginning.’ b. Qué bèni come.1 ‘I am coming.’
couménce begin.3
101
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
102
c. Qué plau rain.3 ‘It’s raining.’ d. Eras mountanhos qu’èron the.. mountain. were ‘The mountains were cold.’
frédos cold..
e. Que’t pàrli =2 talk.1 ‘I’m talking to you.’ As shown in (1a–d), the particle ‘que’ immediately precedes the verb, whatever its argumental structure; the particle ‘que’ can be separated from the verb mainly by clitics, as shown in (1e). The distributional constraints in the use of ‘que’ are rather complex and have given rise to many discussions and various hypotheses among Romanists. I will not discuss the syntactic status of the different ‘particules énonciatives’ found in Gascon, i.e. bè, se, e, jà, etc. (cf. Fossat 2006) nor shall I investigate the supposed correlation between Basque and Gascon constructions and the interference between enunciative particles and other types of modalities (interrogation, negation, etc.). I will just provide evidence for a predicate focus analysis of Gascon ‘que’, basing myself on unpublished dialectological material gathered in the 1970s in the Occitan area. I will show that these data support the view put forth by Dauzat (1906: 234), according to which ‘the patois of a given linguistic mass and of the same origin—for example the Romance varieties— present to us in their infinite geographical variety a simultaneous overview, in space, of the phenomena that have occurred in time at very different periods’ [my translation]. Even though Pusch (2000: 201) observes that ‘any dia- or panchronic account of Gascon enunciatives inevitably remains conjectural, since we lack sufficient textual data from earlier stages of this Occitan variety’, the data of the locality of Les Esseintes, to be discussed, clearly show that the particle ‘que’ is used in the context of a fossilized cleft sentence.
5.2 The controversial nature and origin of Gascon ‘que’ 5.2.1 The geographic extension of Gascon ‘que’ As mentioned, the question of the nature and origin of Gascon ‘que’ is one of the most controversial in Romance linguistics (for a recent overview on the question of enunciatives, see Marcus 2010). Even though this ‘empty’ particle can be traced back to Latin , its functional status and its semantic function are far from clear. Before reviewing some of the interpretations proposed to account for its properties, a few words are in order to delimit its geographical extension.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
103
The existence of a system of preverbal ‘enunciative’ particles is probably the most salient feature of Gascon syntax. Particularly typical of this variety is the use of preverbal ‘que’. Of course, as for any linguistic phenomenon, we should not expect to find rigid delimitations and boundaries. Marcus (2010: 35) points out that ‘the exact geographical boundaries of the énonciatif system are not consistent throughout the literature’. Following Massourre (2005: 295f.), its extension broadly covers most of the Ariège Gascon-speaking area; the southwest of the Haute-Garonne area from Martres-Tolosane; the entire area of Hautes-Pyrénées; the Gascon-speaking area of Pyrénées-Atlantiques; half of the western area of the Gers domain, including the centre of this area; the area of Landes (except Sanguinet); the area of Espiens (Lot-et-Garonne.). Within this domain, a central area is characterized by the absence of the enunciative.⁴ Massourre also observes some non-systematic use of the enunciative variant es que.
5.2.2 Interpretations of the functional value of Gascon ‘que’ According to one of the oldest hypotheses concerning the nature of the enunciative, there is some kind of complementary distribution between preverbal subject clitics and preverbal ‘que’ (cf. Portes 1857: 253). This hypothesis is considered as well by Tesnière (1988: 616f.). On the other hand, Ronjat (1913: 80) holds the view that the development of Gascon ‘que’ has to do with the extended use of asyllabic oblique clitics: these oblique clitics are hosted by the preverbal particle ‘que’ on which they may lean. In example (1e) que’t pàrli ‘I’m talking to you’, the second person singular oblique clitic ‘t’ has no phonological autonomy: that is why it requires ‘que’ as a host.⁵ Ronjat (1913: 80–1) further points out several syntactic restrictions on the use of ‘que’: it cannot be used in the imperative (cf. Pusch 2002: 107), nor can it appear in negative sentences and in unbiased / non-oriented interrogative clauses. The view expressed by Ronjat is also assumed by Rohlfs (1970a: 208), who also points out that que is used exclusively in affirmative sentences, but he observes that in some varieties this particle may be extended to negative and interrogative sentences. Thus, it does not seem that the énonciatif ‘que’ is totally incompatible
⁴ I will not argue here for the existence of a ‘zero énonciatif ’: i.e. the absence of a given sign in the phonetic string must not be interpreted as the presence of a particular (linguistic) sign which is not spelled out. ⁵ Ronjat (1913: 80) points out that once generalized, the enunciative ‘que’ could continue to serve as a host for clitics even though a word appeared in first position (cf. You que’t pàrli ‘As far as I am concerned, I’m talking to you’).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
104
with negation, as witnessed by examples such as (2a–b), taken from Massourre (2005: 304f.): (2)
a. Qui n’a ni boeu ni carriò who .3 neither ox nor cart laure work.3.. ‘He who has neither ox nor cart cannot work.’ b. Que nou cau anar ta need go. to ‘One should not go to mass.’
que
non
missa mass
However, the incompatibility of ‘que’ with the imperative seems to be absolute, and ‘que’ is generally taken to be excluded in subordinate clauses as well, where it is replaced by another énonciatif, the particle ‘e’ (cf. Pusch 1998a, 2000: 192f.). The énonciatif ‘e’ also replaces ‘que’ in unbiased / non-oriented interrogative clauses, as shown in (3b) (cf. Pusch 2002: 106; Massoure 2005: 294), and in jussive sentences (cf. 3c, cited from Bouzet 1963: 27): (3)
a. que n’ ei tanben quan eths of.it is also when .. aulhèrs e pujan tara montanha shepherds climb.3. to.the mountain ‘He also is like this when the shepherds move up to the mountains.’ b. E voletz vos asseitar want.2. sit. ‘Do you want to sit down here?’ c. E demourèsses au mens stay...2 at.the least ‘[I wish] you could stay with us!’
aquí? here dab with
nous! us
From a semantico-functional point of view, early researchers such as Zauner (1896) and Fleischer (1913) held that ‘que’ first appeared as a causal connective or as a complementizer which introduces a clause depending on some implicit main clause; later on, this connective lost its causal and complementing value, and became some kind of asserting preverbal affix (cf. Ronjat 1913: 82).⁶ ‘Emphasis’ and expressivity are often mentioned as a characteristic feature of ‘que’ constructions: such a position is assumed among others by Camproux ⁶ For a synthesis of the main hypotheses which have been put forth in order to explain the nature and origin of the enunciative particle, see Ravier (1991: 90f.).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
105
(1958: 388–90), who mentions in the varieties spoken in Gévaudan the existence of a construction very similar to that found in Gascon (cf. Moun chabal que s’es tuat ‘My horse killed himself ’; Lou charri que rebesset ‘The chariot overturned’; La tranado qu’o tout perit ‘The storm destroyed everything’). Camproux observes that these examples are the manifestation, at the incipient stage, of the same syntactic process that led to the development of Gascon ‘que’. Without entering into the details of his analysis, we note that the crucial fact observed by Camproux is that a syntactic device was selected in virtue of its being ‘emphatic’ or ‘expressive’. Once generalized and installed in the system, the syntactic device in question may lose its original emphatic value. Semantic bleaching and routinization then lead to syntacticization of a previously marked structure which becomes unmarked in the overall architecture of language (cf. Hopper and Traugott 2003). I will show with the data from Les Esseintes that this syntactic device is a focalization strategy, but one which over time has lost its original focus value.
5.3 Gascon ‘que’ and focalization We have seen that the question of the nature and origin of Gascon ‘que’ has given rise to many studies and discussions and many researchers have tried to identify the semantic and pragmatic principles underlying the use or non-use of this particle. Pusch (1999) is probably one of the researchers who has most clearly identified focalization as a possible source of Gascon ‘que’. Raible (2001: 610) rightly observes that the strategy used in Gascon: brings Gascon on a par with Quechua and a couple of African languages like Kikuyu (Heine & Reh 1983). In such cases, cleft sentences no longer serve as tags for the reordering of constituents. Although they do not lose entirely their focalizing effect which, in these contexts, is associated with the highlighting of relevant parts of the utterance, they develop into markers of predication focus, a weaker type of focus that often parallels the unmarked theme/rheme order. Enunciatives are the morphologic remains of cleft constructions in such unmarked declarative sentences.
But if we were to recognize the enunciative as the morphological trace of a cleft construction, we should be able to find historical or dialectological evidence for such a reconstruction. Pusch (1999: 116f.) views this enigmatic enunciative as a morpheme which comes from a focus construction, and which has opened out, as it has been grammaticalized, towards other syntactic contexts. As for the exact nature of the focalization process at issue, Pusch (1999: 117) observes that the process by which the enunciative has become the hallmark of the inner predicative relationships of the proposition statement can be explained as syntactic
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
106
reinterpretation, leading from the focusing of a nominal constituent to the focusing of the verbal node. That is, as illustrated by examples such as (4), the focalization process first had as its domain the preverbal noun phrase, and from there focus passed on to the verb phrase. (4)
a. C’est ton père qu’est .3.=is3 your father that .3 ‘That’s your father that has arrived.’ b. Ton père qu’est your father that .3 ‘Your father arrived.’
arrivé arrive..
arrivé arrive..
In the remainder of this contribution, I will show that if the focalization hypothesis is correct, it is not necessary, however, to postulate the kind of development argued for by Pusch: in effect, the data from Les Esseintes suggest that the ‘que’ construction arises as a pure predicate focus construction, a focus construction which crucially relies on a cleft strategy: it is not necessary to presuppose some previous focalization of the preceding noun phrase.
5.4 The dialect of Les Esseintes 5.4.1 Introduction The dialect of Les Esseintes is interesting for many reasons. Les Esseintes is a locality in Gironde (Aquitaine) 22 kilometres from Marmande and administratively it comes under La Réole. It is a border dialect, whose linguistic features are shared by two contiguous dialectological areas: Languedocian and Gascon. From the point of view of dialectological classification, the variety of Les Esseintes belongs to the Languedocian area, but various linguistic features of this dialect are typical of the Gascon domain: the evolution - > et, as illustrated by examples such as awˈzɛt < *auˈkellu ‘bird’, kuˈtɛt < ‘knife’; the evolution of medial -ll- > r (cf. aˈkerə < + ‘this one’, or the (optional) deletion of initial fricative [f] (cf. the variants la ˈurkə / la ˈfurkə ‘the pitchfork’; lu ˈyk / lu ˈfyk ‘the fire’; ˈfɛjtə / ˈɛjtə ‘done’, etc.). The dialect of Les Esseintes also has a negative indefinite nat / ˈnaδə ‘no [quantifier]’ which is the outcome of Latin : this also is a typical feature of Ibero-Romance. But while Spanish nada ‘nothing’ fundamentally has argumental status, nat / ˈnaδə in the dialect of Les Esseintes appears as a modifier inside a quantified DP. As in French, the inanimate indefinite marker corresponding to ‘nothing’ is an outcome of Latin ‘thing’, r:e:. But the most interesting feature of this dialect is the existence of a particular form of the Gascon enunciative.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
107
Table 5.1. The present indicative paradigms of the verbs surˈti ‘to go out’ and ˈkur:ə ‘to run’ 1 2 3 1 2 3
es kə ˈsɔrti (Q2040d) es kə ˈsɔrtəs es kə ˈsɔrt es kə ˈsɔrtœ̃ ŋ es kə surˈtɛ es kə ˈsɔrtœ̃ ŋ
es kə ˈkur:i (Q2076) es kə ˈkur:əs es kə ˈkur: es kə ˈkur:œ̃ ŋ es kə kuˈr:ɛ es kə ˈkur:œ̃ ŋ
5.4.2 The ‘enunciative’ es ke in the dialect of Les Esseintes A particularly striking morpho-syntactic feature of the variety spoken in Les Esseintes is the use of a special enunciative variant: es ke. The paradigms of the verbs surˈti ‘to go out’ and ˈkur:ə ‘to run’ in (Table 5.1) show that the verb form is preceded by the particle kə ‘that’, and this particle is in turn preceded by the thirdperson singular of the verb əsˈta ‘be’.⁷ Needless to say, the expressions es kə ˈsɔrti ‘I am going out’ (lit. ‘(it) is that I am going out’) and es kə ˈkur:i ‘I am running’ (lit. ‘(it) is that I am running’) show the structure of a cleft sentence with an identificational value. But in the modern state of the language, no particular illocutionary force seems to be associated with it; if the cleft construction in (5) may have had in previous states of the language a strong focalizing value, no such value is available any longer. The strategy illustrated in (5) has just become a way to encode neutral predicate focus: that is to say that it has syntacticized as a predicate focus marker (cf. Lehmann 2002b: 103).⁸ Indeed, it may be observed that the preverbal particle is in no way mandatory;
⁷ The following paradigms are reported from the questionnaire used for the realization of the Atlas Linguistique et ethnographique du Languedoc Occidental (ALLOC). As in other places of this contribution, Q indicates the number of the question given in the questionnaire. ⁸ The question of whether the phenomenon under scrutiny should be considered the result of a grammaticalization process is far beyond the scope of this contribution. Pusch (2000a: 203) rightly observes that ‘The claim that enunciative que is a grammaticalized preverbal marker is made by many authors but generally without any specific theoretical background and without referring to the parameters of grammaticalization processes as outlined by contemporary Grammaticalization theory’; cf. Pusch (1998b: 131–4) for a critical survey. Among the parameters proposed by Lehmann (1995: 121–60) reduced transparadigmatic variability (i.e. high ‘obligatorification’ [op. cit.: 139]) and reduced syntagmatic variability (i.e. fixation to a specific preverbal slot) are most relevant to describe the status of enunciative que as grammaticalized. Unfortunately, Grammaticalization theory is hard put at adequately describing the parameters involved in the development of linguistic elements which already have an exclusively grammatical value at the moment of entering the ‘grammaticalization’ process’. Perhaps a possible solution to accommodate the Gascon data would be to assume that we are dealing with a transgrammaticalization process: the (already grammatical) complementation morpheme refunctionalizes as a predicate focus marker.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
108
variants such as those in (5c–d) may be found, where it is difficult to identify any semantic principle justifying the choice between them: (5)
a. es ke se ˈkuβrə . cover.3 ‘It’s becoming overcast.’ b. lu ˈtẽn se ˈkuβrə art. weather . cover.3 ‘It’s becoming overcast.’
(Q11a)
(Q11a)
c. es ke ˈβuj boil.3 ‘It is boiling.’
(Q1148b)
d. ˈbuj boil.3 ‘It is boiling’
(Q1148b)
It can be hypothesized that the use of the enunciative introduces an event conceived as an unfragmented whole: that is, examples such as (5a) es ke se ˈkuβrə ‘It’s becoming overcast’ and es ke ˈβuj ‘It is boiling’ would be instances of thetic sentences. But synchronically, as illustrated in (5), the fieldwork done in Les Esseintes seems to offer variants of the same states of affairs, with and without the enunciative. In the subjectless sentences reported in (6), there does not seem to be any (synchronic) semantic motivation for selecting in one case a predication without the enunciative (cf. 6c), and in the other a predication with the enunciative particle (cf. 6d): (6)
a. ˈʒɛlə freeze.3 ‘It’s freezing’ b. es ke ˈtunə thunder.3 ‘It’s thundering’
(Q55c)
(Q33b)
c. ˈplaw rain.3 ‘It’s raining’ d. duˈmãŋ es kə tomorrow ‘Tomorrow it will rain’
(Q2366)
plawˈrat rain..3
(Q2369)
The number of negative sentences available in the corpus is rather limited, so that it is difficult to infer anything from the absence of the enunciative in the examples in (7):
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
(7)
109
a. lu kuˈtɛt l ɛj truˈβat œ̃ ˈlɔk (Q1856) the. knife . .1 find.. place ‘The knife, I didn’t find it anywhere.’ b. mũŋ kuˈtɛt kɔs paz my. knife .3.-3 aˈket, ko l ˈawtə laˈbas (Q1819a) (= 6a) this. .3. the. other over there ‘My knife is not this one, it’s that one over there.’ c. n ɛj ʒaˈmɛj diˈβyt də l arˈʒẽŋ a diˈgỹŋ .1 never owe.. of . money to nobody ‘I never owed money to anybody.’
However, the fact that in some complex sentences the enunciative particle appears in the main clause (vs the subordinate clause) can hardly be accidental. As a matter of fact, in complex sentences the subordinate clause is in some sense (informationally) backgrounded, while the main clause brings forth the salient information. No wonder, then, that the enunciative may be found in the main clause of the examples in (8), while it is a priori excluded in the subordinate clause: (8)
a. œ̃ n diˈzẽŋ aˈko es kə buˈlɛβə in saying that 3 want..3 ‘By saying that, he meant to joke.’ b. sə faˈliβə es kə if need..3 3 ‘If necessary, he would leave.’
ˈr:izə laugh.
partiˈret leave..3
(Q2128)
(Q2189)
c. œ̃ m muˈrẽŋ es kə ˈdɛʃœ̃ n ˈtut in die. 3 leave.1. all ‘When we die, we leave everything.’
(Q2345)
d. kur ar:iˈβɛri es kə muˈriβə when arrive..1 3 die..3 ‘When I arrived, he died.’
(Q2347)
e. sə dyˈrãŋ aw saˈβɛβə es kə nœ̃ m muriˈret (Q2350) if Durand it know..3 . die..3 ‘If Durand knew it, he would die from it.’ f. es kə ˈbɔt k 3 want.1 that ‘I want that we open the door.’
ˈuβrəŋ open.1
g. es kə ˈswɛti 3 that wish.1 ‘I wish that he lives.’
ˈbiβə live.3
kə that
la the
ˈpɔrtə door
(Q2200)
(Q2319)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
110
Once again, the information available to us is not sufficient to draw any clear conclusion concerning the use or non-use of the enunciative particle. Suffice it to observe that in languages endowed with complex focus systems, focus markers are suspended in non-assertive contexts.⁹ A possible explanation of the constraints on a sentence such as (8f) es kə ˈbɔt k ˈuβrəŋ (Q2200) (cf. ** ˈbɔt k es k ˈuβrəŋ), then, is to assume that the predicate focus marking expressed by es kə is blocked in some specific contexts.¹⁰
5.4.3 Focus marking and focus marking suspension In many languages focus marking of the predication may interfere with other types of modalities. These interferences can either block the use of a given morpho-syntactic device, or require a heavy syntactic reorganization of the sentence. Such a constraint can be exemplified with the case of Aghem (a Grassfields Bantu language spoken in Cameroon) in (9): (9)
a. m̀ mɔ̂ zì ̵ kí-bέ nέ I P1 ate fufu today ‘I ate fufu today.’ b. m̀ máà zì ̵ bέ-ˈkɔ́ nέ I P1/ ate fufu today ‘I eat fufu today.’
In Aghem, the object noun phrase has different realizations which interact with the focus system (Hyman 1984: 234): in examples (9a–b), the noun ‘fufu’ thus
⁹ Bossong (1998: 1006) views the Gascon enunciative particles precisely as assertive markers. He points out that their existence is unique in the Romance domain and that they developed in recent times. These particles are used only to express the non-marked and non-emphatic assertion and are related to the verb to which they are preposed without merging with it. The category of a specific asserter is found nowhere else in the Romance languages; Bossong observes that it has neither a precedent in Latin nor a parallel in Creole (cf. as well Bernini 2012). In contrast, in their edition of the Cartulaire de Bigorre (eleventh to thirteenth centuries) Ravier and Cursente (2005) hold the view that the ‘énonciatif’ is already present. ¹⁰ In still other contexts, some phonological constraint may be at work that bans multiple occurrences of the same string: Marcus (2010) mentions the following examples taken from Guilhemjoan (2006: 76), where the second occurrence of ‘que’ in the first example is the subordinator, followed by the enunciative e in the subordinate clause; in the second example, the enunciative is retained in the main clause but the subordinator does not surface in the dependent clause, while the enunciative is used before the verb of the subordinate clause: a. Que sabèvi que lo ton pair e vieneré. know..1 .. 2. father come..3 ‘I knew that your father would come.’ b. Que sabèvi lo ton pair que vieneré. ‘I knew that your father would come.’
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
111
shows up either as an ‘on focus’ form—i.e. kí-bέ—or as an ‘out-of-focus’ form— i.e., bέ-ˈkɔ.́ In (9b), the allomorph máà (i.e. the focused variant of the past tense morpheme mɔ́) indicates focus on the validity of the event referred to by the sentence, while no such focus value is conveyed in (9a), causing the noun ‘fufu’ to surface in the on-focus form kí-bέ in the former, and in the out-of-focus form bέ-ˈkɔ ́ in the latter. Interestingly, Hyman (1984: 235) observes that The marker máà cannot occur in non-assertive environments such as relative clauses, if-clauses, and most temporal clauses. Since these environments are considered to be ‘backgrounded’ (or ‘out of focus’) with respect to assertive or ‘foregrounded’ main clauses, we account for this restriction on the distribution of máà by noting the incompatibility of placing a focus marker in an out of focus clause.
So in the environments mentioned by Hyman, the [-focus] form mɔ ́ is required instead of máà. And the same constraints apply in imperative contexts: (10) a. zí ̵ bέ-ˈkɔ́ eat fufu ‘Eat fufu!’ b. zí ̵ bέ-ˈkɔ́ eat fufu ‘Eat !’
nô
In (10a), the object noun is used in its suffixed out of focus form after the imperative verb. As Hyman points out, the imperatives—but this observation holds as well for negatives—trigger the out of focus form of the noun because they bear some (intrinsic) focus value.¹¹
¹¹ In the Kru languages signalled by Marchese (1983: 124), the imperative verb is also incompatible with the assertive focus markers, as shown by the examples below taken from Guéré and Tepo respectively: a. mu ɗɔ̍ɔ̍ go market ‘Go to the market!’ *mu-e ɗɔ ̍ɔ̍ go- market b. di dɛ eat thing ‘Eat !’ *di dɛ eat thing
ni
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
112
Likewise in Grébo and Wobé (two Kru languages), some particles cliticizing on to the verb signal the event as the core of the message: these particles are known as assertive focus markers (AF). Examples (11a) and (12a) show vowel ɛ and syllable ne bring the focus on to the verb (cf. Marchese 1983: 122): ko (11) a. ɔ di-ɛ he ate- rice ‘He ate rice.’ b. ɔ se ko di he rice eat ‘He didn’t eat rice.’ (12) a. ne du-da I pound- ‘I pounded it.’
nɛ it
ne
b. né yi nɛ du I it pound ‘I didn’t pound it.’ / ‘I have not pounded it.’ In contrast, the examples (12a) and (12b) clearly establish that the assertive focus markers are excluded with negation as well: negation suspends or blocks the marking of focus on any other element. Nurse (2006: 195) observes that northeast Bantu languages have a verb-initial ni-, whose main function is often described as assertion. As Nurse points out, ‘[t]he presence of ni is said to represent greater certainty on the part of the speaker about the validity of what is being said, while the absence of ni indicates less certainty. Typically it appears in positive statements and yes-no questions but not in relatives, negatives, or most WH-questions. When the possibility of assertion or certainty is not present, there is no contrast between the presence and absence of ni’. According to Dalgish (1979: 57–63), the morpheme ni in Chaga has as its historical source the copula in constructions originally involving a cleft construction (‘It is that X, it is the case that X’). The parallel with Gascon immediately comes to mind: the enunciative es ke is nothing other than an original cleft construction. Pusch also observed, on the basis of a corpus analysis, that the enunciative ‘que’ very rarely co-occurs with negatives. The distributional constraints on the use of ‘que’ thus now find a possible explanation. As argued in Floricic (2009), the co-occurrence of the two types of exponent triggers some kind of Focus Clash. Of course, the examples from the dialect of Les Esseintes mentioned above do not provide clear-cut evidence for such a constraint in negative sentences, given the scarce data at our disposal. But the imperative sentences found in this dialect seem to obey such a structural
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
113
constraint on their co-occurrence with predicate focus markers. As illustrated in (13), the enunciative particle is excluded in imperatives: ˈbukə! (13) a. ˈuβrə la open the. mouth **es k ˈuβrə la ˈbukə! ‘Open the mouth!’ b. pɔrtəməˈla bring.1.3 **es kə pɔrtəməˈla ‘Bring it to me!’ c. ɛjˈli ỹŋ make.. a. **es kə ɛjˈli ỹŋ pəˈtũŋ ‘Make him a kiss!’
(Q1289)
(Q950)
pəˈtũŋ kiss
(Q1474)
While examples such as (5c–d) es ke ˈβuj ‘It is boiling’ and buj ‘It is boiling’ seem to be free syntactic variants whose choice may be driven by the pragmatic context or the communicative intention of the speaker, on the other hand no alternative is left in the case of imperatives: the only option available is to use the imperative without the enunciative particle.¹² The examples in (13) thus show that a high-ranked constraint of the grammar prohibits the co-occurrence of the predicate focus marker ‘que’ with the imperative. Naturally, the typological relevance of such a syntactic phenomenon is that it sheds a particularly interesting light on the properties and constraints of the Gascon enunciative particles.
5.4.4 A contact-induced phenomenon? We have already observed that the dialect of Les Esseintes is a border dialect, whose linguistic features are shared by two contiguous dialectological areas: Languedocian and Gascon. Although from the point of view of dialectological classification the variety of Les Esseintes belongs to the Languedocian area, various linguistic features of this dialect are typical of the Gascon domain. Concerning the variety spoken in this area, Luchaire (1879: 199) observes that the Gascon dialect
¹² As Aikhenvald (2008) points out, ‘This restriction is intuitively plausible. Imperatives constitute prima facie directive acts to which the verb is central. This centrality of the verb is what makes the imperative mutually exclusive with grammatical focus marking of a nominal constituent within the clause (in Hyman’s 1979: 61 words, “imperatives have an overriding intrinsic focus” which is mutually exclusive with focus-marking of individual constituents)’.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
114
Redon Angers
LOIRE-ATLANTIQUE
Tours Chauvé
Nantes
MAINE-ET-LOIRE
Gorges
INDRE-ET-LOIRE
Cholet Loudon
Thouars
Leugny Châteauroux
La Roche-surYon DEUX-SÈVRES
Chauvigny Niort
VIENNE Civray
La Rochelle
Transition zone (approximate) between Poitevin and Saintongeais varieities
CHARENTEMARITIME
2
Marchois area (half oc, half oïl ) of Saint-Amant-de-Boixe (lost at the start of the 20th century)
Limoges
Angoulême
Saintes
L
Le Verdon
1
CORRÈZE Périgueux
Transition zone between oc and oïl ( = varieties of the La Marche ‘croissant’) Marchois enclave (half oc, half oïl ) of Saint-Eutrope (lost at the end of the 19th century)
CREUSE
CHARENTE
2
Poitevin contact varieties (approximate extent)
1
Guéret
HAUTEVIENNE
Linguistic area of Poitou and Saintonge
Saintongeais varieties
Argenton-surCreuse
Poitiers
VENDÉE
Poitevin varieties
INDRE
Châtellerault
Blaye
Tulle
M
La-Roche-Chalais DORDOGNE
Bordeaux
Libourne LOT
GIRONDE
Les Esseintes
Saintongeais enclave of Monségur Cahors
Villefranche
Map 5.2. The dialect of Les Esseintes in the Occitan domain From https: //journals.openedition.org/mimmoc/docannexe/image/2061/img-2.png
occupies the entire part that extends to the left of the Garonne, from la Réole to the extreme point of Grave, except for the three villages of Verdon, Les Loges, and Royannais, an enclave of langue d’oïl in the Médoc. It also includes the Entre-deux-Mers, except for an enclave that occupies the eastern end of the arrondissement of la Réole, known as the Petite Gabacherie. This gabaï dialect, a mixture of Gascon, Perigourdin and Saintongeois, is taken to embrace most of the cantons of Pellegrue and Monségur (Gironde) and to extend into the canton of Duras (Lot-et-Garonne). Luchaire holds that the extreme points of the gabaï of Monségur are, to the north, Massugas, to the west, Blasimont, to the southwest and, very close to La Réole, Les Esseintes; to the southeast, La Gupie, to the east, Saint-Géraud and Sainte-Colombe (cf. also Suchier 1891: 77; Bourciez 1896: 145). It is thus interesting to observe that the dialect of Les Esseintes (see Map 5.2) belongs, albeit marginally, to the so-called ‘Petite Gabacherie’ (cf. also Ronjat 1930: 25). That is, the locality of Les Esseintes belongs to an area in which, since the fifteenth century, a wave of immigration has taken place involving mainly speakers from the Saintongeais / Poitevin domains who settled in order to farm the land after the Hundred Years’ War (cf. Queyron 1907; Boutruche 1935;
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
115
Lalanne 1951).¹³ Such a situation gave rise to a mixed variety that allowed commercial / social intercourse—Bourciez talks of Sprachmischung concerning this variety. Of course, it is not possible to present in these few pages the properties of the ‘parler gavache’. Observe that among the morphosyntactic features to be mentioned, the Gavache dialects resort to subject pronouns, as illustrated by point 635 Andraut (Monségur) of the ALF, and such a feature is typical of Oïl dialects (cf. map 226 in Brun-Trigaud, Le Berre, and Le Dû 2005: 186). Given the mixed character of this linguistic area—Queyron observes that in the same family some speakers could speak Gascon and others Gavache—the Gascon ‘que’ construction may have been reinterpreted as a cleft construction by nonGascon speakers. Another hypothesis which can be mentioned here only in passing involves Basque. It has been pointed out, at least since Bonaparte (1879), that the use of ‘que’ is an exceptional feature in the Romance domain, a feature which could be due to the contact with Basque. (14) a. Ba dator ‘He is coming.’ b. Ez dator **ba ez dator ‘He is not coming.’ As illustrated by examples (14), the Basque assertive marker ba to the left of the synthetic verb is in complementary distribution with the negative marker ez, a pattern which is reminiscent of the constraints that hold on the distribution of Gascon ‘que’ (cf. Marcus 2010). Needless to say, more historical and geolinguistic investigations should be carried out to identify the common features between the two languages. As mentioned by Rohlfs (1970a: 206) and Haase (1994: 801; 1997: 209), the emergence of the enunciative is rather recent, and old Gascon texts only show a few examples whose interpretation is not always simple. According to Haase (1997: 210), ba in Basque fills the preverbal focus position when it is empty, as typically occurs in thetic sentences. Gascon may thus have functionalized ‘que’ as enunciative under the pressure of the Basque substrate.
5.5 Conclusion The data presented here are a first attempt at describing the special enunciative particle es kə in the dialect of Les Esseintes. Of course, a more detailed analysis ¹³ Boutruche (1935), however, points out that episodic waves of immigration towards the EntreDeux-Mers took place in more remote times.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
116
would be needed to describe accurately the complex system of this dialect. We have seen that in various cases this predicate focus marker does not seem to be mandatory, as also illustrated by some northern Gascon varieties investigated by Massourre (2005: 297). In other cases, however, the particular nature of this marker is revealed in the distributional constraints it is subject to. Even though they need to be confirmed by complementary fieldwork (a difficult task, given the moribund status of this variety), the dialectological data presented so far seem to support the idea that the complex marker es kə is a variant of the simple enunciative particle ‘que’. It has been hypothesized that the former constitutes a possible source of the latter: es kə is a (fossilized) cleft construction that brings to existence a given state of affairs.¹⁴ The same kind of construction may be detected in other Romance languages: the French variants Il fait froid! ‘It’s cold!’ and C’est qu’il fait froid! ‘It’s that it’s cold!’ do not differ from the point of view of the truth conditions of the sentence. They are just two syntactic variants of the conceptualization of the same state of affairs. Interestingly, it had already been noted by Rohlfs (1970b: 14) that constructions like que plau ‘it’s raining’, que cantàm ‘we’re singing’, que m’em boy ‘I’m going’, etc. originally had the function of giving more energy to a sentence or reinforcing a thought, and he pointed out that these constructions were the result of an earlier formula c’est que, e.g. in maritime Gascony (Landes) es que plau ‘it’s raining’, es que ha caloù = Sp. hace calor ‘it’s hot’, in Aragon es que acabo de romperme una pierna ‘I have just broken my leg’. Of course, further research and further fieldwork would be necessary to establish to what extent the dialect of Les Esseintes has syntacticized this original cleft construction. In particular, the question of the interaction between negation, interrogation, focus, and the es kə enunciative is of paramount importance in delineating the exact nature and distributional constraints of this enunciative. Last, given the sociolinguistic situation of this border variety, one might wonder as well whether the cleft construction available in Les Esseintes is not the reinterpretation or reanalysis of a syntactic pattern that was no longer transparent to some (non-Gascon) speakers. Needless to say, this question and many others are left open for further research.
¹⁴ Hetzron (1977: 216f.) has already argued that the sequence es kə was the most probable candidate to account for the origin of the ‘énonciatif ’ que.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
6 Postverbal negators in Gallo-Romance The view from Old Occitan Sandra Paoli and Xavier Bach
6.1 Introduction Postverbal negators that are found in the ‘embracing’ type of sentential negation (e.g. Fr. Il ne mange pas d’escargots ‘He doesn’t eat any snails’) have traditionally been analysed as the result of grammaticalization of elements introduced after the finite verb to strengthen the preverbal negative marker (Jespersen 1917: 19–22), as the first model for many others). The postverbal elements are focused, indefinite nominals, or adverbs whose addition results, at the beginning, in emphatic negation and creates a contrast with the plain negation expressed by the preverbal negator on its own.¹ Cross-linguistically, these postverbal elements are generally derived from a limited set of items that share similar properties, of which scalarity is the most relevant since it interacts with negation giving rise to a scalar implicature: because they denote low points on a pragmatic scale, when within the scope of a negative marker these postverbal elements allow a pragmatic inference of negation of the entire scale (Fauconnier 1975). This easily leads to a generalized, emphatic reading which asserts that a situation does not hold at all. Kiparsky and Condoravdi (2006: 172) identify emphatic negative markers as being either minimizers or generalizers: the former are predominantly nominals indicating a small quantity (e.g. a bit, a wink), the latter generally quantifiers (e.g. not a thing, anything). The negative markers that develop from them have been claimed to be syntactically distinct (Garzonio and Poletto 2008, 2010) on the basis of their diachronic development: while minimizers change position (and syntactic classification) during the process of grammaticalization, quantifiers do not. This study focuses on the diachronic path followed by two old Occitan postverbal negative markers, pas and ges, a minimizer and generalizer respectively,
¹ This research was supported by two small awards to the first author by the Humanities Division at the University of Oxford (REF Strategic Support Fund and Returning Carers’ Fund). Grateful thanks are due to Alice Traisnel and Xavier Bach for painstakingly extracting the data on pas and ges, respectively, from the corpus. Sandra Paoli is responsible for §6.2 and Xavier Bach for §6.3.
Variation and Change in Gallo-Romance Grammar. First edition. Sam Wolfe and Martin Maiden (eds). This chapter © Sandra Paoli and Xavier Bach 2020. First published in 2020 by Oxford University Press.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
118
it discusses the semantic and pragmatic restrictions that are operative in their usage throughout the period from the eleventh to the fifteenth centuries, and suggests that the two have developed in opposite directions along the lines of emphatic negator vs negator of a presupposition. In old Occitan, there was a preverbal negator no(n), the sole sentential negator, that remained the dominant strategy to form negation until at least the sixteenth century; during the period between the twelfth and the fifteenth centuries it could also be reinforced by a variety of postverbal markers, whose presence became increasingly frequent and by the seventeenth century preverbal no(n) was generally lost. The Occitan corpus on which this study is based has been taken from the second part of the Concordance de l’Occitan médiéval (COM2) (Ricketts 2005), and it includes a selection of eleven texts in verse of religious and epic inspiration, broadly from the Languedocian area. There are over 72,000 lines of text, composed between the eleventh and the fifteenth centuries. This mixture of mystery plays and chanson de geste that make up the corpus is a perfect representation of the typical dramatic literature of the time (Fleischman 1995), and, although in verse,² they have the advantage of being of a more discursive and narrative register than the troubadour poetry corpus used by Medina Granda (1999). For the quantitative analysis, the occurrence of each negator has been measured against the total number of negations, either in the given text or for the whole century. Table 6.1 lists the texts, which are referred to by their COM2 code throughout this chapter. Table 6.1. List of texts included in the corpus century
text
COM2 code
length (in lines)
number of negative clauses
eleventh twelfth thirteenth
La Canso de Sancta Fides Roman de Jaufré Flamenca Canso de la Crotzada Guilhem de la Barra Breviari d’Amor Lo Jutgamen General Mystère rouergat de la Passion Résurrection de Lazare Mystère rouergat de l’Ascension Joseph d’Arimathie
CSF JAU FLA CCA GDB BRV JDE MDP2 RDL MDA JAR
593 10,974 8094 9578 5344 34,597 2730 4516 767 1047 1996
67 1585 1432 1201 487 3698 302 855 3 99 0
fourteenth fifteenth
² We are aware that the use of verse to carry out syntactic investigations is usually considered unsuitable. However, in order to compile a genre-coherent diachronic corpus of Occitan, this choice was unavoidable. Furthermore, verse is frequently relied upon for early old French research, and it appears to show a very homogenous syntax which is not markedly different from other medieval varieties where we have early texts in prose. For these reasons, we feel that our choice of verse is motivated and justified as a reliable representation of the language of the time.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
119
6.2 Pas 6.2.1 Introduction In the modern Occitan varieties that use it, pas generally negates a sentence on its own.³ This was already the case by the seventeenth century, and by the end of the nineteenth its co-occurrence with preverbal no(n) was largely restricted to the written language (Schwegler 1990: 163f.). During the period covered by our corpus, we see its first appearance (twelfth century), rise (thirteenth century), and establishment (fifteenth century), shown in Table 6.2 and Figure 6.1.
6.2.2 Pas through the centuries We now turn to a qualitative discussion of the usage of pas in each century. The first occurrences of pas in our corpus are from the end of the twelfth century: Table 6.2. Overview of frequency of pas in the corpus century
no(n) occurrences
pas occurrences
frequency of pas
eleventh twelfth thirteenth fourteenth fifteenth
67 2483 2633 4185 1259
0 2 38 7 238
0% 0.4027% 1.4432% 0.1672% 18.9038%
Century 25
11th
12th
13th
14th
15th
20 15 %
10 5 0
Figure 6.1. Overview of frequency of pas
³ Non can still be used on its own in the spoken language when using formulaic expressions and in some constructions such as ni non . . . ni non . . . .
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
120
(1)
a. ‘Seiner, aintz irai ieu ab vos,’/ Ditz mos seiner Galvan, ‘si·us plas, / Que sol n’en anaretz vos pas!’/ ‘Neps,’ ditz lo rei, ‘no·n parletz plus, / C’ab me non anara negus, / E no m’en sonetz motz oimai!’ (COM2:JAU 200–5) ‘ “Lord, I will go with you”,/ says my lord Galvan, “if you like, / so that on your own you will not go PAS!”4/ “Nephew” says the king “don’t speak anymore, / that with me nobody will come,/ and don’t say any more about it to me now!” ’ b. [ . . . ] e fer del cap a la paretz/ ez laisa·s en terra cazer,/ tan autz con es, per gran poder./ Mas Jaufre non s’es pas mogutz/ per tal, c’aissi es esperdutz,/ c’ades cuja eser feritz. (COM2:JAU 3942–7) ‘[ . . . ] banging his head against the wall/ and letting himself fall on the floor,/ from his full height, with great force./ But Jaufré did not PAS move/ because, lost,/ he now thought he was going to be wounded.’5
Considering first of all its position, in both examples pas follows the finite verb and precedes the past participle, possibly already occupying its modern location (Zanuttini 1997: 4).⁶ The verbs with which it is used are both intransitive, which parallels the observation that in the thirteenth century Fr. pas occurred mainly with être ‘be’ and other intransitive verbs (Foulet 1967: 260); furthermore, they are verbs of motion, ‘go’ and ‘move’, perhaps offering support to the view that in its earlier uses, pas was semantically restricted to verbs of motion (Jespersen 1917: 17). This corresponds to the oft-quoted, but hypothetical, late Latin example non uadere passu(m) lit. ‘not to go a step’, whose structure is generally assumed to be the source of French negation ne . . . pas; however, it has never been actually attested in any Latin or early Romance text (Buridant 2000: 207; Grieves-Smith 2009: 10), and the semantic change of Fr. pas from a nominal into a marker of negation has never been directly documented. It is not possible to draw any firm or general conclusions from two examples; however, the possibility of a link between the first attestations of pas and their co-occurrence with verbs of motion cannot be ignored.⁷ These first instances may be, indeed, the ‘missing link’ in the Jespersen’s chain for Fr. pas. Turning now to its pragmatic import, in both instances pas co-occurs with the preverbal negator n(on), the predominant marker that could, by itself, negate a
⁴ A detailed discussion of these examples is needed in order to establish as closely as possible the semantic or pragmatic properties of pas. In order not to pre-empt the result of the investigation, on first presenting ambiguous cases pas will remain untranslated, and simply rendered as PAS. ⁵ It is typical of the narrative style of the Occitan mystery plays to switch between tenses midsentence (Fleischman 1995: 174); however, the relevant sequence of tenses will be maintained in the translations. ⁶ Considering the richness of the syntactic space existing between finite verb and past participle (cf. Cinque 1999), and given the limitations of the data, this conclusion necessarily needs to be approximative. However, it is likely that pas occupies a position within the expanded T space. ⁷ In the following century, the group of predicates that co-occur with pas is different, suggesting a change in its distribution and lending support to the existence of possible initial semantic restrictions.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
121
sentence: the presence of pas must therefore be understood as ‘adding’ something to the negation. In (1a), king Arthur is in a forest with his men when he hears someone crying for help: he decides to investigate, but he must do so on his own. His nephew offers to accompany him, so that he should not go alone: pas here could have the emphatic meaning of ‘completely’, or it could be interpreted in its literal meaning, ‘not a step’. Additionally, it could negate the ‘discourse-old’ proposition (Hansen 2009; Hansen and Visconti 2009) ‘king Arthur needs to go alone’. In (1b), Jaufré is being held captive in Brunissen’s castle, and allowed to sleep in the great hall where her knights are watching him; at some point the watchman gives a signal and all the knights begin to cry out and mourn. When Jaufré asks the reason for this, the knights start to beat him. When later the same episode is repeated, Jaufré does not move. Since he is lying down, it is unlikely that pas is used here with its literal meaning; rather, it is possible that it has an intensive reading of ‘at all’ or ‘in the least’, or that it is used in contrast with Jaufré’s previous actions, i.e. getting up and asking questions. The latter two interpretations may, in fact, not be mutually exclusive, allowing for an intensive reading perhaps granted by the contextually overt contrast between the negated proposition and the discourse-old scene described earlier. A fuller discussion follows at the end of this section. In the thirteenth-century texts there are a total of thirty-eight examples of pas, with no changes in terms of the relative position of pas and finite and non-finite verbs. There are, however, nine instances of pas preposed to no(n), as shown in (2):⁸ (2)
a. Fin’Amors l’esperit l’en mena/ Lai en la tor on si jasia/ Flamenca, que pas non sabia/ Qu’am fos per leis enamoratz. (COM2:FLA 2147–50) ‘True Love had taken his spirit/ away to the tower where lay/ Flamenca, who PAS did not know/ that someone was in love with her’. b. Pero, si mals fos bes ni mentirs veritatz,/ aqui on es orgolhs fora humilitatz;/ car ieu pas no creiria, si mielhs non o proatz,/ que nulhs hom sia dignes, si no mor cofessatz. (COM2:CCA 162 041–4) ‘But, if evil were good and lying the same as truth,/ then where pride is there should be humility;/ because I PAS do not believe, unless you can prove it better,/ that no man is deserving, unless he dies having been confessed.’
In all these instances, the context confirms an intensive reading; there is also a clear contrast between the intensity of the restlessness and agony of love-sick Guillem and Flamenca’s quiet lying in the tower, completely unaware of his consuming love for her. Similarly, when the Bishop of Nîmes declares that the ⁸ It is interesting to note that Fr. pas and mie were also able to occur in a preposed position in the French of this period.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
122
Count of Montfort’s knight, who died in battle, is a martyr of Christ and that all his sins are forgiven, Folcaus de Berzy reacts strongly against the clergy who are too quick to give absolution where there is no repentance, saying that he does not and will not believe that (the context is that confession is key to salvation). As for the no(n) . . . pas examples, it is impossible to establish with absolute certainty whether an emphatic reading ensues or not. However, each proposition negated by pas is linked to a state of affairs described in the text. This is clear in (3a): the king’s arrival is greeted with great joy by his men who thought he would help them, since they were his people and his friends; however, he had not come to do this. (3)
a. [ . . . ] e tuit sei cavalier, que n’an gran gaug agut,/ que cujan per lui estre ladoncas mantengut,/ que ilh eran sei ome, sei amic e sei drut;/ e si se foran ilh; mas non es pas vengut/ que el non a poder, ni forsa ni vertu,/ mas cant son de preguieira, si el ne fos crezut. (COM2:CCA 27 010–5) ‘[ . . . ] and all his men, who had great joy from it (i.e. seeing the king),/ as they thought they would be supported by him,/ as they were his vassals, his friends and very dear to him;/ and this, they were to him; but he had not come (to help them)/ as he had no power, nor strength nor virtue,/ nothing but prayer, if he were to be believed.’
Pas can also co-occur with a nominal that is in some respect minimal, as in (3b), and the example clearly conveys emphasis; however, this is the result of the presence of una castanha ‘a chestnut’, and not of pas.9 For parallel Old French examples, this co-occurrence has been interpreted as evidence that pas is not a negative polarity item (NPI, Fauconnier 1975) but ‘an integral part of negation’ (Hansen 2018: 284), or that it is used like Eng. even (Eckardt 2006: 142) ‘to signal the minimality of that other object’. b. [ . . . ] vengon a Castelnou rengat per mei la planha./ Mas lo coms, sel de Foiss, ab tota sa companha/ lor es e mei la via e li roter d’Espanha,/ que no les prezan pas per forsa una castanha;/ ans dizon entre lor: “Baros, us non remanha/ que no sian aucis aicela gens estranha, [ . . . ] (COM2:CCA 94 002–7) ‘(Frenchmen from Paris and knights from Champagne) ride across the plain towards Castelnaudary./ But the count of Foix, with his whole company/ they block their way, and (with) their Spanish mercenaries,/ they think the force not worth PAS a chestnut;/ and tell each other: “Barons, do not rest/ until all these foreign peoples have been killed, [ . . . ]’
⁹ In the same text there are also other examples, with poma porria ‘rotten apple’ and gan ‘glove’, which co-occur with preverbal no alone: the resulting intensive reading is clearly conveyed by the two nominals.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
123
It is unlikely that at this early stage pas had already been integrated into plain negation; it is possible that it had an intensive reading, ‘not worth even a chestnut’, but equally that pas negated the value of Bouchard’s army, already described earlier as fewer in number, and only equipped with helmets and hauberks. Turning to the type of verbs that are found with pas, in both texts there are mainly verbs denoting abstract states such as ‘be’, ‘be able’, ‘think’, ‘believe’, ‘wish’, ‘fear’, ‘have the opportunity to’, ‘know’, ‘value’, ‘forget’. A very similar situation, qualitatively speaking, is found in the fourteenth century: pas can co-occur with minimal nominals, (4a), in general it negates a discourse-old proposition, (4b), and it is difficult to establish whether it conveys emphasis or not. There are no examples of preposed pas: (4)
a. [ . . . ] E fon tornatz viassament/ al rey so senhor dir ayssi:/ ‘Senher, lo coms manda per mi/ qu’el no·us presa pas .i. boto/ ni totz aquels ques ab vos so/ ni vostr’aver ni vostra terra,/ ni no tem en re vostra guerra’ (COM2:GDB 4166–72) ‘[ . . . ] And they returned in all haste/ to the king their Lord to say this / ‘Lord, the count sends through me/ that he doesn’t value you a button/ nor of all those who accompany you/ nor of your riches nor of your kingdom/ nor does he fear that you declare war on him.’ b. E·l reys va·l dire: ‘Don est tu?’/ ‘Senher’, diss el, ‘de loc autru/ no son ieu pas, ans suy d’ayci.’ (COM2:GDB 2027–9) ‘And the king says to him: “Where are you from?”/ “Sir”, he replies, “from another place/ I am not, on the contrary, I am from here.” ’
From a quantitative perspective, however, in relation to the great length of the texts, there are very few attestations of pas (a total of seven): the majority of sentential negation is expressed by preverbal no(n) on its own, but there are also some instances of postverbal ges (a total of nineteen). The low incidence of pas is surprising and remains unexplained.¹⁰ In the fifteenth century there is a surge in the use of pas: out of 1259 occurrences of sentential negation across the three texts analysed, 238 are realized by no(n) . . . pas. In many examples pas co-occurs with serta(s), an affirmative adverbial meaning ‘certainly, for sure, indeed’: in these contexts, pas negates a state of affairs that belongs to the common ground (5a and 5b). Pas can still co-occur with words denoting minimal entities, as in (5c), and there are no instances of preposed pas:
¹⁰ One possible (and partial) explanation is that some sources place Breviari d’Amor at the end of the thirteenth century, rather than the beginning of the fourteenth, hence representative of an earlier stage of development. However, this would still not provide an explanation for the other fourteenth century text, Guillem de la Barra, in which there are only three instances of no(n) . . . pas.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
124
(5)
a. Ma bela sor, d’aquo no duptetz pas,/ quar veramen no qual pas,/ se d’aquel fruch vos autres mangatz,/ per aquo sertas no moriretz pas, [...] (COM2:MDP 183–6) ‘My beautiful sister, of that do not doubt at all,/ because in truth it doesn’t matter/ if of that fruit you eat/ for which certainly you will not die, [ . . . ].’ b. [ . . . ] per so que ieu avia presa la querela/ de guardar al tombel lo sant propheta./ Ma ieu no lo podia pas guardar/ quant el es resucitat,/ quar el ha fach tremolar lo tonbel/ que me pensiey que totz periguesen/ [ . . . ] (COM2:MDP2 4238–43) ‘[ . . . ] for this reason I had taken the task/ to stand guard at the tomb./ But I could not guard it / when he rose from the dead,/ because he made the tomb tremble/ and I thought that everyone was perishing/ [ . . . ].’ c. Regardem be que nos qual mori/ he sabem pas cora sera la fy./ Que que sia ni que que no,/ trastot no val pas hun boto. (COM2:MDP 2024–7) ‘Let us remember that we have to die/ and we do not know when the end will be./ Whether it is an ‘or’ or a ‘no’,/ all is not worth a button.’
Similarly to what was found for previous centuries, both an emphatic and a nonemphatic reading are possible, and the context is not sufficient to resolve the ambiguity. The main findings so far are summarized in (6): (6)
a. pas first appears in the twelfth century, and it occurs with verbs of motion; b. its use increases in the thirteenth century, and 9/38 occurrences are a preposed pas; d. pas no(n), which expresses a particularly emphatic negation in which a contrast is also present, is only found in the thirteenth century; e. in the thirteenth, fourteenth, and fifteenth centuries, pas can also co-occur with a nominal expressing minimal value; f. the use of pas increases exponentially in the fifteenth century; g. throughout this period, the proposition negated by pas appears to be linked to previous discourse.
So far, in describing a possible interpretation of pas the term ‘emphatic’ has been used, mainly in an intuitive sense, alternating with ‘intensive’ and ‘strengthen (ing)’, but it has not been defined. Similarly, we have used ‘discourse-old’ without providing a definition. Furthermore, we have claimed that the two functions may not be mutually exclusive: a link between them has already been noticed, since it is a property of all focalizing elements (and negation, here in a scalar intensive reading such as ‘not even’) to imply a presupposition as part of their interpretation (Manzini and Savoia 2005: 154f.), and the two can be mapped onto the same
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
125
lexical item, as with Calabrian mancu discussed by Ledgeway (2017: 111). The data discussed in this chapter point to the same being true for Occ. pas and, as we will see in §6.3, ges. We follow Israel (1996) and Larrivée (2014: 121) in viewing the speaker as central to the interpretation of emphasis: for the former, emphasis is an expression of the speaker’s attitude that the informative strength of their proposition is high, and for the latter it is an ‘unmitigated assertion’ of their views. For ‘discourse-old’ we follow Hansen (2009) and Hansen and Visconti (2009) and understand it in terms of inferential links to previous discourse (Birner 2006) that belong to the common ground. We depart slightly from this stance, though, by proposing that both an assertion and a presupposition evoked by an assertion are relevant,¹¹ and in Occitan they are mapped onto two distinct configurations, pas no(n) and no(n) . . . pas respectively. In the case of pas no(n) a contrast is established with an overt assertion and the preposed pas can be interpreted as an expression of Contrastive Focus, likely occupying a position in the left periphery.¹² No(n) . . . pas could then be understood as the canonical strategy to negate a presupposition, which, following Manzini and Savoia (2005), we assume is always present with focalizing elements. We turn now to the use of pas in negative polar questions.
6.2.3 Pas in negative polar questions In one of the fifteenth-century texts, COM2:JDE, out of a total of 106 occurrences of pas, 31 are in negative polar questions of the type shown in (7): (7)
a. He, diguas, no as pas tu dic / A tota ta posteritat he co[m]paniha,/ Quant tu eras en la granda monarchia / «Veramen ieu soy Dieu/ He soy sertanamen asetiat/ En la cadieyra de Dieu»? (COM2:JDE 202–7) ‘And, tell me, haven’t you PAS said/ to all the members of your tribe and your groups,/ when you were still in the celestial monarchy:/ “In truth I am God and I am seated indeed on God’s throne”?’ b. No vos soy ieu pas estat promés/ En la vostra ley/ En la quala es estat promés/ Que Dieu vos trametria hun propheta/ Del miech de vostres frayres/ He seria senblable a Moyses? (COM2:JDE 310–15)
¹¹ A similar distinction, explicitly activated versus simply inferred, is made by Schwenter (2005: 1450) when explaining the contrast between the embracing não . . . não and the postverbal não negations in Brazilian Portuguese. ¹² The fact that this configuration is not found beyond the thirteenth century may be the result of decreased availability of the Contrastive Focus position. Sichel-Bazin (2011) reports that the preposing of a contrastively focused element is ‘quite rare but possible’ in modern Occitan; however, the preferred strategies to mark focus are either through prosody without any change in word order, or a cleft, suggesting that a Contrastive Focus position in the left periphery is not normally available, and may be subject to diatopic variation.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
126
‘Haven’t I PAS been promised to you/ in your Law/ in which it is promised/ that God will send a prophet/ from among your brothers/ and similar to Moses?’
Significantly, there are no instances of pas in wh-questions: negative wh-questions are not well represented in the text, with only seven out of a total of 59 containing negation, but, crucially, none of these contains pas. Thirty-one out of 42 polar questions, on the other hand, do contain pas; of these, we focus on the root polar questions. In general terms, although interrogatives are the canonical form for seeking information, polar and wh-questions differ in that while the latter are truly ‘open’, and the answer is chosen from an open set, responses to polar questions are chosen from a set of two: an affirmative answer or a negative one (Karttunen 1977, among others). In this respect, we could say that while wh-questions are mainly seeking information, polar questions are seeking confirmation (Bongelli et al. 2018: 31). The set of negative polar questions under investigation are different in that they are not seeking confirmation: in a number of cases, the answer is overtly provided by their utterer. (8)
a. Mossenhor lo jutge, metatz lo cas/ Que en tal prejudici ieu agués pecat,/ He las! no ha pas pecat Adam senblablamen?/ He si a el veraiamen! (COM2:JDE 258–61) ‘Sir the Judge, consider the hypothetical case/ that I have really committed the sins that have been attributed to me,/ Alas! Did Adam not PAS sin in a similar way?/ Indeed, he did!’ b. He no era pas plus rasonable/ De resucitar ho de/ Gerir hun ladre/ Lo jorn del vostre sabat?/ He si era verayamen,/ [ . . . ]. (COM2:JDE 507–11) ‘And was it not PAS more reasonable/ to raise from the death or to/ heal a thief/ the day of your Sabbath?/ And so it was in truth,/ [ . . . ].’
These appear to be rhetorical questions, which formally resemble questions, but have the illocutionary force of a ‘strong assertion’ of the opposite polarity from what is apparently asked (Sadock 1971, 1974). That is, a rhetorical positive question has the illocutionary force of a negative assertion, and a rhetorical negative question has the illocutionary force of a positive assertion. Because they neither seek information nor elicit an answer from the addressee they are also labelled as ‘redundant’ questions (Rohde 2006: 135): the speaker uses rhetorical questions to bring to the fore, for the addressee’s benefit, information already present in the common ground. Their purpose is, therefore, to bring the addressee’s awareness into a state in line with the speaker’s. Rhetorical questions combine a strong assertion by the speaker and a focus on the addressee, since the content of the speaker’s assertion is chosen on the basis of its relevance value for the addressee.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
127
Negative polar questions that do not contain pas behave differently, in that they have a bias towards a negative answer, as suggested by the context, i.e. not having behaved like the shepherd of the Old Testament in (9a), or being forever damned, and hence unable to leave hell in (9b). (9)
a. Tu que parlas enaisi/ No as tu legida la Screptura,/ En Ezechiel en lo ters capitol,/ Que dis enaisi:/ Que ieu iey comandat als pastors/ Que de lors fedas feseso bona guarda/ He que las prediqueso grandamen,/ Ho autramen se gardeso/ De morir eternalmen?/ Mas veramen vos autres/ No avetz pas fach como los pastres/ Del Vieilh Testamen,/ Mas lor avetz mostrat/ Lo cami de malvestat, cosi pe[?]so. (COM2:JDE 685–98) ‘You who speak like this/ haven’t you read the Scriptures,/ Ezechiel chapter 3,/ where it says thus:/ I have ordered the shepherds/ to guard attentively their sheep/ and to preach well to them,/ or they would otherwise/ be condemned to eternal death?/ But in truth you/ have not acted at all like the shepherds/ of the Old Testament,/ but you have shown them (the sheep)/ the way to evil.’ (uttered by the Judge to the Prelates). b. He las paubre! no salhiriey ieu d’aisi/ Huna hora en tota ma vida?/ He las! paubra gen marida,/ Vulhatz penre exemple en mi,/ Quar sertanamen se no ho fachz/ Trastotses seretz sertas dapnatz. (COM2:JDE 1993–8) ‘Alas, poor me! Won’t I be able to leave from here/ one hour in my whole life?/ Alas! Poor miserable people,/ may you consider my example,/ because for sure if you do not learn the moral/ you will be all without a doubt damned’ (uttered by Avaritia to one of the devils).
The two types of questions minimally differ in terms of the presence versus absence of pas. The two functions associated with the use of pas, ‘emphasis’ and ‘negator of discourse-old’, earlier linked to the speaker’s expression of their attitude and the common ground between speaker and addressee, find a perfect match in rhetorical questions, whose goal is to create a synchronized common ground, and the driving force is the strong assertion of the speaker’s views. It is therefore plausible to assume that it is because of its properties that pas coerces the rhetorical reading of a negative polar questions. Incidentally, the complete absence of pas from wh-questions, which are genuinely seeking information, further suggests that pas in not compatible with the function of seeking out information. The properties displayed by rhetorical questions closely resemble Traugott’s and Dasher’s (2002: 30) notions of ‘subjectivity’ and ‘intersubjectivity’. Expressions of (inter)subjectivity ‘are expressions the prime semantic or pragmatic meaning of which is to index speaker attitude or viewpoint (subjectivity) and speaker’s attention to addressee self-image (intersubjectivity)’ (Traugott 2010a: 32). The practice of expressing subjectivity and intersubjectivity leads to the diachronic processes of subjectification and intersubjectification, whose links
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
128
to grammaticalization have been explored in detail in recent years: ‘[ . . . ] it is exactly in cases of intersubjectification, where the highest degree of speaker’s involvement is to be found [ . . . ], that grammaticalization comes into play [ . . . ]’ (Veloudis 2018: 130). This was further aided by the high degree of dialogicity of these question-answer pairs, in which multiple points of view are expressed: dialogic contexts have been claimed to favour the emergence of new syntactic constructions (Schwenter 2000; Traugott 2010b). Only pas is found in rhetorical questions in COM2:JDE: we follow Paoli (forthcoming) in proposing that it is precisely its presence in these highly (inter) subjective and dialogic contexts that contributed to speeding up the grammaticalization of Occ. pas and ensured its establishment as the generalized postverbal negator winning over all the others. It is also interesting to note that negative rhetorical questions do not admit NPIs (Han 2002: 205), and that pas, at this stage, was therefore no longer an NPI. This suggestion may also be relevant to Fr. pas. Price (1978, 1993) discusses polar questions in which point and, later also pas, figure, and how they start to have a negative interpretation from the second half of the fifteenth century. Eckardt (2006: 143f.) points to rhetorical questions as being the first context in which pas occurs as a negator on its own. It is interesting to note that mie (originally ‘crumb’) and goutte (originally ‘drop’) were never recorded in rhetorical questions: neither survived as negators in the modern language. This possible link deserves further research within the Oïl dialects. Let us now turn to ges.
6.3 Ges 6.3.1 Introduction Most of the previous research on the (emphatic) negative ges has been conducted by Medina Granda (1999, 2007). For Medina Granda (1999), ges is an ‘indefinite noun classifier’, and she sets it apart from quantifiers in her introduction, although she sometimes refers to it as a quantifier (Medina Granda 1999: 328). It functions as a syntactic NPI, which means that, contrary to semantic NPIs such as res ‘nothing’ or jamais ‘never’, it only occurs in negative polarity contexts (negative clauses, questions, embedded clauses with a negative main clause, etc.). An equivalent also existed in Old French, giens, but it had very low frequency in texts, and is sometimes used as an equivalent to rien ‘nothing’ and sometimes as an adverb supplementing the negation, often introducing a partitive (Price 1997: 178). Instances of embracing negation involving the marker ges are not frequent. Ges occurs in our corpus in 6.6% of all negative clauses. One cannot say that it may have become a marker of standard negation. If one looks at the proportion of negative sentences involving ges for each century, a very interesting pattern
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
129
0.12 0.1 0.08 0.06
GES
0.04 0.02 0
11th
12th
13th
14th
15th
Figure 6.2. Overview of frequency of ges
emerges (Figure 6.2). There is a regular increase in the usage of negation with ges until the fourteenth century: in the twelfth century only 3.6% of negative sentences involve ges, in the thirteenth century 5.9% and in the fourteenth century 9.8%. But the pattern nearly ceases to be used in the fifteenth century, where only 1.1% of negative sentences use ges. Data for the eleventh century are not reliable for a quantitative study, as there are only two attestations in a very small corpus. One should also note that the texts selected for the fourteenth century are from the earliest part of that century, while texts from the fifteenth century are from the second half. Within this period, the pattern nearly ceases to be used.
6.3.2 Semantics and pragmatics of ges Ges is already attested in some of the earliest textual records for Occitan, and in our corpus in the eleventh century text COM2:CSF, where there are two occurrences, both with ges preposed to non: (10) a. Diana ne Jovi non voil/, ne Minerva gens non acoill/, ne nun lai queir tornar mon oill. (COM2:CFS 266–8) ‘I want neither Diana nor Jupiter,/ nor do I GES welcome Minerva,/ and I don’t want to turn my eye towards them’. b. Gens non lai prendun bon sojorn/; nun voill eu aital deu cabdorn. (COM2:CSF 299–300) ‘They do not GES take a good stay there;/ I do not want such a god’. Both examples express a contrast between a previous assertion and the negative proposition. In example (10a), the young saint Fides has just explained how she will keep to the one true god: there is thus a clear contrast in her not wanting any of the pagan gods of the Romans. Similarly, in example (10b), the poem has just
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
130
described how the old gods burn in hell; the clause containing ges is thus confirming their status by using a contrastive negative sentence. Rather as with pas, the idea of an overt contrast with a previous or following assertion is key in explaining the examples found in the twelfth and thirteenth centuries. The negative clause with ges seems to be expressing that something is really not the case, in that it contrasts with a clause with a somewhat opposite meaning introduced by aintz or enantz ‘but, on the contrary’, or anc ‘instead’, or even sometimes mas ‘but’. This clear expression of a contrast is prevalent in the twelfth and thirteenth centuries, in particular in COM2:JAU and COM2:FLA, as shown in examples (11) and (12). (11) a. Cuida·l consegre denan se,/ mas jes no·l troba ni no·l ve,/ ni sap vas cal part s’es anatz,/ de que s’es fort meravillatz. (COM2:JAU 5299–302) ‘He thinks that he is following him in front of himself,/ but he does not GES find nor see him,/ nor does he know towards which part he has gone, / by which he is greatly astonished’. b. E fes per lo prat un eslais;/ mas sos cavals non es jes fortz,/ antz es fenitz e de fam mortz,/ que .viii. jorn a non manjet blat/ ni alre mais qu’erba de prat. (COM2:JAU 9076–80) ‘And on the meadows he started a run; but his horse is not GES strong, and instead it is finished and dying of hunger, because there for seven days it did not eat wheat or anything other than meadow grass’. (12) a. [ . . . ] mas ges non l’a sola trobada,/ anz fo mout ben acompanada,/ car de las domnas del castel/ avi’ ab si un gran tropel. (COM2:FLA 1009–12) ‘[ . . . ] but he has not GES found her on her own, / she was instead in good company, / because of the women of the castle / she had with her such a great herd’. b. Ges non i demora ni resta,/ ans s’en vai tot dreih a la tor [ . . . ]. (COM2:FLA 1312–13) ‘He does not GES stay nor remain,/ but he goes straight to the tower [ . . . ].’ In a large number of cases, the contrast between the negative clause and the assertion being negated is expressed in the form of a clause introduced by que or car ‘because’. In such cases, a following clause develops the reasons why something is definitely not the case, as in (13): (13) a. [ . . . ] que del re non ai jes poder,/ car lo rei es seinor de me/ e penra venjança de te/ per honor e per seinoria. (COM2:JAU 6156–9) ‘[ . . . ] because I do not have GES the power of the king,/ because the king is my liege /and he will take revenge on you /for honour and seigneury’.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
131
b. E la domna plora mout greu/ e respont: ‘Bel seiner, per Dieu,/ si Taulatz fes tant grantz sobriera,/ non es jes aiso la primiera,/ que ganren en a d’autres faichas’. (COM2:JAU 5003–7) ‘And the lady cries heavily /and replies: “Good lord, by God, /if Taulatz has done such excess, / it is not GES the first time, / because he has done many others’. c. [ . . . ] et ab Amor parlet suau / e dis: ‘Amors, sitot m’estau / de mon amic ara trop luein, / ges mon cor de lui non desluin, / qu’el lo ten, si com dis, en gaje’. (COM2:FLA 7146–50) ‘[ . . . ] and with Love she talked softly / and said: ‘Love, although I am / too far away from my friend, / my heart does not GES go far away from him, / because he keeps it, as he said, as a token’. These contrastive cases leave us with a small number of examples where there does not seem to be a contrast with an assertion of the text, but rather with something that is assumed to be true. It is possible that in such cases what is negated is in fact a presupposition present in the domain of discourse, and which has to be understood from the context rather than from the actual wording of the text. (14) a. [ . . . ] et eis lo jorn Guillems mandet/ a Castillon querre·ls obriers./ Uns vilans ne fon messagiers/ que ges Guillem non conoissia;/ non lur saub dir qui.ls i queria. (COM2:FLA 3720–3) ‘[ . . . ] and that same day Guillem sent someone / to fetch the workers in Castillon. / A villain who did not GES know Guillem was the messenger; / he was not able to tell them who needed them’. b. Guillems davan sidonz estet;/ quan il lo sauteri baiset,/ el li dis suavet: ‘Hai las!’/ Pero ges non o dis tam bas/ ques il fort be non o ausis. (COM2:FLA 3947–51) ‘Guillem was in front of his beloved / when she kissed the Psalter, / and he said to her, very low, “Alas!”. / But he did not GES say it so low / that she wouldn’t hear it perfectly’. In example (14a), one could assume that if a messenger is sent on an errand, he knows who is sending him: the presupposition that the reader makes here is not expressed in the text, but it is negated by using ges. In example (14b), because Guillem speaks very low, one could assume that it is too low to be heard, which would make it a presupposition: as such, the opposite meaning is not expressed in the text previous to the negation, but clarified afterwards. Such examples point to the fact that, even when there is an overt contrast with another assertion in the discourse, ges could in fact be negating a presupposition,
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
132
on the lines of: ‘it is not X (as you may think), but it is Y’, where Y and X are clauses with opposite propositional meaning (factoring out the negation). This is particularly the case in the many examples where the opposite assertion comes after the negated clause in the text, be it introduced by a marker of contrast (aintz, enanz), or by a causal complementizer (que, quar). Two examples give a hint of this possibility, in that the opposition is here with a proposition which is expressed as being commonly believed by people (15a), or what the writer assumes that his readers will think (15b), both being asserted after the negated clause: (15) a. Car cavalliers non es jes miga,/ ni no o par, que que hom s’en diga. (COM2:JAU 8779–80) ‘Because he is not GES at all a knight, /nor does he seem to be one, / whatever people might say’. b. E·l vergier es d’una pulcella,/ que a nom Brunesentz la Bella,/ e son castel a nom Monbrun./ E non cujes ges que sol un/ n’aia, enantz n’a d’autres moutz. (COM2:JAU 3069–73) ‘And the orchard belongs to a girl / named Brunesentz the Beautiful, / and her castle is named Monbrun. / But do not think GES that she has only one, / on the contrary she has many others’. In the fourteenth and fifteenth centuries, although a majority of examples still involve a contrast, it is no longer always present. In those texts, ges often seems to behave like a simple reinforcement of the negation, of the intensifier type, although it is sometimes still possible to derive a presuppositional reading (the two being sometimes present on the same examples, as shown by Ledgeway 2017): (16) a. Per so vos dic ieu que vo·n tornetz/ quar sains vos no intraretz ges. (COM2:MDP2 2741–4) ‘This is why I tell you to go away, / because you will not GES enter here’. b. Senher, nos em vengut ayci,/ e non aiatz ges meravilha. (COM2:GDB 1886–7) ‘Sir, we have come here, / and do not GES wonder’. This is further proven by the fact that ges can be the sole negator in a clause in the case of a rhetorical question in the fifteenth century: (17) Pueys que sains etz vengut,/ an vos volem parlar./ Etz vos ges dels disiplis/ d’aquest malvat truan/ que ha convertida la gen? (COM2:MDP2 3752–6) ‘Because you have come here, / we want to talk to you. / Are you not among the disciples / of that bad crook / who converted people?’
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
133
In this example there is no idea of a contrast, but rather the expectation that the answer to the question is yes. This proves that in the fifteenth century ges no longer acts only as the negator of a presupposition, but as a simple reinforcement (and in this case replacement) of sentential negation.
6.3.3 Syntax of ges Ges is an NPI, as it does not only appear under the scope of sentential negation, but also in negative polarity contexts, for example, in a positive clause embedded under a negative main clause (18a), under the scope of an operator a penas ‘barely’ (18b), or in indirect questions (18c). In such contexts, as in the rhetorical question in (17), ges appears to be the sole negator of the clause. (18) a. Mas tant non quer, que n’atrob ges,/ de que s’es dolentz e iratz. (COM2:JAU 2532–3) ‘But although he searches for it, he finds GES, / about what he is angry and suffering’. b. E ac lo col gros e espes/ e cort, c’a penas l’en par jes. (COM2:JAU 1411–12) ‘And he had a big, thick / and short neck, which nearly did GES appear’. c. Mas demandem a Centurio/, loqual es nostre companho/, se el i volria ges anar. (COM2:MDP2 1903–5) ‘But let us ask Centurio, / who is our friend, / whether he would GES like to go’. There are only two such examples in COM2:JAU, and only one in the following centuries, in which ges seems to be restricted strictly to contexts presenting sentential negation, at least in our corpus. In addition to this, ges can also reinforce a negative answer (19a) or be used for constituent negation in the form of non ges + XP (cf. 19b, with an adjective). (19) a. ‘Es vos aquel que tant d’enueig/ e tant de mal m’a faitz anueig?’/ Jaufre respont: ‘Domna, non jes;/ anc en luec on hom vos fazes/ enueig, non fui ni serai ja.’ (COM2:JAU 3585–9) ‘ “Are you the one who caused me so much trouble / and harm tonight?” / Jaufré answers: “Madam, no GES; / on the contrary in the place where one caused you trouble/ I was not and will not be”’. b. E va s’en lai de mantenen,/ non ges suau, mas tot coren. (COM2:JAU 2675–6) ‘And he immediately leaves that place, / not GES slowly, but all running’. As we have seen, ges is a generalizer, originating in an indefinite noun meaning ‘kind, species’. It functions as a quantifier. This is in particular shown by the fact
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
134
that it can take a partitive complement, either in the form of a full NP introduced by de (20b), or in the form of the clitic pronoun n(e) (20a). (20) a. Senher oste, no nos demandetz/ ponch d’argen, quar no n’avem ges/ quar be vos dic veramen/ que sertas no n’avem ponch. (COM2:MDP2 3289–92) ‘Dear host, do not ask us for / any money, because we do not have any (GES), / because I tell you in truth that we do not have any’. b. e si·s feira s’il non tengues/ davan sa cara ges de benda. (COM2:FLA 2415–16) ‘And he would have (seen her), if she had not held / GES a veil in front of her face’. A particular behaviour sets ges apart from other markers of emphatic negation in Occitan such as pas or ponch:¹³ its ability to be focus fronted, which means that it appears in front of the negative marker no(n). Old Occitan is characterized as presenting a focus position in the preverbal domain particularly used for new information focus (Wolfe 2018). This is particularly interesting because if ges is negating a presupposition, then the negated assertion can be understood as discourse-old, and the only new information is the negation itself. It could explain why focus fronting of ges is so frequent, particularly in the thirteenth century (60% of the occurrences of ges).¹⁴ The availability of focus fronting for ges is reminiscent of the same behaviour for a number of NPIs with added meaning, such as ren ‘something, nothing’ or jamais ‘ever, never’, which tend to be fronted in our corpus: (21) a. Jamais non voil manjar de pera. ‘I never want to eat a pear’. b. que jamais lo reis non veirai. ‘Because I will never see the king’.
(COM2:FLA 2979) (COM2:JAU 4783)
c. Jamais tan grans dampnatjes non s’era receubutz! (COM2:CCA 140 027) ‘Never had anyone received such great damage’.
¹³ Pas can be fronted in the thirteenth century, in COM2:FLA, as shown in the previous section, but the attested cases are marginal, and pas loses its ability to focus fronting after the thirteenth century. In addition, in these examples, pas is still highly correlated to no(n) in that nothing can be placed in between: all examples are pas no(n). On the contrary, ges can be fronted during the whole period, and allows other elements to intervene, although there are some quantitative changes occurring: there is very little fronting in COM2:JAU, but much more in COM2:FLA and COM2:CCA (about 60% of all ges), with a decrease in the availability of fronting in the following centuries, which might correspond to a more general decrease in focus fronting in the language. ¹⁴ New information foci seem to be more common in the thirteenth century (see Wolfe, Chapter 2, this volume).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
- d. Guillems ren als non atendia. ‘Guillem was not waiting for anything else’.
135
(COM2:FLA 1963)
When ges is in focus position, it generally attracts the object of the verb with it in the same position, which indicates a greater syntactic coherence between ges and the object. This, together with the fact that it can take a partitive complement, could be the start of ges becoming only a quantifier of absolute inexistence over an object: in modern varieties of Occitan which preserve ges (north Occitan, Provençal, eastern Languedocian varieties) it is only used to negate a partitive. (22) a. Uns vilans ne fon messagiers/ que ges Guillem non conoissia (COM2:FLA 37223) ‘A villain was the messenger /who did not GES know Guillem was the messenger’. b. Qui·l mi dara/ sapias que devis sera,/ car ges mon cor eu non daria/ a nulla re mais qu’el mon sia/ mas sol a vos, ma douza res [ . . . ]. (COM2:FLA 2875–9) ‘He who will give it to me / know that he will have to be a seer,/ because I would not GES give my heart/ to any other being in the world, / only to you, my sweet being [ . . . ]’. On the contrary, generally but not always, predicative adjectives and past participles are not fronted with ges, although this could come from an independent restriction on focus fronting and not from the interaction with negation. This could indicate that in the case of copular constructions and intransitive verbs, ges has scope over the verb rather than over the object. (23) Tardius sera, so·m cug, mos blatz;/ ges non es madurs al voibre,/ ans atendra lo glas e·l gibre. (COM2:FLA 4676–8) ‘My wheat will be late, I think; / it will not GES be ripe at the harvest season, /but it will wait until there is frost and ice’.
6.3.4 Discussion Ges is inherently a quantifier NPI and can be understood as a generalizer rather than a minimizer, which creates emphatic negation. As such, it always marks a contrast with an assertion or a presupposition present in the domain of discourse. Kiparski and Condoravdi (2006) show how emphatic negation has as a function the negation of discourse old information, assertions, and presuppositions: this confirms that ges is an emphatic negation, as it performs exactly these two
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
136
functions. The fact that ges can be focus fronted in a new information focus position confirms that the rest of the information in the sentence is somehow discourse-old, either from a presupposition or a previous (and in most cases following) assertion. They also identify two types of emphatic markers, minimizers and generalizers (Kiparski and Condoravdi 2006: 172): ges is of the latter type, although it does not initially develop from a quantifier but from a noun meaning ‘type, kind’, which comes to have a quantificational value. But the data on its syntax are too scarce to allow a confirmation in terms of position, along the lines of Garzonio and Poletto (2008). Garzonio and Poletto (2010) divide emphatic negation markers into two types, ‘adverbs which can be considered as equivalent to “at all” in English’, and ‘negative morphemes that are used to express the fact that an explicit or implicit assumption made by the interlocutor is wrong’. Ges clearly started off as the latter type, but there is indication that in the later medieval stages it is moving towards the former. Ledgeway (2017) showed that the same lexical item can be used for both types in some southern Italian dialects. He also shows that intensification of a negation often entails the negation of a presupposition. The case of ges is interesting in that its development seems to involve an opposite entailment: the negation of a presupposition comes to entail an intensive reading. This may have to do with the presence of a strong contrast nearly always expressed in the first half of our corpus, and still very often present in later texts: the contrast emphasizes the reasons why something is not the case, thus making it more likely to be intensifying the negation: ‘X is not the case, because Y’, comes to mean something like ‘X is really not the case’, ‘X is not the case at all, because Y’. This leaves open the possibility for such marker as ges to change function through its grammaticalization. A further indication is the modern behaviour of ges, which in contemporary varieties of Occitan behaves clearly as the first type, an intensifier, only in the negation of a partitive. Another indication is the fact that in modern Catalan, gens is used with the intensifier meaning, while probably having originated in the same type of negation of an assertion (Ledgeway 2017). If the development goes in that direction, one should find more examples of intensive ges in the following centuries, which has to be left for future research.
6.4 Conclusions Following the rise, establishment and, in the case of ges, the decline, of two Occitan postverbal negators, this chapter has highlighted the particular properties that each had at the various stages of their development. Although an ‘emphatic’ use was almost certainly there from the very start for pas, the data also show an early compatibility with another function, i.e. the negation of ‘discourse-old’ propositions, that figures throughout the period. The expression of the highly subjective
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
137
speaker’s view and link to the common ground are both elements found in rhetorical questions in which the strong (and unique) presence of pas may have accelerated its grammaticalization. Ges, on the other hand, seems to have started off as a negator of a presupposition and gradually moved to be an emphatic negator, and towards the end of the period there are signs that it may have become more of an intensifier. Furthermore, by the end of the fifteenth century, while pas is not an NPI, ges clearly is. One cannot but note that in quantitative terms, the use of pas rises exactly when that of ges falls out of usage, in the fifteenth century, suggesting that one strategy may have replaced the other. Further research, considering both later Occitan texts and comparative data from Oïl dialects, will be needed in order to test the hypotheses put forward in this study.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
7 The loss of clitic climbing in French A Gallo-Romance perspective Zack Bekowies and Mairi McLaughlin
7.1 Introduction 7.1.1 Overview The aim of this study is to examine the loss of clitic climbing in French from a Gallo-Romance perspective. Alongside the loss of pro-drop and the evolution of negation, the loss of clitic climbing is one of the best-studied word order changes in the history of French. It was a salient change when it took place: writers removed clitic climbing in new editions of their texts (Fournier 2002: 81), and observers of the language commented on the disappearance of structures displaying climbing (Seijido 2014: 106). The many scholarly investigations that have been carried out have established a clear timeline. In complex verbal predicates where an infinitive is governed by a matrix verb, the order with climbing was preferred until the seventeenth century when the order without climbing became more frequent, eventually becoming the preferred order in the last part of the century.¹ Corpus studies have also revealed a good deal about how the change took place: clitic climbing was lost in a gradual fashion, from complex syntactic environments in the first instance, and the decline in frequency of climbing was accompanied by a reduction in the range of matrix verbs with which it was used.² Our study therefore aims to shift the focus in order to consider this change from a Gallo-Romance perspective. The main interest of this new perspective lies in the fact that Occitan displays a similar evolutionary trajectory. Climbing is dominant until the seventeenth century when the structure without climbing becomes more frequent (Ronjat 1937: 558). In Occitan, however, the structure without climbing did not go on to become the preferred structure and climbing is still dominant today.³ This is perhaps not unexpected since clitic climbing and
¹ See Ayres-Bennett (2004: 218); Galet (1971: 323); McLaughlin (forthcoming: §4.2.2). ² See Ayres-Bennett (2004: 209–19); Ayres-Bennett and Caron (2016: 372–4); Iglesias (2015); McLaughlin (forthcoming: §4.2.3). ³ See Lafont (1967: 417); Oliviéri and Sauzet (2016: 345); Ronjat (1937: 558); Wheeler (1988: 272).
Variation and Change in Gallo-Romance Grammar. First edition. Sam Wolfe and Martin Maiden (eds). This chapter © Zack Bekowies and Mairi McLaughlin 2020. First published in 2020 by Oxford University Press.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
139
pro-drop have been linked in the theoretical literature, and French and Occitan pattern differently with regards to both of these features. As far as we are aware, the only diatopic work to have been done on the loss of climbing in French is Martineau’s (2009) comparative investigation of the loss of climbing in texts from France and from New France, the North American territory colonized by France and held between 1534 and 1763. She finds no evidence of regional variation and the only external factor in operation is date of birth with those born earlier making greater use of the structure with climbing (Martineau 2009: 228f.). We therefore take a different approach in order to explore not oïl varieties outside France but an oïl variety in France in direct contact with Occitan, where we know that clitic climbing remained dominant. This introduction provides an overview of clitic climbing in French, Occitan and the wider Romance context. The corpus and methodology of the current study are described in §7.2, and §7.3 then presents the results and a discussion before we conclude in §7.4.
7.1.2 Clitic climbing in French As Martineau (1991: 235) explains, ‘[i]n Old French, the object pronoun appears as a weak form before the main verb in the infinitival non-prepositional structures’ such as (1). (1)
Mes ele ne la pot veoir (La Chastelaine de Vergi) but she her.= can.3 see. ‘But she cannot see her’
In the case of prepositional structures, it tends to appear ‘as a strong form before the infinitival verb’ as in (2). (2)
Ne fina hui de moi proier au lonc du jor (ibid.: 126–7) finished.3 today of me beg. to.the long of.the day ‘He did not cease to beg me during all day’
As Jensen (1990: 165) points out, however, the version with climbing can also be found. In modern French, on the other hand, ‘the pronoun appears as a weak form before the infinitival verb’ (Martineau 1991: 235) in both types of construction as illustrated by (3) and (4) respectively. (3)
Elle ne peut she can.3 ‘She cannot see her.’
pas
la her.=
voir. see.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
140
(4)
Elle ne cesse de le she stop.3 of it.= ‘She doesn’t stop doing it.’
faire. do.
The only contexts in which climbing is still found in modern French are with verbs of perception and with causatives, as illustrated by (5) and (6). (5)
Elle le voit she him.= see.3 ‘She sees him dance.’
(6)
Elle le fait she him.= make.3 ‘She makes him work.’
danser. danse.
travailler. work.
It is the contrast between examples (1) and (3) that is the focus of the present chapter. Galet (1971) is responsible for the most comprehensive study of the period when the order with clitic climbing (1) ceases to be preferred and the order without climbing (3) becomes dominant in French, namely the seventeenth century. She carried out an extensive survey of the context in which the variants with and without climbing can appear across a wide range of texts and text types. Her data show very clearly that the order without climbing became dominant in the second half of the century: ‘Whatever the text type, tragedy or comedy, letter, funeral oration, fable, whatever the author’s particular style, the construction pronoun + governing verb + infinitive is archaic at the end of the century’ (Galet 1971: 323, our translation). This dating has been supported by subsequent studies. Ayres-Bennett (2004: 218) draws on both Galet’s data and her own analysis of translations into French to pinpoint the 1660s and 1670s as the decades in which the order without climbing became dominant. In her analysis of journalistic texts, McLaughlin (forthcoming: §4.2.2) finds that preferences switched at the same time: the order with climbing was still preferred in 1657 when it was used in 78% of cases but it was no longer the preferred order in 1682 when it appeared in only 30.77% of cases. Corpus-based studies have also shed light on the process of change. First, they all show that it is a gradual change which follows the typical s-curve: even if there are clear preferences for one order or the other, the dominant order was never used to the complete exclusion of the non-dominant one.⁴ Second, the studies have shown that the reduction in frequency of the order with climbing was ⁴ The variation is highlighted by Ayres-Bennett (2004: 210, 218); Iglesias (2015: 98–100); McLaughlin (in preparation: §4.2.3).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
141
accompanied by a reduction in the number of matrix verbs used.⁵ Finally, Iglesias (2015) and McLaughlin (in preparation: §4.2.3) both present data which indicate that the construction with climbing was lost first from more complex constructions. This includes, for example, so-called ‘verbal macrocomplexes’ involving three verbs, the coordination of infinitives, and the interposition of material between the matrix verb and the infinitive.
7.1.3 Clitic climbing in (varieties of) Occitan As in Old French, clitic climbing was the norm in complex verbal predicates in medieval varieties of Occitan (Jensen 1986: 108). This is illustrated by (7). (7)
si ls pogues if= them.= could.1 ‘If I could tame them’
adomesjar tame.
(ibid.)
Unlike in Old French, this also applied in cases where the matrix verb was followed by a preposition, such as (8). (8)
car lo menasava for him.= threatened.3 ‘For he threatened to hang him.’
a to
pendre hang.
(ibid.)
Jensen (ibid.) also notes, however, that already in the medieval language clitic pronouns could break up the matrix verb + infinitive complex as in (9): (9)
et a ta fi devem ti sebelir onradament (ibid.) and at your death must.1 you.= bury. honorably ‘And at your death we must bury you with honour’
According to Ronjat (1937: 558), this variant without climbing became more frequent around the seventeenth century. This is corroborated by Brusewitz (1905: 40–3) who, in a corpus of primarily literary texts, found both constructions to be in use in the seventeenth and eighteenth centuries, while in the sixteenth century and prior the variant with clitic climbing ‘seems to have still been dominant’ (Brusewitz 1905: 40, our translation). Despite this increase in frequency, the variant without climbing did not go on to reach the frequency that it did in French so that clitic climbing tends to remain the norm in varieties of Occitan today. As Ledgeway (2012: 120) notes, the ⁵ See Ayres-Bennett (2004: 219); Galet (1971: 336–49); McLaughlin (in preparation: §4.2.3).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
142
version with climbing is still ‘heavily preferred’ even if both examples such as (10a) with climbing and (10b) without it are found in a single variety. anam ajudar (10) a. los them.= go.1 help. b. anam los ajudar go.1 them.= help. ‘We are going to help them.’
(ibid.) (ibid.)
It is interesting that in Gascon, the alternative without climbing involves encliticization to the infinitive as in (11). plaser(11) Non va go.3 please. ‘It will not please me.’
me =me.
(Massourre 2012: 183)
The position occupied by the clitic in Gascon is not unique to the wider Romance family, since it is also found in varieties of Italo- and Ibero-Romance. It is, however, very rare in Gallo-Romance, with one other notable exception being the dialect of Menton, spoken in the south east of France near the Italian border (Oliviéri and Sauzet 2016: 345). Encliticization to the infinitive is clearly a feature of Italo- and Ibero-Romance that is shared only by border varieties of GalloRomance.
7.1.4 Clitic climbing in other Romance varieties It is not just in (Old and Middle) French and in Occitan that clitic climbing is found: clitic climbing was present across all of the old Romance languages.⁶ Previous scholarship has shown that in each case, the phenomenon became more restricted over time in terms of ‘domain, frequency, and accessibility’ (Wanner 1987: 299). The timing of this restriction varied from language to language. In some languages, such as Spanish, Italian, and Portuguese, clitic climbing began to decline as early as the fifteenth or sixteenth centuries.⁷ In Catalan, as in French and Occitan, it is only in the seventeenth century that the decline becomes apparent (Fischer 2002: 82). Clitic climbing is present in the earliest Romanian texts, which date from the sixteenth century, and its decline can be attributed to the gradual displacement of the infinitive in favour of the subjunctive in many
⁶ See Salvi (2011: 369f.); Wanner (1987: 263, 291f.). ⁷ For Spanish see Davies (1997: 260); for Italian see Wanner (1987: 299f.); and for Portuguese see Davies (1996: 98f.).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
143
complex predicates which took place from the sixteenth through the eighteenth centuries (Zafiu 2016: 15). The decline in the use of clitic climbing did not progress in the same way across all the Romance languages. We have already seen that constructions with climbing were lost almost completely from a language such as French whereas the variants with climbing remain dominant in varieties of Occitan; other languages which retain the variant with clitic climbing include Spanish, Italian, and Catalan.⁸ Portuguese is an interesting case because it presents evidence of diatopic variation. European Portuguese has optional clitic climbing but it is much less frequent in Brazilian Portuguese, particularly in the spoken language.⁹ Clitic climbing is very restricted in Romanian today where it only appears, and obligatorily so, with combinations of a putea ‘to be able to’ + infinitive.¹⁰ This pan-Romance perspective lets us better understand the position of Occitan today: by retaining clitic climbing, varieties of Occitan pattern more closely with varieties of Romance spoken in the Iberian and Italian peninsulas such as Spanish, Italian, and Catalan, than they do with their Gallo-Romance sister variety, French. This is not the only such case. In his study of pronominal placement in medieval varieties of Occitan, French, and Catalan, Hinzelin (2007: 12f.) highlights the fact that Occitan often patterns more closely with its Ibero-Romance neighbours than with French and Franco-Provençal because it is displays more conservative features at a range of different levels. In the concluding chapter, Hinzelin uses the results of his study to call into question the value of the Gallo-Romance grouping in general.
7.1.5 The interest of clitic climbing The present study is certainly not the first to address clitic climbing in French or the Romance languages; in fact it represents a very small contribution to a vast field of research. Clitic climbing is of particular interest in theoretical linguistics where scholars have sought to account for its optionality and its occurrence with a restricted set of verbs. Romance data have featured centrally in accounts of clitic climbing in the general literature, and scholars working on the Romance languages have made important theoretical contributions based on data from this family. Rizzi’s work on Italian (Rizzi 1978, 1982) has been particularly influential because it introduced the notion of restructuring to account for clitic climbing. Other scholars such as Kayne (1989) and Roberts (1997) offer explanations for
⁸ For an overview of the contexts which permit clitic climbing see Ledgeway (2016: 223f.) for Italian; Tuten, Pato, and Schwarzwald (2016: 405) for Spanish; and Alsina (2016: 377) for Catalan. ⁹ See Comrie (1982: 252); Galves, Ribeiro, and Torres Morraes (2005: 150). ¹⁰ See Hill (2001: 307); Monachesi (2005: 207).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
144
clitic climbing that involve head movement. The fact that the theoretical literature on clitic climbing spans more than four decades speaks to the scale of the interest in the phenomenon.¹¹ Despite the richness and value of the theoretical scholarship on clitic climbing in Romance, our study takes a different approach by building on corpus-driven research on historical varieties of Romance. We have already mentioned a number of such studies of French, and Comrie (1982) and Davies (1996, 1997) are responsible for similar work on Portuguese and Spanish. Corpus-based studies certainly have the potential to inform and improve theoretical work on clitic climbing, not least by providing the empirical facts for which theoretical claims must account. However, their main contribution has been to historical linguistics because they have shed light on historical variation and change. The cumulative effect of the work on French, for example, has been a relatively robust dating of the reversal of preferences from the variant with climbing to the variant without climbing in the second half of the seventeenth century. McLaughlin (forthcoming: §4.2) puts the focus on text type by examining the loss of clitic climbing in historical French newspapers and periodicals which have not previously been examined. By replicating the investigations of Galet (1971), Ayres-Bennett (2004: 209–19), and Ayres-Bennett and Caron (2016: 372–4), she is able to show that the reversal of preferences took place at the same time and in the same way in this previously unstudied text type. The aim of the present study is to test a new hypothesis about the loss of clitic climbing in French, namely that it would have taken place more slowly in langue d’oïl texts produced in the Occitan-speaking region because clitic climbing was never lost in the langue d’oc. This is essentially a dialect-contact hypothesis that reverses the traditional perspective by asking whether local dialects had an impact on the way in which a change took place in the standard language. Scholarship on language and dialect contact today supports this hypothesis. For example, Davidson (forthcoming) explores the effects of contact with Catalan on Spanish, reversing the traditional direction of scholarly interest which has tended to focus on the effects of the language of the majority, or the standard language, on minority languages and varieties. In addition, we are not the first to suggest that contact could have affected the placement of pronouns in historical GalloRomance; Hinzelin (2007) tests the hypothesis that contact with French could have affected the placement of object clitics in medieval Occitan. Although the results of his study rule out the possibility of a contact effect, Hinzelin’s work, along with Davidson’s, underscores the value of investigating the effects of internal Gallo-Romance contact in general.
¹¹ See for example Cinque (2004); Gallego (2016); Sportiche (1996); Wurmbrand (2001).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
145
7.2 Corpus and methodology Our initial aim was to create a corpus that would allow direct comparison with the historical journalistic data presented by McLaughlin (forthcoming: §4.2). We chose to examine texts printed in Toulouse because it was one of the three centres of printing alongside Paris and Lyon in the seventeenth century and is firmly located in the Occitan-speaking region. We were not able to compile a directly comparable journalistic corpus for the simple reason that there was no established local newspaper printed in this region in this particular period.¹² Instead, we chose to examine hagiographical texts recounting the lives of saints and church leaders. There are two main reasons for selecting this text type. The first is the volume of works of this type that were produced in the second half of the seventeenth century which let us identify and access texts published close to the years 1657 and 1682 that were studied by McLaughlin (forthcoming: §4.2) and that bracketed the reversal of preferences from the order with climbing to the order without. The second reason for selecting hagiographical texts is the relative stability of the genre in this period: the aims and style of the texts changed little if at all in the intervening years. We prioritized these two factors on the back of McLaughlin’s study of the historical press which shows that the volume of data and the relative generic consistency are two features which make periodical publications a particularly useful source of data for historicallinguistic research. We selected two hagiographical texts printed in Toulouse and two printed in Paris in the second half of the seventeenth century. We tried to match the dates of publication as closely as possible to the years 1657 and 1682, keeping an interval of at least twenty-five years between the two pairs of texts, and ensuring that they fall either side of the reversal of preferences dated to the 1660s and 1670s by previous scholarship. The four texts are listed below. Cueillens, F. (1689). La Vie de Fr. Matthieu Viste. Toulouse: Louis Bosc. De Lacoux, H. (1687). La Vie et les actions de la venerable Mere Marie Alvequin de Jesus, Religeuse de Monmartre, Superieure et Reformatrice des Dames Augustines de S. Magloire dites Penitentes. Paris: Jean-François Dubois. Gondon, G. (1662). L’Imitateur de Jesus Christ, ou la Vie du Venerable Pere Antoine Yvan, Prestre, Instituteur de l’Ordre des Religieuses de Nostre-Dame de Misericorde. Et l’histoire de la fondation du même ordre. Paris: Jean Bouillard. Saint-Martin, R. P. (1659). La Vie de S. Thomas de Vil-Neuve. Toulouse: Jean Boude.
¹² For a basic history of the press in France, see Feyel (1999).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
146
For ease of comprehension, from now on we will refer to the four texts in the corpus by their place of publication and date: ‘Paris-1662’, ‘Paris-1687’, ‘Toulouse1659’, and ‘Toulouse-1689’. As can be seen from the titles of the works listed above, each celebrates the life of a religious figure. They vary in length between 228 and 659 pages. They are all written in what we might characterize as standard high-register French. For the most part, the texts consist of third-person narration although the first person can be found in frame sections introducing either the work as a whole or individual chapters. Unsurprisingly all of the texts are characterized by the frequency with which religious terminology appears. The methodology that we used to extract the tokens parallels the approach of McLaughlin (forthcoming: §4.2) which is based on that of Galet (1971), AyresBennett (2004: 209–19), and Ayres-Bennett and Caron (2016: 372–4). We extracted the first 150 tokens of complex verbal predicates where an infinitive with an object pronoun is governed by a matrix verb. This was possible for all but the shortest text which contained only 86 tokens (Paris 1687). Overall, we collected 536 tokens, which is more than the 272 tokens examined by McLaughlin (forthcoming: §4.2). The tokens were classified according to the position of the clitic, where there are two possibilities, namely a variant with climbing and a variant without climbing illustrated respectively by tokens (12) and (13) from the corpus. leur voulut jamais découvrir le (12) il ne he them.= wanted.3 never discover. the jour de son depart (Toulouse-1659, p. 73)13 day of his departure ‘he never wanted to reveal his day of departure to them’ s’ y arrester long-temps (Paris-1662, p. 27) (13) il ne pût he could.3 self.3 =there= stay. long-time ‘he couldn’t stay there long’ We follow most previous scholars in excluding contexts where the infinitive is governed by a preposition because climbing in those cases had already become obsolete by the seventeenth century (Galet 1971: 33–4).¹⁴ We also exclude verbs of perception and causation which still require climbing today. Both of these
¹³ With the exception of the pairs u/v and i/j whose use we have modernized, the original orthography is retained in all of the examples from the corpus. Each example is followed in parentheses by the text from which it is taken and the page number. ¹⁴ There is one exception to this generalization: venir de can still be found with clitic climbing in the seventeenth century (Galet 1971: 33f.). Iglesias (2015: 101) explains that this is possible because of the high level of grammaticalization of this construction.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
147
excluded cases are illustrated by examples (14) and (15) which are taken from the texts in the corpus but excluded from our token count. avertir de se (14) pour les for them.= tell. of self.3= Parroisse parish ‘in order to tell them to go to the parish’
rendre dans la go. in the (Toulouse-1689, p. 8)
verroit un jour briller avec beaucoup (15) on la one it.= see..3 one day shine. with much d’honneur (Toulouse-1689, Epitre) of-honour ‘we would see it shine one day with great honour’ In addition to classifying tokens according to the position of the clitic pronoun, we also gathered data about the matrix verb and the complexity of the environment in which the construction is found (presence/absence of interposed material, coordination of infinitives, verbal macrocomplexes).
7.3 Results and discussion This section presents the results of the study followed by a discussion. In §7.3.1, we examine the data regarding the frequency with which clitic climbing is found in the different texts in the corpus. These are the data which let us address the main hypothesis, namely that climbing will be more frequent in texts printed in Toulouse. In §7.3.2, we use the data from the corpus to address more general questions about the process by which the order with climbing was replaced by the order without climbing in French. A general discussion of the results of the study is offered in §7.3.3.
7.3.1 Frequency of clitic climbing Table 7.1 presents the frequency and relative frequency with which clitic climbing is present in complex verbal predicates in the four hagiographical texts in the corpus. It shows that just over half of all examples display climbing: 280 out of a total of 536 tokens, or 52.24%. The fact that the frequency is near the 50% mark suggests that we are looking at the right period since the change is evidently underway. This figure is nothing like the figure found by McLaughlin (forthcoming: §4.1) for the year 1632, where 84% of examples display climbing, and it is nothing like the figures
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
148
Table 7.1. Frequency of variants with and without climbing in the corpus place
year
with climbing
without climbing
total
Paris
1662
Paris
1687
Toulouse
1659
Toulouse
1689
45 (30%) 30 (34.88%) 97 (64.67%) 108 (72%) 280 (52.24%)
105 (70%) 56 (65.12%) 53 (35.33%) 42 (28%) 256 (47.76%)
150 (100%) 86 (100%) 150 (100%) 150 (100%) 536 (100%)
total/average
she finds for the eighteenth century when on average only 14.29% of examples display climbing. Looking more closely at Table 7.1, we see that there is considerable variation in the frequency of the constructions with and without clitic climbing in the corpus. At its lowest, clitic climbing appears in only 30% of cases in Paris-1662 and at its highest, it appears in 72% of cases in Toulouse-1689. The texts all actually display a fairly even one-third: two-thirds ratio with the order either with or without climbing being dominant. This is presumably a chance effect but it makes comparison relatively straightforward. The data in Table 7.1 let us examine the effects of two different variables: place of publication and date of publication. If we compare the rates of climbing in the texts printed in Toulouse to those printed in Paris we see a very clear split whereby the variant with clitic climbing is preferred in Toulouse and the variant without climbing is preferred in Paris. Climbing is found in 64.67% and 72% of cases in Toulouse-1659 and Toulouse-1689 respectively, whereas it is found in just 30% and 34.88% of cases in Paris-1662 and Paris-1687 respectively. These results appear quite clearly to confirm the main hypothesis of the present study, namely that the variant with clitic climbing remained for longer in texts printed in the south of France. The design of the study prevents us from concluding with certainty that this difference is the result of influence from Occitan but the data certainly support this idea. If we examine date of publication, however, the data are considerably less easy to interpret. Rather than the reduction in the rate of climbing that would naturally be expected, both pairs of texts display an increase in use of climbing. In Paris, the frequency of climbing increases from 30% in 1662 to 34.88% in 1687, and in Toulouse it increases from 64.67% in 1659 to 72% in 1689. The very small size of the increases means that the most important point here is not that there is an increase but that the decrease in the rate of climbing that would be expected according to the previous literature is not in evidence. For example, McLaughlin (forthcoming: §4.1) reports a reduction of almost 50% between 1657 and 1682 when the rate of climbing falls from 78% to 30.77%. Such a drop is far from
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
149
anomalous; Ayres-Bennett (2004: 215–18) dates the turning point in the history of the loss of clitic climbing to the 1660s and 1670s. The texts under analysis in the present study were selected precisely on the grounds that they fall on either side of this window. There are several possible explanations for the absence of the expected reduction in the frequency of clitic climbing in the corpus. The most obvious reason is that the texts under analysis were produced too close in time for the change to be shown. We know that at some point in time, clitic climbing would have been the dominant order in hagiographical texts in French and that at some later point, the order without climbing would have become dominant. A future study could presumably capture the change by expanding the time frame. A second possible explanation for the absence of the expected reduction in frequency in the two pairs of texts relates to the nature of the genre. One of the main advantages of the journalistic texts investigated by McLaughlin (forthcoming) is the precision with which they can be dated. In contrast, there can be variation in the interval of time that lies between the date of composition of a hagiographical text—or part thereof—and its eventual publication. Both of the explanations proposed above, however, are complicated by the fact that there is a clear contrast between the texts printed in Paris and those printed in Toulouse, which suggests that the change was simply further along in the oïl region than in the oc region. A future broader study of the loss of climbing in hagiographical texts will be needed in order to get a fuller picture of the dating of the change in this particular text type.
7.3.2 The process of change The aim of this section is to explore what the data provided by the four hagiographical texts in the corpus can tell us about the process by which the variant with climbing was lost in French. As indicated in §7.1.2, previous studies have already revealed a certain amount of information about the process by which this change took place. The data gathered from the hagiographical texts let us explore two different aspects: first, the range of matrix verbs with which clitic climbing can occur, examined in §7.3.2.1, and second, the complexity of the environment in which the variant with clitic climbing is found, examined in §7.3.2.2.
7.3.2.1 Matrix verbs Table 7.2 presents the frequency with which each of the matrix verbs occurs with and without clitic climbing in all four texts in the corpus. Table 7.2 shows that in total 22 different verbs are used in complex verbal predicates either with or without climbing across the corpus as a whole. As we would expect, modal verbs such as devoir, pouvoir and vouloir are very frequent, with other verbs such as aimer mieux, être (in the sense of aller), and témoigner appearing only once. The data in Table 7.2 can be used to calculate the type/token
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
150
Table 7.2. Frequency of variants with and without climbing by matrix verb (whole corpus) matrix verb aimer mieux ‘prefer’ aller ‘go’ courir ‘run’ croire ‘believe’ daigner ‘deign’ désirer ‘desire’ devoir ‘must’ espérer ‘hope’ être ‘be’ falloir ‘be necessary’ oser ‘dare’ penser ‘think’ pouvoir ‘be able’ prétendre ‘claim’ retourner ‘return’ revenir ‘come back’ savoir ‘know’ sembler ‘seem’ témoigner ‘testify’ venir ‘come’ venir de ‘to have just’ vouloir ‘want’ total
with climbing 26
1 44 1 21 1 1 96 1 1 1 6 16 64 280
without climbing 1 3 1 4 3 3 15 2 23 9 3 80 1 3 4 13 1 17 1 69 256
total 1 29 1 4 3 4 59 2 1 44 10 4 176 2 1 4 10 13 1 33 1 133 536
Table 7.3. Type/token ratio for variants with and without climbing by matrix verb (whole corpus)
types tokens type / token ratio
with climbing
without climbing
overall
14 280 20
20 256 12.8
22 536 24.36
ratio for the matrix verbs in the variants with and without climbing. Those results are presented in Table 7.3. The fact that the type/token ratio is lower for the variant without climbing means that these data conform with those presented in previous studies. The change is quite clearly under way here and already clitic climbing is found with a smaller range of verbs than is the variant without climbing. It is possible to explore the corpus from a different perspective by grouping the texts according to place of publication. Since the variant with climbing is dominant in both of the texts printed in Toulouse and the variant without climbing is dominant in both of the texts printed in Paris, in so doing, we are effectively
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
151
Table 7.4. Frequency of variants with and without climbing by matrix verb (Toulouse) matrix verb aller ‘go’ croire ‘believe’ devoir ‘must’ espérer ‘hope’ être ‘be’ falloir ‘be necessary’ oser ‘dare’ penser ‘think’ pouvoir ‘be able’ prétendre ‘claim’ retourner ‘return’ revenir ‘come back’ savoir ‘know’ sembler ‘seem’ venir ‘come’ venir de ‘to have just’ vouloir ‘want’ total
with climbing 23 26 1 20 1 67 1 1 1 6 9 49 205
without climbing 2 3 4 2 15 4 2 30 1 3 2 1 7 1 18 95
total 25 3 30 2 1 35 4 3 97 2 1 4 8 1 16 1 67 300
comparing texts which represent two different stages of the change. It is certainly unusual to use a diatopic criterion to access diachrony in this way but it is not totally without precedent because it in some ways parallels what is done when change in apparent time is used to capture change in real time.¹⁵ Table 7.4 presents the data for the texts printed in Toulouse, and Table 7.5 presents the data for the texts printed in Paris. Comparing Tables 7.4 and 7.5 shows quite clearly that a reduction in the frequency of the variant with clitic climbing is accompanied by a reduction in the range of matrix verbs with which it is found: clitic climbing is used with twelve different verbs in the texts from Toulouse but with only eight different verbs in the texts from Paris. And as was the case in Table 7.2, in both Tables 7.4 and 7.5, the variant without climbing is found with a wider range of verbs than the variant with climbing. The results of this section align very clearly with those of previous studies in showing that the loss of the variant with clitic climbing was accompanied by a reduction in the range of matrix verbs with which it was found.
7.3.2.2 Complexity of the environment The aim of this section is to explore whether the corpus illustrates another aspect of the loss of the variant with clitic climbing that was proposed by Iglesias (2015) ¹⁵ For a discussion of real time and apparent time, see Cukor-Avila and Bailey (2013).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
152
Table 7.5. Frequency of variants with and without climbing by matrix verb (Paris) matrix verb aimer mieux ‘prefer’ aller ‘go’ courir ‘run’ croire ‘believe’ daigner ‘deign’ désirer ‘desire’ devoir ‘must’ falloir ‘be necessary’ oser ‘dare’ penser ‘think’ pouvoir ‘be able’ savoir ‘know’ sembler ‘seem’ témoigner ‘testify’ venir ‘come’ vouloir ‘want’ total
with climbing 3
1 18 1 1 29
7 15 75
without climbing
total
1 1 1 1 3 3 11 8 5 1 50 2 12 1 10 51 161
1 4 1 1 3 4 29 9 6 1 79 2 12 1 17 66 236
and preliminarily confirmed by McLaughlin (forthcoming: §4.2.3). Iglesias carried out a quantitative study of rates of clitic climbing in different contexts and his results indicate that there is a difference between contexts which he labels simple and complex. Simple contexts are those in which ‘V1 is a simple verb form and V2 is an infinitive with no interposed elements, no coordination of V2 etc., and involving the most frequent constructions: devoir, pouvoir, and vouloir + infinitive’ (Iglesias 2015: 98, our translation). All other contexts display some level of complexity. The data presented by Iglesias support his hypothesis that the variant with climbing was lost first from more complex constructions. He finds the most variation in simple contexts where the three writers that he examined make frequent use of both orders (Iglesias 2015: 98). He finds no such variation when a preposition is present: climbing is always avoided with à and almost always with de (Iglesias 2015: 100). The coordination of verbal predicates also favours the variant without climbing (Iglesias 2015: 103), as does, at least for two out of the three authors in his study, the interposition of an element between the two verbs (Iglesias 2015: 106). What he calls ‘verbal macrocomplexes’ (our translation) where three verbs are present represent a slightly more complex case because there are three possible positions in which the clitic could appear: before the first verb, between the first and second verb, and between the second and third verbs. Iglesias (2015: 104) finds no examples of the clitic appearing in the highest position but he does find that the clitic will systematically appear to the left of the second verb if that verb normally allows climbing (Iglesias 2015: 105). Iglesias investigates a range of other linguistic factors but none of them can so
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
153
clearly be correlated with either of the variants as can the ones discussed above. McLaughlin’s (forthcoming: §4.2.3) journalistic data appeared to confirm Iglesias’ (2015) general claim that the variant with climbing is progressively lost starting with more complex environments. She found that the order without climbing is first used for the most part in contexts which display some level of complexity. She also found that, vice versa, the order with climbing remains the longest in simple contexts. In order to test whether the change is taking place in the same way in our corpus of hagiographical texts, we examined two texts at the extreme ends of the continuum that is created by the four texts in the corpus, namely Toulouse-1689 where clitic climbing occurs in 72% of cases, and Paris-1662 where it occurs in 30% of cases. We examined only those factors which Iglesias (2015) found to have the strongest effect, namely the interposition of elements between the two verbs, the coordination of the verbal predicates, and the presence of a third verb to form a verbal macrocomplex. Each case is illustrated in turn by tokens (16), (17), and (18). crû que je le (16) & j’ ai I=have believed that I it.= faire do. ‘I believed that I had to do it in this way’
devois ainsi must..1 thus (Paris-1662, p. 128)16
ne témoigna jamais plus de constance à (17) Il He show..3 never more of commitment to y vouloir demeurer, travailler & souffrir (Paris-1662, p. 136) there= want. stay. work. suffer. ‘He never showed any more commitment to staying, working and suffering there’ le vouloient faire parler (18) Lorsque les Religieux When the religious.. him.= want..3 make. speak. alors de la devotion (Toulouse-1689, p. 113) so of the piety ‘So when the men of the cloth wanted to have him speak about piety’ We also analysed the tokens which represent the most simple environment, that is to say tokens such as (19) which contain a finite simple verb form with no negation, inversion or interrogation.
¹⁶ This page is misnumbered as 812 in the original.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
154
(19) une one
seule alone
passion passion
nous us.=
pouvoit could.3
damner éternellement damn. eternally (Toulouse-1689, p. 45)
‘just one passion could damn us for eternity’ That means that a group of tokens from each text such as those where the matrix verb is in a compound tense is excluded from the analysis because they can be considered neither fully simple nor completely complex. Tables 7.6 and 7.7 present the frequency of the variants with and without climbing in the four different environments in the texts Toulouse-1689 and Paris1662. If we recall that the average frequency of the variant with climbing in the text Toulouse-1689 is 72%, then we can see that the data in Table 7.6 are in line with Iglesias’ (2015) findings: the frequency of the variant with climbing is under 72%
Table 7.6. Frequency of variants with and without climbing according to environment (Toulouse-1689) environment
with climbing
without climbing
total
interposition
20 (62.5%) 3 (60%) 3 (50%) 65 (79.27%) 91 (72.8%)
12 (37.5%) 2 (40%) 3 (50%) 17 (20.73%) 34 (27.2%)
32 (100%) 5 (100%) 6 (100%) 82 (100%) 125 (100%)
coordination macrocomplexes simple total / average
Table 7.7. Frequency of variants with and without climbing according to environment (Paris-1662) environment
with climbing
without climbing
total
interposition
3 (13.64%) 3 (20%) 0 (0%) 29 (38.16%) 35 (28.93%)
19 (86.36%) 12 (80%) 8 (100%) 47 (61.84%) 86 (71.07%)
22 (100%) 15 (100%) 8 (100%) 76 (100%) 121 (100%)
coordination macrocomplexes simple total/average
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
155
in all three complex environments (interposition, coordination, and macrocomplexes), and it is over 72% in the simple environment. Although the figures are small, they are consistent with the idea that the variant with climbing was lost first from the more complex constructions. Exactly the same remarks apply to the data from Paris-1662 where the average frequency of the variant with climbing is 30%. Table 7.7 shows that the variant with climbing is present at a lower rate in all three of the complex environments and at a higher rate in the simple environment. Once again, the results are consistent with Iglesias’ (2015) hypothesis that the construction with climbing was lost first from more complex environments. As was the case in §7.3.2.1, the results of this section align with those of previous studies that have examined the process by which the variant with climbing was lost in French. The interest of these results is that they underscore the relative stability of the process of change across text type even in cases where the results regarding the date at which the change takes place display more variation.
7.3.3 Discussion In this section, we offer a discussion which places the results presented in §7.3.1 and §7.3.2 in their wider disciplinary context. We consider what the results tell us about clitic climbing, Gallo-Romance syntax, language change, and text types in historical-linguistic research. Unsurprisingly, this study does not make any direct contribution to theoretical scholarship on clitic climbing as a linguistic phenomenon. However, it is possible that some of the findings will be of use to scholars working in different theoretical frameworks in the future. This is most obviously true of the work presented in §7.3.2 which supports Iglesias’s (2015) suggestion that the variant with climbing was lost progressively from more complex environments. If corroborated by future empirical studies, this information may shed light on the nature of the clitic climbing construction itself. In addition, we could imagine a future project which would be inspired in part by generative scholarship. Kayne (1980: 40; 1989) famously posited a link between clitic climbing and pro-drop. French and Occitan pattern with regards to pro-drop as they do with clitic climbing: both allowed null subjects in the medieval period and this possibility was lost in French but not in Occitan. Pro-drop is therefore an obvious choice for a future investigation. The loss of pro-drop pre-dates the advent of printing,¹⁷ but if sufficient textual evidence survives, a future study could examine French manuscripts produced in the langue d’oc region to see whether pro-drop was retained for longer than in manuscripts produced in the langue d’oïl region.
¹⁷ See for example Balon and Larrivée (2016).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
156
As regards Gallo-Romance syntax, the results of this study support Hinzelin’s (2007) previous work on pronoun placement in the medieval varieties in underscoring the distinction between French and Occitan. As regards both pronominal placement relative to the finite verb and clitic placement in complex predicates, varieties of Occitan clearly align more closely with varieties of Ibero-Romance than they do with other varieties of Gallo-Romance. The importance of syntactic differences between old French and old Occitan is also highlighted by Vance, Donaldson and Steiner’s (2009: 313–17) comparison of word order after fronted clauses in old French and old Occitan. Despite the growing number of studies highlighting the syntactic differences between French and Occitan, both the present study and Hinzelin’s also draw attention to their interconnexion because they both posit the existence of contact effects operating within Gallo-Romance. The results of this study contribute more significantly to historical linguistics. The primary aim of the investigation was to test whether contact with varieties of Occitan where clitic climbing was retained affected the progression of the change in French texts produced in the Occitan-speaking region. The comparison of two texts produced in Toulouse with two texts produced in Paris presented in §7.3.1 suggests that this hypothesis is correct: the frequency with which clitic climbing appears is higher in the texts from Toulouse than in the texts from Paris. This suggests that dialect contact can affect the progression of a change in the standard language even if its only effect is to slow the rate of change in texts produced in the relevant regional context. The fact that dialect contact had this effect is not in itself surprising; we see similar mechanisms in operation in situations of language contact today. For example, Otheguy and Zentella’s (2012) research on the Spanish spoken in New York City has shown that both dialect contact and contact with English have affected the rate of, and constraints governing, pro-drop. The authors single out ‘variable morphosyntactic features such as subject pronouns’ (Otheguy and Zentella 2012: 219) as being susceptible to this kind of contact effect. It is probably therefore only during the process of change that dialect contact is likely to have had an effect on the placement of clitic pronouns; once clitic climbing was lost in French, it is far less likely that the influence of Occitan could have affected the placement of the clitic in standard French texts produced in the langue d’oc region. The study also contributes to our understanding of text types in historical linguistics. Recent research has attempted to improve our understanding of the history of individual languages such as French by broadening the range of text types used to elucidate the process and dating of linguistic changes. A group of chapters in Ayres-Bennett et al. (2018) aims to respond to the following question: ‘Given that descriptions of previous stages of the language are generally based on literary texts, what sources have until now remained underutilized in diachronic studies and to what extent do they allow us to improve our understanding of the history of French?’ (Ayres-Bennett et al. 2018: 7, our translation). The present
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
157
study is intended to serve as a follow-up to McLaughlin’s (forthcoming) work on historical newspapers and periodicals which showed that changes tend to take place in the same way across different text types but that there can be important differences as regards the date at which they take place. Both of these conclusions are supported by the research presented here: the variant without clitic climbing does not become dominant in all text types—nor indeed in all texts of each type— at the same time, but the process by which the variant with climbing is lost appears to be consistent across both text and text type. Once again, therefore, an empirical study has illustrated the advantages of diversifying the text types used in historical-linguistic research. Hagiographical texts such as the ones studied here are not frequently the object of linguistic investigation but the frequency with which they were produced, at least in the seventeenth century, and their relative generic constancy, makes them a useful source of data for diachronic work.
7.4 Conclusion The aim of this study was to examine the loss of clitic climbing in French from a Gallo-Romance perspective. The analysis of a corpus of four hagiographical texts published in Paris and Toulouse in the second half of the seventeenth century showed that clitic climbing was lost more slowly from texts produced in the Occitan-speaking region than in texts produced in the north of France. The investigation does not prove that contact with varieties of Occitan where climbing remains to this day was the cause of its longer retention, but the data are consistent with this hypothesis. The research demonstrates the value of incorporating text types that are not frequently used in historical-linguistic research and it underscores the importance of dialect contact even in the medium of printing and in the case of a language with a relatively high level of standardization. There is a growing body of comparative work on French and Occitan in the medieval period,¹⁸ but this study indicates that there is the potential for work on this interface in the early-modern period as well. The fact that French is so often at the extreme end of the continuum of variation displayed by the different Romance languages while Occitan tends to pattern with the more conservative core makes the Gallo-Romance branch a particularly interesting case for study.¹⁹
¹⁸ Examples include Hinzelin (2007); Vance (2009); Wolfe (2018). ¹⁹ Bossong (2016: 68–70) examines three other features which place French at the extreme end of the continuum of variation: phonetic reduction and possible positions for stress, the partitive, and the replacement of the synthetic perfect by the analytic perfect. See also Smith (2016: 317).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
PART II
T H E V E R B CO M P L E X
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
8 Motivating the North–South continuum Evidence from the perfects of Gallo-Romance Bridget Drinka
8.1 Introduction In his comprehensive examination of the development of Romance morphosyntactic features, Ledgeway (2012: 314) claims that the Romance languages should be classified not on the traditional east–west axis or according to central vs peripheral status but rather on a north–south continuum. Following Zamboni (2000), Ledgeway points to a number of morphosyntactic features which diverge in the Romance languages of the north (French, northern Italo-Romance, the Raeto-Romance varieties) and the south (central and southern Italo-Romance, the languages of the Iberian Peninsula, Romanian). Important among these features is the / dichotomy in the north, which contrasts with a generalized auxiliary in the south. Ledgeway attributes this distribution to the retention or loss in northern and southern varieties, respectively, of active / stative alignment distinctions established in late Latin. According to this schematization, the early destabilization of the nominative / accusative balance represented by the periphrastic deponent perfects was either restabilized by the introduction of a transitive periphrastic counterpart, + participle, resulting in a / dichotomy as in the north, or by the removal of the imbalance through the generalization of , as in the south. The brief but incisive analysis of the north–south continuum by Smith (2016) brings the distribution of perfect features among the northern oïl varieties into focus. While Smith does not call Ledgeway’s north / south typology into question, he does point out that the northern Gallo-Romance varieties show considerable variability with regard to several key features, inviting ‘further research into the nature of the correlation between the various characters of this typology’ (Smith 2016: 318). It is in the spirit of this recommendation, and in recognition of the exceptional contribution that J. C. Smith has made to the field, that the present chapter is conceived and dedicated.
Variation and Change in Gallo-Romance Grammar. First edition. Sam Wolfe and Martin Maiden (eds). This chapter © Bridget Drinka 2020. First published in 2020 by Oxford University Press.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
162
The organization of the chapter is as follows: §8.2 presents the theoretical background for this analysis, focusing especially on the claims of Zamboni (2000) and Ledgeway (2012) concerning the north–south orientation of the Romance languages, with regard to the development of the periphrastic perfect. Section 8.3 discusses several key implications of these claims. Section 8.4 points to the need for a stratified areal approach in the analysis of the perfects of Romance and other European languages. Section 8.5 presents evidence from the GalloRomance varieties of southern France, contrasting evidence from Occitan and Catalan. Section 8.6 compares evidence from the perfects of Wallon, Lorrain, Norman, and other varieties of northern France, with special focus on the contributions of Smith (2016). Section 8.7 provides conclusions and suggestions for the refinement of the characterization of the North–South Continuum.
8.2 The North–South Continuum 8.2.1 Zamboni In his classic work, Alle origini dell’italiano, Alberto Zamboni (2000: 86f.) makes a significant observation about the shared tendencies of the Romance languages: rather than being arranged along an east–west axis (Wartburg 1950) or according to a central to peripheral diffusion (Bonfante 1968), the morphosyntactic patterns of the Romance languages point to a north–south distribution of features (Map 8.1). Zamboni presents evidence such as that listed in Table 8.1 to support his claim. Of special interest to us here is the fact that the ‘northern’ Romance varieties— Gallo-Romance, Raeto-Romance, the northern varieties of Italo-Romance—all form both and perfects, mark gender and number agreement in the participles of these perfects, and tend towards the increased use of perfects as preterites, with the concomitant decrease in the use of synthetic perfects; in the ‘southern’ Romance varieties—Ibero-Romance, central and southern Italo-Romance, Table 8.1. Key Features of the North–South Continuum according to Zamboni (2000) N -s plurals obligatory subject S-marking / perfects past participle agreement preterite weakened
S vocalic plurals null subject O-marking perfect past participle non-agreement preterite strengthened
Northern Romance Southern Romance
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
Map 8.1. North–South Continuum of the Romance languages
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
164
and southern Romanian¹—these tendencies are not in evidence. Zamboni regards this distribution of features as indicating a shift in alignment: he assumes that late Latin had developed active-inactive tendencies, as witnessed especially by the unaccusative vs unergative values of vs auxiliaries, respectively; these active / inactive features were, according to Zamboni, maintained in the northern varieties but given up in the south. Especially attractive in this characterization is Zamboni’s recognition (Zamboni 2000: 87) that the northern varieties provided evidence of a ‘conservative innovation’ (‘un’innovazione conservativa’), the construction of a new system of periphrasis based on two systems already in existence: in this scenario, the construction and the construction, not construed as one category in earlier Latin, came together to share agreement patterns and semantic value in late Latin and early Romance, and it was this unified system that the northern varieties retained. In the south, this contrast was ‘cancelled’, according to Zamboni (2000: 104), with only the perfect persisting.
8.2.2 Ledgeway Ledgeway (2012: 317f.) likewise points to the split-intransitivity of the late Latin perfects, as witnessed by transitive / unergative predicates with A / SA subjects taking a auxiliary, and unaccusative and passive predicates with SO subjects taking a auxiliary. He goes on to point out that the integration of and constructions into a single perfect category was probably not fortuitous, but, as claimed by La Fauci (1988: 46–50; 2011: 103–14), represents a dichotomy already in existence in Latin, since periphrastic forms already existed as an integral part of the passive and deponent (i.e. middle) paradigms of the perfect system. These forms were already viewed in contrast to active / transitive perfect forms (e.g. ‘made..3’: ‘was made’ ‘said.3’; ‘was said’), so that the analogical extension of to active verbs would simply have reinforced the contrast between active and passive / deponent, already in existence in the perfect system. According to Ledgeway, the / contrast and other signs of active / stative contrast were maintained in northern Romance (1), but not in southern Romance (2), except in particular varieties—Alguerès and Balearic Catalan (3, 4), Aragonese (5), Sardinian (6), etc.—where the dichotomy is maintained. ¹ Spoken Daco-Romanian and Istro-Romanian, however, also participate in the perfect > preterite shift, as do several other northern Balkan varieties such as northern Albanian (Gheg) and northern varieties of Serbian and Croatian, as well as Slovenian, suggesting late areal adoption of this trend. The retention of the preterite vs perfect distinction in several southern Romanian dialects (e.g. in Oltenia, where the preterite refers to recent, hodiernal past) also supports this explanation (Gvozdanović 1995; Thomas 2008, 2010; Zafiu 2013; Drinka 2017). While minimizing the role of areal influence in his explanation of the development of Romanian auxiliaries, Ledgeway (2014b: 7, 27) recognizes the heavy reliance of Romanian on analytic structures as a Balkan feature.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
- sem arribats al we.are arrived at.the ‘we arrived at the market’
(1)
avèm susat we.have sweated ‘we sweated’
(2)
Ja havia fumat already I.had smoked ‘I had already smoked’
(3)
antuɲeta es tunara Antonietta is gone ‘A. has gone’
(4)
Què has trobat? what you. have found ‘What have you found?’
(5)
¿Has trobau as claus? You.have found the keys ‘Have you found the keys?’
(6)
s akkwa a bbuɖɖía ɛ nnáššiu su zɔli (Campidanese Sardinian) the water has boiled is risen the sun ‘the water boiled’ ‘the sun rose’ (Ledgeway 2012: 320, 341; Loporcaro and Putzu 2013: 207; Loporcaro 2016: 813)
Ja havia arribat already I.had arrived ‘I had already arrived’
mercat market
165
(Occitan)
(standard Catalan)
la munɛra, no l’as preza the money., not it. =you.have taken. ‘the money, you haven’t taken it’ (Alguerés Catalan) som tornada de Barcelona I.am returned from Barcelon ‘I have returned from Barcelona’ (Balearic Catalan) l’augua yera bullida the water was boiled The water had boiled.
(Aragonese)
The presence or absence of agreement of the participle is also regarded as telling evidence as to the alignment patterns of varieties: while northern varieties tend to mark agreement of the participle (7), southern varieties tend to lose this agreement (8), a trend which Ledgeway (2012: 348) views as ‘concomitant attrition’ to the loss of auxiliary variability. (7)
la mela, l’avevo mangiata the apple it I.had eaten. ‘I had eaten the apple.’
(8)
la manzana la había comido / **-a the apple it I.had eaten. / ‘I had eaten the apple’
(Italian)
(Spanish)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
166
He notes, however, that standard Catalan, in spite of its adoption of a auxiliary for the perfect, as in Spanish and Portuguese, does show such participial agreement (9), as in the northern varieties (10). (9)
(10)
aquella revista ja l’he that magazine already I.have ‘that magazine, I’ve already read it’
llegida read.
quella rivista l’ ho già that magazine it I.have already ‘that magazine, I’ve already read it’
letta read.
(standard Catalan)
(Italian)
Ledgeway, following La Fauci (1988: 102f.), claims that this presence of agreement in standard Catalan does not represent retention from a previous active / stative stage of the language, but represents a newly-established active / stative alignment. This claim, along with several others, requires closer examination, and will be returned to in §8.3.² Romanian, like other Romance languages, is assumed to have inherited active / stative alignment from late Latin, but to have lost this feature by the time of the first records. The / division of labour in present-day Romanian is manifest not in alignment variability but rather in the contrast between finite present perfect forms (with ) (11a) and future perfects, conditionals, and subjunctives (11b–d) (with , fi < ‘become’)³ (Loporcaro 2016: 813). (11) a. am I.have
mâncat eaten
/ plecat / left
(Romanian)
b. voi fi cântat / fugit I.will be sung / fled ‘I will have sung / fled’ ² See also Ledgeway’s arguments (Ledgeway 2012: 333–40) concerning the distribution of case and syntactic tendencies across the northern and southern varieties, a distribution which also suggests greater adherence to an active / stative orientation in the north. The role of the perfects in establishing this orientation remains central, however. ³ In some Romanian dialects (especially located in the north and west, in Transylvania, the Banat, Crişana, the Maramureş, and part of northern Moldavia), the analytic surcomposé pluperfect can be formed alongside or in place of the synthetic pluperfect (mâncasem / plecasem ‘I had eaten / left’), utilizing the finite auxiliary (am ‘I.have’) and the non-finite form (fost ‘been’) in one construction, incorporating, that is, an ordinary perfect of the verb ‘be’ ‘I have been’ as part of the new construction: am fost mâncat / plecat ‘I had eaten/left’. (Smith 1989c: 311f.; Ledgeway 2012: 347).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
167
c. aş fi cântat / fugit I.would be sung / fled ‘I would have sung / fled’ d. nu cred să fi cântat / fugit not they..beieve that be sung / fled ‘They don’t believe that I / you / (s)he / we / you / they have (/ has) sung / fled’ (examples from Ledgeway 2012: 346; Loporcaro 2016: 813) Like the other Romance languages, Romanian has moved towards a nominative / accusative alignment, but retains traces of its former split-intransitivity, according to Ledgeway. Some of the key features of Ledgeway’s characterization of the North–South continuum are presented in Table 8.2. Ledgeway’s most important claims can be summarized as follows: • Classical Latin was an accusative language. • While this accusative alignment was maintained in the imperfectum, active / stative alignment arose in the perfectum, especially in connexion with / auxiliation of the perfects. • The active / stative alignment was maintained most firmly in the north, as seen in the retention of / aux distinctions. • Through time, some southern languages gave up this distinction, and began to use only as an auxiliary; at the same time, they ceased to show agreement. These languages have moved towards accusative alignment, a tendency found across the Romance languages. Table 8.2. Some key features of the North–South Continuum, according to Ledgeway (2012: 314) N V2 syntax kept longer A/S marking; subject clitics / perfect auxiliaries agreement kept perfect replaces preterite
vs
S V2 syntax lost early O marking; subject pro-drop perfect auxiliaries agreement lost preterite maintained⁴
⁴ See footnote 1 for the variable distribution of Romanian.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
168
8.3 Discussion of Ledgeway’s characterization While the existence of a North–South continuum appears convincing and largely credible, several assumptions require further consideration. Assumption 1: The development of the perfect is dependent on the primacy of the perfect, and is built upon it. The perfect is indisputably older than the perfect in Latin: even in the earliest texts, the passive and deponent perfects were formed periphrastically, with + participle. Deponents are originally middles, often identified as passive in morphology but active in meaning. As shown in (12) and (13), the passives and deponents are structurally identical, forming periphrastics (14, 15) with a auxiliary in the perfect and pluperfect: (12) perfect passives present laudor perfect laudatus sum imperfect laudabar pluperfect laudatus eram (13) perfect deponents present sequor perfect secutus sum imperfect sequebar pluperfect secutus eram
‘I am praised’ ‘I was / have been praised’ ‘I was being praised’ ‘I had been praised’
‘I follow’ ‘I followed / have followed’ ‘I was following’ ‘I had followed’
(14) perfect passive Ita proelium restitutum est thus engagement renew. be ‘thereupon the battle was renewed’ (Caesar, Commentarii de Bello Gallico 1, 53) (15) perfect deponent Ex libris Punicis[ . . . ] interpretatum nobis est From books Punic translate. by us is ‘We have translated from the Punic books.’ (Sallust, Bellum Iugurthinum 17, 7) Ledgeway (2012), following La Fauci (1988) and Zamboni (2000), assumes that the perfect was formed as a counterpart to the perfect, and that the two
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
169
together established a split-intransitive system in late Latin and early Romance which all the Romance languages inherited. What should be noted, however, is the fact that the construction is not embedded into a paradigm as the construction is. It does not substitute for the synthetic perfect of Latin, and does not show high frequency in most stages of Latin. While the construction does show some signs of grammaticalization at various points throughout the history of Latin, it frequently maintains its possessive value, at least to some extent, in appearing in an objective complement construction (16). Only rarely does it appear without an object (17). cinctum habebant fīlō (16) (Flamines) . . . caput Flamines head... girt... had.3. woollen_fillet... ‘The flamines had their hair girt with woollen fillet’ (Varro, De Lingua Latina 5,84) re supra scriptum (17) quemadmodum de ea in_what_manner of this thing... above write.... habemus have.1.. ‘as we have written above on this matter’ (Vitruvius, De Architectura book 9, chapter 1, section 14) Adams (2013: 646–51) goes so far as to suggest that the construction should probably not be regarded as a single construction but several different ones, that these constructions were not grammaticalized during most of the history of Latin, and only showed some sign of grammaticalization in the late Republican era, in the work of authors such as Cicero (18). intersit, (18) hoc quid here which... lie_between.3... si tuos digitos novi, certe habes if your fingers know.1.. certainly have.2.. subductum subtract. .. ‘What a wide difference this implies, if I know your fingers, you have certainly reckoned’ (i.e. you have [habes] what difference it makes [hoc quid intersit] reckoned [subductum] ) (Cicero, Epistolae ad Atticum 5, 21, 13) The perfects were not fully grammaticalized, either: as mentioned above, their function was circumscribed, since they were used only as perfect passives and
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
170
perfect deponents. In the end, we can conclude that the early construction and construction do not behave as one category during most of the history of Latin, are not recognized by speakers as one category, and, thus do not provide a convincing unified source for a split intransitive auxiliary system.⁵ Assumption 2: The establishment of split-intransitivity occurred in late Latin, at a time when all of the Romance languages could be affected. What is suggested here is that this dichotomy occurred not in late Latin, but that it came to full articulation only in Carolingian Latin, applying only to those varieties within the Carolingian realm, the so-called Charlemagne Sprachbund. All the Romance languages would have had access to the earlier and forms of Latin discussed above (the periphrastic perfect passives and deponents and the less-fully grammaticalized objective-complement perfects, respectively). But it was in the Carolingian area where the and constructions came to be construed as one category, the perfect, and came to be fully grammaticalized. As will be explored more fully in §8.4, it was the expansion of the deponents in this region that helped establish their periphrastic variant, the perfect, in this role. This expansion did not occur in the non-Carolingian areas— e.g. in Castilian, Portuguese, southern Italo-Romance, Romanian—and, as a result, alone came to predominate there. It should also be noted that some Germanic languages were spoken within the area of the Carolingian empire, and those within this area—for example, Old High German (19 a,b) and Old Saxon—also developed productive / auxiliation. Those Germanic varieties lying outside the Charlemagne Sprachbund—for example, Swedish (20 a,b) and English—did not. perfect (19) a. wizist thaz, thiz wib firworaht habet ira lib knowest that this woman forfeit. has her life ‘you know that this woman has forfeited her life’ (Old High German, Otfrid III, 17.13) b. perfect ‘Thu bist’, quad er, ‘herasun queman’ thou art quoth he hither come ‘You have’, he said, ‘come here.’ (Old High German, Otfrid II 7,67) (20) a. hon ‘she
har has
förverkat forfeited’
b. han ‘he
har has
kommit come’
(modern Swedish)
⁵ See Drinka (2015) for additional evidence and arguments concerning the multiple sources of the European perfects.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
171
Assumption 3: The resurgence of a nominative / accusative alignment, especially in the ‘southern’ Romance varieties, caused these languages to give up the / alternation and focus on as the sole auxiliary. The establishment of a single-auxiliary system is viewed as a mechanism for resolving the imbalance that had occurred when the new alignment pattern was introduced (Ledgeway 2012: 351f.). The difficulty with this claim is that it does not take into account the surprisingly similar solution that all the ‘southern’ languages reached: they all greatly reduced or removed the use of a auxiliary, and established as the only auxiliary for the perfect. It is assumed that this reduction occurred separately in all of these varieties. Romanian, for example, did not have contact with the other Romance languages for a millennium, and yet would have lost its original split-intransitive system, much as Spanish or southern Italo-Romance did, ending up with a -prominent system. What I will claim here is not that the rise of a split-intransitive system is incorrect, but that the full implementation of a split-intransitive system should be seen as localized to the ‘northern’ area, where a more robust establishment of the dichotomous auxiliary system occurred. The ‘southern’ varieties did not engage in this expansion upon incipient tendencies from earlier Latin and did not construct the dichotomy between auxiliaries that the ‘northern’ varieties did. These languages fell back on resources already within the language, and came to rely almost exclusively on the perfect. Their unified predominance must represent not innovation but archaism—a non-participation in the Carolingian innovation. Ultimately, then, only the unified areal distribution of the northern varieties calls for explanation; the southern varieties can be explained as not having adopted the innovative alignment as thoroughly. For example, the (partial) reversal of the active / stative innovation in Romanian (Ledgeway 2012: 346; Ledgeway 2014b) may be better seen as a less complete adoption of the innovation in the first place. I claim that the contrast represented by the and perfects is not just a retention of an ancient pattern but a reinvigoration of that pattern, not unlike the ‘innovazione conservativa’ proposed by Zamboni (2000: 87). As will be explained in the following section, the geographical expanse of this realm corresponds exactly to Ledgeway’s northern territory.
8.4 Motivating the North–South Continuum: a stratified model of the development of the perfect in Europe 8.4.1 Why should the Carolingian realm be the locale of the innovative expansion of the / dichotomy? In previous work (Drinka 2017, 2019), I have claimed that the development of the perfects in western Europe took place in three major stages.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
172
S
S
S
The widespread use of the perfect across western Europe represents the diffusion of the Latin construction, inherited in Romance, calqued into Germanic. Within the area, a division of labour between auxiliaries and auxiliaries later developed, corresponding to the increased use of deponents seen in Carolingian documents. This reinforcement only occurred in the core area; in the peripheral area, prevails. Within the / area, anteriors began to take on preterital value. First witnessed in the vernacular of twelfth-century Paris and its environs, it spread to areas influenced by French culture, such as western and southern Germany and northern Italy, and eventually into contiguous areas such as the Slavic territories under the rule of the Habsburgs.
If these stages are charted on maps which are then stacked one atop the other, we obtain a stratified, three-dimensional view of how this complex linguistic area developed. The three stages are depicted schematically in Map 8.2, a slightly revised map from the World Atlas of Language Structures (WALS).⁶ = []+[ /]+[ > ] = [/]+[ > ] = [ > ] =
Peripheral area with less typical perfects S I. Present-day area where perfects occur S II. Within the area, division of labour between perfects and perfects S III. Within the / area, anterior meaning of perfect has developed into past or perfective
Adapted from B. Drinka (2017) Language Contact in Europe. The Periphrastic Perfect through History. Reproduced with permission of Cambridge University Press through PLSclear. It should be noted that the boundaries of Stages II and III do not coincide exactly. This lack of overlap clearly demonstrates the shifting of the centre of influence over time. Sociolinguistic pressures will promote feature-sharing during times of intense contact, but centres of influence will shift when these pressures change.
⁶ See Drinka (2017: 150) for details of the revisions.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
173
HAVE HAVE HAVE/BE
HAVE/ BE
P
PERF>PRET
HAVE/ BE
HAVE P>P
HAVE
P
Map 8.2. Revision of WALS map of the Have Perfect Based on Dahl and Velupillai (2013)
Focusing especially on Stage II, we note that, remarkably, the distribution of the / auxiliaries coincides precisely with the boundaries of Charlemagne’s reign at the time of his death (marked in bold on Map 8.3). The only exceptions are Danish, which, though not a Carolingian territory, has adopted the / dichotomy through intense contact with German (Johannisson (1945: 226; Askedal 1995: 115f.), and Breton, which has, in similar fashion, undergone extensive influence from French (Orr 1992).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
174
HAVE HAVE HAVE/BE
HAVE/ BE
P
PERF>PRET HAVE/ BE
HAVE P>P
HAVE
P
Map 8.3. Revised WALS map of Perfect compared to Charlemagne’s Realm (814 ) = []+[ /]+[ > ] = [/]+[ > ] = [ > ] =
Peripheral area with less typical perfects S I. Present-day area where perfects occur S II. Within the area, division of labour between perfects and perfects S III. Within the / area, anterior meaning of perfect has developed into past or perfective
Adapted from B. Drinka (2017) Language Contact in Europe. The Periphrastic Perfect through History Reproduced with permission of Cambridge University Press through PLSclear
Why should the innovative increase in periphrastic perfects correspond so well with the territory of Carolingian rule, and what role does Latin play in this distribution? An examination of the evidence for and auxiliaries will help us to address these questions.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
175
8.4.2 H perfects in Carolingian Latin While Latin legal language, that is, laws, juridical writings, and legal formulae, frequently contained + participle, this construction retained the possessive meaning of ‘keep’, and cannot be construed as a perfect (Benveniste 1962: 57f.; Adams 2013: 635–7). However, the construction underwent a noteworthy increase in productivity in the juridical formulae of early medieval documents, especially those produced in Gaul (21): (21) Utrum ille homo hoc homicidium perpetratum haberet Whether that man this homicide perpetrated ...3 ‘Whether that man had committed this murder’ (Formula of Lindenbrogius p. 280, 18) Noteworthy above all in this innovative expansion is the more frequent use of the perfects in the Carolingian capitularies (Thielmann 1885: 545–7) (22): (22) secundum quod iudicatum habemus following what judged ..1 ‘in conformity with what we have ruled’ (Capitulary of Charlemagne 175, 25; 183, 33) This expansion of the perfect in Carolingian legal documents is, in turn, reflected in legal and literary corollaries in the Romance and Germanic languages.
8.4.3 Deponents and perfects in Carolingian Latin Not only did the perfects flourish in particular contexts in Carolingian Latin, but the perfects also grew in prominence. Deponents came to be especially productive in late Latin, beginning especially in the sixth century, and increasing in frequency to a climax in the late eighth and ninth centuries, precisely at the time of Charlemagne’s reign. Their use was so pervasive that virtually all deponents had active counterparts with which they could be freely interchanged. And just as the synthetic passives were replaced by periphrastic forms, the growing category of deponents was likewise paraphrased by the perfects. It appears that Latin and the vernacular shared the responsibility for the development of the periphrastics in their influence of one another. As noted by Flobert (1975: 589), the trends seen in Carolingian Latin must correspond directly to changes in the vernacular, and vice versa. The result of these developments was the firmer establishment of periphrastics, both in Carolingian Latin and in the early vernaculars of the ‘northern’
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
176
varieties. The effects of the Charlemagne Sprachbund persist to this day in the distribution of the and dichotomy.⁷
8.5 Evidence from southern Gallo-Romance varieties 8.5.1 Introduction If the stratified approach presented above accurately depicts the development of the ‘northern features’ of the Romance languages as originating first in the Charlemagne Sprachbund (see Stage II above), and second, through Parisian influence (see Stage III above), we should be able to find evidence for this stratification in the distribution of features in the present-day Romance languages. I will examine the morphosyntactic features of several varieties of Gallo-Romance to assess whether their perfects provide evidence for this stratification. I will begin with a look at the southern varieties, comparing especially the histories of the perfect in Occitan and Catalan, and will then turn to several northern oïl varieties, especially Wallon, Lorrain, and Norman.
8.5.2 Occitan Vernacular langue d’oc, or Occitan, was written down as early as the eleventh century, well before the northern French vernaculars were committed to writing, and, as the language of the troubadours, gained high prestige as a courtly language in the late eleventh to the late thirteenth centuries. Administrative texts also appeared as early as the twelfth century, surviving in large numbers, especially in Toulouse (Bec 1967: 74f.). Old Occitan spread as a language of administration across the Midi in the thirteenth and fourteenth centuries before being subsumed by French in the fifteenth century (Lodge 1993: 111–13). Occitan offers especially valuable evidence for the validity of the stratified chronologization of the Romance languages presented here, since it participated in the innovative formation of the / dichotomy as part of the Carolingian Empire (Stage II), but did not take part in the shift of perfect to preterite (Stage III), lying outside the realm of influence of Paris at the time. In example (23), from Boeci, the earliest extant literary text of old Occitan, transitive verbs formed the perfect with aver ‘have’. In other early texts, unaccusatives (24) and reflexives (25) appear with eser ‘be’.
⁷ See Drinka (2017: 158–66) for additional evidence for this development from the Annales Regni Francorum and the Strasbourg Oaths.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
- (23) Las mias musas, qui ant perdut lor cánt, de sapiencia anava eu ditan; ‘My poems, which have lost their song, I had written about wisdom’
177
(Old Occitan; Boeci, c.1000)
(24) Ar non puesc plus soffrir lo fais, (Old Occitan; Guilhem IX of Aquitaine ‘Pos de chantar m’es pres talenz’, early-12th century) tant soi aprochatz de la fi. ‘Now I can no longer bear the burden, so near I have [lit. am] come to the end.’ (25) ‘Toza’ fi m’ieu, ‘cauza pia,
(Old Occitan; Marcabru ‘L’autrier just’una sebissa’, mid-12th century)
destoutz me soy de la via . . . ’ ‘Girl,’ I said, ‘you pious creature, I have turned off the road’ [lit. turned myself am] The contrast between for unaccusatives (26) and for transitives (27) is well illustrated in the variable auxiliation of morir ‘die’: (26) Unaccusative (‘die’) e lai es mortz Wilelmes
(La Chanson de la Croisade albigeoise, 13th century) ‘and Wilelme has [lit. is] died there’
(27) Transitive (‘kill’) trops homes a mortz ses glavi
(Marcabru ‘Dire vos vuelh ses duptansa’, mid-12th century) ‘[Love] has slain many men without a sword.’
Thus, the / dichotomy is productive and well established in Occitan, and is in evidence as early as the twelfth century. As mentioned above, however, the perfect in old Occitan is almost always anterior, that is, referring to an event which began in the past but which has present relevance, as illustrated in (23)–(27). The Occitan perfects do not show signs of productive participation in the semantic shift of perfect > past in the twelfth and thirteenth centuries.
8.5.3 Catalan Like Occitan, Catalan provides clear support for the role of the Charlemagne Sprachbund in establishing the / dichotomy in western Europe,
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
178
furnishing crucial evidence of Carolingian-style auxiliation during the time of its connexion with the realm, and a decided realignment toward Castilian-style auxiliation when that connexion ceased. When Charlemagne’s son, Louis the Pious, captured Barcelona in 801, Catalonia was incorporated into the Carolingian holdings as part of the Spanish march, a military buffer between Al-Andalusia and the Carolingian Empire. While there are no extant narrative historical sources written in Catalan before the twelfth century to document the history of the area and its connexion with the Carolingian Empire, an abundance of administrative documents in Latin do exist, the richest collection of such documents in ninth century medieval Europe outside of northern Italy (Collins 1990: 172). This profusion of documents reflects Catalan’s complex history: unlike the rest of the Carolingian realm, Catalonia continued to adhere to Visigothic Law, the Lex Gothorum, which, ultimately, represented a continuation of late Roman practice. This tradition required written documentation of all legal proceedings, and a constant need to seek the king’s confirmation for all land entitlements. These two tendencies together explain why so many Catalan abbots, bishops, and landowners travelled to the Carolingian court over the ninth and tenth centuries, and why the ties with the Carolingian realm remained so strong (Collins 1990: 181f., 184f.). While synthetic preterites (‘perfects’) predominate in these Latin documents, examples like (28), a precept from the time of Charlemagne himself, do exist. (28)
Charlemagne’s Præceptum pro Hispanis, 2 April 812, concerning land grants in the unsettled lands of the Spanish Marches Dicunt etiam quod aliquas villas, quas ipsi laboraverant, say.3. also that some farms which self work.3.. laboratas [ab] illis eis abstractas work.... from those these separate.... habeatis [ . . . ] have.2. ‘They also say that some farms, which they themselves worked, you have worked and separated from these’ et quicquid contra justiciam eis vos aut juniores vestri and whatever against justice these you or younger your factum habetis do.... have.2. ‘and whatever you or your heirs have done against justice’ (Abadal i de Vinyals 1952: 314)
That Catalan took part in Carolingian trends early in its history is witnessed by the use of auxiliary with a reflexive in (29):
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
179
(29) après que Cató se fou gitat sobre la sua after that Cato was throw. on the espasa (Fr. Antoni Canals 14th century) his sword ‘after C. had thrown himself on his sword’ In the twelfth century, political affiliations began to change: the count of Barcelona became the King of Aragon, so that Catalonia was separated from France; in the fifteenth century, Aragon, which controlled Catalonia, was united with Castile, with a subsequent realignment of cultural and linguistic allegiances (Moll 1952). As a result, many varieties have gravitated toward the Castilian pattern of predominance. Catalan’s reorientation towards other varieties of the Iberian Peninsula can also be seen elsewhere in its verb system: in the semantic extension of tenir from ‘hold’ to its use as the unmarked verb of possession, with the concurrent relegation of haver to auxiliary-only status, and, likewise, in the later parallel establishment of estar as copula, impinging upon ser (Steinkrüger 1995). Other evidence is also available that supports this approach. First, the language of isolated communities, such as Balearic and Algherese Catalan, provide telling insights as to earlier stages of the language, pointing to a ‘northern’ orientation in earlier times. The Catalan of Alghero, in northwestern Sardinia, still had the / contrast in place in the fourteenth century at the time of its settlement. This contrast was maintained and reinforced in contact with Sardinian and northern Italo-Romance varieties (30) (Blasco Ferrer 1984: 162f.). (30) Algherese
sọ́ anát I.am go. ‘I have gone’
vs
standard Catalan
he anat I.have go.’ ‘I have gone’
Likewise, the Balearic varieties of Mallorca and Isola Rossa keep some vestiges of the sixteenth century Catalan, whether in the use of the auxiliary in unaccusative contexts (31) or in the exclusive use of , as in Catalan and Castilian, but with more extensive participial agreement (32) (Alcover 1903: 470f.; Moll 1952: 33; Alsina 2016: 380). trobat? (31) a. Què has what have.2 find. ‘What have you found?’ b. som tornada de Barcelona am.1 return. from Barcelona ‘I have returned from Barcelona’
(Balearic Catalan (=4))
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
180
(32) La gràcia e la mercè que Deus m’ha The grace and the mercy that God me has feta (Balearic Catalan, Mallorca) make.. ‘The grace and mercy that God has made for me’ Secondly, in some communities in northwestern Catalonia, older rural speakers maintained the more conservative / auxiliation, especially in varieties in close proximity to Aragonese (33a), but younger speakers, especially male urbanites working in industrial or service jobs, are now adopting the exclusive use of the auxiliary, as in standard Catalan (33b) and Castilian (Alturo Monné 1995: 228f.). s’ era casat (33) a. Paquita ia Paquita already was married ‘Paquita had already gotten married’ b. Paquita ia s’ havia casat Paquita already have married ‘Paquita had already gotten married’
(NW Catalan dialect of Alta Ribagorça) (Standard Catalan)
Finally, as mentioned in §4.2.2, participial agreement patterns persist in some Catalan varieties, including standard Catalan, even with a auxiliary, as in ‘northern’ varieties (34). ja l’he llegida (standard Catalan (=9)) (34) aquella revista that magazine already I.have read.. ‘that magazine, I’ve already read it’ (example from Ledgeway 2012: 348) Ledgeway (2012: 348), following La Fauci (1988: 102f.), denies that the agreement seen in (34) implies retention, but La Fauci’s suggestion that this agreement was constructed later based on a newly-established active-statement alignment seems inconsistent with other data presented here, and indifferent to sociolinguistic and areal considerations. Keeping in mind the stages proposed for the development of the Romance perfect, we note a surprising fact which suggests, once again, that the stages represent different innovations, with borders that do not necessarily coincide with one another: by the beginning of the sixteenth century, some varieties of Catalan had taken part in the semantic shift of perfect to preterite similar to that which occurred in French and northern Italy (Eberenz 1977) (Stage III), implying an ongoing relationship with languages outside the Iberian Peninsula. To summarize, Occitan and Catalan, together, provide us with strong evidence as to the role of the Charlemagne Sprachbund in establishing a ‘northern’
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
181
linguistic tradition of / auxiliation in the Romance languages. In so doing, they also provide us with a chronologically fine-grained perspective of how speakers align themselves socially by means of linguistic choices. As part of the Carolingian realm, Occitan took part in these ‘northern’ innovations, as did early Catalan; when political realignment occurred for Catalan, this fact is recorded in the new ‘southern’ realignment of the perfects. Traces of previous connections persist, for example in participial agreement. The adoption of this new orientation was surely gradual, as exemplified by the present-day movement of young Catalan speakers of the northwest from / towards exclusive -usage. As documented by scholars such as Loporcaro and associates (e.g. Loporcaro 2007; Loporcaro and Putzu 2013), Giancarli (2011), Smith (1989c, 2016), and others, this renegotiation of identity occurs through the gradual and systematically variable adoption of innovative features by speakers at linguistic and dialectal borders like these. Catalan, then, is extremely valuable as an example of a language which has undergone the alignment shift proposed by Ledgeway, but in the light of history. Traces of a previous adherence to a split-intransitive system are evident in the agreement patterns of the modern participles, which appear even with a auxiliary, and in the archaic retention of / auxiliation in isolated enclaves. At the same time, the gradualness of the shift and the permeability of borders is witnessed in the present-day realignment towards the standard by young urban speakers in north-western Catalonia. In its movement away from splitintransitivity and towards a nominative / accusative alignment, then, Catalan points to the essential role of contact and social motivation for this change: it has not occurred in a vacuum, but coincides with social realignment.
8.6 Evidence from northern oïl varieties 8.6.1 Introduction As mentioned in §8.1, Smith (2016: 318) regards the variable distribution of perfect features among the northern oïl varieties as somewhat challenging for the North / South typology. Some of the variability that Smith describes turns out to accord well with the stratified approach presented here; some of it creates difficulties, just as it does for the North–South hypothesis, and will require further explanation. I will first examine the data from several varieties, and will then attempt to draw conclusions. Comparing Ledgeway’s features of ‘northern’ and ‘southern’ types, Smith points out that standard French provides clear evidence of its aptness as a ‘northern’ language. The three key features of ‘northern’-style perfects— / auxiliation,
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
182
Table 8.3. Key Features of the North–South Continuum in Standard French, according to Smith (2016: 317) N
S
V2 syntax kept longer A/S marking; subject clitics case system kept longer / perfect auxiliaries agreement kept perfect replaces preterite
V2 syntax lost early O marking; subject pro-drop cases lost early perfect auxiliaries agreement lost preterite maintained
participial agreement, and preterital use of the perfect—are all in evidence there (Table 8.3).
8.6.2 Northeastern oïl varieties In several northern and northeastern oïl varieties, such as Wallon, Lorrain, Champenois, and Picard, these ‘perfect’ features tend to follow so-called ‘southern’ patterns. For example, tends to appear as the only auxiliary (35a,b), and agreement with the past participle does not occur (35b). (35)
dialect of Wallon in La Gleize a. dj’a v’ni I have come ‘I have come / came’
cf. standard French je suis venu I am come ‘I have come / came’
b. il a mouri îr he has died yesterday ‘he died yesterday’
il est mort hier he is died yesterday ‘he died yesterday’
Furthermore, in eastern Wallon and to some extent in southern Wallon, the preterite is still in productive use. Sporadic preterites are also found in Lorrain, Franc-Comtois, and Bourguignon, but not in Picardy (Smith 2016: 314f.)⁸ (Table 8.4).
⁸ The exclusion of Picard from these two micro-areas may be due to the linguistic history of the region: from the eighth century until the time of the French Revolution, the Pas-de-Calais and Nord regions of northern France were Flemish-speaking. The cities of Lille, Cambrai, Calais, and Dunkirk were part of the County of Flanders in the Middle Ages (Treffers-Daller and Willemyns 2002: 3); beginning in the thirteenth century, Picard varieties began to gain prestige, and spread first to towns and later to surrounding areas, especially through commerce and education (Ryckeboer 2002: 22–6).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
183
Table 8.4. Variability in Eastern and Northeastern (oïl) varieties (Smith 2016: 314f.) N
S
V2 syntax kept longer A/S marking; subject clitics case system kept longer / perfect auxiliaries agreement kept perfect replaces pret: w. Wallon
V2 syntax lost early O marking; subject pro-drop cases lost early perfect auxiliaries agreement lost preterite maintained: e.Wallon
8.6.3 Northwestern oïl varieties In northwestern oïl varieties, on the other hand, such as in western and southern Normandy, the / dichotomy and participial agreement are found as in standard French, but, in contrast to the standard, the perfect does not, in general, replace the preterite. Following Lepelley (1974), Smith notes that speakers from the Norman dialects of Val de Saire will use the preterite in their own Norman variety (36a,b) but will switch to the perfect for preterital contexts when speaking French. In the Norman varieties of the Channel Islands, likewise, the preterite is used for past tense (Table 8.5). j akati sté vak iló (36) a. kẽã when I bought that cow there ‘when I bought that cow there’ (cf. St.Fr. quand j’achetai cette vache-là) b. ól a vnœ͂ she has come../ ‘she has come’ (cf. St.Fr. elle est venue).
(Norman dialect of Val de Saire (Manche))
(Lepelley 1974: 123)
Table 8.5. Variability in western northern French (oïl) varieties (Smith 2016: 314f.) N V2 syntax kept longer A/S marking; subject clitics case system kept longer / perfect auxiliaries agreement kept perfect replaces preterite
S V2 syntax lost early O marking; subject pro-drop cases lost early perfect auxiliaries agreement lost preterite maintained
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
184
8.6.4 Comparison of northeastern and northwestern oïl varieties 8.6.4.1 Introduction Smith’s (2016: 315) careful categorization of dialectal features thus illustrates how closely tied this variability is to geographical locale. He notes, for example, that the northern oïl varieties show a split distribution with regard to preterite use: in the western varieties, the preterite is used productively; in the central area, its use has disappeared; in the east, the preterite persists, but is only found in isolated areas, and only with particular verbs and / or persons. What this categorization also reveals is that the introduction of these features— / auxiliation and a preterital value for the perfects—may well have occurred at different times and have emanated from different sources. An examination of evidence from Lorrain, Wallon, and Norman will allow us to assess Smith’s claims, and consider the implications for the North–South continuum. Examples are provided of the three key perfect features—auxiliary use, agreement patterns, and the perfect > preterite shift—across the northern oïl varieties. 8.6.4.2 Exclusive use of auxiliary, especially in the northeastern oïl varieties In the Lorrain dialect of Ranrupt, is used as the auxiliary in all compound forms except with the verb ‘come’, which takes the auxiliary (37) (Aub-Büscher 1962: 84f.). falu k èl fœ̇ s venou (Lorrain) (37) il erã ̊ it have. require../ that she be. come.. ‘It would have required that she have come’ (cf. St.Fr. il aurait fallu qu’elle soit venue) In the Wallon dialect of La Gleize, is used for all verbs (36) (Remacle 1956: 40). à Courbevoie (Wallon) (38) nous avons arrivé we have arrived at Courbevoie ‘we arrived at Courbevoie’ (cf. St.Fr. nous sommes arrivés à Courbevoie) (example from Grognards 31) Remacle (1956: 41) notes that this exclusive usage appears also in the French of the area, not just in the Wallon dialect. Other -only varieties include the Ardennes dialect of Gespunsart (Descusses 1986: 126f.), and the Picard dialect of Vimeu (Somme) (Vasseur 1996: 52). Significantly, the popular language of Liège, a larger, industrial city in the eastern Wallon-speaking territory of Belgium, also uses exclusively in the formation of the perfects (39), and forms only a preterite (passé simple) for a large number of verbs (e.g. èsse ‘be’ aler ‘go’, avanci ‘advance’) (Stasse 2004: 577–637).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
185
8.6.4.3 Non-agreement of past participles In the Lorrain dialect of Ranrupt, the past participle is invariable when used with (39a). This is also the case with reflexive pronouns, which likewise take (39b) (Aub-Büscher 1962: 79). etrapè l ṑy è j l ḕ (39) a. j ḕ I have caught the goose and I it have ‘I have caught the goose and I have cooked it’ (cf. St.Fr. j’ai attrapé l’oie et je l’ai cuite) b. èl s è lā̀vē`
kæ̅ ́ (Lorrain) cook../
she has wash../ ‘she has washed herself ’ (cf. St.Fr. elle s’est lavée) Wallon perfects likewise do not show agreement (40a), and appear with reflexives (40b) (Remacle 1956: 40–4). jusqu’au 6 (Wallon) (40) a. nous avons resté we have stayed until 6 ‘we stayed until the 6th’ (cf. St.Fr. nous sommes restés jusqu’au 6) (example from Grognards 31) b. vos v’sav trompé you have mistake. ‘you have made a mistake’ (cf. St.Fr. vous vous êtes trompé) In the Lorrain dialect, however, when the past participle is used as an adjective, referring to the result of an action, the participle does show agreement (41a,b): à kǽt (41) a. èl she is cook.. ‘it is cooked’ (cf. St.Fr. elle est cuite) b. èl à lā̀vā́y she is wash.. ‘she is washed’ (cf. St.Fr. elle est lavée)
(Lorrain)
Notably, this same construction appears in Wallon, but does not show gender agreement, even when formed with (42) (Remacle 1956: 147f.).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
186
(42) lès foyes sont mouyis the leaves are soak.. ‘the leaves are soaking wet’ (cf. St.Fr. les feuilles sont mouillées)
(Wallon)
What this development may represent is the innovative recombination of + participle in the wake of the ongoing loss of the perfect, in what could be regarded as a skeuomorphic development (Smith 2011: 314), that is, a co-opting of old structures for new uses, based on different principles of organization. A similar example can be found in the popular French vernacular, described by Guiraud (1969): as many perfects in this vernacular have switched to , (e.g. baisser ‘lower’, changer ‘change’, divorcer ‘divorce’), a new + participle construction, which Remacle and others refer to as a ‘present perfect’, has grown up to refer to states in the present which result from past actions, to contrast with perfects referring to punctual events in the past (Guiraud 1969: 38f.). This new stative resultative construction capitalizes on the resultative value of the past participle, and resembles similar constructions in other varieties with similar profiles. Giraud notes that this innovation creates a new temporal-aspectual construct as others fall away. Bertinetto and Squartini (2016: 944) suggest that such developments, also to be found in Portuguese, Piedmontese, and in the Lombard varieties of Switzerland, may represent the grammaticalization of ‘actional’ distinctions (stative vs eventive). The similarity of the Wallon and Lorraine varieties, both in exclusive use of the auxiliary and in the lack of participial agreement, suggests that an areal explanation would best explain this distribution. Such an assumption is bolstered by the similar profile of perfects in the vernacular variety of Liège, Belgium: in the paradigms of verbs collected by Stasse (2004), all perfects use a auxiliary (43a,b). For a number of verbs, no passé composé is listed at all, but only the passé simple (43c). (43) a. dj’a intré I have enter. ‘I have entered’ (cf. St Fr je suis entré)
(Wallon Liègeois, Stasse 2004: 586)
b. nos-avans intré we have enter. ‘We have entered’ (cf. St.Fr. nous sommes entrés) c. dj’ ala I went (cf. unmarked St.Fr. je suis allé) Because Liège is the largest urban centre in the area, it appears likely that it is the centre of the innovative use of the auxiliary. The city is located in an area of
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
187
intense, long-lasting, three-way contact, between French, Dutch, and German (Nelde 1984), and it does not seem unlikely that this contact with non-Romance vernaculars could have played a role in the development of this structure. The suggestion that contact is responsible remains only conjecture, however, since I was not able to find other evidence of the exclusive use of the perfect in surrounding vernacular varieties. If Liège does turn out to be the source variety of this innovation, we cannot help but note the irony of this outcome, that the purported birthplace of Charlemagne himself was the source of a feature which does not conform to the Charlemagne Sprachbund.
8.6.4.4 Persistence of the preterite especially in the northwestern oïl varieties While the tendency to maintain the old contrast between perfect and preterite is to be found, if more sporadically, in northeastern oïl varieties (43a,b, 44), it remains much more productive in the northwestern oïl varieties (45). pwa lo jå̃ (43) a. jè lo sṑ I it learned from the Jean ‘I learned it from Jean’ (cf. St.Fr. je’l’ai appris par Jean) b. l òm kè vnœ́ ermẽ the man who came yesterday ‘the man who came yesterday’ (cf. St. Fr. l’homme qui est venu hier)
(Lorrain dialect of Ranrupt)
(Aub-Büscher 1962: 81f.)
(44) çu fout dîmègne à l’ nut’, dju m’ aveû mètou so l’ gazète it was Sunday at the night, I had set on the newspaper come çou-volà, èt tot d’on côp, dju sinta k’ dj’ like this, and all of a sudden, I felt that I èsteû pris (Wallon dialect of La Gleize) was taken ‘it was Sunday evening, I was sitting with the newspaper like this, and all of a sudden, I felt that I was taken’ (cf. St.Fr. ce fut dimanche soir, je m’étais mise sur le journal comme ceci, et tout à coup, je sentis que j’étais prise) (Remacle 1956: 57) (45)
(= 36a) kẽã j akati sté vak iló (Norman dialect of Val de Saire) when I bought that cow there ‘when I bought that cow there’ (cf. St.Fr. quand j’achetai cette vache-là) (Lepelley 1974: 125)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
188
Just as noted with regard to Occitan and Catalan, the sociohistorical context of the distribution of perfect features is essential to understanding the outcome. Taking all of these facts about the oïl dialects of northern Gallo-Romance into consideration, we can, with Smith (2016), conclude that these varieties do not conform to all the features predicted for ‘northern’ Romance varieties. Rather, these dialects share tendencies according to geographical location, with the easternmost dialects of the northeastern region showing the special development of exclusive perfects and lack of participial agreement, while the northwestern varieties continue to use the preterite much more productively than other vernaculars or than standard French. The northeastern development appears to represent innovative loss of the perfect; the northwestern situation may simply be due to conservative tendencies. This would make its non-participation in the perfect > preterite innovation, Stage III above, similar to the non-participation, or only recent, sporadic participation, of Occitan and Catalan in this innovation. Once again, we note that earlier established isoglosses will not necessarily be replicated as new centres of influence and the innovations that issue from them are successively established.
8.7 Conclusions This brief examination of the distribution of the periphrastic perfects in Europe and in several Gallo-Romance varieties has provided evidence in support of the views of Zamboni (2000) and Ledgeway (2012) concerning the existence of a North–South Continuum in the Romance languages. However, the evidence presented also points to the need for several modifications to the theory. 1. It is claimed here that the creation of a dichotomous category, the amalgamation of the + participle and + participle constructions into a single perfect category, occurred not in late Latin, when all the Romance languages are assumed to have had access to the innovation, but rather much later, in Carolingian Latin. The ‘northern’ innovation is thus conceived of as a Carolingian innovation: the Romance and Germanic varieties within the Carolingian realm (the ‘northern’ varieties) underwent this expansion of the perfects, but those outside the influence of the Carolingian empire (the ‘southern’ varieties) did not. Rather than adopting this innovation, the southern languages would have followed trajectories already set in motion, retaining and developing the inherited perfect. 2. The establishment of split-intransitivity through the creation of this fully articulated contrast in the auxiliaries thus corresponds well with the insights of La Fauci (1988), Zamboni (2000), and Ledgeway (2012), but positions the innovation in another time and place. The lateness of the construction of the
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
3.
4.
5.
6.
189
perfect category per se coincides in some ways with the claims of Adams (2013), who views the perfects as ungrammaticalized throughout most of the history of Latin. It also excludes the ‘southern’ varieties from the innovation. Ledgeway’s claim that the predominance of perfects in the ‘southern’ varieties represents a resurgence of nominative / accusative alignment in all of these varieties does not sufficiently account for the similarity of the solution that many Romance varieties would have had to devise in establishing the use of a auxiliary alone. While some evidence exists for the adoption of certain active / stative tendencies in the ‘southern’ varieties, predominates, signaling archaism. The most important piece of evidence for the correctness of this claim is the distribution of languages with / perfects on the map of present-day Europe: those which lay within the Carolingian realm are precisely those which keep the / contrast. Central to this proposal is a recognition of the role of areal influence in the innovative invigoration of the dyadic / construction in a limited geographical territory, coinciding with the Carolingian realm. This mapping is conceived of as three-dimensional, with one layer of innovation placed upon another, thus constructing a ‘Stratified Convergence Zone’ (Drinka 2017). The establishment of the / contrast did not take place at the same time or emanate from the same centre of influence as the perfect > preterite shift. Evidence from two regions of Gallo-Romance was provided in support of these claims. Old Occitan and old Catalan were both part of the Carolingian realm, and can thus be seen as ‘northern’ varieties; they did not, however, undergo the Paris-initiated perfect > preterite shift mentioned above, a fact which reinforces the separateness of the operation of these two innovative strata. Catalan then moved from being squarely positioned among the ‘northern’ languages to being, at least to some extent, a ‘southern’ variety, as a result of its realignment with Castilian. It provides strong evidence for the role of sociohistorical pressures as a key motivation for such changes, and underlines, once again, the importance of areal factors in the development of this distribution. With regard to the oïl dialects of northern Gallo-Romance, evidence has been presented which illustrates the incongruities detected by Smith (2016): while standard French displays all of the ‘northern’ traits listed, the vernacular oïl varieties do not. Instead, these dialects share tendencies according to geographical location, with the easternmost dialects of the northeastern region illustrating exclusive auxiliation and a lack of participial agreement, while the northwestern varieties continue to use the preterite more productively than other vernaculars. It is suggested that the northeastern development may represent innovative loss of the perfect, while the
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
190
northwestern distribution may simply reflect conservative tendencies in maintaining the anterior value of the perfect, as seen in Occitan and Catalan. 7. In sum, in this attempt to provide some explanation for why the ‘North– South Continuum’ appeared, it is suggested that sociohistorical pressures were at work, in conjunction with internal processes, on several levels in the Gallo-Romance languages, and in Romance in general: the prestige of Latin, acting as a ‘roof ’ language, fostered innovations connected with the formal register, such as the use of deponents and their periphrastic equivalents, the perfects. These perfects clearly made their way into the vernaculars, as witnessed by the distribution of / perfects on the present-day map of Europe. Sociohistorical motivations have continued to operate within these varieties, as political power has been shuffled and new allegiances have been constructed. The Gallo-Romance varieties thus provide an invaluable opportunity to explore speakers’ ‘responses to macrohistorical processes’ (Gal 1989: 357), and, conversely, the macro-effect of micro-variations.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
9 Active-middle alignment and the aoristic drift The North–South divide in the Romània on evidence from northern Gallo-Romance Delia Bentley
9.1 Introduction A striking typological difference between Latin and the Romance languages is found in the respective roles of morphology and syntax in the expression of syntactic functions and grammatical categories.¹ Whereas Latin heavily relied on morphology to express these, the grammar of the Romance languages is innovative in developing new syntactic structure. One need only think of the impoverishment of the case inflexion (which played a key role in the marking of the subject and other syntactic functions in Latin), and the rise of a wide range of periphrastic structures in the nominal and verbal systems. Traditional scholarship viewed these profound changes in terms of successive, and in part cyclic, alternations of synthesis and analysis in grammar, the Romance languages being more analytic than their mother, Latin (e.g. Harris 1978; Posner 1996; Schlegel 1818; Vincent 1982). In light of modern syntactic theory, the same changes have instead been analysed as aspects of the development of configurational and functional syntax (Ledgeway 2012, 2016a; Lyons 1999; Vincent 1993, 1997), with a resultant increase in head marking, in the sense of Nichols (1986). Among the grammatical categories which came to be expressed analytically in Romance we find the perfect, which is an aspectual (experiential, resultative, or perfective) category. The perfect periphrasis originated from two sources (Vincent 1982), one being a structure with ‘have’ taking as its object a noun phrase modified by an adjectival, resultative, participle (pecunias magnas collocatas habent ‘they have large sums invested’ Cicero Pro Lege Manilia 18). The other source was the perfect of passive and deponent verbs, which was analytic in ¹ A JC dedico questo lavoro, nel ricordo dei tempi di Manchester, con rinnovata stima e simpatia. Thanks are due to Martin Maiden, Nigel Vincent, and Sam Wolfe for very helpful comments on a first draft of this article. Variation and Change in Gallo-Romance Grammar. First edition. Sam Wolfe and Martin Maiden (eds). This chapter © Delia Bentley 2020. First published in 2020 by Oxford University Press.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
192
Latin and exhibited the auxiliary ‘be’ plus a past participle (mortuus est ‘he (has) died’). The vicissitudes of the perfect periphrasis constitute an extremely important chapter in the history of Romance and have also figured prominently in syntactic and typological literature since Perlmutter (1978). The debate has focused, on the one hand, on a cluster of interrelated patterns and their diachronic developments, including the alternation of the reflexes of ‘have’ and ‘be’ as the auxiliaries figuring in the periphrasis and the inflexion carried by the participle (e.g. Burzio 1986; Kayne 1989a, 1993; La Fauci 1988, 1989; Loporcaro 1998; Perlmutter 1978, 1989; Smith 1991b, 1995b). On the other hand, scholars have paid attention to the enrichment and expansion of the functional load of the perfect periphrasis in many Romance languages, with the concomitant decline of the synthetic past tense, which was the direct continuation of the Latin synthetic preterite. This diachronic process, known as aoristic drift, had been previously noted by Meillet (1909/1982: 149–58) for German, Armenian, and a number of Slavonic and Indo-Iranian languages. For Romance, I refer to Squartini and Bertinetto (2000) and Bertinetto and Squartini (2016) (see also Comeau, King, and Butler 2012; Fournier 1998; Smith 2016 for Gallo-Romance). The observation that a number of the languages that have undergone the aoristic drift have also maintained the ~ alternation (henceforth ~ alternation) and, to a lesser extent, the agreement of the participle with the subject of the perfect formed with played an important role in the development of a classification of the languages of the Romània along a northernsouthern continuum (Ledgeway 2012: 314; Zamboni 1998: 129–31, 2000, drawing on La Fauci 1988). In accordance with this classification, Gallo-Romance, RaetoRomance, northern Italo-Romance (Gallo-Italian, Venetan) and Tuscan belong to the North, whereas Ibero-Romance and southern Italo-Romance belong to the South.² As well as having undergone the aoristic drift, and retained the ~ alternation, the majority of the northern languages are characterized by the presence of subject clitics. The southern subfamilies of Romance, instead, never developed subject clitics, nor have they undergone the aoristic drift. In fact, some southern Romance varieties even provide evidence of developments in the opposite direction (see Fløgstad 2016 on Porteño Spanish). Many southern languages have also generalized as the sole perfect auxiliary and, to various extents, lost morphological agreement marking on the participle in the perfect. Incidentally, was replaced by a reflex of ‘hold’ in some such languages, for example in European Portuguese, where the synthetic preterite is the main perfective past tense form.
² Sardinian and Daco-Romance were originally included among the languages of the South due to properties that are not central to the present discussion (see note 4).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
193
Smith (2016: 317f.) notes that northern Gallo-Romance challenges the areal classification based on the North–South divide, in that the northwestern varieties have not undergone the aoristic drift, while, at the same time, retaining the ~ alternation and the agreement marking on the perfect participle. By contrast, some northeastern varieties have generalized the perfect auxiliary and lost the agreement marking on the perfect participle, thus patterning with southern Romance in these respects, while also exhibiting mere vestiges of the synthetic preterite, as a result of the aoristic drift. The evidence of northern Gallo-Romance, therefore, ‘invites further research into the nature of the correlation between the various characters of this typology’ (Smith 2016: 318). In this chapter I address the question of whether the northern Gallo-Romance evidence challenges the North–South classification or the typology upon which this classification is based. The discussion focuses on the verb system and on its relation to the morphosyntax of the clause. I note that the said typology conflates the expression of temporal–aspectual notions, namely the aorist past tense and the perfect, with the marking of major syntactic functions, the function of subject being the one that is relevant to the history of Gallo-Romance. While being in principle orthogonal to each other, both types of marking converge on a single functional projection, in the transition from Latin to Romance. I shall refer to this projection as the inflexional phrase (see the notion of IP, or TP, in syntactic theory). Importantly, the ~ alternation on the inflexional phrase marks the subject differentially in accordance with active-middle alignment (in the sense of La Fauci 1988).³ The subject clitics, instead, mark the subject in accordance with the more conservative nominative-accusative alignment type. Northeastern Gallo-Romance differs from other northern varieties in having drastically reduced the active-middle marking of the subject, thus generalizing nominative alignment. I argue that the developments that occurred in the syntax of the Gallo-Romance subject, which combines facets of head and dependent marking (Vincent 1993, 1997), with a preponderance of nominative marking on the dependent (the subject clitic), provided the structural conditions for the weakening of active-middle marking on the inflexional head in the northeastern varieties and, less conspicuously, throughout Gallo-Romance. As for northwestern Gallo-Romance, it would seem solely to differ from other northern varieties in having maintained the preterite, thus proving not to rely on functional structure to the same extent as other northern varieties for the expression of the past tense. However, we also uncover evidence suggesting that the ~ alternation is no longer particularly strong in this subfamily of Gallo-Romance.
³ The agreement of the participle with the gender and number features of the subject of the perfect formed with can also be conceived of as a differential marking strategy, albeit of a slightly different kind, as will be pointed out later.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
194
After discussing northern Gallo-Romance, I briefly consider the southern Romània. I argue that, in this area, the return to nominative-accusative alignment was not triggered by a change in the syntax of the subject. Rather, this return is to be characterized as a broader and earlier drift, which, as noted by Zamboni (1998), also involved the differential marking of the object. I conclude that, rather than being characterized by a North–South divide, the modern Romània exhibits a stronghold of active-middle alignment in a central area (Occitan, northern Italo-Romance, Tuscan, Corsican, Sardinian, and RaetoRomance), where the inflexional head, including the auxiliary of the perfect, licenses the subject, whether alone or in combination with a subject clitic.
9.2 The North–South divide in the Romània The areal classification of the Romània along a northern-southern continuum (Zamboni 1998, 2000; Ledgeway 2012, 2016a) relies on a number of structural properties, some being distinctly innovative, by comparison with Latin, others being conservative. In Table 9.1 I provide a list of those among such properties which are relevant at this point in our discussion. They are arranged in accordance with their areal distribution.⁴ The development of subject clitics is interesting because it raises the question of whether northern Romance is innovative. The Romance subject clitics derived from weakened nominative pronouns in the context of frequent fronting of a constituent triggered by the V2 constraint of medieval Romance (Harris 1978: 112f.; Poletto 1995; Pope 1952; Vanelli, Renzi, and Benincà 1985; Wolfe 2018). This change occurred earlier in Gallo-Romance than in Gallo-Italian and had different outcomes in different parts of the northern Romània (Poletto and Table 9.1. The North–South divide in the Romània and its typology northern Romance (Gallo-Romance; RaetoRomance; north. Italo-Romance) subject clitics ~ alternation inflexion on perfect participle aoristic drift
southern Romance (Ibero-Romance; south. Italo-Romance) no subject clitics no ~ alternation no inflexion on perfect participle no aoristic drift
⁴ Another property that is relevant to this classification is the differential marking of the object (e.g. Bossong 1998; Iemmolo 2010). In accordance with this property, Sardinian and Daco-Romance are traditionally claimed to be part of the South, even though Sardinian lost the preterite at a very early stage of its history and it has maintained ~ alternation (Jones 1993; Mensching and Remberger 2016). As for Daco-Romance, standard Romanian, most Daco-Romanian dialects, with IstroRomanian have lost the preterite. Furthermore, although the ~ alternation is found in Romanian, the two auxiliaries are not distributed in accordance with active-middle alignment.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
195
Tortora 2016; Wanner 1987: 16).⁵ Abstracting away from in fact substantial cross-dialectal variation (see, e.g. Cardinaletti and Repetti 2010; De Cia 2019), the subject clitics of the dialects of northern Italy can be considered to be ‘erstwhile dependents that have migrated towards the inflexional head of the clause’ (Ledgeway 2016a: 770; see also Brandi and Cordin 1981; Poletto and Tortora 2016; Rizzi 1986a). These clitics behave like extended inflexion in licensing null subjects. Despite evidence suggesting that the subject clitics of some Gallo-Romance varieties may now be following suit (Culbertson 2010; Roberts 2010; Villeneuve and Auger 2013), these subject clitics historically exhibit a nonnegligible degree of independence from the inflexional head. This is still manifested in the syntax of French and other oïl varieties nowadays. Thus, these subject clitics are not repeated in VP coordination (Rizzi 1986a), unlike the personal subject clitics of the northern Italo-Romance dialects (cf. 1a vs 1b and see also the authentic example in 8 below). They precede the preverbal negation (cf. 2a) and they can be conjoined, albeit under very limited circumstances (cf. 2b). (1)
a. Tu manges des pommes et (**tu) bois du vin. (French) you eat.2 potatoes and you drink.2 wine b. Ti magni patate e **(ti) bevi vin. (Venetian) eat.2 potatoes and drink.2 wine ‘You eat potatoes and drink wine.’ (Poletto and Tortora 2016: 775 for 1b)
(2)
a. Il ne viendra he come..3 ‘He will not come.’
pas.
b. Ili ou ellei viendra. he or she come..3 ‘He or she will come.’
(French)
(French) (Smith 2016: 311 for 2b)
In addition, the Gallo-Romance subject clitics have reference (cf. 3a), and, indeed, they cannot co-occur with a non-dislocated subject (cf. 3b), thus contrasting with the subject clitics of many northern Italo-Romance and Tuscan dialects (cf. 3c). (3)
a. Il viendra demain—‘il’, c’est à dire, he come..3 tomorrow he i.e. ‘He will come tomorrow—‘he’, i.e. Pierre’
Pierre. (French) Pierre (Smith 2016: 311)
⁵ Harris (1978: 113), citing Price (1971: 147), notes that by the thirteenth century two-thirds of the relevant contexts exhibit subject pronoun insertion in French. In Gallo-Italian the subject clitics appear later and cannot be said to have developed until the fifteenth–sixteenth century.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
196
b. Personne (**il) ne no one he ‘No one will come.’ c. Nissun gl a no one have.3 ‘No one said anything.’
viendra. come..3 detto said
(French)
nulla. (Florentine) anything (Brandi and Cordin 1981 for 3c)
As a matter of fact, these clitics have been analysed as full phrases (Kayne 1975) or weak pronouns (Cardinaletti and Starke 1999). For a discussion of the strengths and weaknesses of the above analyses, I refer to Poletto and Tortora (2016) and Smith (2016). The mutual exclusiveness and the interchangeability of the Gallo-Romance subject clitics with noun phrase subjects, as well as their relative independence from the verb, challenges an analysis in terms of head marking.⁶ Crucially, the Gallo-Romance subject clitics occur obligatorily in every grammatical person, whereas the subject clitic paradigms of the northern Italian dialects are often defective (see Renzi and Vanelli 1983 for details). Importantly, the requirement of an expletive subject clitic in constructions that lack a referential or preverbal subject (cf. 4a–b) is a facet of positional licensing, in the sense of Kiparsky (1997). Since the subject clitics are inflected for nominative case, the subject is licensed both inflexionally and positionally on the dependent. (4)
a. Il pleut. . rain.3 ‘It is raining.’ b. Il est arrivé . be.3 come.. ‘There arrived somebody.’
(French)
quelqu’un. somebody
(French)
Whereas the subject clitics were a Romance innovation, in Gallo-Romance they engendered a return to conservative nominative marking on the dependent. In the final analysis, the development of subject clitics was an innovative property of the increasingly configurational syntax of the northern Romance languages, which, at least in Gallo-Romance, signalled the emergence of a subject position that needed to be overtly filled. This requirement is still valid throughout the oïl varieties nowadays (Smith 2016). The syntax of subject clitics will be relevant to our account of the northern Gallo-Romance facts noted by Smith (2016) (see §9.4).
⁶ See Vincent (1993: 153) for evidence from other constructions (for example, negation), indicating that French is ‘basically D[ependent]M[arking]’. Such constructions are also present in the other GalloRomance varieties.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
197
The alternation of two perfect auxiliaries is an innovative strategy of argument marking, which differentiates two subclasses of subject. (5)
a. Elle est morte. she be.3 die... ‘She has died.’
(French)
b. Elle a she have.3 ‘She has eaten.’
(French)
mangé. eat..
In the syntactic literature, the subject classes exemplified in (5a–b) are normally characterized as subjects that are underlying objects (cf. 5a), and subjects that are not (cf. 5b). This analysis originates from Perlmutter’s Unaccusative Hypothesis (Perlmutter 1978), which relies on the co-patterning of the subclass of intransitive subjects exemplified in (5a) with transitive objects. Alongside the ~ alternation, other diagnostics of unaccusativity in Romance are the agreement features carried by the perfect participle (see below) and the cliticization with an outcome of Latin (e.g. Perlmutter 1989; see Loporcaro 2016 for relevant discussion), which is also constrained to the transitive object (cf. 6a) and a subclass of intransitive subjects, namely the selecting ones (cf. 6b vs 6c). (6)
a. Elle en a acheté she have.3 buy.. ‘She has bought one (of them).’ b. Il en est . be.3 ‘One (of them) died.’
mort die..
un. one un. one
(French)
(French)
c. **Il en a mangé un. (French) . have.3 eat.. one ‘One (of them) ate.’ [ungrammatical in the intended reading] The ~ alternation and the other diagnostics of unaccusativity have also received syntactico-semantic analyses, whereby they are captured in terms of lexicosemantic or macro-thematic properties shared by the unaccusative subject and the direct object (e.g. Bentley 2006; Bentley and Eythórsson 2003; Cennamo 2001a; Centineo 1986; Sorace 2000; Van Valin 1990). The key issue for the purposes of our discussion is that the alternation illustrated in (5a–b) differentiates two subclasses of the same syntactic function. These are sometimes called SA and SP/O. SA subsumes the subject of transitives and of a subclass of intransitives, specifically, those that take . Contrastingly, SP/O designates the remaining subclass of intransitive subjects. This alternation thus
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
198
departs from the generalized nominative marking of the subject in Latin (though see Zamboni 1998: 127–9 and, for late Latin, Cennamo 2001b, 2009), as well as from the undifferentiated licensing of the subject with subject clitics. La Fauci (1988: 34) refers to the subclass of subjects which are flagged with (SP/O) as middle subjects and to the alignment that emerges with the ~ alternation as active-middle alignment. The alternation of the perfect auxiliaries was once attested on a much larger scale than nowadays (see, e.g. Benzing 1931; Aranovich 2003; Rosemeyer 2013 for Spanish; Buridant 2010: 372f. for French; Mateu Fontanals for Catalan 2014). The synchronic investigation of Gallo-Romance, both within and outside the Romània, is a cogent testimony of this. Standard French only selects with twenty to thirty verbs (Giancarli 2011: 373f., cited in Loporcaro 2016: 812), while other varieties of French exhibit even smaller sets of -selecting verbs, which are ordered subsets of the standard French set (Loporcaro 2016, and references therein, for Acadian and New Brunswick French; Sankoff and Thibault 1977 for Canadian French; Smith 2016: 315; see also §9.3). In the diachrony and the synchronic variation of Gallo-Romance, we thus find evidence of the rise of an innovative active-middle marking pattern, which later gives way to a gradual return to the more conservative undifferentiated nominative pattern. As for the agreement marking on the perfect participle (Loporcaro 1998, 2016; Smith 1991b, 1995b), originally, this carried the gender and number features of the direct object of transitives and of the same subclass of intransitive subjects as is flagged by the selection of in the perfect. As such, the participial agreement is one of the principal diagnostics of unaccusativity in Romance, although a key difference must be noted between this pattern and the ~ alternation. In agreeing with both transitive objects and selecting subjects, the participle of the perfect is not marked in accordance with middle alignment, but rather in terms of activeinactive alignment, in the sense of Harris and Campbell (1994: 240f.) (see also Bentley 2006; Van Valin and LaPolla 1997). Over time the agreement of the participle of the perfect has also drastically receded across Romance. However, it is still attested to various extents in languages of the North, as well as in central Italo-Romance dialects (e.g. Anderson 2016: 177 for Romansh; Benincà and Vanelli 2016: 152 for Friulian; Loporcaro 1998 for Italo-Romance; Salvi 2016: 162 for Ladin). To return to the properties that characterize the North of the Romània (see Table 9.1), the expansion of the domains of the perfect, and the concomitant decline of the synthetic preterite (Anderson 2016: 175; Benincà, Parry, and Pescarini 2016: 193, 203; Salvi 2016: 160), is indicative of a tendency towards the expression of grammatical categories, in this case the temporal–aspectual dimensions relevant to the past, on functional syntax. To give but one example, RossiGensane (2016: 95) notes that the perfect can describe past completed events in contemporary spoken French, thus replacing the preterite as a definite past.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
- (8)
199
Le père a essayé de le rattraper et n’ a the father has tried of him catch. and have.3 pas pu. (spoken French) can.. ‘The father tried to catch him, but wasn’t able to.’ (Rossi-Gensane 2016: 95)
A point of clarification is in order. The preterite has not completely disappeared from the northern Gallo-Romance languages, but rather it is found in certain genres, such as fairy tales (Rossi-Gensane 2016: 95), or narratives, with the propulsive function of indicating a further stage in a succession of events (Bertinetto 2003; Bertinetto and Squartini 2016: 942). Its distribution in some oïl varieties of Gallo-Romance is sensitive to grammatical person. Thus, Alex (1965: 115) notes that, in the variety of Naisey (Besançon, Franche-Comté), the preterite normally occurs in the first and third person of ‘certains verbes indiquant une action qui peut être localisée avec précision à un moment donné, par example arriver’ (‘some verbs, which describe an action that can be localized at a precise moment in time, for example arrive’). (9)
Kã: y érivé: sù lè fwèr, ì rãkõtré: ju:zè k when I arrive. on the fair - meet. Joseph èvá: j vãdù sá: bú: (Naisey, Franc-Comtois) had already sold his oxen ‘When I arrived at the fair, I met Joseph, who had already sold his oxen.’ (Alex 1965: 115)
In sum, the North–South divide has its roots in two innovative tendencies, which set Romance apart from Latin: on the one hand, the encoding of syntactic functions and grammatical categories on a newly developed functional projection; on the other hand, a drift towards the marking of syntactic functions in accordance with active-middle and active-inactive alignment. Although the development of configurational syntax engendered a number of head-marking strategies throughout the Romània (Ledgeway 2012, 2016a; Vincent 1997), the oïl varieties of GalloRomance developed new dependent-marking syntax in the licensing of the subject.
9.3 The case of northern Gallo-Romance Smith (2016: 317f.) notes that although the northern oïl varieties exhibit subject clitics, alongside the prolonged retention of the nominal case system and of a V2 syntax, which are historical properties of northern Romance, they do not readily fit in the northern-southern classification with respect to perfect auxiliary
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
200
selection, past participle agreement and the survival or disappearance of the preterite. Indeed, most eastern oïl varieties pattern with the southern Romance languages in having generalized the perfect auxiliary and lost participial agreement with the subject in perfective periphrases. However, they differ from southern Romance in having extended the perfect periphrasis to the domains of the preterite, the latter being a mere relic feature [sic]. These properties are illustrated here with evidence from Picard of Valenciennes (10), Wallon (11), and spoken French from the Nord-Pas-De-Calais area (12). For Alsatian, I refer to Aub-Büscher (1962: 78–80), cited in Loporcaro (2016: 812), for Lorraine, see Esch (2002). Note in passing that the co-occurrence of the perfect with the adverb îr ‘yesterday’ in (11b) supports Smith’s view on the obsolescence of the preterite in Wallon. I will briefly return to this point in §9.4. (10) J’ai quéu / j’ai arvénu à pied. (Picard) I have.1 fall.. I have.1 return.. at foot ‘I have fallen / I came back on foot.’ (Dauby 1979: 35) (11) a. Dj’a v’ni. I have.1 come.. ‘I have come.’ b. Il a mouri he have.3 die.. ‘He died yesterday.’
(Wallon)
îr. yesterday
(Wallon) (Remacle 1956: 40)
(12) a. Il a venu voir mon frère. (spoken French) he have.3 come.. see. my brother ‘He has come to see my brother.’ b. Elle avait she have..3 ‘She had falled’
tombé. fall..
(spoken French) (Dawson et al. 2016: 152)
Importantly, is even reported to occur with reflexives in the variety of Naisey, Besançon (13a–b), and in Wallon (Remacle (1956: 44) (14a–b), reflexives being a stronghold of selection in French.⁷
⁷ The results of a survey of very limited scope in the area of Ittre / Oisquercq (Walloon Brabant, Belgium), instead, suggested that the ~ alternation is still attested, although the preterite has indeed been replaced by the perfect periphrasis. While I do not illustrate these results here, as they are hardly representative, I note that there may still be some variation in the relevant grammatical domains in some northeastern areas. Thanks are due to Yann De Lombaert for conducting the survey in Ittre / Oisquercq.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
201
(13) a. Ì sè ku:pá: lu dè. (Naisey, Franc-Comtois) he .have.3 cut.. the finger ‘He has cut his finger.’ b. Ì sè he .have.3 ‘He left.’ (14) a. I s’ a he have.3 ‘He burned himself.’
ã:n
àlá: go. . (Alex 1965: 118)
broûlé. burn..
b. Vos v’s -av trompé. you have.2 mistake.. ‘You have made a mistake.’
(Wallon)
(Wallon) (Remacle 1956: 44)
As is the case with Romance languages of the North and the South alike, can join with a participle to form a resultative periphrasis, whereby the participle is a result-state adjective. This is illustrated in the Wallon example in (15a), where è ‘is’ is a copula and mwêrt ‘dead’ is an adjectival participle. This example contrasts with (15b), which illustrates the perfect with auxiliary (see a ‘have’) and the perfect participle mouri ‘died’. (15) a. Il è mwêrt. he be.3 die.. ‘He is dead.’ b. Il a he have.3 ‘He has died.’
mouri. die..
(Wallon)
(Wallon) (Remacle 1956: 42)
The pattern shown in (15a) is irrelevant to the issue of the generalization of , since the latter generalizes as an auxiliary of the perfect, and not as a copula (Abeillé and Godard 2000). And yet, where is reported solely to occur with one verb (e.g. aller ‘go’ in the Picard of Valenciennes, see Dauby (1979: 35), and mourir ‘die’ in the Picard of Nibas, see Vasseur (1996: 52)), the question has been raised whether is a copula in the relevant structures (Loporcaro 2016: 812 and references therein). The hypothesis that is a copula, when it occurs with a single participle, requires proper testing against the evidence and ultimately poses more questions than it answers, in that there is no obvious principled reason why the resultative construction, unlike the perfect, should be limited to a single result-state adjectival participle. In fact, the pattern in (15a–b) is very productive in many Romance languages, including those of the South (Bentley and Ledgeway 2014; Bentley 2018a).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
202
(16) a. L’ ascensori jè the lift. be.3 ‘The lift is broken.’ b. L’ascensori un s’ the lift ‘The lift did not break.’
ruttu. break... a have.3
(Sicilian)
rumputu. break..
(Sicilian)
Turning now to the western oïl varieties, the preterite has not receded here, the perfect auxiliaries are claimed to alternate, and the agreement marking on the perfect participle is said to be robust (Smith 2016: 318). Evidence of the expression of the past, including the recent past, with the preterite in Gallo is readily found online.⁸ I illustrate the ~ alternation in (17a–b), which is an excerpt of an online document designed to help candidates prepare for the Gallo baccalauréat examination.⁹ (17) a. Était-y le chien [ . . . ] qu’ avait was it the dog have..3 les galettes? the biscuits ‘Was it the dog that had eaten the biscuits?’
mangé eat.. (Gallo)
b. Le lecteur est [ . . . ] ben en peine de savèr tchis qui (Gallo) the reader is well in pain to know what that s’ est passé. be.3 happened ‘The reader is at a loss to know what has happened.’ Comparable evidence is found in Guernsey and Jersey Norman French. Liddicoat (1994: 203) reports the alternation of ave ‘have’ and e:tr ‘be’ both in Jèrriais and in Sercquiais, e:tr being selected with a dozen verbs of change of state and directed change of location (18a–b). Reflexives also select e:tr in both varieties (18c). The participle of the perfect agrees in gender and number with the subject of ‘be’ selecting verbs (18a–c). (18) a. Il eit arrivε. he be.3 arrive... ‘He has arrived.’
(Jèrriais)
⁸ See Le Gallo, la langue romane de Bretagne, http://www.cndp.fr/crdprennes/crdp/crdp_dossiers/ dossiers/gallo/methode/12.htm (accessed 10 March 2019). Thanks are due to Prudence du Breil de Pontbriand for pointing this out to me. ⁹ Le Gallo à Fulgence Bienvenüe, http://welcome2bienvenue.free.fr/LV_LYCEE/gallo%20LY/ gallo_accueil.html (accessed 10 March 2019).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
- b. Ul ε:t mɔrt. she be.3 die... ‘She has died.’10
203
(Sercquiais)
c. I ly: sõ mẽ:ð a trava:ji a õ:ð œð (Jèrriais) they be.3 put.. at work. at eleven hours ‘They started work at 11.’ (Liddicoat 1994: 203) The perfect is not only a past tense (18c), but also an aspectual form with the function which is usually called existential or experiential (cf. 19a), in Jèrriais and in Sercquiais, and these varieties also have a healthy preterite (cf. 19b). (19) a. ʒ ei treʒu fei l pɛ̃ sje mei. (Jèrriais) I have.1 always make.. the bread at me ‘I have always made my bread at home.’ (Liddicoat 1994: 204) b. ʒ l dmãdɪ kjεsk u l a dit [ . . . ]. (Sercquiais) I him ask..1 what she him has tell.. ‘I asked him what she has told him.’ (Liddicoat 1994: 182) The Norman variety of the Val de Saire is also reported to have maintained the preterite. In particular, Lepelley (1974: 125) remarks: ‘Les formes du passé simple sont beaucoup plus vivantes dans le parler du VdS qu’elles ne le sont en français. Un locuteur patoisant les emploiera naturellement pour exprimer le ponctuel du passé, alors que, parlant français, il utilisera les formes correspondantes du passé composé’ (‘The forms of the simple past are much more alive in the speech of VdS than they are in French. A speaker of the patois will use them to express the punctual past tense, whereas, speaking French, he will use the corresponding forms of the compound past’). An example is provided here. (20) Ke˜ã j akati sté vak when I buy. cow ‘When I bought that cow . . . ’
iló . . .
(Norman French) (Lepelley 1971: 125)
With respect to the perfect, however, Lepelley states that it is generally formed with the auxiliary avó ‘have’ plus the past participle.
¹⁰ Liddicoat’s (1994: 203) translation of this example is ‘she is dead’, which would suggest a copular analysis (see above). The point that is relevant to the present discussion is that Liddicoat (1994) explicitly states that the perfect exhibits the auxiliary e:tr ‘be’ in alternation with ave ‘have’, in both Sercquiais and Jèrriais.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
204
(21) a. J i acató. I have.1 buy.. ‘I have bought.’ b. Il éra tœ̃ bó. he have..3 fall.. ‘He will have fallen.’
(Norman French)
(Norman French) (Lepelley 1971: 127)
This suggests that even in some northwestern Gallo-Romance varieties the process of generalization of is well under way, contrary to expectations.
9.4 Making sense of northern Gallo-Romance Let us now reflect on the three patterns which have been claimed to set northern Gallo-Romance apart from the rest of the northern Romània: the preterite, the ~ alternation, and the agreement of the perfect participle with a subset of subjects, specifically SP/O ones. These three structures are only interrelated insofar as the ~ alternation and agreement marking on the perfect participle are exhibited by the periphrastic perfect, which, crucially, can take over the past tense function of the preterite. Functionally, the preterite is not related in a principled way with the ~ alternation or the agreement marking on the perfect participle. Indeed, the preterite expresses a grammatical category, whereas the other two patterns are strategies for the differential marking of the syntactic function subject. If, as I assume, there is no principled correlation between the differential subject marking strategies and the maintenance or loss of the preterite, the case of northwestern Gallo-Romance hardly requires an explanation. All there is to note is that the aoristic drift is not as advanced in these varieties as in other varieties of the North (spoken French (see 8), Romansh, Ladin, Gallo-Italian, northern Italian, etc.), which in turn means that northwestern Gallo-Romance does not rely on functional syntax to the same extent as the other northern languages in the expression of the past tense. It is worth noting that a similar situation is not only found in the higher and written registers of French and Italian, but also in modern Tuscan, central and upper-southern Italo-Romance, and Corsican (Ledgeway 2016b: 224), these varieties having maintained the ~ alternation, as well. Although in some northeastern Gallo-Romance varieties the preterite still exists (see example 9 and Alex 1965: 115, as well as Remacle 1956: 56f.), we have also seen clear evidence of the aoristic drift (see example 11b), in support of Smith (2016: 317f.). In addition, the ~ alternation and the agreement marking on the perfect participle have been virtually lost in this area. Again, once we differentiate
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
205
between the expression of grammatical categories and the marking of syntactic functions, these combined outcomes are neither puzzling nor problematic. If a principled relation exists, this is between the advancement of the perfect periphrasis and the decline of the preterite. I refer to Schaden (2012) for a proposal on the parallel process which has taken place in German. An important consequence of this development, noted in Vincent (2014: 110f.), is that the grammatical function comes to be expressed non-compositionally by the combination of a present-tense auxiliary with a perfect participle (22a). A further development may then combine the non-compositional past periphrasis with a perfect participle in the expression of the function , thus restoring compositionality (22b). This is what we find in the temps surcomposés or double compound tenses. (22a)
present auxiliary + perfect participle =
(22b)
(present auxiliary + perfect participle) () + perfect participle () =
It could hardly be a coincidence that in northern Romance, and especially in northeastern Gallo-Romance, we find abundant evidence of the latter development.
(23)
(Naisey, Franc-Comtois) É: vu èvù kuŋù jã: yó:d? have.2 you. have.. know.. Jean Claude ‘Have you already / Had you met Jean Claude?’ (Alex 1965: 116)
In Wallon, every compound tense, in the indicative, subjunctive, and conditional, as well as in the non-finite forms of the verb, has a double compound counterpart in the active voice (Remacle 1956: 71f.). (24) a. Il a avou he have.3 have.. ‘He has done/finished.’
fêt. do..
b. Qu’ il âye avou that he have..3 have.. ‘that he should have done.’
(Wallon)
fêt. do..
c. (Duvant d’) aveûr avou fêt. before of have. have.. do.. ‘before doing’ (lit. before having done)
(Wallon)
(Wallon) (Remacle 1956: 71)
The double compound tenses are also found in the passive voice, where provides the compound auxiliary.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
206
(25) Cwand k’ il arint avou stou when that they have..3 have.. be.. batous. (Wallon) beat... Lit. ‘when they would have been beaten.’ (Remacle 1956: 72) There is, therefore, undeniable evidence of analyticity in the expression of grammatical categories in northeastern Gallo-Romance. The generalization of as the only perfect auxiliary and the erosion of the participial agreement marking are manifestations of the loss of active-middle and active-inactive alignment (see §9.2).¹¹ Although these processes are entirely compatible with the expression of the past by the perfect periphrasis, and, therefore, there is internal consistency within the linguistic systems under scrutiny, the northeastern Gallo-Romance varieties do stand out in the northern Romània in having generalized perfect . Therefore, the question arises whether there are any properties that are specific to Gallo-Romance that may have played a role in this development. It was highlighted above that the syntax of subject clitics is a matter of debate and that the contrast between Gallo-Romance subject clitics and northern ItaloRomance ones may be more nuanced than was once thought (Cardinaletti and Repetti 2010). Be that as it may, it is in the syntax of the subject that the said specific properties of Gallo-Romance are to be found. Not only do Gallo-Romance subject clitics exhibit a certain degree of independence from the verb (see 1a, 2a–b, 3a–b), but the subject position must be filled overtly, regardless of grammatical person, referentiality, or information structure (see 4a–b). Therefore, the primary expression of the subject is on the dependent in Gallo-Romance. While being at the same time inflexional, by nominative case on the clitic, and positional (Kiparsky 1997), this dependent marking pattern does not differentiate between SA and SP/O. I propose that the obligatory marking of the subject on the dependent in accordance with the nominative principle is what first engendered the drift towards the reduction and, in some cases, the eventual loss of the ~ alternation on the head in Gallo-Romance. The erosion of the finite inflexion, with the complete levelling of the three singular persons, may also have been a factor in the weakening of middle alignment, in that it contributed to the loss of the role of the inflexional head as the licenser of the subject.¹²
¹¹ Observe, incidentally, that the inflected participle batous ‘beaten.’ in (25) is the passive, not the perfect, participle. The passive participle has generally maintained the subject agreement inflexion in the Romance languages which have lost the agreement on the perfect participle. ¹² Note that the developments which the Gallo-Romance subject clitics are undergoing at present, and which are suggestive of a tendency towards affixation (Culbertson 2010), postdate the decline of in the perfect.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
207
Our hypothesis is entirely independent of the traditional view that the rise and development of subject clitics may have been favoured by the erosion of the verbal inflexion. That the latter development is not the cause of the former has been convincingly argued on chronological and comparative grounds (see, e.g. Harris 1978: 112f.). What is proposed, instead, is that the erosion of the verbal inflexion, which post-dated the emergence of subject clitics, altered the fine balance between head and dependent marking even further in favour of the latter. Since the marking of the subject on the dependent exclusively follows a nominative principle, this principle became predominant with the consequent regression of activemiddle alignment by means of the ~ alternation on the head. The partial erosion of the agreement marking on the perfect participle also added to the return to consistent nominative-accusative alignment. Significantly, the pattern found in northeastern Gallo-Romance is only a very clear instantiation of a very general trend, which has affected a great deal of the Gallo-Romance varieties, including some northwestern ones (see §9.2, §9.3, and examples 21a–b). This trend, I suggest, is ultimately motivated in terms of the renewed and strengthened dependent marking system that developed in these varieties. Importantly, by contrast with the oïl languages, the Occitan languages do not exhibit this trend, in that they are fully or partly null-subject languages, they have richer verbal inflexion, and, crucially, they exhibit the ~ alternation (Schlieben Lange 1971: 131; Ledgeway 2012: 324; Oliviéri and Sauzet 2016).
9.5 The North–South classification reconsidered Having discussed northern Gallo-Romance, in this section we return to the classification of the Romance languages in terms of a northern-vs-southern continuum. We begin by observing that the subject clitics are not the only exponent of person and number in northern Italian dialects and, therefore, they supplement, rather than replace, the verbal inflexion in the licensing of the subject (Bossong 2016: 67). In fact, as was noted in §9.2, these clitics are not necessarily present in all grammatical persons or in all syntactic contexts. In presentational constructions, which normally have the argument of an intransitive verb in post-verbal position, only some northern Italo-Romance dialects exhibit an expletive subject clitic, whereas a larger number of these dialects require an etymologically locative clitic whose distribution is constrained in lexico-semantic terms (Tortora 1997, 2014; Bentley 2018b). Put crudely, the majority of the northern Italo-Romance dialects really do exhibit the hallmarks of head-marking and are thus distinct from Gallo-Romance in the subject properties which were claimed to be at the root of the gradual return to nominative-accusative alignment in Gallo-Romance.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
208
As for Romansh, it did develop obligatory subject clitics and, in fact, it is classified as a non-null-subject language. Yet, it still exhibits the ~ alternation (though not in all reflexives) and the agreement marking on the perfect participle. Romansh has also undergone the aoristic drift and, therefore, it fits the northern Romance profile squarely (Anderson 2016: 175–9 for details). The evidence of Romansh indicates, unsurprisingly, that there is no deterministic correlation between the licensing of the subject on the dependent and the loss of the perfect auxiliary alternation on the head. At the same time, it should be noted that Romansh has a healthy verbal inflexion and, indeed, it has been cited as comparative evidence that challenges the view that the French subject clitics might have developed as a result of the loss of this inflexion (Harris 1978: 113). Therefore, Romansh clearly differs from Gallo-Romance in marking the subject obligatorily both on the dependent and on the head. The latter licenses the subject inflexionally and marks it differentially by means of the ~ alternation. The syntax of subject clitics is characterized by a great deal of variation across the Ladin varieties, with many such varieties exhibiting, in a range of syntactic contexts, the doubling of the clitic with a noun phrase subject. This is suggestive of lack of independence or even affixal status of the clitic (Haiman and Benincà 1992: 187–97). In Friulian the doubling of the subject clitic is unconstrained (Haiman and Benincà 1992: 192). For our current purposes, therefore, these RaetoRomance languages are comparable with northern Italo-Romance, and, indeed, they too exhibit the hallmarks of the northern Romance typology. Turning now to the South of the Romània, the process of return to generalized nominative-accusative alignment was not triggered in these languages by a change in the licensing of the subject. As for the lack of aoristic drift, although these languages have maintained the preterite, which in some of them (notably, European Portuguese) is, in fact, the principal perfective past tense form, it is a moot point whether the existence of a synthetic past tense may have favoured the weakening of the ~ alternation. We have seen that the preterite has been maintained elsewhere (northwestern Gallo-Romance, Tuscan, etc.) and yet this has had no effect on the alternation. In the South, the drift towards active-middle and active-inactive alignment was never as profound as it was in the North. Clear indications of this are the absence of -cliticization (see §9.2 and examples 6a–c) in Romanian and modern Ibero-Romance, and the infrequency of -cliticization of intransitive unaccusative subjects in the Sicilian vernacular texts of the fourteenth and fifteenth centuries (Bentley 2004: 539). The same texts, while providing evidence of the ~ alternation, also bear a clear testimony that the domains of were already limited at that diachronic stage (La Fauci 1992). In some of the southern languages that have had—or indeed still have—a type of ~ alternation, this is also partly or wholly governed by semantic parameters that are orthogonal to active-middle and active-inactive alignment, notably, the realis-irrealis opposition
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
209
(see Ledgeway 2003 for Neapolitan and Avram and Hill 2007 for Romanian, which we have left out of the classification, see Table 9.1 and note 4). At this point, Zamboni’s (1998) characterization of the North-vs-South divide in terms of a split between languages oriented towards the marking of the subject or, respectively, the object becomes relevant. Zamboni (1998: 135–44) explains that the languages of the South neutralized the case oppositions, extending the domains of accusative case, at an early stage. Object orientation is also testified by the differential marking of the object. Admittedly, this is also attested in some of the northern areas: Gascon, Languedocian (Occitan), Engadine (Romansh) (Bossong 1998: 220), and, in specific discourse contexts, northern Italo-Romance (Iemmolo 2010 and references therein). However, the differential marking of the object is also, and, in fact, primarily, widespread in the South. In particular, it is found in Ibero-Romance, Catalan, southern-Italo-Romance, as well as Romanian. (26) a. Lo has visto a mi hermano? (Spanish) .. have.2 see.. my brother ‘Have you seen my brother?’ b. La adus nenea pe copil. (Romanian) . have.3 bring. uncle child ‘The uncle has brought the child.’ (Bossong 1998: 221, 226) The differential marking of the object operates within nominative-accusative systems, singling out pronouns and, more broadly, noun phrase objects which are high on a D-hierarchy (Kiparsky 2008). In addition, this type of marking of the object may be related to the doubling of a topicalized object with an accusative clitic (see 26a–b), although its relatedness to clitic doubling is not equally strong in all languages. Therefore, the differential marking of the object is a manifestation of nominative-accusative alignment on a dependent: not the subject, in this case, but rather the object. To conclude, the loss of the ~ alternation and of the agreement marking on the perfect participle in the languages of the South is to be conceived of as a drift which started from weaker active-middle syntax and was different from, and unrelated to, the drift that can be observed in Gallo-Romance. The odd one out, among the languages which Zamboni (1998) classifies as part of the South, is Sardinian, where we find both the differential marking of the object, following nominative-accusative alignment (see 27), and the differential marking of the subject, following active-middle and active-inactive alignment (28a–b, 29a–b) (Jones 1993; Mensching and Remberger 2016). (27) Apo vidu a have.1 see. ‘I have seen Carminu.’
Carminu. Carminu
(Orgosolo, Nuorese Sardinian)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
210
(28) a. Juanne at John have.3 ‘John has danced.’
ballatu. dance..
(Lula, Nuorese Sardinian)
b. Frantziscu est semper vénnitu Francis be.3 always come... ‘Francis has always come to my house.’ (29) a. At pappau. have.3 eat.. ‘S/he has eaten.’
a domo mea. to house my (Jones 1993: 140)
(Villacidro, Campidanese Sardinian)
b. Funt mortas. be.3 die... ‘They (F) have died.’ Sardinian also expresses the past tense (in fact, most of the grammatical categories pertaining to the verb system) periphrastically, although this is not the result of a recent drift, but rather a feature that is already, almost without exception, attested in the chancery language of the eleventh to the fifteenth centuries. In the final analysis, Sardinian is a telling illustration of the complexity of the typology upon which the North–South classification is based: it simply does not fit into this typology. In the light of the above evidence and discussion, the classification of the Romance languages in terms of alignment as well as the aoristic drift requires careful rethinking. The same can be said of the contrast between, on the one hand, languages characterized by a variety of patterns of subject marking, including the differential marking by the perfect auxiliary alternation (the North), and, on the other hand, languages characterized by the differential marking of the object (the South). The scenario which emerges in the modern Romània is one in which there is little evidence that the aoristic drift correlates with the maintenance of active-middle alignment.¹³ More importantly, only a group of central languages with clear patterns of head marking of the subject (Occitan, northern Italo-Romance (Gallo-Italian and Venetan), Tuscan, Corsican, Sardinian, and, Raeto-Romance) have maintained the primary manifestations of active-middle alignment. By contrast, the languages of the South (Ibero-Romance, Catalan, southern Italo-Romance), with dependent marking of the object, and northern
¹³ The aoristic drift also occurred in German, which exhibits the ~ alternation. While this fact might seem to suggest that there may after all be a principled correlation between the periphrastic expression of the past and the maintenance of the auxiliary alternation, and further research is no doubt needed on this issue, it is undeniable that the Romance evidence is not conclusive with respect to such a correlation.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
-
211
Gallo-Romance, with dependent marking of the subject, have undergone separate gradual processes of return to undifferentiated nominative-accusative alignment.
9.6 Conclusion The gradual return to nominative-accusative alignment, after a shift towards active-middle and active-inactive alignment which originated in Latin (Zamboni 1998: 127–9), has characterized the history of Romance (La Fauci 1988). Two clear manifestations of this return are the weakening and eventual loss of the alternation of the two perfect auxiliaries derived from and , with the consequent generalization of , and the loss of the agreement of the perfect participle with the gender and number features of inactive (S/) subjects. This process of gradual loss has occurred most conspicuously in a number of Romance languages which have traditionally been regarded as the languages of the South (Ibero-Romance, Catalan, southern-Italo-Romance). A comparable, if later, process has also characterized the history of Gallo-Romance, a subfamily of Romance which is traditionally classified as belonging to the North. As noted by Smith (2016), this process has reached a very advanced stage in the eastern oïl varieties (e.g. Picard, Wallon, etc.), which, however, fit the northern profile in having undergone the decline of the preterite. In this article, I have suggested that the maintenance or loss of the preterite is in principle unrelated to the issue of alignment, witness the case of northwestern Gallo-Romance, where both the preterite and the alternation of the two perfect auxiliaries are still attested, though not without exception (see §9.4). This outcome does not single out northwestern Gallo-Romance uniquely, as both the preterite and the perfect auxiliary alternation are also attested in Tuscan, central and uppersouthern Italo-Romance and Corsican. The fact that the marking of active-middle alignment by means of the perfect auxiliary alternation is parasitic on the periphrastic expression of the perfect is, of course, undeniable. Indeed, building upon Ledgeway (2012, 2016a), and references therein, I have argued that the main pattern of active-middle alignment in Romance ensued from a process of acquisition of functional structure. I have also claimed, however, that the developments observed in northeastern Gallo-Romance, and, more generally, the fact that, within the northern Romània, it is Gallo-Romance that has followed the southern languages in gradually losing the primary manifestation of active-middle alignment (the ~ alternation) is not trivial. On the contrary, this process has its roots in the emergence of subject clitics which, unlike the subject clitics that developed elsewhere in the northern Romània, had features of proper dependents (obligatoriness, relative independence from the verb, case marking, etc.). Dependent marking on the subject clitic system did not differentiate between SA and SP/O, but rather followed a generalized
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
212
nominative principle. The partial erosion of the verbal inflexion, which postdated the emergence of subject clitics, further weakened the marking of the subject on the inflexional head. As a result, the tendency towards the re-establishment of the undifferentiated marking of the subject on the dependent was strengthened. I conclude that the classification of the Romance languages in terms of a contrast between the aoristic drift plus active-middle making of the subject (the North) and the maintenance of the preterite plus generalized nominativeaccusative alignment (the South) must be reconsidered. The scenario that emerges in the modern Romània is one in which, I claim, head marking of the subject has played a role in the maintenance of the key manifestation of active-middle alignment, viz. the ~ alternation, in some central Romance languages. The drift towards the return to undifferentiated nominative-accusative alignment in the South occurred independently from the Gallo-Romance facts discussed in this chapter and, by and large, at an earlier stage. Sardinian stands out as a Romance subfamily with both differential head marking of the subject and differential dependent marking of the object.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
10 A comparative analysis of French auxiliation, with new evidence from Montréal Béatrice Rea
10.1 Introduction In modern French, as in many Romance and Germanic languages, the periphrastic verb tenses are formed by combining a past participle with the conjugated auxiliary avoir ‘have’ or être ‘be’ (or their equivalents), depending on the verb and the context.¹,²,³ In written standard French, the distinction between verbs that take avoir and those that take être appears relatively clear: while all nonpronominal transitive verbs require avoir, a small subset of intransitive stative and motion verbs combine with être. The full list of intransitive verbs that are conjugated with être (henceforth ‘intransitive Ê-verbs’) traditionally comprises advenir ‘to happen’, aller ‘to go’, arriver ‘to arrive’, décéder ‘to die’, (re)devenir ‘to become (again)’, échoir ‘to fall due’ or ‘to expire’, (r)entrer ‘to go (back) in’, intervenir ‘to intervene’, mourir ‘to die’, naître ‘to be born’, (re)partir ‘to leave (again)’, parvenir ‘to achieve’, provenir ‘to come from’ or ‘to be the result of ’, rester ‘to stay’, retourner ‘to return’, (res)sortir ‘to go out (again)’, (re)tomber ‘to fall (again)’, (re)venir ‘to come (back)’, and survenir ‘to take place’, ‘to arise’, or ‘to arrive unexpectedly’ (Grevisse 2011: §782). Since a few of these verbs are employed fairly rarely in everyday speech, second-language learners of French usually resort to mnemonic devices for remembering the list of verbs that take auxiliary être. However, many studies seem to suggest that in various parts of the Frenchspeaking world native Francophones employ both auxiliaries, avoir and être, in the spoken language with verbs that, officially, require solely être. This phenomenon plausibly points towards a progressive generalization of the avoir auxiliary in these areas (for Québec French, see Sankoff and Thibault 1977; for Ontario ¹ My thanks go to André Thibault and Mathieu Avanzi for generously sharing their data with me. I also thank the two anonymous reviewers for providing helpful comments on an earlier version of this chapter’s abstract. ² The verbs appearing in French will only be glossed the first time they occur in the text. ³ All translations are my own, except where otherwise indicated.
Variation and Change in Gallo-Romance Grammar. First edition. Sam Wolfe and Martin Maiden (eds) This chapter © Béatrice Rea 2020. First published in 2020 by Oxford University Press.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
214
ˊ
French, see Canale, Mougeon, and Bélanger 1978; for Acadian French, see King and Nadasdi 2005; for New England French, see Russo and Roberts 1999; a more exhaustive list can be found in §10.3.1). This usage of the auxiliary avoir in contexts where, prescriptively, it ‘should not’ appear occurs as something notable in every variety of North American French, as well as in certain regions of France and Belgium, and challenges the typology established by Zamboni (1998: 128; 2000: 86, 104f.), and later corroborated by Ledgeway (2012: 314), which divides Romance languages according to a northern–southern continuum (see also Chapter 9). Among various structural features supposedly opposing northern language groups to southern ones, it serves us here to highlight the binary contrast in auxiliation between reflexes of ‘have’ and of ‘be’, characteristic of northern languages (to which French belongs according to this typology) versus the generalization of a single auxiliary, either or ,⁴ contingent ‘on the variety and/or the syntactic context’ (Ledgeway 2012: 314). This chapter will introduce French auxiliation by providing a brief outline of theoretical approaches, both semantic and syntactic, put forward in order to synchronically describe French split intransitivity and by explaining why a sociolinguistic framework might yield more accurate results (§10.2). I will then provide an evaluation of previous work on (socio)linguistic variables that have been shown to influence avoir-generalization in North American varieties of French (§10.3).⁵ The last section of the chapter will discuss auxiliary alternation in new Montréal French data, collected in the context of sociolinguistic interviews (Rea forthcoming) and through crowdsourcing questionnaires (Thibault and Avanzi 2015), and will analyse how it relates to previously studied factors, both external / social and internal / linguistic (§10.4).
10.2 Theoretical approaches to split intransitivity Among the Romance languages that make use of two auxiliary verbs are French with various oïl dialects, Sardinian, Raeto-Romance, standard Italian and many Italo-Romance dialects, Catalan dialects, Occitan, Corsican, and Romanian. Syntactic analyses of split auxiliary selection are based on evidence that the subject of -verbs (verbs selecting the BE auxiliary) shares properties with direct objects of transitives, an analysis supported by evidence from other constructions (impersonal passivization, verbs with reflexive clitics, etc.; see Aranovich 2007: 2), while the subject of -verbs (verbs selecting the HAVE auxiliary) shares them with the subject of transitives. More evidence for a syntactic analysis of auxiliary ⁴ See Tuttle (1986) on the generalization of in central Italo-Romance, although generalization of in Romance languages is more common. ⁵ Since no sociolinguistic variationist study of auxiliary alternation has yet been conducted on European varieties of French, North American studies are the only data available. However, the Sociolinguistics Lab of the University of Ottawa is currently conducting a study on auxiliary alternation in Parisian French.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
215
selection comes from pronominal verbs in languages like Italian and French. In these languages, reflexive verbs (i.e. languages with reflexive clitics) select essere / être ‘be’ as the perfect auxiliary. The argument rests on the hypothesis that the subject of a reflexive verbs is an object at some abstract syntactic level. The existence of variation in split auxiliary systems across languages is part of the evidence syntactic theories of auxiliary selection use to argue against a semantic basis for the selection of BE or HAVE (Rosen 1984; Burzio 1986; Perlmutter 1989; Kayne 1993).
10.2.1 The interface of syntax and semantics: the Auxiliary Selection Hierarchy Sorace (2000) introduces the Auxiliary Selection Hierarchy (ASH), an aspectual framework aiming to predict how Romance (mostly French and Italian) and Germanic languages (mostly Dutch and German) that have a two-auxiliary system distribute their perfective auxiliaries in intransitive verbs.⁶ The hierarchy predicts how the level of telicity of certain verbs dictates their auxiliary choice. Verbs at the extremes of the hierarchy (‘core’ verbs, as opposed to ‘peripheral’ verbs in the centre of Table 10.1 which can display more variation) are change of location verbs at the BE end and non-motional process verbs at the HAVE end. They are characterized by the following properties: ‘categorical and consistent syntactic behaviour across languages’ (see examples 1 and 2); ‘consistent behaviour within individual languages’; ‘insensitivity to compositional properties of the predicate’; ‘determinacy of native speakers’ intuitions’;⁷ ‘primacy in acquisition’ and ’diachronic stability’ (Sorace 2004: 256). The verbs that select BE tend to be unaccusative and the ones selecting HAVE unergative: Table 10.1. The Auxiliary Selection Hierarchy (Sorace 2000: 863) Change of location Change of state Continuation of a pre-existing state Existence of state Uncontrolled processes Controlled processes (motional) Controlled processes (non-motional)
Select E (least variation)
Select A (least variation)
⁶ Her theory has also been further discussed and developed in Sorace (2004), Legendre and Sorace (2003), Legendre (2007), Legendre (2017), Ackema and Sorace (2017), among others. ⁷ Loporcaro (2016: 818) notes that the study of Legendre and Sorace (2003) ‘propagates some dubious data such as It. la popolarità del governo è scesa/ha sceso notevolmente “the government’s popularity is/has dropped notably” and sono/?ho rimasto solo “I am/?have remained alone”—judgments reported by Legendre and Sorace (2003: 195f.), despite HAVE being totally ungrammatical in both examples in standard Italian.’
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
216
ˊ
Sorace (2004: 243) explains that the distinction is also systematically related to the semantic characteristics of the predicate: agentivity correlates with unergativity and patienthood correlates with unaccusativity (Perlmutter 1978; Dowty 1991). According to her (2004: 256), verbs such as arriver, venir, partir, belong to the top of the hierarchy and verbs like travailler ‘to work’ and parler ‘to speak’ belong to the bottom: (1)
Ma soeur est arrivée / **a arrivé hier (Fr.)8 ‘My sister is/has arrived yesterday’
(2)
Les délégués ont parlé / **sont parlés toute la nuit (Fr.) ‘The delegates have/are spoken all night.’
The ‘core’ verbs also select their auxiliary according to their inherent meaning, even when they appear alongside contradictory elements. This is illustrated in (3–5) (Sorace 2004: 257): (3)
a. Des plaintes sont arrivées continuellement. ‘Complaints have arrived continuously’
atelic predicate
b. Des imitations sont apparues depuis des années. ‘Copies have been appearing for years’ (4)
a. Maria est tombée volontairement pour ne pas aller travailler. agentive predicate ‘Maria has fallen on purpose in order not to have to go to work.’ b. Le verre est tombé de la table. ‘The glass has fallen from the table.’
(5)
Les policiers ont travaillé jusqu’à l’aube. ‘The police officers have worked until dawn’
non-agentive predicate
telic predicate
Legendre and Sorace (2003) develop the hierarchy further, in Table 10.2, by showing the difference in the cut-off point where French and Italian select avoir / avere: for French it seems to be somewhere in the ‘change of state’ category, much higher in the hierarchy than it is in Italian (% indicates variation).
⁸ It is worth pointing that such sentences are found frequently in the Rea 2016 Montréal Corpus (discussed in §10.4.1). This corpus comprises forty-eight sociolinguistic interviews with native speakers of Montréal French.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
217
Table 10.2. Auxiliary selection in (standard) French and (standard) Italian (adapted from Legendre and Sorace 2003: 200) French Italian E
E
E E%
E E
E% A A
E E% E%
A A
E E%
A A A A A
A% A A% A% A
change of location: arriver/arrivare ‘to arrive’, venir/venire ‘to come’, etc. change of state: a. change of condition: mourir/morire ‘to die’, etc. b. appearance: apparaître/apparire ‘to appear’, etc. c. indefinite change in a particular direction: monter/salire ‘to go up’, descendre/scendere ‘to go down’ faner/appassire ‘to wilt’, empirer/peggiorare ‘to worsen’, etc. continuation of a pre-existing state: durer/ durare ‘to last’, etc. existence of state: a. être/essere ‘to be’ b. exister/esistere ‘to exist’, suffire à/bastare ‘to be enough’ uncontrolled processes: a. emission: résonner/risuonare ‘to resonate’, etc. b. bodily functions: suer/sudare ‘to sweat’, etc. c. involuntary actions: trembler/tremare ‘to shiver’, etc. motional controlled processes: nager/nuotare ‘to swim’, etc. non-motional controlled processes: travailler/lavorare ‘to work’, etc.
10.2.2 Syntactic approach to split intransitivity 10.2.2.1 French as an outlier among Romance auxiliation systems? Ledgeway (forthcoming) analyses auxiliation in Romance languages in terms of a syntactic hierarchy involving four mesoparametric patterns of variation.⁹ Among the various Romance varieties that he studies, some show no alternation at all and generalize a single auxiliary.¹⁰ Others varieties display ‘free’ variation between HAVE and BE in all contexts, as is the case of the southern Calabrian dialect of ⁹ Ledgeway (forthcoming) conceives linguistic variation in a scalar fashion modelled in terms of ‘parametric hierarchies’: Macroparameters, the simplest and least marked options that uniformly apply to all functional heads, are placed at the very top of the hierarchy, but, as we move downwards, variation becomes progressively less ‘macro’ and, at the same time, more restricted with choices becoming progressively more limited to increasingly smaller subsets of features. [ . . . ] More specifically, functional heads increasingly display a disparate behaviour in relation to particular feature values which may, for example, characterize: (i) a naturally definable class of functional heads (e.g. [+N], [+finite]), a case of mesoparametric variation; (ii) a small, lexically definable subclass of functional heads (e.g. pronominals, proper nouns, auxiliaries, unaccusatives), a case of microparametric variation proper; and (iii) one or more individual lexical items, a case of nanoparametric variation. ¹⁰ For instance, Pescolanciano, like many central-southern dialects of Italy, generalizes BE, whereas Castilian / Portuguese / urban Catalan / Sicilian, and many other Ibero-Romance and extreme southern Italian varieties, generalize HAVE.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
218
ˊ
Saline Ioniche, a district of Montebello Ionico. Among the varieties that have a structured two-auxiliary variation (Piemontese, Raeto-Romance, Occitan, Sardinian, standard Italian), auxiliary distribution appears to be sensitive to mood, tense, person, or argument structure. It is worth noting that the Romance varieties whose auxiliary distribution is sensitive to person are all non-standard and display this sensitivity solely in the present perfect tense (Ledgeway forthcoming). For instance, in Ariellese, an eastern Abruzzese dialect, the present BE is used for first and second persons and HAVE is employed for third persons. This specific pattern of auxiliation is observed quite often because it distinguishes the basic discourse participants by their saliency: the first and second persons having a different form from the third ones (see Štichauer 2018 for a detailed analysis of person-based perfective auxiliation systems in Italo-Romance). Another perspective on grammatical person is discussed by Croft (2003: 140–2), who observes that, globally, singular forms occur with a greater frequency than plural ones, and that third-person forms are more frequent than first- and second-person ones. This could lead us to conclude that first and second persons would be rarer forms in natural speech, and one could expect them to undergo auxiliary generalization more easily than third persons. However, at the local level, grammatical person frequencies can also depend upon the different types of lexical verbs. Ledgeway adds that with personsensitive auxiliary selection various layers of restriction (mood / tense, number, saliency) can be progressively superimposed or even combined.
10.2.2.2 Argument structure and unaccusativity In modern French, as in other languages such as Italian, Dutch, and German, it has been argued that auxiliary selection patterns with unaccusativity (cf. §10.2). The Unaccusative Hypothesis specifies that there exist two types of intransitive verbs: unaccusatives and unergatives, which carry different semantic and syntactic features (Perlmutter 1978: 160; Burzio 1986). With regard to auxiliary selection, it is assumed that in their periphrastic tenses, especially the perfect tense, the unaccusative verbs take the BE auxiliary, thus generally patterning with pronominals and passives,¹¹ and the unergative verbs require the HAVE auxiliary (Burzio 1986: 53; Loporcaro 2007: 187). The surface subject of unergative verbs behaves like the subject of a transitive verb, while the surface subject of unaccusative verbs acts rather like the object of a transitive verb (Burzio 1986: 30, 56, 74; Mackenzie 2006: 6). Thus, this theory suggests that while the underlying subjects of unergative verbs are also their surface subjects, the surface subjects of unaccusative verbs are in fact underlying objects. Clear examples of unaccusative verbs in French
¹¹ Utheim (2013) shows that passive constructions correspond to an unaccusative structure. The same appears to be true for pronominal constructions (Buchard and Carlier 2008: 2426).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
219
include tomber, mourir, and naître. Mackenzie (2006: 117), however, notes that the membership of certain French verbs in the unaccusative category does not necessarily match their auxiliary, as opposed to standard Italian for example, so that obvious unaccusative verbs such as manquer ‘to be missing’, exister ‘to exist’, or surgir ‘to arise’ surface in the standard language with the auxiliary avoir. Auxiliary selection in French can thus not exclusively correspond synchronically to semantic or lexical features, but rather ‘contributes grammatical [ . . . ] meaning to a compound verb form’ (Loporcaro 2007: 175). Ledgeway (forthcoming) also discusses auxiliary distribution sensitive to argument structure, which all Romance varieties once exhibited across all paradigms and persons. Various layers of restriction (mood / tense, person / number, saliency) can also be progressively added or combined to this dimension. A typical case of this auxiliation pattern is Lengadocien Occitan, which has a binary active-stative split: HAVE surfaces in conjunction with the subjects of transitive and unergative verbs (A/SA), and BE surfaces in conjunction with the subjects of unaccusative verbs (SO). Ledgeway explains that this pattern ‘represents the least constrained option and historically is also is the most widespread pattern in Romance which survives today in numerous Gallo-Romance and Italo-Romance varieties’, where unaccusative verbs align with BE and transitive/ unergative verbs align with HAVE for historical reasons (Vincent 1982; Ledgeway 2012: 130–4).¹² Ledgeway (forthcoming) also discusses an additional dimension of variation with regard to BE-selection: unaccusativity being lexically fossilized as in the case of many contemporary langue d’oïl varieties. In such languages, the selection of BE is now reduced to a ‘synchronically opaque, small number of intransitive predicates. This nanoparametric state of affairs, in which a once productive auxiliary distinction has all but fallen out of the system today precariously surviving in association with particular predicates as a lexical idiosyncrasy’. Ledgeway then cites the example of standard French, which, as was noted above, now has only between twenty and thirty unaccusatives that still select BE (Benveniste 1965: 181; Giancarli 2011: 373f.) and correspond to a small subset of the verbs that still select BE in Italian, Occitan, and Corsican (Maiden and Robustelli 2007: 262; Giancarli 2011). Additionally, it appears that ‘popular’ varieties of European French and Occitan, such as Gascon (Bouzet 1963), display even fewer instances of BEselection with most unaccusatives and in pronominals (Frei 1929: 68, 166; Bauche 1946: 105; Guiraud 1969: 40f.), as is the case in many eastern langue d’oïl varieties (Remacle 1956: 39–48; Descusses 1986: 126; Hendschel 2012: §166b p. 177, for Wallon French). Moreover, some Picard dialects (Flutre 1955: 59;
¹² For a more detailed diachronic account of auxiliary alternation in French, see Nyrop (1930: 212f.3), Levitt (1979), Tailleur (2007), and Rideout (2011).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
220
ˊ
Vasseur 1996: 52; Dauby 1979: 35; Pooley 1988) also show retention of BE in just one or two unaccusatives, mourir ‘die’ and aller ‘go’, and some generalize avoir even in pronominals (Auger 2003). Similarly, in the Lorrain variety of Ranrupt all unaccusatives take HAVE with the sole exception of ‘come’, which still licenses BE (Aub-Büscher 1962: 84 §107). Smith (1989c: 320) notes that in some Occitan dialects, such as Limousin, Languedocien, and Provençal (as well as in Sardinian), if the subject follows the verb, the auxiliary that surfaces is HAVE, and if the subject precedes the verb, the selected auxiliary is BE (see Ronjat 1937; Roux 1895; Sicre 1909 for data on Occitan dialects; see Jones 1988 for Sardinian data). Ledgeway does not fail to mention the ‘free’ alternation of HAVE~BE in unaccusatives and reflexives in certain Canadian French varieties, even though he notes that this variation can be partially attributed to sociolinguistic factors (Sankoff and Thibault 1977; King and Nadasdi 2005; Rea 2014). However, in these varieties a subset of core unaccusatives (e.g. aller ‘go’) tends to select BE more often than the others. Ledgeway refers to all these cases as ‘synchronically unpredictable cases of lexical exceptions which residually reflect formerly more widespread and regular patterns of variation’. Using the work of Manente (2008: 42f.), Ledgeway (forthcoming) claims that, in the recent history of Québécois French, ‘HAVE has been extended to unaccusatives of change of location/directed motion’. Ledgeway’s use of the verb ‘extend’ is rather problematic here because it suggests that BE was underlyingly present or used to be found more commonly in Québécois French. Even though HAVE has replaced BE to mark punctual events, as in (6a), BE appears to have survived with these same verbs when they have a resultative reading, as in (6b): (6)
a. Je suis rentré. /J’ai rentré. ‘I came back’ b. Je suis rentré./ **J’ai rentré. ‘I am back’
However, Manente’s claim about Québécois French is not particularly substantial since her Québécois French data appear to come exclusively from Sankoff and Thibault’s (1977) seminal paper. Moreover, more recent work by Rea (2014, 2017, 2018a) has shown that auxiliary BE is much more prominent than in 1971. Since Manente also mentions the work of Canale, Mougeon, and Bélanger (1978) on auxiliary alternation in Ontarian French in this section of her thesis, it is also unclear whether she uses the term ‘Québécois French’ as an equivalent of the umbrella term ‘Laurentian French’, which would include both Québécois and Ontarian French. This is also problematic since studies (Canale et al. 1978, among others) have shown that, in the 1970s, Ontario French was ahead in the process of avoir-generalization in comparison to Québécois French.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
221
10.2.3 Issues with theoretical approaches and the problem of ‘free’ variation I will not endeavour to form a detailed critique of the ASH because many scholars (including Bentley and Eythorsson 2003; Legendre 2007; Giancarli 2011, 2015; Utheim 2013; Kailuweit 2015; Loporcaro 2016) have already tried to improve and / or question the hierarchy. However, it is worth pointing out that one important flaw of the ASH is that it does not take into consideration pronominal verbs (Giancarli 2015: 86), even though they take BE both in Italian and in standard French (Legendre and Sorace 2003: 217). Regarding the pronominal verbs, Chaudenson et al. (1993: 24) explain that, on top of the diachronic explanation, the use of avoir with pronominal structures also has a motivation intrasystémique ‘intrasystemic motivation’ insofar as it can be observed in situations of language acquisition and language learning. Hallion (2000: 355) also confirms that French children (3- to 4-year-olds) sometimes elicit the following pronominal constructions with avoir: Je m’ai fait piquer ‘I have been stung’ and Regarde ce que je m’ai fait ‘Look what I have done to myself ’. Moreover, the semantic theoretical framework cannot fully explain French auxiliation, neither can it explain the immense variety of Romance auxiliary selection patterns based on mood, tense and person, outlined in §10.2.2. For example, the ASH cannot account for the use of être with verbs like demeurer ‘to remain, to live somewhere’ and rester which are ‘continuation of a pre-existing state’ verbs. For Giancarli (2015: 83f.), telicity does not seem to play a role with regard to these two verbs, and Legendre and Sorace (2003: 213) noticed their paradoxical atelicity: ‘de manière inattendue, quelques verbes atéliques dénotant l’absence de changement sélectionnent être’ (‘unexpectedly, a few atelic verbs that denote the absence of change select BE’), but according to Giancarli (2015: 83), this should have made them question ‘telicity as their main factor’. Indeed, the ASH does not explain why monter (7) and descendre (8) behave differently from other ‘change of state’ verbs which normally require the auxiliary avoir, like exploser ‘to explode’ (9) and fondre ‘to melt’ (10), for example: (7)
Je suis montée vers la colline. ‘I have gone up/climbed toward the hill’
(8)
Je suis descendue vers la plaine. ‘I have gone down toward the plain’
(9)
La bombe a explosé. ‘The bomb has exploded’
(10) La neige a fondu. ‘The snow has melted’
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
222
ˊ
In addition to analysing only standard (usually written) languages,¹³ and hence reducing language to arbitrary systems rather than analysing ‘real language in use’ and ‘non-standard’ varieties, proponents of these formalizing semantic theories, outlined in §10.2.1, treat the languages under consideration as stable synchronic entities. They do not take into account the fact that the distribution of avoir and être is historically determined (thus no synchronic explanation could be truly valid), how their auxiliation systems have evolved, and how much variation there was before the implementation of codification processes. Moreover, these approaches do not explain how different auxiliaries can be used with the same verb, in the active voice, in the same language, and by the same speaker, as we observe in contemporary speech in many varieties of French. They also leave no place for sociolinguistic considerations (intra- and inter-speaker variation in terms of age, gender, socioeconomic background, etc.) and linguistic constraints (apart from lexical effect), when numerous studies (see §10.3.2) have shown that auxiliary selection can vary along those lines. As to the syntactic analyses mentioned in §10.2.2, they rigorously cover an exhaustive number of non-standard dialects, however, there is little or no mention of how the data for each of the studies quoted was collected. It also seems highly improbable that there would only be one Romance variety, i.e. Canadian French, showing social, otherwise ‘free’, variation. As Tristram (2014: 6) points out: Even if phonological and morphosyntactic variation turn out to pattern in significantly different ways (and there is some suggestion that this may be the case; cf. Armstrong 2001), it seems almost inconceivable, given what is now known about phonological variation and change, that social factors would not play some role in morphosyntactic variation and change as well.
Since free variation is extremely rare,¹⁴ I would argue that the latter issue stems mainly from a difference in data collection methodology, because very few metadata are collected and analysed in these dialectological studies. Given the various auxiliary distributions summarized in §10.2.2, why would Canadian French pattern so differently from the other Romance varieties? It has been suggested by Thibault and Sankoff (1997: abstract) that: although there is a tendency for the non-standard use of avoir [in Montréal, Canada] to be associated with working class speakers, this applies differentially to ¹³ For example, Giancarli’s (2011) French data comes from a corpus of French plays. ¹⁴ Regarding free variation of HAVE / BE, Loporcaro (2016: 816) mentions that he knows ‘of no cases where it occurs across the board in the verb paradigm in all clause types’, but adds that ‘the varieties coming closest to this are spoken in northern Apulia: Minervino Murge (Manzini and Savoia 2005, III: 27f.) has full free variation in the pluperfect, whereas an auxiliation contrast persists in the present perfect just in the third singular, in contrast with free variation in all other persons exemplified with the first singular’.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
223
the various verbs, and the social class based pattern is far from categorical. A semantically driven association between auxiliary use and verb meaning is weakly motivated at best. Rather, there seems to be a lexically-focused distribution.
This would imply that the auxiliary variation recorded in Montréal in 1971 (Sankoff and Thibault 1977) was lexically arbitrary, but no other Romance languages seem to display this pattern. Smith (1989c: 311) suggests that a system requiring the use of two different auxiliary verbs is opaque, i.e. ‘is at variance with the norm for forms with a similar function’, and this characteristic might favour the progressive extinction of BE as an auxiliary in Romance languages. It therefore does not come as a surprise that HAVE should be the sole auxiliary verb used to generate the ‘transparent forms’ since it is the most unmarked and frequent auxiliary. The fact that there are two auxiliary verbs sharing one function is not transparent because from a functional perspective only one auxiliary is needed. And yet, this ‘opaqueness’ approach would suggest that split systems would not persist for millennia, even though they seem to in certain cases. However, it appears that speakers of various Romance varieties try to make sense of the distribution by mapping it into formal distinctions, as seen in §10.2.2. In psycholinguistic terms, it could be argued that some of the patterns described earlier can be explained through the concept of synonymy avoidance or by the Principle of Contrast theorized by Clark (1987, 1993: 64), which states that ‘speakers take every difference in form to mark a difference in meaning’. In this section, I have provided a brief overview of theoretical frameworks, both semantic and syntactic, put forward in order to explain French split intransitivity. I have also outlined the most important gaps in such approaches.
10.3 Recorded variation in French auxiliary selection I now illustrate why a variationist sociolinguistic methodology can best account for the types of auxiliary alternation observed in spoken / non-standard varieties of French, by first referring to the studies listed in §10.3.1. It is well known that today all over the Francophonie, some native French speakers employ both auxiliaries, avoir and être, in the spoken language with verbs that prescriptively require être. Illustrative examples (11–13) are taken from Rea’s Montréal French 2016 corpus: (11) Mon frère, il descendait une côte pis sa roue d’en avant a parti. (Rachel) ‘My brother, he was going down a hill and his front wheel came off ’. (12) On a juste resté là, pis on a bu, pis on se promenait pis c’était le fun. (Charles-Antoine) ‘We just stayed there, and we drank, and we were walking around and it was fun’.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
224
ˊ
(13) Mais après une semaine, j’ai, j’ai retourné travailler. ‘But after a week, I, I went back to work’.
(Gaëtan)
10.3.1 Variationist sociolinguistic studies on North American French auxiliary alternation A. Laurentian French a. Québec, Canada i. Montréal: Sankoff and Thibault (1977, 1980) and unpublished papers (Thibault and Sankoff 1997; 2004 with Wagner) ii. Québec: Canale, Mougeon, and Bélanger (1977) iii. Ottawa–Hull: Willis (2000) iv. Chicoutimi–Jonquière, Saguenay: Renaud and Villeneuve (2008) v. Televison interviews (formal): Bigot (2011) vi. Montréal pilot study: Rea (2014)¹⁵ vii. Televison interviews (informal): Villeuve (2016) b. Ontario, Canada i. Welland, Sudbury Rayside-Balfour: Canale, Mougeon, and Bélanger (1978) ii. Sammons, Nadasdi, and Mougeon (2015) iii. Béniak and Mougeon (1989) iv. Ottawa–Hull: Willis (2000) v. Immersion students in Toronto: Knaus and Nadasdi (2001) c. Western Provinces, Canada i. Alberta: Papen (2016) ii. Mitchif French, Alberta French, and Manitoba French: Hallion Bres (2006) iii. Manitoba French: Hallion (2000) d. New England, United States of America i. Vermont: Russo and Roberts (1999) ii. Massachusetts and Rhode Island: Stelling (2011) B. Acadian and Cajun French a. Maritime Provinces, Canada i. Prince-Edward Island: King and Nadasdi (2005) ii. New Brunswick: Roussel (2016); Balcom (2008); Péronnet (1989) iii. Nova Scotia: Gesner (1978) iv. Saint-Pierre-et-Miquelon: Brasseur and Chauveau (1990) v. Newfoundland: Brasseur (2000)
¹⁵ The italicized studies are not strictly speaking variationist studies but adopt a more descriptive approach.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
225
b. Louisiana, United States of America: i. Papen and Rottet (1997) ii. Picone and Valdman (2005)
10.3.2 Factors of influence 10.3.2.1 Sociolinguistic factors Gender, age, and socioeconomic status are thought to be the three basic macrocategories in sociolinguistics (Sankoff 1988: 902). Their study is crucial to understand whether a variable is undergoing change (Labov 2001). Willis (2000: 104) concludes that sociolinguistic factors play a more important role in conditioning auxiliary choice than linguistic ones, indicating the ‘neutralization in the meaning of avoir and être in composé contexts’. However Russo and Roberts (1999: 67) establish that external social factors do not play a significant role in the replacement of être by avoir in Vermont French, which indicates ‘a later stage of linguistic change, supporting patterns found in the literature of language death’. Gender. According to previous studies on Montréal French (Sankoff and Thibault 1977), male speakers use the non-standard form (conjugation with avoir) slightly more often than women. This evidence would be in line with the first half of the Gender Paradox, observed by Labov (2001: 293), in which ‘women conform more closely than men to sociolinguistic norms that are overtly prescribed, but conform less than men when they are not.’ Age. Younger speakers are known to be a vector of language change (Labov 2001), and Sankoff and Thibault (1977) as well as Canale et al. (1978) both observed that younger speakers tended to generalize avoir slightly more than their elders, which would potentially indicate a change in progress at the time. However, Sankoff and Thibault (1977) believed that this change in progress was blocked by the pressure to conform to the standard. Willis (2000: 77) also observed more retention of être in older speakers in Ottawa–Hull and her results suggested a change in progress towards increased use of avoir with the être verbs in Ottawa–Hull. In contrast, in their study of Chicoutimi-Jonquière in the Saguenay region of Québec, Renaud and Villeneuve (2008) found that older speakers are further away from the standard, namely generalize avoir more often than young people. Stelling (2011) also found that New England French speakers below the age of 70 were most likely to use avoir, and first-generation speakers were the least likely to use avoir, along with the third generation.¹⁶ ¹⁶ Stelling (2011: 1f.) defines ‘Franco-Americans’ as the individuals whose families came from French Canada during the nineteenth and early twentieth centuries and settled in the northeastern United States.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
226
ˊ
Socioeconomic status / Level of education. According to previous studies (Sankoff and Thibault 1977; Willis 2000; Stelling 2011), speakers belonging to the lowest socioeconomic class select avoir more often than the others. Many studies make use of the socioeconomic scale from Thibault and Vincent (1990) that combines both occupation and education levels, so their individual effects become almost indistinguishable. Willis (2000: 19) however had two different scales for socioeconomic class and level of education, and demonstrates that less educated speakers tend to favour avoir more. Linguistic Market Index (LMI). Sankoff and Laberge (1978) developed an alternative way to look at socioprofessional status, which instead involves examining the importance of the standard language in one’s economic life: In many communities, people steer their speech toward more standard varieties than would be predicted of their socioeconomic status if their occupation warrants it. Indeed, this is so commonplace that some sociolinguists have argued that the extent to which the standard language variety is valued in people’s daily life—that is, their position in the linguistic market—plays a far greater role in shaping patterns of language variation and change than their class background (e.g. Sankoff & Laberge 1978). (Schilling 2013: 48)
Sankoff and Thibault (1977) noticed that the LMI was one of the best predictors to explain auxiliary alternation: the use of avoir correlating with lower use of the standard language in their economic life. Contact with English and with other French dialects. While some studies have shown that auxiliary alternation exists in a few French-speaking regions of Europe (see §10.2.2), most studies on auxiliary alternation in French have been carried out in primarily English-speaking communities. Since HAVE is the sole auxiliary verb of the present perfect tense in English, many linguists have tried to elucidate whether structural borrowing could be at play. In Ontario French, Sammons et al. (2015) show that English bilingualism is actually a significant factor of influence for avoir selection with aller in the past infinitive (viz. avoir allé), and Béniak and Mougeon (1989) observed that language restriction played a significant role: highly restricted users of French showed a preference for avoir. Since Canale et al. (1977b) suggested that ‘structural interference may take place in non-optimal areas of grammar where language internal simplification processes may already be at work’ (Canale et al. 1977a: 3), they believe that the use of the English auxiliary HAVE ‘may have served to reinforce or accelerate the leveling process that was already underway in Ontarian French’ (Canale et al. 1977a: 3). In contrast, Willis (2000) showed that fluency in English did not have an influence upon a speaker’s auxiliary selection in Ottawa–Hull French, and Knaus and Nadasdi (2001) noted that ‘while the number of tokens available for
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
227
individual verbs are low, we see once again that the immersion students’ behavior for the variable is “native-like” ’. The results of Stelling (2011: 11) ‘lend support to the notion that contact with varieties of French other than the source dialect’ disfavours the use of avoirgeneralization: informants who had French only as a ‘foreign’ language used avoir with much lower frequency (18%) than those who had bilingual schooling (42%–46%). Attention to speech / registers. Bigot (2011) aimed to verify whether Québécois French had its own distinctive grammatical standard, irrespective of pronunciation. His corpus is composed of televised journalistic interviews from a famous newsmagazine¹⁷ on Radio-Canada focusing on current affairs. He analysed fourteen variables in total, including the generalization of avoir with Ê-verbs. He found that only 5% (7/128) of his Ê-verb tokens surfaced with avoir. He concludes his paper by stating that standard oral Québec French thus resembles the Bon usage of Grevisse and Goosse (2008) very closely, and recent work by Villeneuve (2016), on televised interviews discussing private matters, corroborates this result regarding auxiliary alternation. However, Stelling (2011) shows that daily use of French was the only ‘frequency of use of French’ factor that favoured use of avoir as an auxiliary verb in the speech communities of Rhodes Island, where French is now used solely between family members and friends. For Stelling, this would seem to indicate that attention to speech plays a key role.
10.3.2.2 Linguistic factors Lexical effect. Sankoff and Thibault (1977: 96) noted that the various Ê-verbs did not select avoir in the same proportions, and created a ranking of verbs based on the selection rate of avoir. In an unpublished paper from 1997, Thibault and Sankoff conclude that in Montréal French, the distribution of avoir and être was both lexically and socially determined, partially semantically motivated, but was also to a considerable extent arbitrary. In their study of Chicoutimi-Jonquière, Renaud and Villeneuve (2008) found a total of 24% of auxiliary alternation in the speech of eighteen speakers, which varied, among other things, according to lexical verbs: être was always used with décéder, mourir, naître, and aller, but avoir was used the most with passer, sortir, rentrer, (re)tomber, and monter. In their study of auxiliary alternation in the speech of French-speaking students from Ontario, Canale et al. (1977a) compared their results with the auxiliary selection patterns of students from Québec City. They found that the ranking of the verbs which selected avoir in Québec City was roughly the same as the one in ¹⁷ Le Point, which aired from 1983 to 2006, explored the news in depth with interviews and documentary reports.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
228
ˊ
Ontario, but the Québécois students displayed a consistently lower selection rate of avoir. The percentages of AVOIR use are given first for the Québec students, followed by those for the Franco-Ontarians: passer (100%, 100%), descendre (67%, 91%), tomber (62%, 92%), rentrer (50%, 83%), monter (33%, 100%), retourner (33%, 75%), rester (used in the sense of ‘remain’, 33%, 82%), sortir (29%, 74%), venir (24%, 63%), arriver (10%, 60%), partir (9%, 41%), revenir (7%, 43%), reflexive or pronominal verb (6%, 31%), aller (0%, 22%) (Canale et al. 1977a: 2). Canale et al. (1978) examined this variation within nine intransitive Ê-verbs in the spontaneous speech of 170 Franco-Ontarian students in French-speaking high schools of three Ontario communities. They also noticed that even some reflexive verbs surfaced at times with the auxiliary avoir in Ontario French. They observed that the verbs tomber, rentrer, rester, and sortir were conjugated exclusively with avoir, and the verbs venir and arriver showed strong tendencies in that direction (more than 50% of the utterances), while the verbs partir, revenir, aller as well as the reflexives verbs showed weaker tendencies of avoir selection (Canale et al. 1978: 48f.). In her study of Manitoba French, Hallion (2000: 365) records auxiliary alternation in verbs arriver (3%), partir (26%), tomber (59%), and venir (7%) with a lot of interspeaker variation, as well as in passer, déménager, retourner, rester and sortir. Interestingly, the verb rester, when meaning habiter ‘to live’ was always used with avoir. In Prince Edward Island Acadian French, King and Nadasdi (2005) observed 99% of avoir selection (avoir used almost always with entrer, (re)venir, devenir and arriver), but some variation in mourir and naître. For mourir, it was observed that the auxiliary selected the morphological structure of the past participle: if the avoir auxiliary is selected the past participle is mouri, but the être auxiliary appears alongside the past participle mort. King and Nadasdi (2005) also detect that those two conjugations of mourir denote two different aspectual interpretations: use of avoir is reserved for the act of dying and être for the state that results from it. Picone and Valdman (2005) echoed these conclusions and found that avoir may be used with all verbs, including intransitives and pronominals in Cajun French. Russo and Roberts (1999: 83) argue that in Vermont this process of replacement might even be starting to affect the verbs that have traditionally shown the strongest resistance to conjugation with avoir, such as the high-frequency verbs aller, venir, and arriver. What is clear is that every variety studied has a different number of verbs showing alternation, ranging from 9 to 21. Verb semantics. Some of the intransitive verbs under study have both ‘core’ meanings as well as ‘lexicalized’ meaning. The examples of ‘lexicalized’ meaning in Sankoff, Thibault and Wagner (2004) are sortir avec quelqu’un ‘date someone’ and venir au monde ‘to be born’ (lit. ‘to come into the world’). Willis (2000: 39) reports that this factor was significant to variant selection with the verbs partir, rester, and venir in Thibault and Sankoff (1997). In her study (Willis 2000: 62), the ‘core’,
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
229
motional, meaning of five intransitive motion verbs (rentrer, tomber, partir, repartir, and venir) and the ‘core’ continuing state interpretation of the stative verb demeurer showed the lowest percentage of avoir. Since different meanings of the same lexical verb yield different avoir-percentages, Willis (2000: 63) concludes that lexical semantics influences auxiliary selection, at least with these six verbs. Frequency of use. In terms of analogical change, it is known that more frequent forms are less likely to undergo change (Martinet 1969; Bybee 2010), and frequency has been shown to have an impact on auxiliary alternation in previous studies (Sankoff 1977; Russo and Roberts 1999; except not statistically significant in Canale et al. 1978, who used a frequency dictionary of European French): verbs such as aller, venir, and arriver very rarely surfaced with avoir in Montréal French in 1971 (Sankoff and Thibault 1980: 334f.). Knaus and Nadasdi (2001) found that the use of avoir of immersion students of French from Toronto is relatively low with the high-frequency verb aller,¹⁸ and fell ‘somewhere in between that of Francophones in the majority setting of Montréal and those of the minority communities within Ontario and Vermont’. This was also true of verbs partir, passer, rester and tomber, which also had ‘several’ tokens, but not of arriver for which the immersion students never made use of avoir. Kailuweit’s paper on the residua of the semantic motivation behind the use of two auxiliaries in contemporary standard French analyses Sorace’s ASH and concludes that other factors have to be taken into consideration to explain French auxiliation: change of location verbs which are at the top of the hierarchy, thus expected to take être categorically, also happen to be very frequent verbs— usually verbs of motion (Kailuweit 2015: 272). Transitive counterpart. Some of the verbs at issue also have transitive uses, and the existence of a parallel transitive use has also been considered to have a significant influence on auxiliary alternation because the transitive counterparts are homonymic and always require the auxiliary avoir. Variationist studies mentioned earlier (Sankoff and Thibault 1977; Canale et al. 1978; Russo and Roberts 1999; Willis 2000; Stelling 2011; Roussel 2016) have observed the correlation between the possibility of using some of these verbs transitively and the selection of avoir as an auxiliary when they are used intransitively. Amongst the verbs which have transitive equivalents, we can find: (re)partir ‘to start up x (again)’ (Québécism), (re)monter ‘to bring (back) up x’, ‘to assemble x (again)’, ‘climb (back) up x’, ‘to pull up x’, or ‘to wind up x’, (re)descendre ‘to bring (back) down x’,
¹⁸ Various English-speaking Canadian provinces offer French immersion programmes at elementary and secondary school levels for students who are not from French-speaking homes. In these intensive programmes, French is sometimes the sole medium of instruction for the first few years or might represent 50% of the medium of instruction. Typically, the proportion of teaching in English increases as the students get older. Notably, as Knaus and Nadasdi (2001) explain, the use of French of these students ‘is limited almost entirely to a scholastic environment where Standard French is upheld as a model’.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
230
ˊ
entrer ‘make fit x’ or ‘key in x’, (res)sortir ‘to pull out x (from a place)’ or ‘to take out x (again), (re)passer ‘to cross x (again)’, ‘to iron x’, ‘to spend (time)’, ‘to pass on x (again)’, ‘to (re)sit (an exam)’, or ‘to pass (an exam)’ (Québécism), rentrer ‘to bring in x’, or ‘to pull in x’, déménager ‘to move’, and retourner ‘to turn x over, inside out’ or ‘to return x’.¹⁹ Parallel pronominal use. The existence of a parallel pronominal usage (sortir vs se sortir ‘to get oneself out of, to escape’) has been tested by Willis (2000: 65), Roussel (2016) as well as by Renaud and Villeneuve (2008), who all found that the influence of this factor was statistically significant in the Ottawa–Hull, New Brunswick, and Chicoutimi–Jonquière data, respectively, in that it contributed positively to the probability of avoir-selection. Parallel adjectival use. Another factor that seems to have some impact on auxiliary alternation is the availability of the past participle of the relevant verbs for use as a resultative adjective. Sankoff and Thibault (1977) used a type of adjectival construction without copula for their tests, as exemplified in (14): (14) a. Après minuit, les pensionnaires sortis ne peuvent plus rentrer. ‘Boarders out after midnight cannot get back in.’ (Sankoff and Thibault 1980: 335) b. ? Un voyageur arrivé (d’un long voyage) est souvent content de retrouver son pays. ‘A traveller, arrived (from a long journey), is often happy to return to his own country.’ (Sankoff and Thibault 1980: 334) c. **Un enfant allé (à l’école après une longue maladie) est souvent content de retrouver ses copains. ‘A child gone (to school after a long illness) is often happy to rediscover her friends.’ (Sankoff and Thibault 1980: 334) Their results showed that the intransitive Ê-verbs that could have an adjectival use without copula were conjugated with avoir more often than the others. This is not entirely surprising since Labelle (1992) includes parallel adjectival use as one of the various tests for unaccusativity. The verb fondre is therefore unaccusative, even though it takes the auxiliary avoir, as in (15): (15) a. La neige a / **est fondu(e) pendant la nuit. ‘The snow has / is melted during the night’ b. La neige fondue, toutes les stations de ski ont fermé. ‘(With) the snow (having) melted, all the ski resorts closed down’ ¹⁹ It should be pointed out that the transitive use of tomber, meaning for ‘to beat’, ‘to seduce’, ‘to take off ’, or ‘to drop’ in popular Metropolitan French, is not attested in Québécois French (Sankoff and Thibault 1977: 99; Villers 2009: 1592).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
231
In some cases, the sentences were made acceptable by adding what Sankoff and Thibault name ‘adverbial support’, found in parentheses in (14b) and (14c) (e.g. d’un long voyage ‘from a long journey’ and à l’école après une longue maladie ‘to school after a long illness’). In addition, the influence of parallel adjectival use seemed to be greater than that of the existence of parallel transitive use in the findings of Sankoff and Thibault (1977: 336). However, they do not mention who participated in the grammatical judgement task, determining which verbs could be used as adjectives, and how it was put together. For Canale et al. (1978: 51), potential adjectival use with copula (exemplified in 16) was only acceptable with sortir, partir, arriver, rentrer, revenir, and tomber. Copula + participial stative adjective: (16) Marie est sortie / partie / arrivée / rentrée / revenue /tombée maintenant. ‘Marie is out / gone / here / back / fallen now.’ The ability or inability to be used adjectivally with the copula did seem to correlate with the percentage of avoir or être use in the cases of aller, arriver, rentrer, sortir, tomber and the reflexives (Canale et al. 1978: 52). However, this factor alone could not account for the high frequency of avoir with rester and venir, nor the high frequency of être with partir and revenir (Canale et al. 1978: 52). For Willis (2000), as well as for Roussel (2016), admissibility of parallel adjectival use (combined with transitive use for Roussel 2016) was the greatest determinant of variant choice. In order to code this factor, Willis (2000: 36) examined each of the être verbs in her Ottawa–Hull corpus to determine which past participles had also been used as adjectives (most were used with copula). Morphologically derived forms. According to Kiparsky (1973), morphologically derived forms often seem to be regularized more easily than non-derived ones. In their study of auxiliary alternation in Ontario, Canale et al. (1978: 54) test whether morphologically derived forms (e.g. revenir) behave in the same way as non-derived forms (e.g. venir) and, on the contrary, they found that morphologically complex verbs showed resistance to the spread of avoir: for instance, revenir is conjugated much less frequently with avoir than venir in their data. Again, this is most probably because change occurs first in unmarked contexts (Smith 1999b; Andersen 1990; Stein 1989; Timberlake 1977). For Canale et al. (1978: 54), pronominal verbs are themselves morphologically derived (e.g. sortir > se sortir, my example) and, in their study, they did not appear to be as easily affected by this levelling process as other non-derived verbs. Person and number. In their study of Chicoutimi-Jonquière, Renaud and Villeneuve (2008) found that auxiliary alternation varied according to the subject pronoun: avoir was most often selected with indefinite ce / ça, with the thirdperson plural, and with the first-person singular je.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
232
ˊ
With pronominal verbs in Manitoba French, Hallion (2000: 357) observes that avoir selection is most common with the first-person singular (15/21 tokens, 71%). With restricted speakers of French, she notes that with venir-tokens the third-person plural of the passé composé is always conjugated with avoir (Hallion 2000: 367). Tense and mood. Roussel (2016), with data from the Corpus de français acadien du Nord-Est du Nouveau-Brunswick (Beaulieu 1995), showed that the present tense of the auxiliary affected variant choice in favour of avoir. Renaud and Villeneuve (2008) found that verb tense had an influence on auxiliary alternation: more avoir-generalization with passé composé than with plus-que-parfait. Animacy. This factor was tested by Thibault and Sankoff in a 1997 paper (unpublished) that looked at the 1971 and 1984 Montréal data, but the results were only statistically significant in the case of partir ‘to leave’, whose inanimate subjects favoured avoir. This is consistent with the principle that linguistic change tends to occur first in unmarked contexts: inanimate and impersonal subjects are potentially more likely to favour avoir since they are less marked (i.e. they are the default option, are more common, basic, regular, etc.). Type of complement. Willis (2000) also empirically tested for the influence of the type of complement, based on various prescriptive grammar works of the 1800s. She tests whether verbs with a locational complement might favour avoir-selection, as predicted by Blondin (1823), while those with a following infinitive one might prefer être, as prescribed by Barthélemy (1839). Examples of locational and infinitival complements, taken from the Ottawa–Hull corpus (Willis 2000: 67), are shown in (17) and (18) respectively: (17) Locational complement Après ça quand j’ai sorti de l’hôpital, c’était fini ça. ‘After that when I left the hospital, it was over.’ (18) Infinitival complement Non, non, j’ai appelé le prêtre, il a venu me voir puis il a pris mon nom [ . . . ]. ‘No, no, I called the priest, he came to see me and he took my name [ . . . ].’ This study of auxiliary alternation in Ottawa–Hull French confirms the historical claims that a locational complement will favour avoir-selection and that an infinitive one will instead favour être (Willis 2000: 67). This section has discussed the various factors, both external / social and internal / linguistic, that have been shown to influence avoir-generalization in previous studies of North American varieties of French. What emerges from this survey is that patterns of auxiliary alternation in French are much more complex than what purely syntactic or semantic approaches are able to capture.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
233
Further analyses are required to test whether the same factors of influence on auxiliary alternation are still at play in newly collected data and whether additional ones can be uncovered with the help of more advanced statistical tools. The combination of these two research goals will be the focus of the following section.
10.4 Evidence from new Montréal French data 10.4.1 Sociolinguistic corpus of Rea (2016) Rea (forthcoming) revisits the now famous 1977 Sankoff and Thibault paper ‘L’alternance entre les auxiliaires avoir et être en français parlé à Montréal’ (see Sankoff and Thibault 1980 for the English version). After having analysed the Sankoff–Cedergren corpus of spoken Montréal French from 1971 including sociolinguistic interviews with 120 speakers (150 hours of recording), Sankoff and Thibault recorded variation in the auxiliary selection of 17 intransitive verbs (see Figure 10.1). On the sociolinguistic level, their results reveal that women, people who have greater exposure to the standard language, speakers of a higher socio-economic status, older speakers, and speakers who completed higher levels of schooling were all likely to display a more conservative use of the auxiliary, namely greater retention of être. Sankoff and Thibault state in the conclusion of their paper that (1977: 107) greater exposure to the standard would significantly inhibit the progression of the auxiliary levelling with avoir in the compound tenses of intransitive Ê-verbs. 100 90 80 70 60 50 40 30 20 10
% in 1971
r
pa ss er
e)
ste
(r
r re nt
re
r m
)to (r e
(r )e
be
r ag e
én
dé m
(r e)
(r e
)m
on
te
r
tir
e
)s or
dr
(r
es
r
sc en
(r
e)
de
to
ur
ne
rti r re
r
er
)p a
iv
(r e
ar r
ni ve
(r
e)
all e
r
0
% in 2016
Figure 10.1. Distribution of avoir-selection (in %) per lexical verb in 1971 and 2016
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
234
ˊ
Rea’s trend study re-examines the state of this alternation, given the great sociodemographic changes that have taken place in Montréal in the last forty-five years: the province of Québec has witnessed an important wave of nationalism (two referenda on independence, in 1980 and 1995) that strongly correlated with the officialization of French as the only language of Québec, with the promulgation of the Charter of the French language in 1977, also known as the controversial Loi 101. Moreover, in 1988, it became a legal requirement for all children to attend school at least until the age of 16. Rea built a corpus consisting of 48 sociolinguistic interviews (conducted in spring / summer of 2016) with native speakers of Montréal French (see Rea 2017, 2018a, 2018b, forthcoming, for a detailed account of how the corpus has been compiled).²⁰ Every interview also included grammaticality judgements on auxiliary selection. Rea (forthcoming) also investigates auxiliary variation in pronominal verbs, since there is no doubt that Montréal French presents certain irregularities in this regard, as in example (19), and since Sankoff and Thibault (1977) had excluded from their research this verb category, assumed to be more conservative in its auxiliary selection. (19) Faqu’ j’avais’ m’avais déguisé en, pour euh, le Père Noël, pour une famille, mais mes enfants étaient là aussi. (Guy) ‘So I had, I had dressed up as, for uh, Santa Claus, for a family, but my children were there too’. Intransitive verbs. Despite the rules stated in prescriptive grammars, a remnant (10.7%, namely approximately a third of the rate that was recorded in 1971) of auxiliary alternation, at least with the intransitive verbs, persists in the 2016 Montréal data, as shown in Table 10.3. Table 10.3. Distribution of auxiliaries in compound tenses of intransitive Ê-verbs in Montréal French in 1971 and in 2016 Distribution of auxiliaries in intransitive Ê-verbs in Montréal French corpus 1971
corpus 2016
variant
%
N
%
N
avoir être total
32.8 67.2
719 1474 2193
10.7 89.3
253 2121 2374
²⁰ Blondeau, Frenette, Martineau, and Tremblay (2012) have also studied Montréal French but by focusing on individual neighbourhoods, in particular Hochelaga-Maisonneuve and Saint-Michel. Since the aim of the Rea’s research is to reproduce as closely as possible the study of Sankoff and Thibault (1977), Montréal has been studied as a single speech community.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
235
The data for this table include most but not all the verbs studied by Sankoff and Thibault (1977): changer ‘to change’, demeurer, and entrer have been excluded from the analysis because of the very small quantity of recorded tokens. Even with the 1971 data, Sankoff and Thibault (1977: 99) did not want to discuss in detail the occurrences of entrer because they were too few, since most Montrealers systematically replaced it with rentrer. In order to express the iterative aspect of rentrer, Montrealers tend to use the construction re-rentrer, as shown in (20). (20) T’es-tu re-rentré dans le magasin? ‘Did you go back in the store?’
(Jean-François)
Additionally, the verb devenir, a canonical Ê-verb, had to be excluded from the present analysis because it did not show any alternation: all 42 tokens collected, including one of redevenir, were conjugated with être. Interestingly, (re)devenir had also been excluded from the 1977 Montréal analysis, conceivably for the same reason, even though it is not explicitly mentioned, but it was included in the Ottawa–Hull study and displayed a 12% rate of avoir-selection (Willis 2000: 55). The data in Figure 10.1 show the distribution of avoir-selection rates for the various lexical verbs studied in 1971 and in 2016. Interestingly, 46 out of 48 speakers had at least one token of auxiliary alternation, and the verb (re)déménager is the only one that shows an increase in avoiruse over time. It is also worth noting that déménager takes avoir in European varieties of French. Among the Ê-verbs that do exhibit auxiliary alternation in the corpus (shown in Figure 10.1), the distribution of avoir-selection appears to be quite widespread: the percentages span from 0.4% to 80.3%. With descriptive statistics, it appears that past infinitives, a perfective non-finite form and therefore neither a tense nor a mood, correlate with avoir-extension, and grammaticality judgements collected during the sociolinguistic interviews confirm these results. Moreover, these judgements indicate that auxiliary alternation is accepted even with aller and many speakers have indicated the acceptability of avoir with venir, revenir, and arriver. In terms of quantitative analysis, the development of the mixed-effects logistic regression statistical software package Rbrul (Johnson 2009) has improved the accuracy of statistical significance with respect to software like GoldVarb, which had been in use in variationist linguistics for over thirty years. It allows us to analyse which factors significantly influence linguistic variables, and in what direction and to what degree the effects are observed. Johnson (2009: 376f.) shows with simulations and real data that GoldVarb tends to overestimate ‘the statistical significance of external factors such as age and gender, whenever individual speakers vary in their behavior over and above the factors considered in an analysis’. This is because GoldVarb does not take speaker grouping into account and as a result, ‘GoldVarb necessarily ignores the grouping and treats
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
236
ˊ
Table 10.4. Rbrul results: statistical distribution of the significant factors on auxiliary alternation in the 2016 Montréal data by p values avoir-selection input probability total rate total N R² deviance significant factors: animacy pronoun before auxiliary socio-professional status intervening element (between auxiliary and past participle) hodiernal pronominal use speaker (48) lexical item (13)
0.158 10.6% 2341 0.726 885.465 p values: p. < 3.19e-06 p. < 1.62e-03 p. < 3.04e-03 p. < 5.49e-03 p. < 8.03e-03 p. < 0.0143 random random
each token as if it were an independent observation. This leads the programme to overestimate—potentially drastically—the significance of external effects, those of social factors like gender and age’ (Johnson 2009: 363). In order to fix this sort of Type I error, misidentifying a chance effect as a real effect, Rbrul can treat individual speakers as random effects which ‘takes into account that some individuals might favor a linguistic outcome while others might disfavor it, over and above (or “under and below”) what their gender, age, social class, etc. would predict’ (Johnson 2009: 365).²¹ For this statistical model to converge, the effects of both the speaker and of the lexical item have been set as random and factor groups which had very few tokens in them have been removed from the analysis. It is notably the case of secondperson plural subjects, past infinitives, past subjunctives, and some types of subjects (qui est-ce qui ‘who’, quelqu’un ‘someone’, and tout ‘everything’). For the same reason, other factor groups have had to be collapsed: lexical verbs and their iterative forms, as well as hodiernal tokens and ones in which the action took place within forty-eight hours. Unexpectedly, it was possible to collapse the middle and high socioeconomic classes into a single category, because their factor weights, i.e. the impact they had on the data, were almost identical (0.402 and 0.436 respectively). The total number of tokens analysed with Rbrul therefore went from 2374 to 2341. Remarkably, the statistical analysis reveals that most factors of influence listed in previous studies on auxiliary alternation in French (see §10.3.2) are not at play in the 2016 Montréal data. Table 10.4 reveals that the predictors of avoir-selection ²¹ However, Johnson (2009: 365) notes that this more conservative behavior has a trade-off: in some situations, Rbrul is more likely than GoldVarb to make a Type II error by failing to identify an effect that really does exist.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
237
are mostly linguistic (depending on whether the subject is animate, whether a pronominal clitic appears before the auxiliary verb, whether there is an intervening element between the auxiliary and the past participle, whether the action of the verb took place in the last twenty-four hours, and whether a verb can be used pronominally) apart from the socioprofessional status. Additionally, the fact that ‘age’ is not a significant social factor might suggest that auxiliary alternation in Montréal is a stable variation. When the factor group ‘tense’ was added to this model, it yielded a significant p value of 9.93e-03, but the model would not converge. It is probably therefore safe to assume that tense might also have a small effect in the alternation. Table 10.5 shows in detail how each factor group influences the data: inanimate subjects, when a pronominal clitic precedes the auxiliary verb, membership of the lowest socio-professional status, when there is an intervening element between the auxiliary, actions that have not taken place within twenty-four hours, as well as permissibility of pronominal use, are the only statistically significant factors that favour the use of avoir. The strongest predictor, (in)animacy, had been tested by Thibault and Sankoff in 1997 (cited in Willis 2000: 30) on the 1971 and 1984 data, but the results were only Table 10.5. Rbrul results: detailed statistical distribution of the significant factors on auxiliary alternation in the 2016 Montréal data by factor weights avoir-selection factor weights
%
N
animacy: inanimate animate
0.657 0.343
p. < 3.19e-06 23.1 9.4
216 2125
pronoun before auxiliary: pronoun before aux no pronoun before aux
0.716 0.284
p. < 1.62e-03 20 10.4
55 2286
socio-professional status: low mid-high
0.617 0.383
p. < 3.04e-03 14.7 8.9
707 1634
intervening element (between auxiliary and past participle): intervening element no intervening element hodiernal: not hodiernal hodiernal pronominal use: pronominal use allowed pronominal use not allowed pronominal use allowed only with en
p. < 5.49e-03 0.6 0.4 0.61 0.39 0.903 0.543 0.0828
17.7 10 p. < 8.03e-03 12.2 3.4 p. < 0.0143 24.4 11.8 0.5
186 2155 1933 408 745 526 1070
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
238
ˊ
statistically significant in the case of partir ‘to leave’, whose inanimate subjects favoured avoir. Willis (2000: 66) had also tested for this factor in her Ottawa–Hull study but had not found it to be significant. The second highest predictor, when a pronominal clitic precedes the auxiliary verb, can be illustrated in sentences such as (21). (21) Le vélo a glissé sur le côté, le guidon m’a rentré dans la cuisse. (Martine) ‘The bicycle slid on the side, the handlebars hit me hard in the thigh’ As to the presence of an intervening element between the auxiliary and the past participle, this factor was tested by Thibault and Sankoff in 1997 (cited in Willis 2000: 42) but the results were only significant for the verb rester. An example of an adverb acting as an intervening element in the 2016 data is presented in (22): ²² (22) J’ai tout de suite tombé sur un, un pâtissier français mais qui est très jeune. (Caroline) ‘I immediately met a, a French pastry chef but who is very young.’ The permissibility of pronominal use, which favours avoir-selection, reflects the fact that some verbs, like sortir, can be used pronominally, as in se sortir ‘to get oneself out of’, whereas some verbs, such as rester, cannot. Some other verbs can be used reflexively but exclusively when appearing with the pronoun en: for example, this is the case of aller—s’en aller ‘to go away’. Pronominal verbs. In total, 226 different lexical pronominal verbs surfaced in compound tenses in the corpus, out of which 110 have only been uttered once. They exhibit an average avoir-selection rate of 1.3% (17 tokens out of 1263). Rbrul revealed that the highest social predictors for avoir-auxiliary extension in pronominals are the age (p value of 2.30e-07, where older speakers favour avoir more) and the socio-professional status of the participants (p value of 6.33e-07, where once again membership to the lowest socioeconomic status favours the avoir variant more).
10.4.2 The crowdsourcing platform Les français de nos régions (2017) The blog Les français de nos régions launched by André Thibault and Mathieu Avanzi in 2015,²³ aims to map linguistic variation through postal codes by using
²² Negation adverbs were also included in this factor group, but polarity of the sentence was not found to be a significant factor of influence on the data. ²³ See https://francaisdenosregions.com/.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
239
Table 10.6. Distribution of auxiliary selection data from the Les français de nos régions survey based on education levels Education Levels Je suis monté / J’ai monté sur le toit de la maison elementary, secondary, professional school CÉGEP university Je me suis lavé/J’m’ai lavé les mains elementary, secondary, professional school CÉGEP university
avoir or both
être
total
%
11 43 165
11 42 244
22 85 409
50.0% 50.6% 40.3%
8 17 37
14 68 372
22 85 409
36.4% 20.0% 9.0%
crowdsourcing surveys to track various usages in the French-speaking world. In 2017, they were able to collect more than 5000 native grammaticality judgements in North America on the two pairs of (written) sentences Je suis/J’ai monté sur le toit de la maison ‘I have climbed on the roof of the house’ and Je me suis/ J’m’ai lavé les mains ‘I have washed my hands’. Of these, 516 participants were from the Montréal area (including 71 from the Laval suburb), and consisted of 369 women and 143 men (4 preferred not to reveal their gender). The corpus is also skewed in terms of education levels: 21 participants had only completed elementary, secondary or professional school, 85 had attended CÉGEP (higher school leaving examinations) or an equivalent,²⁴ and 409 had had university-level education. The survey asked whether the participants would say the two sentences with être, with avoir or with both indiscriminately. The results show that 42.4% of participants would use avoir or both auxiliaries with the sentence Je suis monté / J’ai monté sur le toit de la maison. As with the pronominal sentence, the results are much less balanced: only 12% of participants would use avoir or both auxiliaries with the sentence Je me suis lavé/J’m’ai lavé les mains. A more detailed analysis of social factors shows, in Table 10.6, that education levels play a role in the alternation. Statistical analysis confirms the results obtained by Rea (2018a), namely that the influence of education levels is highly significant (p value of 2.34e-04) for the use of avoir with the aforementioned pronominal sentence.
²⁴ Collège d’enseignement général et professionnel ‘general and vocational college’ (CÉGEP): Québécois students who wish to pursue post-secondary education in a Québec university must attend this two-year college before enrolling. CÉGEP can also prepare students for a technical profession (three-year programme).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
240
ˊ
10.5 Discussion By carefully examining the data analysed, extracted from both standard and nonstandard varieties of French, it seems obvious that auxiliary selection is not at all straightforward, as prescriptivists, semanticists, and syntacticians would have it. While semantic and syntactic factors can help us make sense of standard French split-intransitivity, they are not adequate or sufficient to explain the variation observed all over the French-speaking world. In the 2016 Montréal data, it appears that an analysis combining both semantics (animacy and hodiernal effects) and syntax (influence of the presence of a pronominal clitic before the auxiliary and of intervening material between the auxiliary and the past participle) must also incorporate social variables, since the membership to the lowest socio-professional status is a major predictor for avoir-selection. The fact that new statistical tools yield different results from previous studies in terms of predictive factors is not entirely surprising, but one key element is undeniable: auxiliary alternation has significantly decreased in Montréal French in the last forty-five years. However, this decline appears to evolve in the opposite direction of a functionally and morphologically ‘transparent’ trend (Lightfoot 1979: 121; Smith 1989c: 311) displayed by many Romance varieties to use a single auxiliary, namely HAVE, in the periphrastic tenses of active verbs. Since it is unlikely that linguistic structure is responsible for this change, one must look more closely at social dynamics, such as increased access to education as well as exposure to the standard language. One could even postulate that the progress of this alignment with the metropolitan standard is an expected consequence of the stigmatization to which Laurentian French has been subjected for decades, and thus might result from the linguistic insecurity of its speakers.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
11 Présent inclusif and passé composé à valeur de présent accompli in modern French and Occitan Ingmar Söhrman
11.1 Introduction The two constructions dealt with in this study are the présent inclusif (‘inclusive present’; henceforth PI) and the passé composé à valeur de présent accompli (perfect with the value of a fulfilled present action; henceforth VPA).¹ It would seem that the tense consistency is lacking in these two morphologically contrary constructions in French and Occitan (and other Romance languages) that share the idea of referring to an event that starts in the past and goes on until present time. Louis de Saussure (2013: 47) describes this illusory temporal inconsistency in the following manner. In a number of cases [ . . . ] tenses do not or not only refer to the time(s) they should refer to according to their intuitive semantics: some present tense utterances tell about a past or a future; some utterances with a past tense tell about a present or a future; some future tense utterances are about the present or the past.
The temporal relations of these constructions are not very often discussed, especially not from a comparative perspective, it would thus seem appropriate to undertake such a comparison now. There has always been a relation between the present and the perfect tenses, and the two usages of the two tenses that are described here illustrate this close temporal relation. The purpose of the present chapter is to discuss and differentiate what, at a quick glance, would seem to be two sides of the same coin. We will also carry out a brief comparison with some Germanic languages in order to see the syntactic differences from a typological perspective.
¹ I have seen the term antéprésent used, but this does not seem to have gained currency in French, so here it will not be used. An extensive overview of the perfect tense in Europe can be found in Drinka (2017). Variation and Change in Gallo-Romance Grammar. First edition. Sam Wolfe and Martin Maiden (eds) This chapter © Ingmar Söhrman 2020. First published in 2020 by Oxford University Press.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
242
It is well known that the present tense not only refers to the present moment but also to ‘situations which occupy a much longer period of time than the present moment, but which nonetheless includes the present moment within them’ (Comrie 1976: 37); such a situation may start long before the moment of enunciation and go on into the unknown future as well as include just one of these moments. It is also important to attend to the role of the temporal adverbial in expressing the present time. Adverbial modifiers of time could specify a temporal interval and include a certain amount of time that is part of the past or of the future. They might also indicate the moment when an event starts or ends as can be seen in a phrase like Il neige depuis vingt-quatre heures ‘It has been snowing for twenty-four hours’, where we know the beginning of the action but not the end (Riegel, Pellat, and Rioul 1994: 531). I will return below to the importance of these temporal adverbials that specify the limits of a tense that is as semantically open as the present tense. This semantic openness is often interpreted as a semantic emptiness of the present indicative (‘une vacuité sémantique du présent de l’indicatif ’; cf. Riegel, Pellat, and Rioul 1994: 529f.). With regard to the perfect, we may quote Comrie’s succinct definition: ‘[T]he perfect indicates the continuing present relevance of a past situation’ (Comrie 1976: 52), and the type of perfect that is dealt with in this article must thus be the ‘perfect of a persistent situation: a situation that started in the past and persists to the present.’ (Comrie 1976: 56; cf. Drinka 2017: 48). According to Drinka (2017: 53), these so-called anteriors could also be called present perfects and describe ‘situations occurring prior to reference time but having relevance to it’ (ibid.) although the process expressed is not necessarily finished. It can still go on as in the following sentence Annie habite Paris depuis vingt ans ‘Annie has been living in Paris for twenty years’ (cf. example 3), and Riegel, Pellat, and Rioul (1994: 536) add that the passé composé is never cut off from the present moment. From the moment of enunciation, the speaker regards it as an action of a certain psychologic closeness. Benveniste (1966: 244) sees the link between the two tenses as a ‘living’ link between the past event and the present where the passé composé revives the action, i.e. the tense used by someone who has witnessed an action and wants to tell us what has happened. Riegel, Pellat, and Rioul (1994: 534) summarize well the differences between the present and the passé composé in their relation to the enunciation. They consider the passé composé as a tense that permits the past action as a process that is finished at the moment of enunciation insisting on the difference between what is achieved and what is still taking place. As Jensen and Fleischman point out, tense usage has been extremely variable since the Middle Ages ‘Both Old French and Occitan are characterized by an
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
243
extreme variation in tense use. Writers do not hesitate to move abruptly from the present to the perfect or the imperfect and vice versa’ (Jensen 1990: 342), and he adds the following Occitan example that clearly shows this temporal variation (first imperfect and then present) where both verb forms refer past tense. (1)
Poiss mes aguait molt fraudulens qi Constantin facza prendent (Sainte Foi 539) ‘Then (he) laid a trap in a very deceitful manner which would capture Constantine.’2
But this does not mean that we do not find examples similar to modern usages, as we find more consistent uses in medieval French texts as can be seen in (2). (2)
Pere, fait Aucassins, ves ci vostre anemi qui tant vous gerroie et mal fait ; vint ans ja dure cette guerre ; . . . (Aucassins et Nicolette x, 37–8) ‘Father, says Aucassin, here you see your enemy who fights you so much and does bad [things]; this war has been going on [lit. goes on] already twenty years.’
Fleischman (1985: 851) points out that this tense switching is a stylistic resource. ‘[T]ense switching, in particular the insertion of present tense into past narration, functions in Old French as a strategy for “narrative subordination” ’. That this phenomenon is not something unique to Old French or Romance languages has been shown very clearly by Josephson (2006), who traces this from Hittite and other ancient Indo-European languages. Even in modern Russian with its two aspectual alternatives for every verb and only one past tense, the present can also be used in past and future narratives. The tense switching is thus quite common— at least in Indo-European languages. Temporal adverbs, such as depuis ‘since’, déjà ‘already’ or as in (2) ja ‘already’, play an important semantic role as they ‘entail the idea of a beginning, and are used for periods of time up to the present and beyond’ (L’Huillier 1999: 100). However, it is reasonable to limit this study to the contemporary uses in the two languages where tense variation is more stable—and to the written language (even if some examples reflect written direct speech), as this register tends to favour more complicated tense-time relations. To summarize the problem, let us look at three examples: (3)
Annie habite Paris depuis vingt ans.3 ‘Annie lives [has been living] in Paris for twenty years’
² The translations to English are transliterations, i.e. they represent the original sentence as closely as possible, which means that they are not very idiomatic. ³ Here I will not discuss the use or not of the preposition à with habiter, since they are both accepted and do not seem to have anything to do with the temporal usage.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
244
(4)
Un vétéran adopte le chien qui a travaillé à ses côtés en Afghanistan.4 ‘A veteran adopts the dog that has been working at his side in Afganistan.’
(5)
Je le connais depuis longtemps. II a toujours perdu son temps. ‘I know him since long ago. He has always wasted his time.’ (so this is what can be expected of him. i.e. a clear reference to the present) (L’Huillier 1999: 93)
These could be compared to the following Occitan examples (6–8). (6)
Annie viu a París dempuèi vint ans. ‘Annie has been living in Paris for twenty years.’
(7)
Eth Govèrn d’Aran a trabalhat pendent es darrèri ans entà melhorar es servicis de transpòrt intèrn ena Val d’Aran, ara, quan s’a ja hèt balanç des resultats deth darrèr an, se pòt afirmar qu’es cambiaments que s’an produsit en aguest servici an provocat uns boni résultats. (Aranese)5 ‘The Aran government has worked these last years to improve the local transport system in the Aran Valley, and now when the balance of this last year has been made, it is possible to confirm that the changes that have been carried out in this service have given good results.’
(8)
Lo coneissi de i a longtemps. A totjorn perdut son temps. ‘I know him since long ago. He has always wasted his time.’ (with a possible continuation now and in the future)
Both constructions, the present and the passé composé, are thus used with actions that start at a specific time in the past, but, as is illustrated here, the time limit/s/ of the action may be explicitly stated (as in examples 1 and 3), less precisely stated (as in 4 and 7) or left vague (as in 5 and 8). It may also be implied, and from a syntactic point of view, there is no difference between vague adverbial expressions, such as longtemps ‘long, a long time’, toujours ‘always’, and more precise ones, for instance, cinq ans ‘five years’: Je le connais depuis longtemps ’I know him since long ago (I have known him for a long time)’ vs Je le connais depuis cinq ans ‘I know him since five years ago (I have known him for five years)’. Significantly, the first sentence of example (5) is a case of PI, which gives the idea to what is said in the second sentence. Here the speaker refers to a person who has always wasted his time and implicitly will keep on doing so, and thus this last ⁴ https://wamiz.com/chiens/actu/veteran-adopte-chien-travaille-cotes-afghanistan-14371.html. ⁵ http://www.conselharan.org/eth-transport-public-intern-dera-val-daran-incremente-es-usatgerspendent-eth-2014/.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
245
sentence is supposed to be the obvious consequence of the first one. However, there is possibly a slight difference between the tense usage in (4) and (5). In (4) the activity has stopped, but as a result of this friendly experience the soldier had had with the dog, he decided to adopt the dog, so it is still relevant, while in (5) it is more the possible continuity of the action that is in the speaker’s mind. For that reason, it seems appropriate to see the passé composé à valeur de présent accompli as a concept that is divided in two kinds of subcategories, one referring to what is fulfilled or finished—terminé, (hence VPAT) such as examples (4 and 7) and one referring to what will or might probably continue—non-terminé, (VPANT) such as (5 and 8). This difference will be discussed in detail in a later section. The examples found in this chapter are taken from handbooks or Linguee,⁶ which present a good collection of modern French and which include examples found on the Internet and also from Jules Verne, Le Château des Carpathes (1892). The Occitan examples are, if not otherwise marked, kindly given by Xavier Bach, a native speaker of Occitan (Languedoc variety).⁷ If there is no indication of the origin of a French example, it is a creation of the author of this chapter but always checked with a native speaker. Although the analysis is based on real examples, it is a qualitative and not a corpus-based, quantitative study.
11.2 Theoretical considerations The classical Reichenbachian (Reichenbach 1947) model, which describes the relationship between the event, its point of reference and the enunciation, has later been developed and refined by Gosselin (2005: 31–40), along with other linguists (e.g. Klein 1994). Gosselin’s model successfully summarizes his findings from a previous work (Gosselin 1996), and also takes aspectual complications into consideration. He presents the temporal intervals, i.e. the ingressive moment (beginning) and the egressive one (end) of an event or process, as well those of a period or interval of reference and of an interval of enunciation and does not only describe the temporal relation between pastpresentfuture (cf. Söhrman 2013). Gosselin’s descriptions have made it possible to differentiate more clearly the past tenses in French (and other Romance languages; cf. Gosselin 1996, 2005, 2013). Gosselin discusses the intervals of an action, i.e. the ingressive and egressive
⁶ Linguee (https://www.linguee.fr/francais-anglais/) is a multilingual dictionary with millions of examples. ⁷ Since there is a vast variety of Occitan verb morphology I have limited this study to one dialect. An overview of this morphological variation is found in Kremnitz (1974).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
246
moments of an action on the temporal axis based on the three topic times, however vague, and short these moments might be. These intervals are: interval of enunciation (01, 02), interval of process (B1, B2) and interval of reference (I, II); see Figure 11.1.⁸ In the two cases described in this article the interval of reference goes from the beginning of the interval of process, which will go on until some (unknown) time in the future, while the interval of reference stops at the end of the interval of enunciation. Therefore, there is a question mark below B2 as its relation to 02 (end of the enunciation) is yet to be seen. However, there still exists the possibility that someone is making an utterance but knowing that the process will end (because Jacques in the example will move to Paris next month or something similar). In this case the question mark must disappear from the model, but this does not alter the model as such. It simply modifies this model as it is shown in Figure 11.1 when indicating (or just knowing) that B2 is going to take place at a determined moment in the future, although the relations between the intervals remain as described. The second construction that is discussed here could seem similar but somehow opposite, as is shown in Figures 11.2 and 11.3 which describe two possible and somewhat different situations since one (Figure 11.2) refers to an event or a
I B1 Jacques [begins to
01
II 02
B2
(utterance) → ?
live in Nancy (and goes on doing so)]
Figure 11.1. Jacques habite Nancy ‘Jacques lives in Nancy’
I B1
II B2a
01
02
(utterance) Anne worked until
→ her retirement
Figure 11.2. Anne a travaillé durement pendant plusieurs années jusqu’à sa retraite. ‘Anne worked hard until her retirement (B2a).’
⁸ In this article I will not discuss why Gosselin has called the different intervals 01–02, B1–B2 etc. Here I just take them for granted since this is what concerns us here, and I leave the discussion about what terms to use to the reader’s discretion; cf. Gosselin 1996.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
I B1
II B2b ?
Anne has worked
01
247
02
(utterance) Pr → ?
Figure 11.3. Anne a toujours travaillé durement [et donc il est fort probable que cela va continuer]. ‘Anne has always worked hard [and so it is very likely that will continue] (B2b).’
process that is terminated but that still has relevance for the present moment (VPAT) and the other one that is evaluated at the present moment but it is at the same time a presupposition or prediction (VPANT) of what is likely to happen as an almost deontic consequence of the first event. Thus, even if they show the same locations of the egressive B2, this is egressive only in the sense that the speaker has a limited interval of reference which does not mean that the process has stopped. Thus, in Figure 11.2 this B2 moment has taken place before the enunciation has begun, although it is connected as relevant to the ingressive moment of the interval of enunciation, while in Figure 11.3 the speaker has chosen to evaluate an event or a series of repeated events, indicating that there is a strong possibility that B2 will take place after the egressive moment of the interval of enunciation. In Figure 11.3 the uncertainty of when the end of the interval of process (B2b) will happen is indicated by a question mark. This differences between the two egressive moments of this interval is shown by two different symbols B2a and B2b. As we have just seen both situations relate to the interval of enunciation. In the first case (Figure 11.2) the process has been going on until very recently when it has changed for bad or for good, which is stated at the same time as the change is presented as a contrast to the present situation (Figure 11.2), VPAT, or as a consequence or evaluation based on earlier events (4 and 7). It does not necessarily mean that this ends before the interval of the enunciation, but that this is the moment when the evaluation of the process is made, VPANT. The process itself might go on for a long time. Sometimes it only suggests that, because of a certain process, we can assume a logical probability (Pr. in the figure) of the realization or continuous repetition of an event such as is shown in (5 and 8). It is thus a question of an evaluation based on ‘the total number of relevant equipossibilities’ (Audi 1999: 743) as Reichenbach (1935) has suggested. What we see in (5) and (8) are good examples of how a repeated event leads to a logical conclusion, while (9) is similar but ‘only’ refers to the present situation and the subject’s obvious qualification without necessarily regarding the probability of repeated actions. What this amounts to is a past situation with present relevance (cf. Plungian and van der Auwera 2006: 317).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
248
(9)
Elle a travaillé pendant 40 ans à la Banque Royale et [par conséquent]9 a de l’expérience dans le domaine des finances.10 ‘She has worked for forty years at the Royal Bank and [consequently] has some experience in the world of finance’.
Thus, theoretically, the main difference between these two constructions, PI and VPA, is that in the first case, the interval of reference is extended and passes the interval of enunciation while the two VPA cases refer to past experiences of a process that ends just before the beginning of the interval of the enunciation or that are considered by the speaker in a process of evaluation that takes place in the interval of enunciation as a possible consequence or probable prediction. The beginning of this process has started in the past, but it is extended and almost reaches or even passes into the posteriority of the interval of enunciation. Although in the VPA examples the verbs are either perfective or imperfective, the verbs in the PI examples in this study are mainly imperfective. Thus, I would suggest that this is mainly the case in this kind of construction, but with one exception. Repetition of actions is, of course, also possible with perfective verbs. Then it is the ‘endless’ repetition that gives the verb an imperfective aspect. This imperfectiveness of the present tense used for partly past events seems confirmed by the Bronckart and Sinclair study (1973) where French children under 6 years of age used the passé composé for past actions with end points but used the present for inherently durative actions in the past; the imperfect, on the other hand, was hardly ever used (Bardovi-Harlig 2012: 483).
11.3 Présent inclusif The PI indicates that there must be some kind of a logical starting point, but it does not have to be specified in the text, just implied in some way which shows a basic difference between the general atemporal usage of present tense for ongoing or general statements that do not have any detectible beginning (and/or end) since they are general and refer to atemporal statements that have to do with more or less eternal processes (10) or the idea of an ongoing activity regardless of the time reference (11). (10) Donc, le fer se dilate sous l’influence de la chaleur. ‘So, iron expands under the influence of heat.’ (Abbé A. Robert, Leçons de logique, [1914] 2009 Quebec, p. 115)
⁹ My insertion. ¹⁰ https://www.linguee.fr/francais-anglais/traduction/elle+a+travaille+pendant+4+ans.html.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
249
(11) Ils affirmaient, « avec preuves à l’appui », que les loups-garous courent la campagne, que les vampires, appelés stryges, parce qu’ils poussent des cris de strygies [, . . . ] (Verne 2013: 28) ‘They asserted “with supporting evidence” that the werewolves haunted the countryside, that the vampires, called stryges, since they like cried like stryges’ (Verne 2013: 28) From a logical point of view the Romance usage of present in similar cases seems quite adequate as they refer to an ongoing situation, but in Germanic languages perfect and present tenses may ‘compete’, which is never the case in Romance languages. In English the following two examples (12) and (13) are possible: (12) I live in London / I’m living in London. (13) I have lived in London for 10 years / now/. In the latter case (13) the adverbial now creates a preliminary end of the interval of process in order to evaluate the present situation that is still ongoing, and emphasizes a more retrospective stance, looking back over the past while the same sentence lacks this emphasise. ‘[M]odification with a temporal adverbial can shift the VP [verb phrase] from the description of an achievement (a punctual event) to an activity or ongoing process whose termination is not implied, [ . . . ]’ (Asher 2013: 15). In a northern Germanic language such as Swedish (examples 14 and 15 are translations of 12 and 13), we find an even more ambiguous situation as the perfect can be used, and often is, with reference to the present situation or process. However, in French and Occitan there is only one alternative—present (cf. 16 and 17). (14) Jag bor i London. (15) Jag har bott i London i 10 år /nu/. (16) Je travaille depuis 15 ans sur ces sujets, qui ne sont pas au départ ceux de mon travail d’historien, mais m’intéressent tout à fait.11 ‘I work since 15 years on these issues (I have been working on these issues for 15 years) which were (are) not in the beginning things of my profession as a historian but that interest me immensely.’
¹¹ htt3ps://www.linguee.fr/francais-anglais/traduction/je+travaille+depuis+10+ans.html.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
250
(17) Quin païs òrre ! E dire que dempuèi dos ans lo camin de fèr i passa. ‘What a horrible country! And to say that in two years the railway will pass here.’ The first sentence (12 and 14) with the verb in present tense is clearly indicating an ongoing process in the Germanic examples while the second sentences (13 and 15) without the adverbial now/nu remain ambiguous, and even more so in Swedish than in English. It may mean that this took place many years ago or that it reflects the present situation. The adverbial indicates the continuity of the process and should be seen as a reflection of how fast (or slowly) time goes by. However, while now in English and Swedish implies a continuity (13 and 15), without this adverb the action has perhaps come to an end or maybe it is just enigmatic (and possibly still ongoing). However, in French and Occitan there is no doubt. In (18) and (20) we find PI and in (19) the use of passé compose clearly indicates that this has come to an end, and it is probable that we are dealing with VPAT, and that the following sentence contains a conclusion or result such as, for instance, ‘so I know the place well’. (18) J’habite Grèz-sur-Loing depuis 14 ans. I live in Grèz-sur-Loing since 14 years.’ (19) J’ai habité Grèz-sur-Loing 14 ans. ‘I have lived in Grèz-sur-Loing for 14 years.’ (20) Mais, quelle qu’ait été sa constitution politique il est resté le commun habitat de diverses races qui s’y coudoient sans se fusionner, . . . ‘But whatever has been its political constitution it =this country, Transylvania] has remained the common habitat of different races who live side-by-side there without fusing, . . . ’ (Verne 1892: 10) In spoken French the passé simple and the perfect have merged into one tense— perfect as it used to be in Latin, albeit it is the compound form that won and not the synthetic form as was the case in Latin. In Occitan we find similar examples such as (21), but Occitan would or could use the passé simple (22), which shows that this tense is used also in the spoken language and in order to show the remoteness of an event as it would in southern Italian varieties and in many Spanish varieties, especially in Latin-America (Hendersson 2010). We thus see a return to the Latin system with just one morphological realization of the synthetic perfect tense (23). While in many parts of Spain only the perfect would be acceptable with a temporal adverb with a clear reference to the interval of enunciation or present moment (24). Catalan, possibly the closest language to Occitan, uses a periphrastic construction in these cases, anar (=go) +infinitive
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
251
meaning a past action, i.e. vaig pescar means that I fished not that I am going to fish as its equivalents would in modern French (or Spanish). (21) Vivi a Venés dempuèi 14 ans. ‘I live in Vienna since 14 years.’ (22) Visquèri 14 ans a Venés. ‘I lived in Vienna for 14 years.’ (23) Compré el periódico esta mañana. ‘I bought the newspaper this morning’ (24) He comprado el periódico esta mañana. ‘I have bought the newspaper this morning.’ We can thus see that French and Occitan are clearer than the Germanic languages in indicating the beginning of a process that goes from a moment in the past and goes on into present time if they want to indicate that it is still valid at the interval of enunciation and stays that way for some time. The end of this process might be known, but mostly it is not. It is just a way of saying that something started in the past and it is still going on.
11.4 Passé composé à valeur de présent accompli ‘The French PC [= passé composé] has two main realizations: (i) past [ . . . ] and (ii) perfective resultative. Resultative PC utterances convey the information that a resulting state of a past eventuality relevantly holds at S¹² [ . . . ] (Saussure 2013: 55). The very notion of a passé composé à valeur de présent accompli indicates that it is a past action in perfect but with relevance for the present situation (25); it is, therefore, a resultative utterance. (25) Les Roms en Roumanie ont souffert beaucoup en Hongrie, en Bulgarie et en République Tchèque ils ont beaucoup souffert pendant la période communiste des politiques qui les obligeaient à prendre des logements permanents et à travailler dans des usines. (Linguee)13 ‘The Roma in Romania have suffered a lot in Hungary, in Bulgaria and in the Czech Republic they suffered much in the Communist period from policies which obliged them to take permanent dwellings and work in factories.’ ¹² Moment of Speech. See Reichenbach supra. ¹³ https://www.linguee.fr/francais-anglais/search?source=auto&query=beaucoup+souffert+pendant.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
252
(26) Tous deux sont écrits par des êtres qui ont beaucoup souffert, qui ont, en quelque sorte, cru tout perdre, et ont donc la [ . . . ] pureté et le détachement nécessaires pour parler dignement de la souffrance14 ‘Both are written by beings who have suffered greatly, who have believed, in some way, that they would lose everything, and therefore have the necessary purity and detachment to speak with dignity about suffering’. However, when there is no clear reference to the present time in the statement, the passé simple can substitute for the passé composé as in example (27), although there seems to be a logical consequence in (26), donc ‘and therefore’, which generalizes this statement, and thus it is necessary to use the general tense passé composé with a reference also to the present time as it would seem that the logic makes the statement atemporal. It would then seem to be a kind of presupposed content of the future that is expressed in (26). (27) Ma vie jusqu’à cinquante ans fut simple. ‘My life up to 50 was simple’. (Folle de Chaillot 14; quoted by Togeby 1982: 413) Reinterpreting Reichenbach’s well-known scheme, Saussure (Saussure 2013: 56) suggests that [r]esultative PC can be schematized as follows, where ‘s’ stands for the resulting state, and ‘O’ for overlaps. ER; S & E ! s O S [This] formula reads: E[moment of event] is anterior to R [moment of reference], simultaneous to S [moment of speech] and implies a new resulting state (perfect) overlapping with S (Saussure 2013: 56). The resultative character of the verb is clear in examples 28 and 29, but as we have seen it is quite normal to use the preterite instead of a perfect in Occitan. (28) Jean a vécu plusieurs années au Maroc. Donc il connaît très bien le pays. ‘Jean lived several years in Morocco. Thus, he knows the country very well.’ (French) (29) Jean visquèt pendent d’annadas al Maròc. Tanben coneis plan lo país. ‘Jean lived several years in Morocco. Thus, he knows the country very well.’ (Occitan) ¹⁴ https://www.linguee.fr/francais-anglais/search?source=auto&query=beaucoup+souffert).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
253
VPA includes a final phase in that the comment ends at the point of enunciation, even if the action may continue (VPANT). Such comments tend to be made when things have changed or are about to change, which means that there is often an expression such as but, now explicit or implicit in the utterance which has an evaluative character. (30) Il a souffert beaucoup dans sa vie / mais maintenant (cela a changé)/ ‘He has suffered much in this life, /but now (this has changed)/’ It is not possible to use the passé simple in French, because this would give the impression that the person is now dead (31 and 32), and we are talking about his whole life and not making an evaluation of how things have been until now. (31) Il souffrit beaucoup pendant sa vie. ‘He suffered a lot during his life.’ (32) Jean a vécu une vie difficile—Jean vécut une vie difficile. (Il est mort) ‘Jean has lived a difficult life—Jean lived a difficult life. (He is dead)’ In this latter case, (32), the futur composé is also frequently used—Jean aura souffert beaucoup, ‘Jean will have suffered a lot (during his life)’, also called futur de bilan. The use of futur composé modalizes the sentence possibly in order to suggest an opinion. The French future tense also expresses possibility or probability without affirming the statement, however sure it may seem (see Sundell 1991; Saussure 2013: 60f.). This leads us to assume that a past event is not only resultative, but it also triggers another event or reaction which should be seen as a modal indication, i.e. what Saussure calls a ‘deontic-practical modality with epistemic future’ (Saussure 2013: 60). In these clauses the perfect/preterite ‘entails an epistemic modality of certainty [ . . . ] and as a pragmatic consequence, a deontic-practical modality about how to behave in the actual present in the perspective of the eventuality happening with certainty in the future’ (Saussure 2013: 60). We often give statements in the perfect indicating an assumption that they will have an effect not only on the present time (33–5) but also on the future as can be seen in the Occitan example (36). These statements can sometimes be hard to differentiate, but as can be seen in (36) where we would find a present also in French, and the repetition of action Luc achète turns it into a PI with a perfective verb. (33) Marie a toujours aimé les pêches. ‘Mary has always liked peaches.’ [and therefore might it be a good idea to buy her some?]
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
254
(34) J’ai vu monsieur Dupont presque tous les jours dans le parc. ‘I have seen Mr Dupont almost every day in the park. (35) Marc l’a vu tout à l’heure. ‘Marc has seen him just now.’ (36) Luc cròmpa una autò nòva cada an. ‘Luc buys a new car every year.’ In these cases it is quite clear that to the speaker it seems reasonable to conclude that the frequent repetition of events (Marie’s passion for peaches, Mr Dupont’s frequent walks in the park, and Luc buying a new car every year) indicates a certainty in (33) and strong probability in (34–6). In (35) there is proof of the person (= him) being around a little while ago, so he is probably around somewhere even now. And in (37) and in the Occitan example (38) we can assume that the speaker is now rested and fit to do something now or in the near future. As has been shown on several occasions the temporal adverbs play a very important role in our interpretation of the utterances. (37) Maintenant j’ai dormi. ‘Now I have slept.’ (38) Ara ei dormit. ‘Now I have slept.’ One also sees that the negation has an influence on the interpretation and temporal reference of a sentence. In the French example (39) and the Occitan (40) we can see that, with the negation, one could easily alternate between the present and the passé composé as both describe something that has not happened, as there is no clear semantic difference between has not happened and does not happen. (39) Ce n’est point une vapeur, c’est une fumée qui va se confondre avec les nuages . . . Et pourtant le burg est abandonné . . . Depuis bien longtemps, personne n’a franchi sa poterne qui est fermée sans doute et le pont-levis qui est certainement relevé. ‘That is no steam, that is smoke that will merge into the clouds . . . And still the castle is abandoned . . . For a long time nobody has passed its gate, which is closed without doubt and the drawbridge which is certainly raised.’ (Verne 1892: 31) (40) Lo dalús, quo es ‘na bèstia feramina que degun l’a jamai vista. [Limousin] ‘The dalús which is a horrible beast that nobody has ever seen.’ (Bec 1970: 149)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
255
11.5 Passé composé with future projection Apart from using the passé composé à valeur de présent accompli, the passé composé can also have a future projection (see Riegel, Pellat, and Rioul 1994: 535). This usage is related to the two syntactic constructions discussed here, but since its projection is the future and not necessarily the past or the present, which are the fundamental foci of our interest, the presentation of this third category will only be dealt with in a cursory way. With reference to Damourette and Pichon (1911–1936), Saussure (2013: 61) states that ‘this amounts to saying that epistemic futures also trigger a perspectival effect: they do not denote the state of affairs but they do represent the event of its acknowledgement in the future’. Saussure (2013: 57) sees this usage as an allocentric representation, i.e. ‘a representation which is attributed to some third-party (the speaker in the future, normally) viewpoint’. At this point in time a process or event has taken place and serves as a background process, linked to this imagined moment of enunciation. This interpretation seems very reasonable, but in order to connect this construction with my Gosselinian analysis, I will propose an analysis of this kind (see Figure 11.4), which does not gainsay Saussure’s theory. It rather supports it. This is a construction that competes with the futur antérieur (41) and (42). (41) J’aurai fait cela demain ‘I will have done / made this tomorrow’ (42) J’ai fait cela demain ‘I have done / made this tomorrow’ and presupposes a continuation such as et puis je peux faire autre chose ‘and then I can do something else’. Without going into detail, I present a preliminary explanation based on Gosselin’s intervals. I 01
B1 → ?
02
II B2
(utterance)
Anne began to write the novel
Figure 11.4. Anne a fini son roman dans un an, je crois ‘Anne finished her novel in a year, I believe’
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
256
In this case the ingressive moment of the interval of reference starts at the same time as the ingressive moment of the interval of enunciation, i.e. the focus lies on the moment of saying something and the projection or focus is directed towards the future, and this future needs a temporal adverbial such as demain ‘tomorrow’ dans un an ‘in a year’, etc., but the ingressive moment of this process is more uncertain. This uncertainty explains why there is a question mark and furthermore an arrow that indicates that B1 might start later, at any moment until right before the utterance. It could thus, as is the case in Figure 11.4, start before the interval of utterance or it could start at the same time or very soon after this point (see examples 43 and 44). (43) Demain vous me montrerez le tableau que vous avez (aurez) choisi. ‘Tomorrow you will show me the picture that you have (will have) chosen.’ (44) Françoise a (aura) terminé le travail dans deux jours, si tu le lui dis maintenant. ‘Françoise has (will have) finished the work in two days, if you tell her that now.’ My point here is simply to show that this construction is related to, but not part of, the subject of this study, and that it merits further research.
11.6 Conclusions As we have seen, both present and passé composé can be used with a projection towards the present time and the future, although the more exclusive use of passé composé to focus on a coming event is fairly restricted and is not really the theme of this article; it has only been dealt with superficially. Gosselin’s model helps us to understand better when and how to use the present that starts in the past as we have seen in Figure 11.1, where the egressive moment of the interval of reference lies in the present moment (interval of enunciation) while the egressive moment of process lies in the future, mainly at an unknown point of time. His model also helps us to see how the use of the passé composé normally indicates that we have an interval of process that lies in the past, but is still relevant to the interval of utterance (= present moment), and this can be seen in Figure 11.2. Nevertheless, the process need not end before the enunciation. It may well keep going on after the enunciation, but the speaker has chosen this moment in order to evaluate what is known of a process up until that moment. The speaker may modalize the statement by projecting the future as something epistemic that
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
257
will happen with a well-founded certainty, almost as if it were deontic—a value which it could assume (see Figure 11.3). The rare use of the passé composé with a future projection competes with the futur antérieur, an infrequent tense in its own right. As PI always refers to an ongoing action, it would seem impossible that this meaning could be expressed by the use of a perfective verb, while the VPA could be expressed by both an imperfective and a perfective verb. In both cases, however, a perfective verb can (only) be used if it refers to a multiple repetition or a habitual action. Thus, the two (or three) constructions and their internal relations can be seen and explained by this Gosselinian interval model that shows how the different semantic interpretations could and should be seen in relation to other events and how the tenses reveal the intentions of the speaker. The fact that the first construction seems syntactically simpler than the second one in Romance languages does not mean that the use of PI is generally the case (see the Germanic examples quoted). The second (passé composé) is more complex and has more options as to where the egressive moment of the point of process ends; its relevance for the present moment and/or the future is shown and discussed in Figures 11.2 and 11.4. We could thus summarize these tense usages in the following example (45) in a way that clarifies the usages. (45) Il a mené une vie très agréable, comme il habite Oxford depuis longtemps, et il a vu le monde entier, et maintenant il va prendre la retraite. ‘He has led a very pleasant life, since he has long lived in Oxford, and he has seen the whole world, and now he is going to retire.’
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
12 Future temporal reference in French and Gascon Aller / anar + infinitive periphrasis and structural transfer in the bilingual grammar Damien Mooney
12.1 Introduction This chapter examines variation and change in the expression of future temporal reference (FTR) in two typologically distinct varieties of Gallo-Romance which find themselves in contact in the region of Béarn, southwestern France. Gascon, a variety of the langue d’oc or Occitan, has largely been ousted from its traditional domains by a process of language shift to French, a langue d’oïl. In rural areas of Béarn, however, those native to the region display varying levels of Gascon– French bilingualism in the over-65 age group; the languages of these bilingual speakers will be the primary object of study in this chapter, focusing in particular on the verb morphology they use in French and Gascon to refer to future events. In both languages, the inflected future tense is used to express FTR, as in, for example, Fr. je ferai and Gsc. que harèi (‘I will make/do’). Alternatively, FTR may be expressed using the periphrastic future, or an analytic construction of the type seen in Fr. aller or Gsc. anar (‘to go’) + infinitive: e.g. Fr. je vais faire or Gsc. que vau har (literally, ‘I am going to make/do’). While FTR may, of course, be expressed in other ways, the analysis presented in this chapter will focus on these two variants, and will investigate the bilateral transfer of structural constraints on the distribution of these variants, and how these constraints may have resulted from long-term language contact. The chapter begins by providing an external overview of the sociolinguistic situations of both French and Gascon in Béarn (§12.2), before presenting the theoretical framework (§12.3), and a detailed discussion of research on the FTR variable (§12.4). The data collection and analysis is outlined in §12.5 and the FTR variable is modelled statistically for both French and Gascon in §12.6. Finally, §12.7 seeks to evaluate the implications of the quantitative analysis for a theoretical model that posits a single bilingual grammar where cognate variables in each language and constraints on variability are stored in the same abstract mental representation. Variation and Change in Gallo-Romance Grammar. First edition. Sam Wolfe and Martin Maiden (eds) This chapter © Damien Mooney 2020. First published in 2020 by Oxford University Press.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
259
12.2 Research context: French and Gascon in Béarn By the tenth century, the varieties descended from the Latin spoken in Gaul had become strongly diversified along regional lines, leading to the development of three broad dialect areas for Gallo-Romance. The most significant division within Gallo-Romance is between the langue d’oc (often referred to as ‘Occitan’) in the south, and the langue d’oïl, in the north. The third, eastern area, francoprovençal, shares features of both northern and southern dialects, but also has many distinctive features of its own. The modern langue d’oc area is commonly divided into six main dialects: Gascon in the southwest, including the Béarnais sub-dialect; central Lengadocian; Lemosin and Auvernhat in the north of the oc region; Provençal in the southeast, including the Nissart (Nice) sub-dialect; VivaroAlpin above the Provençal region (Bec 1963: 37). The Occitan standard language, as promoted by the Institut d’Estudis Occitans, aims to represent all dialects of the langue d’oc by focusing on linguistic features they have in common, and adopting a diasystemic approach, which functions as an abstract, standardized linguistic reference point and allows for the integration of the main localized linguistic forms in each dialect. The present study will focus on the variety of the sub-dialect of Gascon spoken in the region of Béarn and commonly referred to in French as le béarnais or in Gascon lo biarnés. While Gascon is the indigenous language of Béarn, French has been present in the region (at least in written form) from as early as the fourteenth century; until the turn of the twentieth century, however, Gascon remained the language of daily communication for the majority of the region’s habitants, at least in rural areas. The eventual shift from Gascon to French was supported by the Lois Jules Ferry (1881–6) and the resultant introduction of national free and compulsory education. The period immediately following the Loi Jules Ferry was not necessarily characterized by large levels of fluency and literacy in French among the general population. Moreux and Moreux (1989) note that although the first signs of disaffection with Gascon, in rural areas, appeared in the generation born during the First World War, the predominant daily use of French in Béarn began only with the generation born during the Second World War. At the turn of the twenty-first century, Moreux (2004: 29) stated that 16% of people over the age of 14 said they spoke Gascon well (fluent speakers); this amounted to approximately 40,000 speakers, over half of these being over the age of 60 and rural dwellers, and only 3.5% between the ages of 14 and 24. These younger speakers are most likely to have acquired the language via one of the limited options in the education system to do so, since the chain of intergenerational transmission has largely ceased since the Second World War. Since the 1980s, some public schools have offered bilingual education in French and Occitan and a number of private Escolas Calandretas offer immersion education through Occitan alone at primary school level, while there is one second-level
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
260
Collègi Calandreta in Pau, the region’s capital; the Calandreta schools are attended by approximately 2000 of the region’s children. Additionally, the association Centre de formation professionelle en langue et culture occitane offers classes and certification to adults across the southwestern region. We can therefore make a distinction between ‘native speakers’, for whom Gascon is their first language, and ‘new speakers’, who have learned Gascon in an education context and for whom French is their first language. These new speakers are commonly defined as speakers ‘with little or no home or community exposure to a minority language but who instead acquire it through immersion or bilingual education programs, revitalization projects or as adult language learners’ (O’Rourke et al. 2015: 1).
12.3 Theoretical framework: the bilingual grammar The bilingual speakers in this study speak two languages that are related but typologically distinct and this chapter will seek to investigate the extent to which the future tense systems of their languages interact, both in terms of the frequency of occurrence of different morphological variants that express FTR and in terms of the constraints which condition variability. Bilinguals are largely seen to be the locus of linguistic change because competence in two languages is a precondition for the adoption of material from one language into another (Hickey 2010: 9). Transfer will normally arise when a bilingual identifies a linguistic feature of the secondary system with one of the primary system and, in reproducing it, subjects it to the (variable) rules of the primary language (Weinreich 1968: 14). Flege (1988, 1990) formalized the mechanisms involved in phonetic and phonological transfer from one language to another in his Speech Learning Model. Flege (1995: 98) proposes that, during the acquisition of a second language, the process of ‘interlingual identification’ causes L2 sounds to be associated initially with L1 categories. This (initial) association of L2 sounds with L1 categories is the basis for the central tenet of the Speech Learning Model: the categories making up the L1 and L2 subsystems of a bilingual exist in a ‘common phonological space’ and so have the potential to influence one another (Flege 2007: 366). Within this model, phonemes in the L1 and L2 that correspond structurally are termed ‘cognate phonemes’ and are said to be equated with each other in the common phonological space, which may lead to bilateral transfer (see Mooney 2016, 2019 for an application of this model for Gascon–French bilinguals). The present study will examine the applicability of Flege’s model in the case of grammatical variation. If cognate phonemes, for example, are stored in a common phonological space, we might expect that cognate grammatical structures (such as inflected and periphrastic future morphologies) would equally be equated in the process of interlingual identification, leading to their storage in a common
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
261
grammatical space and thus allowing the grammatical systems of each of the languages to mutually influence one another. If this is the case, we might expect to see similar frequency distributions for the morphological variants used for FTR in both Gascon and French and, crucially, similar language internal constraints on variant choice.
12.4 Linguistic background: future temporal reference 12.4.1 Historical evolution of future temporal reference variants In both French and Gascon, the inflected future tense is descended from a late Latin infinitive verb + periphrastic construction (Comeau 2015: 344). In the transition to old French (and old Occitan), reduced reflexes of the auxiliary were reanalysed as suffixes (Fleischman 1982), as shown in (1). (1)
> >
*kantar-ˈajo Fr. (je) chanter-ai; Gsc. (que) cantar-èi (‘I will sing’)
The periphrastic future tense, which is composed of the present tense of French aller and Gascon anar and an infinitive verb, as in (2), did not originally function as an expression of FTR (in French). It originally functioned to indicate spatial movement but eventually grammaticalized as a future marker in the middle French period because of its near categorical co-occurrence with temporal adverbs (Fleischman 1982: 84; Comeau 2015: 345; Blondeau and Labeau 2016: 242), with the future meaning spreading from the adverbial to the verbal phrase (Comeau and Villeneuve 2016: 232). It is not clear when this reanalysis took place in Gascon, but the periphrastic construction is certainly productive in modern varieties. (2)
Fr. (je) vais chanter; Gsc. (que) vau cantar (literally, ‘I am going to sing’)
Early grammarians, such as Maupas (1607) and Antonini (1753), described variation between the two forms in terms of temporal distance, with the periphrastic future used for imminent events, or events occurring shortly after the moment of speech, and the inflected future used for distal events (Comeau 2015: 341); this has led to the frequent designation of the periphrastic future as le futur proche (‘the near future’). Flydal (1943: 39f.) argues that the change from imminence to futurity took place in the sixteenth century and that, by the seventeenth century, the periphrastic construction appeared without temporal adverbs and functioned to express futurity, bringing it into competition with the inflected future.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
262
(3)
Fr. La chanson que je chante demain; Gsc. La canson qui canti doman (‘The song that I will sing tomorrow’).
FTR may also be expressed in French and Gascon using the present tense, often referred to as the futurate present (FP), usually accompanied by a temporal adverb or adverbial phrase, as in (3). It was possible to use the present tense to express FTR as far back as in Classical Latin and therefore its use as a marker of futurity is common across the Romance languages (Comeau and Villeneuve 2016: 233).
12.4.2 Circumscription of the variable context The study presented in this chapter will focus on variation between the inflected future (IF) and the periphrastic future (PF) in French and Gascon, as these are the two primary morphological forms used in both languages to express FTR (Comeau 2015: 340). The futurate present (FP) is commonly excluded from studies of FTR (e.g. King and Nadasdi 2003; Blondeau 2006; Grimm 2010; Grimm and Nadasdi 2011; Wagner and Sankoff 2011; Roberts 2012; Comeau 2015; Roberts 2016) for two main reasons: (i) it occurs much less frequently than the other variants (IF and PF) in spoken French; (ii) it occurs almost categorically with future adverbials or adverbial phrases (Roberts 2016: 287). The frequency of occurrence of the FP variant is particularly problematic for studies with smaller corpora; indeed, Poplack and Turpin (1999) is the only major variationist study to include the FP, using as its basis the Ottawa–Hull corpus, which contains approximately 3.5 million words. As such, the futurate present may be considered a marginal variant, occurring less that 10% of the time, and there is evidence to suggest that the frequency distribution of the FP remains unchanged, at least in Laurentian French (Poplack and Dion 2009: 572). Therefore, if there is a change underway in the FTR system of French (and other Romance languages, such as Gascon), this primarily involves the replacement of the synthetic future tense (the IF) by a composite or analytic future tense (the PF) (Wales 2002: 75, §12.4.4). The present study will, for these reasons, not consider the FP variant, focusing instead on the distributions of and constraints on the IF and PF in both languages.
12.4.3 Previous studies of future temporal reference The FTR variable has frequently been the object of study in variationist research on French but, to my knowledge, has never been systematically examined in Gascon (or other varieties of the langue d’oc). There is a large body of work on FTR in Canadian varieties of French and, in almost all cases, the PF was found to be the majority variant, particularly for Laurentian varieties. For example,
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
263
Emirkanian and Sankoff (1986) and Zimmer (1994) found that, in Montréal, the PF was used 79% and 83.3% of cases of FTR, respectively, Poplack and Turpin (1999) found that the PF was used 73% of the time in the Ottawa–Hull corpus, and Grimm and Nadasdi reported a PF rate of 86.5% in Hawkesbury, Ontario. In Acadian varieties, the frequency distributions of the IF and PF were somewhat more variable: 75.2% (Chevalier 1996) and 77% (Chiasson-Léger 2014) in New Brunswick; 47.2% in Newfoundland and Prince Edward Island (King and Nadasdi 2003); 62.3% in the Baie Sainte-Marie, Nova Scotia (Comeau 2015). Studies on Metropolitan varieties attest much lower rates of occurrence of the PF: 27.3% (Kahn 1954) and 38.3% (François 1974) in Paris; Roberts (2012) examined the variable at four sites in northern France, using the Beeching Corpus, and found that the PF occurred 58.8% of the time to express FTR. Robert (2014) examined data from Martinique and showed that the frequency distribution was 72.3% PF and 27.7% IF. In the southwest of France, Gudmestad and Edmonds (2016) showed that native speakers used the PF most often (44%), followed by the IF (38.1%), and the FP (17.9%). All of these studies examined speech data and it is worth noting that in written sources the frequency distribution of the variants is markedly different. For example, Lesage and Gagnon (1993) reported only 3% PF use in their Québec French dataset, while Wales (2002) found 10% PF in his study of the regional newspaper Ouest-France. Early grammarians (e.g. Maupas 1607 or Antonini 1753, as cited in Comeau 2015: 341) frequently described variation between the IF and PF as being influenced by the temporal distance of the event: the PF was seen to be used for proximate events, while the IF was seen to be used for distal events. More recent studies of Laurentian French have shown the linguistic system underlying the variation observed to be quite different from these early descriptions; in Laurentian French, sentential polarity has been consistently shown to be the strongest factor governing variation between the PF and IF, with the IF favoured in negative constructions and the PF favoured in affirmative constructions. This polarity effect has been shown, by the studies cited above, to exert a much stronger effect on variant choice than temporal distance in Laurentian French. Studies of Acadian varieties, however, have shown the polarity effect to be absent and the temporal distance effect to significantly constrain the variation, broadly in line with the descriptions of early grammarians (King and Nadasdi 2003; Comeau 2015). While other factors have been shown to influence the choice between IF and PF in varieties of French (see §12.5.3), the bipartite distinction between varieties for which sentential polarity exerts the strongest effect and varieties for which temporal distance exerts the strongest effect on FTR is an important one, as varieties that show the effect of one tend not to show the effect of the other (Villeneuve and Comeau 2016: 320). Outside the Canadian context, Roberts (2012: 96) has shown the polarity effect to be significant in northern varieties of metropolitan French but, in contrast, Villeneuve and
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
264
Comeau (2016: 314) found only temporal distance to constrain the variation in Vimeu (northern France); in Martinique French, the FTR variable was shown to be influenced by the temporal distance factor, showing a systemic pattern akin to that of Acadian French (Roberts 2016). In the literature on Canadian French, in particular, the increasingly low rates of occurrence of the IF have frequently been interpreted as evidence of a linguistic change in progress by which the PF is ousting the IF and will eventually replace it in speech. For example, Emirkanian and Sankoff (1985) reported that the inflected future occurred at a rate of 21% in Montréal, while Zimmer (1994) reported the rate of the IF to be lower, at 17%. Some studies have, however, examined individual speakers’ linguistic behaviour in real time in Montréal (Blondeau 2006; Wagner and Sankoff 2011): both of these studies noted that for some speakers, their behaviour over time is in the opposite direction of the community pattern in that they increase their use of the IF as they get older. This pattern of age-grading is interpreted by the authors as indicative of the IF’s formal stylistic value; this interpretation is further supported by the fact that age-grading of this type was only found for speakers with a higher socio-economic class (Sankoff and Wagner 2006: 212). In metropolitan French, Roberts (2012) found that all speakers except those with a university degree favoured the use of the PF, again lending weight to the argument that use of the IF is associated with higher socioeconomic classes and, by extension, with more formal stylistic registers. These real-time findings suggest that change during the lifetime, indicative of agegrading, appears to act as a ‘brake’ (Wagner and Sankoff 2011: 305) on the progression of the synthetic to analytic linguistic change, rather than a reversal of the change, because the general community pattern is towards a reduction in the use of the IF over time. Sankoff, Wagner, and Jensen (2012: 107) tested this hypothesis further using a trend study, examining the same population at a more recent point in time, and found that ‘older speakers were much more likely than their juniors to use IF, and among these older speakers, the highest SPS group was considerably more likely to do so’, thus confirming the findings of previous studies. The effect of social class was extremely important here, as the age-grading change was only significant within social groups: ‘social class dominated, with younger speakers using fewer IFs than their elders in each SPS group’ (Sankoff, Wagner, and Jensen 2012: 111).
12.5 Methodology 12.5.1 Data collection The data sets used in this study of FTR in French and Gascon all consist of spontaneous speech elicited during 45–60-minute sociolinguistic interviews with
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
265
the author, a native English speaker and fluent speaker of both French and Gascon, and are structured as follows. (i) 2012 French Corpus: 8 older (65+) Gascon–French bilinguals (speaking in French), 5 female, 3 male; 10 middle-aged (30–50 years) French speakers (5 male; 5 female), of which one male informant was a Gascon–French bilingual; 10 younger (16–18 years) French speakers (5 male; 5 female). (ii) 2016 Gascon Corpus: 11 older (66–92 years) native Gascon–French bilinguals (speaking in Gascon), 6 male, 5 female. (iii) 2017 Gascon Corpus: 8 ‘new’ Gascon speakers, speaking in Gascon, who acquired ‘Occitan’ in educational contexts and who were native speakers of French, 5 male, 3 female. For the purpose of the study presented below, the 2016 Gascon Corpus and the 2017 Gascon corpus are considered together, though ‘speaker type’ (native or new) is examined as a factor that may influence variation in FTR.
12.5.2 Coding the dependent variable All instances of FTR were coded as either ‘inflected’ (IF) or ‘periphrastic’ (PF). Following initial token-extraction, there were 450 tokens of the FTR variable for French (2012 French Corpus) and 139 tokens for Gascon (2016 and 2017 Gascon Corpora). The following tokens were excluded, following an expanded adaptation of the methodology developed by Poplack and Turpin (1999) for excluding tokens with future morphology but which do not refer to a future eventuality (so-called ‘false futures’). • Habitual usages, e.g. Mon père . . . il parle, il va commencer en français, il enchaîne en béarnais, il va revenir en français (Speaker T) (‘My father . . . he speaks, he usually starts in French, he continues in Bearnese, and then he’ll go back to French’); • Timeless truths and hypotheticals (so-called ‘gnomic futures’), e.g. Que’t vas diser “oh que ço qui’m pouch tirar? (Speaker Z) (‘You’d say to yourself “what can I get from that?” ’); these are non-temporal because they are assumed to hold at all times (Wagner and Sankoff 2011: 280); • Tokens in which aller is used as a verb of movement as it is not clear whether these have present or future temporal reference (Comeau 2015: 352), e.g. Que plauva; ok, que vam anar dehens, que vam anar dehens (Speaker T) (‘It’s raining; OK, we’re going to go inside, we’re going to go inside’); • Reported speech of others, e.g, Après je pense que les parents ont choisi ces classes parce que les- ils- les é- ils étaient très peu nombreux, du coup ils se sont
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
266
dit ils vont être mieux suivi (Speaker N) (‘So, I think that the parents chose these classes because the- they- the—there were so many of them, so they said to themselves that there will be a better uptake’). • Fixed expressions, e.g. La bergogna de parlar patoès, . . . c’est ço qui van diser lo monde (Speaker AO) (‘The shame of speaking patois, . . . that’s what people usually say’); • Imperatives/pseudo-imperatives, e.g. Non, moi, j’aimerais bien le faire (Author) (‘No, no, I’d like to do it’ [travel to Montréal]), vous allez le faire ! (Speaker J) (‘You will do it!’); • Hesitations or false starts, e.g. N’auri pas . . . aquiu que dit . . . dab lo parlar mielhoù sus Auloron mes lo parlar d’origina, qu’ei . . . , qu’ei aute causa (Speaker AR) (‘I will not have . . . here they say it . . . with the Oloron dialect, but my original dialect, that’s . . . that’s something else entirely’). Following the exclusion process, 217 tokens of the FTR for French and 91 for Gascon were retained for analysis; this is due to an extremely high rate of future morphology being used with habitual aspect in French in the 2012 French Corpus.
12.5.3 Independent variables This section presents the linguistic (§§12.5.3.1–12.5.3.10) and extralinguistic (§12.5.3.11) factor groups included in the variable rule analysis in order to determine the variable effect that they have on variant choice. Not all factors are included in all analyses, due to collinearity between factor groups and / or low token counts within factor groups. For the statistical analyses presented in §12.6, the factors included in the modelling process are made explicitly clear. The linguistic factors presented below have been shown in previous studies to have a variable effect on variant choice, not a categorical one (King and Nadasdi 2003: 324).
12.5.3.1 Sentential polarity From the earliest studies of FTR in Laurentian French (e.g. Deshaies and Laforge 1981), sentential polarity has been consistently shown to be the most significant linguistic constraint on variability in FTR, with the IF occurring almost categorically in negative contexts and the PF occurring near categorically in affirmative contexts. In Blondeau and Labeau’s (2016: 254) examination of FTR in weather forecasts in Québec, however, the polarity constraint was shown to be absent, leading them to argue that it is this factor that clearly distinguishes ‘l’oral spontané’ from ‘l’oral préparé’ (see also Wales 2002). Equally, the polarity effect has been shown to be absent in some Laurentian varieties spoken in minority francophone communities such as Pembroke (Grimm and Nadasdi 2011: 174).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
267
In Acadian varieties, the polarity effect has been shown to be absent (King and Nadasdi 2003; Comeau 2015), with ‘temporal distance’ emerging as the most influential factor affecting variant choice (see §12.5.3.2). In Roberts (2012) study of FTR in metropolitan varieties (Brittany, Lot, Minervois, and Paris), polarity was shown to be a significant factor, showing northern varieties of French in France to pattern largely with Laurentian varieties in Canada, though the rates of occurrence were very different in metropolitan varieties (e.g. 62.5% IF in negative contexts; 61.5% PF in affirmative contexts). Therefore, the PF appears to be productive in negative contexts, in a way that it is not in Laurentian varieties (Roberts 2012: 101). The fact that a polarity constraint is present, however, raises the question of whether this constraint was present in the French varieties spoken by the Canadian settlers, even though it was never the subject of comment by the grammarians of the time (Poplack and Dion 2009: 574). All tokens were therefore coded as having either negative or affirmative polarity.
12.5.3.2 Temporal distance When the polarity constraint is absent, the most influential factor on variant choice is commonly temporal distance, with proximate events favouring the PF and distal events favouring the IF. In Acadian varieties of French, the PF has been shown to be favoured for events occurring within an hour from the moment of speech (Comeau 2015: 357) and up to week following the moment of speech; similarly, Robert’s (2016) analysis of Martinique French identified a temporal distance effect, but the PF was shown to be favoured up to a year following the moment of speech. Comeau and Villeneuve’s analysis of Vimeu French, from Picardy, also revealed a temporal distance effect, contrary to the findings of Roberts (2012), which found no temporal distance effect in northern French varieties in France. Equally, Gudmestad and Edmonds (2016: 269) found that in southwestern varieties of metropolitan French, the IF was favoured by events occurring more than a week from the moment of speaker and the PF was favoured for those occurring within the week. In Blondeau and Labeau’s (2016: 252) analysis of prepared speech in Québec, they found a weak temporal distance effect, with the PF favoured by events within the day or coming days, again highlighting the distinction between spontaneous speech and prepared speech in Laurentian French. In line with these studies, each token of future morphology retained for analysis was coded as an event expected to occur within the next minute, the next hour, the next day, the next week, the next month, the next year, or longer than a year, following King and Nadasdi (2003) and Comeau (2015). Tokens with a continuous or indeterminate FTR were also coded. 12.5.3.3 Imminence Until the sixteenth century, the PF functioned as a marker of imminence (see §12.4.1), in that it was used to refer to future eventualities that would occur within
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
268
moments of the speech act. By the seventeenth century, in many varieties of French, the imminent function became grammaticalized as a marker of futurity, giving rise to the variability we see today between PF and IF. Comeau (2015: 343) notes, however, that ‘as forms grammaticalize into their new functions, they can nevertheless retain vestiges of their earlier source meanings’ and, using data from Acadian French, he argues that ‘imminence’ is actually a stronger predictor of variation between the IF and PF than temporal distance. I thus also coded each French and Gascon token as imminent or non-imminent (following Villeneuve and Comeau 2016: 324f.), with imminent events expected to occur within the minute and non-imminent events to occur within the hour, day, week, month, year, and farther into the future. In Comeau (2015: 359), the imminent context was shown to account for 89% of PF variants; the IF was used 39% for nonimminent events. Similarly, the study of Vimeu French showed that this variety ‘marks imminence through periphrasis’ (Villeneuve and Comeau 2016: 331). The ‘temporal distance’ and ‘imminence’ predictors cannot be considered in the same statistical model as they are collinear in that ‘within the minute’ factor is the same as ‘imminent’ factor; the modelling process assumes that independent variables are not related in this way and so separate models are run with ‘temporal distance’ and ‘imminence’ respectively; the model that accounts most for the variation observed is reported in §12.6.
12.5.3.4 Verb stem regularity Blondeau and Labeau (2016: 250f.) showed that the IF is significantly favoured by irregular verb stems and that the PF is favoured by verbs that would have a regular stem in the inflected future. They interpret this result as a tendency for irregular forms to favour conservative variants, which has been demonstrated for other morphosyntactic variables such as the subjunctive (Poplack 2001) and the passé simple (Labeau 2015). In a written corpus, however, Wales (2002: 81) found that the regularity of the verb stem had no effect on variant choice. Nonetheless, tokens in each language were coded as ‘regular’ and ‘irregular’. 12.5.3.5 Certainty King and Nadasdi (2003) found that, in Acadian French, an eventuality deemed more likely to occur favoured the PF; they operationalized this constraint in terms of how certain a speaker was that a given event would take place. They did this by using a procedure of adding sans aucun doute ‘without any doubt’ to each token; if this addition rendered the token more certain, the token was coded as ‘uncertain’. If the eventuality was no more certain after the addition of sans aucun doute, it was coded as ‘certain’. Gudmestad and Edmonds (2016) coded certainty as (i) presence of an uncertainty marker (e.g. probablement pas ‘probably not’), (ii) presence of a certainty marker (e.g. certainement ‘certainly’), (iii) no marker.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
269
For native speakers, they found that the IF was most strongly associated with uncertain contexts, while the FP (and not the PF) was favoured in certain and unmarked contexts (cf. King and Nadasdi 2003). For parity with Comeau (2015) and Roberts (2016), I adopted the King and Nadasdi methodology and coded each token as ‘certain’ or ‘uncertain’.
12.5.3.6 Adverbial specification Poplack and Turpin (1999: 152) have showed that the FP is favoured by the presence of specific temporal adverbials (e.g. demain ‘tomorrow’), that the IF is favoured by non-specific temporal adverbials (e.g. plus tard ‘later’), and that the PF is favoured by the absence of an adverbial phrase. Other studies of Canadian and European varieties (King and Nadasdi 2003; Grimm and Nadasdi 2011; Roberts 2012; Gudmestad and Edmonds 2016) have found that adverbial specification does not govern variant choice, while Grimm (2010) has shown the PF to be favoured by the presence of both specific and non-specific adverbials (Comeau 2015: 354) in Hawkesbury, Ontario. Despite the fact that it is the FP that is favoured by the presence of specific temporal adverbials, and that the effect has never been found in a variety of European French, I coded the IF and PF data for whether a specific time adverbial was present, a non-specific time adverbial was present, or no adverbial was present. 12.5.3.7 Grammatical person The IF has been shown to be associated with vouvoiement (using the polite form of second person singular address, vous) and, by extension, with more formal styles in Canadian varieties of French (Poplack and Turpin 1999; Poplack and Dion 2009); the correlation between formal styles and higher socio-economic classes has also been reported (see §12.4.3). Roberts (2013: 146) found that impersonal il strongly favoured the selection of the IF in Martinique, with the PF favoured by other personal pronouns. In light of this, all tokens were coded for grammatical persons: je, tu, vous sing., il/elle, nous, vous, ils/elles for French; jo, tu, vos sing., eth/ era, nosautes, vosautes, eths/eras for Gascon (‘I, you, he/she, we, you, they); nonpronominal forms were coded as having a nominal subject. 12.5.3.8 Subject type Wagner and Sankoff (2011: 295) have argued that 1 nous and 2 vous, and nominals constitute a set of subject types that favour the IF; on the contrary, Blondeau and Labeau (2016: 252) found that, in Quebec, pronominal subjects favoured the IF and that nominal subjects favoured the PF. In order to investigate these claims further, each token was coded as having a nominal or pronominal subject. The ‘grammatical person’ and ‘subject type’ predictors cannot be considered in the same statistical model as they are collinear in that the ‘nominal’
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
270
factor is the same for each factor group and the ‘pronominal’ factor for the ‘subject type’ factor group collapses all of the grammatical person categories in the ‘grammatical person’ factor group; the modelling process assumes that independent variables are not related in this way and so separate models are run with ‘grammatical person’ and ‘subject type’ respectively; the model that accounts most for the variation observed is reported in §12.6.
12.5.3.9 Contingency Previous studies of Laurentian French have shown that the IF is favoured when the future event is dependent on the outcome of another (Poplack and Turpin 1999; Wagner and Sankoff 2011); in most cases, ‘contingency is indicated by a conditional si + present + future sequence’ (Roberts 2016: 296) and, as such, tokens were coded as being the main clause (apodosis) of the conditional sentence or not. 12.5.3.10 Presence of ‘quand’ The presence of quand ‘when’ has been shown to favour the IF in Canadian French varieties (King and Nadasdi 2003; Wagner and Sankoff 2011); the presence and absence of quand was coded for each token but there was only one token of quand in each of the French and Gascon data sets and, therefore, this factor group was excluded from the statistical analyses (see §12.5.4; §12.6). 12.5.3.11 Extralinguistic factors Each token of the FTR variable was coded for the speaker’s year of birth in order to investigate potential differences in the rate of use of each variant in apparent time. Additionally, the speaker’s biological sex was also coded for each token; while some research on Canadian varieties has suggested that female speakers are more likely to use the PF (Comeau 2011; Grimm and Nadasdi 2011), others have shown speaker sex not to constrain the variation observed (Poplack and Turpin 1999; Roberts 2012). For the French data, the bilingual status of each speaker was coded for each token (as ‘bilingual’ or ‘monolingual’); all Gascon speakers are, of course, bilingual speakers. The investigation of the effect of bilingualism on variant choice is based on the hypothesis that bilingual speakers may behave differently from their monolingual French counterparts (Comeau and Villeneuve 2016: 237); it is worth noting, however, that bilingualism is partially, but not completely, bound up with age in situations of language death. In their analysis of Picard–French bilinguals, Comeau and Villeneuve (2016) found that, in fact, these speakers behaved in the same way as French monolinguals from the same area (Vimeu), despite the fact that the periphrastic future is not productive in Picard (from a traditional perspective). This supports previous evidence that suggests that, even in the modern era, closely related varieties, such as French
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
271
and Picard, represent separate grammars for bilingual speakers (Villeneuve and Auger 2013) (see §12.7 for discussion). While Picard has adopted PF structure from French over time, its preference for the IF was shown not to favour higher rates of the IF in French for bilingual speakers. In line with the distinction between ‘new’ and ‘native’ speakers, outlined in §12.2, the Gascon data were coded separately for speaker type: ‘native’ (2016 Corpus) or ‘new’ (2017 Corpus).
12.5.4 Statistical analysis For each token of the FTR variable, speaker ID and the lexical verb were coded in the token file. Linear regression analyses were then carried out on the dependent variable (FTR) in the R environment (§12.3.3.3) using the Rbrul (version 2.3.2) text-based interface (Johnson 2008). Rbrul is a variable rule programme which evaluates the effects of multiple language-internal and/or social factors on a binary linguistic ‘choice’. The specific technique employed for all statistical models was mixed-effects regression. Mixed-effects models make a distinction between two types of factor that can affect a response (dependent) variable. First, fixed effects are factors that are replicable in another study, for example, speaker sex (male / female), adverbial specification, etc. Random effects, on the other hand, are factors drawn from a larger population which are not completely replicable (Johnson, 2009: 365), such as individual speakers and different lexical items. Coding for individual speakers allows the speaker variable to be included as a ‘random effect’ in mixed-effects statistical models which ‘takes into account that some individuals might favour a linguistic outcome while others might disfavour it, over and above (or “under and below”) what their gender, age, social class, etc. would predict’ (Johnson 2009: 365). The mixed model will only return a significant result for a given factor, such as speaker sex, if the effect is strong enough to rise above the inter-speaker variation in the model. Likewise, including lexical verb as a random effect takes into account the variation introduced into the model by individual words and only returns a significant result for language internal independent variables when their effect is large enough to outweigh inter-lexical-item variation. If individual speakers and individual lexical verbs are not included in the regression model as random effects, the results of the analysis will only be relevant for the individuals and words sampled and p-values may be too small and misleading to generalize to the larger population (Tagliamonte and Baayen 2012: 7). For collinear predictors that cannot be included in the same mixed-effects regression model (‘temporal distance’ versus ‘imminence’; ‘grammatical person’ versus ‘subject type’), the analysis was run separately for each combination and the model with the highest log likelihood, indicating the model which best fits the data, is reported in §12.6.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
272
12.6 Results 12.6.1 Frequency distributions Table 12.1 presents the overall frequency distributions for the IF and PF in French and Gascon. For French, these rates (48.8% IF; 51.2% PF) are much more evenly distributed than in studies of Laurentian French, where rates of occurrence of the PF are between 73% and 86.5% (see §12.4.3), with more recent studies showing a marked decline in the use of the IF. In Acadian French, the PF ranged from 47.2 in Newfoundland and Prince Edward Island (King and Nadasdi 2003) to 77% in New Brunswick (Chiasson-Léger 2014). In northern varieties of metropolitan French, Roberts showed the PF to occur 59% of the time; in the southwest, Gudmestad and Edmonds (2016) report 44% PF, 38.1% IF and 17.9% FP. The southern regional French results in this study are therefore much more in line with other studies of metropolitan French, the near 50/50 split between the IF and PF mirroring the results of Gudmestad and Edmonds in the same region of France. The Gascon data shows a preference for the PF across informants, indicating a preference for the analytic form in the obsolescent language. While the sample size is more limited for Gascon, it is common for obsolescent languages to undergo increases in transparency or in the use of more analytic than synthetic structures (Jones 1998: 249).
12.6.2 Modelling future temporal reference in French The results of the mixed-effects regression analysis for FTR in French are presented in Table 12.2. In this model, ‘temporal distance’ (and not ‘imminence’) and ‘grammatical person’ (not ‘subject type’) were included in the modelling process as the inclusion of these factors resulted in a higher log likelihood, and therefore a better fit of the model to the data. Two fixed-effect predictors were returned as statistically significant in the model: verb stem regularity (p > 0.01) and temporal
Table 12.1. Frequency distribution of IF and PF variants in French and Gascon French
inflected future periphrastic future total
Gascon
%
N
%
N
48.8 51.2 100
106 111 217
35.2 64.8 100
32 59 91
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
273
Table 12.2. Regression output for PF versus IF in French response variable = periphrastic N = 217
random effects: speaker; word
factor group
factor weight
factor
N
p-value
verb stem regularity
regular 0.704 108 0.00 irregular 0.296 109 temporal distance within the year 0.947 11 0.01 within the month 0.789 10 within the minute 0.784 23 within the week 0.560 5 within the day 0.542 4 continuous 0.332 48 indeterminate 0.201 68 within the hour 0.174 8 over a year 0.093 40 Non-significant factor groups were certainty (p = 0.183), grammatical person (p = 0.22), bilingual status (p = 0.223), adverbial specification (p = 0.294), contingency (p = 0.493), sentential polarity (p =0.584), speaker sex (p = 0.922), and speaker year of birth (p = 0.987).
distance (p < 0.05). For ‘verb stem regularity’, verbs with regular future stems favour the PF, while irregular stems favour the IF, mirroring the findings of Blondeau and Labeau (2016: 251). They interpret this result as part of a wider tendency for conservative morphological variants to occur at a statistically significantly higher rate with irregular verbs, as has been shown for the subjunctive (Poplack 2001) and the passé simple (Labeau 2015). As discussed above, previous studies of metropolitan French (e.g. Roberts 2012) have shown variation to be constrained by sentential polarity, and not by temporal distance, as is the case in most Laurentian varieties. In southern regional French, however, the second highest predictor is temporal distance, much like the results for Acadian and Martinique varieties. The PF is favoured for events that will occur within the minute (imminent events), within the month, and up to a year from the moment of speech. For events within the week and within the day, the IF and PF are equally probable, and the IF is favoured by continuous future events, those with indeterminate future temporal reference, events within the hour and, in particular, events that will take place over a year after the moment of speech. In the Acadian varieties of Newfoundland and Prince Edward Island (King and Nadasdi 2003), the PF is favoured for events that will occur within the day and within the week, whereas in the Baie Sainte Marie (Comeau 2015), the PF appears to be a marker of imminence, favoured primarily for events that will occur within the minute. In Martinique, Roberts (2016: 304) showed that all events that would occur up to a year following speech time favoured the PF to the same degree, with
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
274
only the most distal time context (over a year) strongly favouring the IF. This contrasts with Acadian varieties where the PF acts as a marker of imminence or proximity, with the IF favoured for events occurring after a week from the moment of speech. In the southern regional French data of this study, there appears to be a similar temporal split to Martinique French, with only events that occur in the most distal time context strongly favouring the IF. The ‘within the hour’ factor seems somewhat anomalous in this ranking, but anecdotal evidence from the interviewing process suggests that there is a tendency to use the IF for events that will occur within the interview period (usually 45–60 minutes in length) which contrasts with the use of the PF for events that are just about to happen (imminent events). Speaker sex and speaker year of birth were returned as non-significant in the analysis, showing that these extralinguistic factors do not appear to constraint the variation observed. Equally, the bilingual status of the speaker had no effect, indicating that bilingual Gascon–French and monolingual French speakers pattern in the same way, much like the bilingual Picard–French and monolingual French speakers examined by Comeau and Villeneuve (2016).
12.6.3 Modelling future temporal reference in Gascon The results of the mixed-effects regression analysis for FTR in Gascon are presented in Table 12.3. In this model, ‘temporal distance’ (and not ‘imminence’) and ‘grammatical person’ (not ‘subject type’) were included in the modelling process as the inclusion of these factors resulted in a higher log likelihood, and therefore a better fit of the model to the data. The following fixed-effects Table 12.3. Regression output for PF versus IF in Gascon response variable = periphrastic N = 91
random effects: speaker; word
factor group
factor weight
temporal distance
factor
N
p-value
within the month 0.947 2 0.01 within the year 0.789 4 within the minute 0.784 32 within the day 0.560 3 indeterminate 0.542 29 continuous 0.332 11 within the week 0.201 2 within the hour 0.174 7 over a year 0.093 1 Non-significant factor groups were certainty (p = 0.238), regularity (p = 0.352), speaker sex (p = 0.427), grammatical person (p = 0.514), speaker type (p = 0.887), and speaker year of birth (p = 0.896).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
275
predictors were excluded from the modelling process because of rank deficiency (uneven distribution of data): sentential polarity, adverbial specification, contingency. For sentential polarity, there were only two tokens with negative polarity (and 89 tokens with affirmative polarity); when this predictor was considered alone, it was returned as significant by the regression analysis (p = 0.0389), with the two negative contexts favouring the IF. Due to the uneven distribution of the data, however, we cannot interpret this result as a polarity constraint without considering a larger data set. Indeed, the size of the Gascon data set means that all results must be interpreted with extreme caution; it is hoped, however, that this analysis will constitute a useful point of reference for future studies of FTR in Occitan varieties. The regression analysis presented in Table 12.3 returned only one fixed-effect predictor as statistically significant: ‘temporal distance’ (p < 0.05). Noting the low token counts in some cells, we can make the following observations: events occurring within the minute (imminent events), within the month, and within the year favour the PF. For events with indeterminate future temporal reference and those occurring within the day, the IF and PF are equally probable, and for continuous events and those occurring within the hour, within the week, and over a year from the speech time, the IF is favoured (noting that the token counts per cell are particularly low for these factors). Despite these low token counts for some cells, it is striking that, with the exception of the ‘within the week’ factor, this constraint ranking is very similar to the constraint ranking presented for French in Table 12.2; in particular, the seemingly anomalous distinction between the PF being used for imminent (‘within the minute’) events and the IF being used for event within the hour is found to be significant in both languages. Speaker sex and speaker year of birth did not constrain the variation in any way, as was the case for the French data. The distinction between ‘native’ and ‘new’ speakers was not returned as a significant predictor, indicating that the grammatical systems of L1 and L2 Gascon speakers are not significantly different for the FTR variable.
12.7 Discussion: structural transfer in the bilingual grammar The regression models of FTR variability in French and Gascon have shown that, in both languages, the choice between the IF and PF is constrained by temporal distance, with the PF favoured by imminent and proximate events, and the IF favoured by more distal events and those occurring within the hour. Additionally, FTR variability in French is conditioned by the regularity of the verb stem, with the PF favoured by regular verb stems. The temporal distance effect appears markedly similar in both languages, though the constraint rankings are not identical. Poplack (2011) has compared the expression of FTR across the Romance languages
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
276
(French, Portuguese, and Spanish) and shown different constraints and constraint rankings to influence the choice of the PF across the languages, despite the fact that the languages are related. Differences between the language varieties may be taken to indicate different stages of grammaticalization of the PF: The PF has fully grammaticalized into a general futurity marker in some varieties (i.e., Laurentian varieties, which show little to no temporal distance effects), appears to be in an intermediate stage in others, clearly functioning as a marker of general proximity (i.e., PEI and NL varieties), and in a third group, retains functions associated with the early stages of grammaticalization, marking imminence (i.e. Vimeu French and Baie Sainte-Marie). (Villeneuve and Comeau 2016: 331)
The grammaticalization of the PF as a marker of imminence into a marker of futurity is not always temporally consistent across varieties; in French and Gascon, however, the proximate–distal distinction appears to hold, putting both varieties in the ‘intermediate stage’ cited above, much like varieties of French found in Prince Edward Island and Newfoundland (King and Nadasdi 2003), where the PF functions as a marked of general proximity, rather than uniquely as a marker of imminence. In contrast to previous studies of European French, the FTR system of southern regional French (in Béarn) does not pattern with that of the (supralocal) French monolinguals examined in Roberts (2012), where there was no effect of temporal distance but a strong sentential polarity effect. Comeau (2015: 360) speculated that ‘if temporal distance ever did condition the future variable in European French, it may have been neutralized in contemporary varieties’; this data from Béarn shows that the temporal distance effect has not been completely neutralized in European French. This finding also supports the hypothesis that the temporal distance effect was present in European varieties before colonization and that it may have been transplanted to Martinique and the Atlantic Provinces in Canada by the input varieties of the first settlers (Roberts 2016: 308). This, in turn, implies that the temporal distance effect has been lost from supralocal and Laurentian varieties, in favour of the polarity constraint, making southern regional French more conservative with respect to this variable than the spoken standard. In Villeneuve and Comeau’s (2016) examination of Picard–French bilinguals in Vimeu, they found that bilingual status did not appear for effect variant choice for the FTR variable; this contradicted their hypothesis that Picard–French bilinguals would exhibit lower rates of the PF because of Vimeu Picard’s general preference for synthetic (IF) forms for FTR. They note, however, that this lack of a bilingualism effect may be due to the significant rise in the use of the PF in Picard since the 1960s (Auger and Villeneuve 2013) due to transfer from French. Given the pervasive nature of the contact between French and Gascon, we might expect
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
277
them to display similarities in terms of the distribution of and constraints on the FTR variants. For the bilingual speakers, this would imply that the FTR systems of each of their languages are equated and function in the same way and that, for monolingual French speakers in younger generations, this is the French system that they acquired in the transmission process; the Gascon–French bilinguals in this study did not behave any differently from French monolinguals when speaking French, as evidenced by the non-significant result for bilingual status in the regression analysis (p = 0.223). When speaking Gascon, however, the bilinguals’ variant choice was not constrained by verb stem regularity, as it was in French, but the temporal distance constraint was similar in both languages, as we have seen. The overall frequency of the FTR variants appears different in French and Gascon, with higher rates of the analytic form in Gascon, but more Gascon data are needed to confirm this. What is not clear is whether the PF was adopted in Gascon due to pervasive contact with French (as in Picard) or whether it was pre-disposed to this particular synthetic-to-analytic change. In situations of language obsolescence such as these, it is not always possible to make a distinction between internal and external motivators for change, due to the lack of a monolingual control group (for Gascon). From the perspective of the Speech Learning Model, there is not sufficient evidence in these data to say that the FTR systems of French and Gascon are stored in a common grammatical space for bilingual speakers, resulting in convergence of their grammatical systems for each language. Indeed, the French system used by Gascon–French bilinguals and French monolinguals is the same but the Gascon system lacks one constraint used in French, indicating that the languages have separate grammatical systems and that variability in FTR is not constrained in the exact same way in both languages. Overall, the grammatical systems of French and Gascon are relatively aligned with respect to the temporal distance constraint on FTR choice, though we cannot posit that they share the same grammatical system for the bilingual speakers in this study or that the variants are equated as cognates in both languages, without exploring a larger Gascon data set. For these bilingual speakers, it does not seem appropriate, in the case of the FTR variable at least, to speak of a single bilingual grammar, in which the grammatical systems of French and Gascon are equated in the same abstract mental representation (a ‘common grammatical space’); the similarities between the grammatical systems, however, certainly warrant further investigation, ideally including other varieties of Occitan and southern French from other regions.
12.8 Conclusion This chapter has examined the expression of future temporal reference (FTR) from a variationist perspective in two related varieties of Gallo-Romance, nonstandard southern regional French and Gascon, a dialect of the langue d’oc.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
278
By examining both ‘native’ and ‘new’ Gascon–French bilinguals and monolingual French speakers, the analysis found that, in both languages, FTR variant choice is constrained by an effect of temporal distance, with the periphrastic future (PF) used for imminent and proximate events, and the inflected future (IF) used for more distal events. While the grammatical systems of both languages in contact shared the temporal distance effect, the French system used by both monolinguals and bilinguals also favoured the use of the PF for verbs with regular verb morphology in the future stem; this independent variable was not returned as a significant predictor of variant choice for Gascon. As such, the grammatical systems used for FTR in the contact language are not the same, contradicting to some extent the hypothesis that the variables are equated in a single bilingual grammar (by bilingual speakers) and that constraints on the variability in each language are stored in the same abstract mental representation, or a ‘common grammatical space’. The constraint ranking for the primary ‘temporal distance’ effect in each language is markedly similar; however, it warrants further investigation using an expanded data set in future studies.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
13 Mainland and insular Norman Pronoun sharing and pronoun sparing Mari C. Jones
13.1 Introduction Studies of language contact often include discussion of borrowing ‘hierarchies’ in which the presence of a specific type of loan is correlated with a likely sociolinguistic setting (cf. Thomason 2001: 70f.). Typical statements include, for example, that, in more casual contact situations, borrowing tends to be restricted to openclass items (such as content words) and, the more intensive the contact situation, the more likely it is that a language’s grammar and phonology will also be affected (cf. Haugen 1950: 224; Muysken 1981; Thomason and Kaufman 1988: 74–6; Matras 2007, 2009: 153–63). What follows focuses on the effect of language contact on personal pronouns which, as Thomason and Everett (2001: 301) highlight, are often described as ‘[p]erhaps the most commonly mentioned hard-to-borrow lexical feature’. Although disputing this claim, they suggest that the reasons behind it include the fact that personal pronouns ‘comprise a closed set of forms situated between lexicon and grammar; they form a tightly structured whole and are so deeply embedded within a linguistic system that borrowing a new personal pronoun, and in particular a new pronominal paradigm, would disrupt the workings of the system’. ‘Therefore’, they continue, ‘the argument goes, it is extremely unlikely that any language [ . . . ] would borrow pronouns’. Such arguments are by no means new: Weinreich (1964: 37) also commented that ‘[s]ome morpheme classes of a languages (like inflectional endings or pronouns) do indeed seem less hospitable to newcomers than others’, since ‘those classes are more self-sufficient in the face of cultural innovation’; cf. Backus (1996), Berk-Seligson (1986), and Bentahila and Davies (1995), who also list pronouns as relatively low on borrowing hierarchies. Nonetheless, as many linguists have pointed out, one need look no further than English to find an instance of a borrowed pronoun (cf. Thomason and Kaufman 1988: ch. 9.8; McMahon 1994: 204; Thomason 2001: 83).¹ Indeed,
¹ For a different view see Cole (2018). Variation and Change in Gallo-Romance Grammar. First edition. Sam Wolfe and Martin Maiden (eds) This chapter © Mari C. Jones 2020. First published in 2020 by Oxford University Press.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
280
.
Dixon (1997: 22) and Aikhenvald (2006: 19), among others, take care to distinguish between the borrowing of an individual pronoun, which is described as not so rare, and the borrowing of a whole pronominal paradigm, which seems to be considerably less common. Thomason and Everett (2001: 303–12) discuss studies that suggest pronoun borrowing may not in fact be as rare as might once have been suspected (cf. also Trudgill 1986; Woolhiser 2005; Mithun 2012: 15–36; Rose 2015). Matras (2009: 204–6) comments on how, in certain contexts, pronoun borrowing can be explained by ‘special circumstances’, such as when the pronouns in question are considered to be honorifics; and when borrowing occurs for purposes of levelling in order to match the system of categorization of the two languages that are in contact or to maintain a single categorization system for those two languages across the linguistic repertoire (cf. van der Voort 2000: 158; Voorhoeve 1994; Carlin 2006; Heine and Song 2010). Law (2014), following Jake (1994), argues that the nature of the pronoun may be relevant, with disjunctive pronouns more likely to be borrowed than clitic pronouns. Thomason (2001: 71) draws attention to the importance of typological distance between the languages in contact: ‘It is easier to introduce borrowings into typologically congruent structures than into typologically divergent structures’. This has also been highlighted by Weinreich (1964: 33): ‘The transfer of morphemes is facilitated between highly congruent structures’ (cf. also, among others, Vildomec 1971: 78; Thomason and Kaufman 1988: 52; Woolhiser 2005). This chapter aims to contribute to the growing body of literature on pronoun borrowing by presenting a case study from northern Gallo-Romance. It examines four different varieties of Norman, one spoken in mainland Normandy (specifically, in the Cotentin peninsula), which is in contact with French, and three in insular Normandy (the Channel Islands), namely Jèrriais (Jersey), Guernesiais (Guernsey), and Sercquiais (Sark), which are in contact with English (Jones 2015). The study’s aim is twofold: first, it seeks to make available original new data on the pronouns of mainland Norman (hereafter, MN) and insular Norman (hereafter, IN); second, to explore the effects of structural congruence on pronoun borrowing, the fact that Norman is typologically closer to French than it is to English making it an ideal testing-ground. Norman’s daily co-existence with, respectively, French and English has produced similar outcomes on the mainland and in the islands, namely that all varieties of MN and IN have undergone significant territorial contraction and speaker reduction and may now be described as obsolescent. Current estimates put speaker numbers at 29,000 in mainland Normandy (roughly 1% of the population); fewer than 1000 in Jersey (1%); a few hundred in Guernsey (0.5%) and some 10 individuals in Sark. Norman has, in both territories, witnessed the breaking down of its former diglossic relationship with the local superstrate language as similar social factors, including demographic movement,
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
281
intermarriage, and stigmatization, have contributed to the loss of its former strongholds, such as the family domain. According to Thomason and Kaufman’s typology (Thomason and Kaufman 1988: 83–95), MN and IN would both be described as existing in situations of ‘intense contact’, in which moderate to heavy structural borrowing would be expected. Many features of these varieties do indeed display evidence of contact-related change, and their different superstrates have caused MN and IN to diverge structurally and lexically (Jones 2015). It is important to clarify that French is not ‘in contact’ with Norman in the Channel Islands in the same way as it is on the Norman mainland. Although French served as the de facto standard language in the archipelago until the twentieth century, it always functioned as an exoglossic standard and, despite enjoying exclusive use in so-called ‘High’ domains from the Middle Ages up to the time when English began to predominate (Brasseur 1977; Jones 2001, 2008, 2015), it was never spoken there as a native language. For most contemporary speakers of IN, therefore, their linguistic relationship with French is akin to that which one would have with a ‘foreign’ language: being encountered—if at all—solely via the education system and having little relevance for their daily lives, although its structural similarity to Norman means that most speakers of IN can understand French reasonably well. All speakers of MN speak French fluently and usually as a second native tongue.
13.2 Methodology This study investigates whether the linguistic divergence between MN and IN observed in Jones (2015) may also be observed in the pronominal system. Specifically, it examines whether the close typological distance between MN and its superstrate, French, has led to pronoun borrowing from French into MN (this situation will be termed ‘pronoun sharing’) and whether the greater typological distance between IN its substrate and English, leaves the IN pronominal system more structurally ‘intact’, or ‘safer’ from the reach of contact (this situation will be termed ‘pronoun sparing’).² The data were gathered via free conversations conducted by myself and in the local variety of Norman. In an attempt to lessen the Observer’s Paradox (Labov 1972: 32), I was accompanied by a native speaker who was well known to the interviewees and who often took the lead in the conversation (Milroy and Gordon 2003: 75; Bowern 2010: 351). Since Norman is a non-pro-drop variety, it was predicted that pronouns would occur sufficiently frequently during such ² The sharing / sparing framework was first developed in an agricultural context (viz. land sharing / sparing) (Phalan, Onial, Balmford, and Green 2011). I am grateful to Lydia Collas for introducing me to this dichotomy.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
282
.
naturalistic speech to yield enough tokens for analysis. Interviews were conducted with 17 speakers of MN from the northern half of the Cotentin peninsula, 26 in Jersey, 41 in Guernsey, and 7 in Sark.³ The virtual cessation of intergenerational transmission of Norman in the Cotentin peninsula and its total cessation in the Channel Islands means that all speakers were aged over 60 and, for this reason, age-related variation is not considered. All speakers had been bilingual in the local variety of Norman and French (Cotentin peninsula) or English (Channel Islands) since their childhood. Norman still represented an everyday language for these speakers (although often no longer their main everyday language) and they were all sufficiently proficient in both the local variety of Norman and in the local standard language to be able to engage in monolingual discourse in either. Speakers were of a broadly similar socio-economic grouping, usually with close connexions to agriculture and farming. In order to provide the study with a diachronic dimension, the synchronic data are supplemented, where possible, with historical data from the Atlas Linguistique de la France (Gilliéron and Edmont 1902–10; hereafter, ALF), which were collected in mainland Normandy between 1897 and 1899 and in the Channel Islands in 1898. Given the unreliability of the ALF Guernsey data (see Collas 1931; Jones 2015), the pronouns of Guernesiais are examined in relation to Collas (1931) rather than to the ALF. The Atlas Linguistique et Ethnographique Normand (Brasseur 1980, 1984, 1997, 2010; hereafter, ALEN) includes more recent data (collected between 1970 and 1976), with pronominal forms elicited as part of the ALEN fieldwork also documented in Brasseur (1995). The different types of data presented herein make transcription a complex matter. All original data obtained via my own fieldwork are transcribed using the International Phonetic Alphabet. Where relevant, provenance is indicated as follows: mainland Norman of the Cotentin peninsula (MN), Jèrriais (J), Guernesiais (G), Sercquiais (S). Where individual varieties do not form the central focus, the relevant lexeme (which, for reasons of accessibility, is given in standard French) is represented in small capitals. Such representations are not intended to suggest any form of etymon, Latin or otherwise. All written data are left in their original spellings.
13.3 The pronouns of Norman The pronouns of the four varieties of Norman examined in this study share many features and are given in Table 13.1 in Jèrriais.⁴ For details of the phonetic
³ This sample includes, and adds to, the speakers interviewed in Jones (2015). ⁴ For ease of accessibility, the Jèrriais spelling is included alongside the transcribed forms, using the orthographic system of Le Maistre (1966).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
283
Table 13.1. The pronominal system of Jèrriais i. Subject pronouns ‘I’ ‘you’ ‘he’⁵ ‘she’ ‘we’ ‘you’
jé / j [ʒ] / [ʒə] tu / t’ [ty] i[i] (pre-vocalic form: il [il]) ou / alle (post-verbal) [u / al] (pre-vocalic form: oulle [ul]) jé / j [ʒ] / [ʒə] ou [u:] (also used as the second person singular polite form) (pre-vocalic form: ous [u:z]) ‘they’ i (masculine and feminine) [i] (pre-vocalic form: il [il]) impersonal nou [nu] (pre-vocalic form: nou-z [nu:z]) ii. Direct object pronouns (conjunctive)
‘me’ ‘you’ ‘him’ ‘her’ ‘us’ ‘you’ ‘them’
m’/ mé [m/ me] t’/ té [t / te] l / lé [l / le] l / la [l / la] nouos / nos [nwɔ] (pre-vocalic form: [nwɔ:z]) vouos /vos [vwɔ] (pre-vocalic form: [vwɔ:z]) les [le:] (pre-vocalic form: l’s [lz])
iii. Direct object pronouns (disjunctive) ‘me’ ‘you’ ‘him’ ‘her’
mé té li lyi
[me] [te] [li] [ji]
‘us’ ‘you’ ‘them’
nouos / nos vouos / vos ieux
[nwɔ] [vwɔ] [jœ:]
iv. Indirect object pronouns (different from the conjunctive direct object pronouns only in the third person) ‘to him’ ‘to her’ ‘to them’
lî [li] li [ji] lus [ly:] (prevocalic form: [ly:z]).
v. Reflexive pronouns (different from object pronouns only in the third person singular) m / mé t / té s / sé nouos vouos lus
[m] / [me] [t] / [te] [s] / [se] [nwɔ] (pre-vocalic form: [nwɔ:z]) [vwɔ] (pre-vocalic form: [vwɔ:z]) [ly:] (pre-vocalic form: [ly:z])
For full details of the positional variants of these pronouns, see Liddicoat (1994: 241–7).
⁵ Like French, all the third person singular masculine and feminine pronouns of Norman can have animate or inanimate referents.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
284
.
variation between Jèrriais and the other three Norman varieties, see Lepelley (1974: 112–14) (MN); Tomlinson (1981: 55–62) (G), and Liddicoat (1994: 240–9) (S).⁶ It will be immediately apparent that, as a langue d’Oïl variety, Norman shares several pronominal cognates with French, which are either phonetically identical or extremely similar. Since this is not an acoustic analysis, measuring fine-grained phonetic features, the pronouns examined are limited to those which contrast more distinctly between Norman and French (§13.4.1); and Norman and English (§13.4.2). The study’s focus is therefore the presence or absence of pronoun sharing rather than the narrow phonetic detail of the pronominal forms (for these details, see Brasseur 1995).
13.4 Results 13.4.1 Pronoun sharing with French 13.4.1.1 Conjunctive pronouns 13.4.1.1.1 Conjunctive third-person singular feminine subject pronouns Different forms of the conjunctive third-person singular feminine subject pronoun are recorded across the Norman territory: [ɔ], [ɑ] [ɑl], [u] (pre-vocalically, [ɔl], [ɑl], and [ul]) (MN); [u] ([ul]) (S, J), (ALF maps 462, 900); [a] ([al]) (G) (Collas 1931: CLXXXII, LXVIII). Pronoun sharing is not at all prevalent, although the presence of the form [ɛ] (< ) (pre-consonantally) at two ALF data-points (386 and 395) towards the north of the central Cotentin peninsula, suggest that, even if it is not currently widespread across Normandy, pronoun sharing has been present at these specific locations for over a century. It is, however, impossible to determine from the ALF data whether or not these data points represent merely idiolectal usage. No evidence of pronoun sharing is recorded for the Cotentin peninsula in Brasseur (1995: map 1). In my data, the French pronoun [ɛl] was present in the speech of most MN informants although it was not used systematically by any speaker (see (1)–(4) and Table 13.2), with some speakers alternating between MN and French pronouns within the same utterance (3).⁷ Since the French pronoun may be inserted into an utterance of Norman without any concomitant change being necessary in the underlying syntactic structure (2), (3), (4), the pronoun sharing has parallels with lexical borrowing (cf. Hornsby 2006: ch. 4 for insertion of Picard third person singular feminine pronouns in the regional French of Avion). That no evidence of such sharing is present in any variety of IN, where third person singular feminine ⁶ For the pronouns of Haute Normandie (Pays de Caux), which are not examined in this study, see Hébert (1984: 36–9). ⁷ In this and all tables in this study, the number of speakers who produced at least one token of a shared pronoun is given in brackets.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
285
pronouns remain ‘intact’ (5)–(7) suggests that this development may be facilitated by the typological proximity between MN and French. Table 13.2. Conjunctive third person singular feminine subject pronouns in modern Norman
mainland Norman Jèrriais Guernesiais Sercquiais
Norman pronoun
French pronoun ()
141 263 455 51
37 (14 speakers) 0 0 0
(1)
[ɔdiœ̃ ptiʁɛkɛepɑœ̃ ptiʁɛkaw] ‘She [ɔ] says a little “warm up” ’ [of coffee] (i.e. pronounced [ʁɛkɛ] rather than [ʁɛkaw]).’ (MN)
(2)
[ɛlasyzy:lʒvɑ] ‘She [ɛl] followed the horse.’
(3)
[ɔkɔ̃nɛsemilmo:meɔnpʁɛ:ʃɛpɛʔɛlnfɛzɛpɛdfʁɑ:z] ‘She [ɔ] knew a thousand words [i.e. of Norman] but she [ɔ] didn’t speak, she [ɛl] didn’t make any sentences.’ (MN)
(4)
[ɛlkʁejegɑɲiœ̃ mjo] ‘She [ɛl] thought she would earn some (money).’ (MN)
(5)
[udvi:zləʒɛ:rjej] ‘She [u] speaks Jèrriais.’
(6)
[anvœrpɑsnalɑj] ‘She [a] doesn’t want to leave.’
(G)
(7)
[ufinirabwɛtɒut] ‘She [u] will finish before long.’
(S)
(MN)
(J)
13.4.1.1.2 Conjunctive first person plural subject pronouns In traditional Norman, [ʒ(ə)] functions as both the first person singular and first person plural subject pronoun, hence j’pâlons [ʒpɑ:lõ] (J) cf. Fr. nous parlons [nupaʁlɔ̃] ‘we speak’ (Robin et al., 1879: 241; Romdahl 1881: 67; Fleury 1886: 61; Guerlin De Guer 1901: 157; Barbe 1907: 71; Hébert 1984: 36; Birt 1985: 16; Liddicoat 1994: 240; UPNC 1995: 60–1). This ‘dual’ function of is described by Lepelley (1999: 91) as being very much alive in the Norman dialects. Norman therefore differs from standard French, which uses as its first person singular subject pronoun and as its first person plural subject pronoun.⁸ In modern ⁸ The use of as the first person plural subject pronoun was common in the spoken French of the sixteenth century (Fleury 1886: 61; Brunot 1967 II: 335) and, though stigmatized, was still widespread at the end of the nineteenth century (Flikeid and Péronnet 1989; King and Butler 2005: 177).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
286
.
colloquial French, 1 is frequently replaced by the impersonal pronoun + third person singular verb: meaning that can denote ‘we’ as well as ‘one’ (Price 2003: 208). The data were analysed to determine whether contact with French is affecting the distribution of the MN impersonal pronoun (no).⁹ Although details vary slightly between ALF maps, the general picture that emerges is that, at the end of the nineteenth century, 1 was present in all varieties of MN and IN.¹⁰ Moreover, contemporary metalinguistic works list as the (sole) first person plural subject pronoun for MN (UPNC 1995: 60f.; Marie 2012: 412f.). Yet, in the present study, 1 was only used by three speakers of MN and on only one occasion each (cf. (8) and Table 13.3), with most MN speakers expressing first person plural by extending the use of the impersonal pronoun + third person singular verb (cf. Jones 2015, 2017: 125–8).¹¹ In the Norman of the Cotentin peninsula, therefore, no has a dual function (impersonal / first person plural), as (9)–(11) illustrate: (8)
[ʒavɔ̃py:dfɑ̃] ‘We [ʒ] are not hungry any more’ (first person plural) (MN)
(9)
[iladmɑ̃w̃ desinopʁɛ:ʃɛnɔʁmɑ̃adjɛp] ‘He asked if one [no] spoke Norman in Dieppe’ (impersonal) (MN)
(10)
[kɑ̃no:knɑlsõilonopʁɛ:ʃfʁɑ̃se:] ‘When our children are there we [no] speak French’. (first person plural) (MN)
(11)
[no:zaynfilno:zakɑtpətjo:eladœ:zime[X]enomadikʃegalwɑ:] ‘We [no:z] have a daughter, we [no: z] have four little children, and the second is called [X], and I’ve been told [no] that it is [a] Welsh [name]’. (first and second occurrences first person plural, third occurrence impersonal) (MN)
It seems likely that the use of the MN impersonal pronoun to convey a first person plural meaning (10) has arisen from contact with French. In other words, MN seems to be sharing the ‘French’ + third person singular structure but maintaining the Norman surface form. Examples such as (11) demonstrate how the different meanings of no may contrast within the same utterance. Different varieties of IN show different patterns of usage in this context (12)–(19). In Jèrriais and Sercquiais, although some pronoun sharing is present, 1 is still widely ‘spared’. By contrast, in Guernesiais, 1 is virtually ⁹ Despite its formal resemblance, the Norman impersonal pronoun no is not generally believed to be related etymologically to the French first person plural pronoun (cf. Fleury 1883, 1886: 65; Lepelley 1999: 88f.: although see Guerlin De Guer 1901: 162). ¹⁰ ALF maps 27, 91, 97, 100, 360, 506, 512, 515, 518, 522, 806A, 1154A, 1201. ¹¹ To facilitate the analysis (Table 13.3), all tokens where the impersonal pronoun does not carry a clear first person plural inclusive meaning are classified as ambiguous.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
287
Table 13.3. Conjunctive first person plural subject pronouns in modern Norman
mainland Norman Jèrriais Guernesiais Sercquiais
3 451 18 79
761 (17 speakers) (all) 291 (26 speakers) (all) 1375 (41 speakers) (all) 28 (6 speakers)
obsolete, with the impersonal pronoun used almost categorically to convey a first person plural meaning, in a way that parallels MN usage (Tomlinson 1981: 93, 2008: 39; De Garis 1983: 322).¹² Collas notes for Guernesiais that 1 was categorical for the speaker he interviewed in the southwest of Guernsey as long ago as 1931, unlike in the north and the southeast where, at that time, 1 was generally still present (1931: XI, XVIII, XIX, LXXII, LXXVI, CXLII, CL, CLII, CLIX, CLXXI, CLXXIV, CLXXXI). The ALEN (map 1360) records 1 as present in most varieties of MN and in IN. These findings are therefore not consistent with the present study and suggest that either this change has proceeded quickly in MN and in Guernesiais (i.e. during the last forty years) or that some variation was missed by the ALEN because of its smaller sample sizes (cf. Collas 1931 for the same point with regard to the ALF). Brasseur (1995) also records both 1 and 1 in Guernesiais. (12)
[ʒərɑmɑsi:mmɑ̃kwɔzẽɛmɛʒərɑmɑsi:mdɛjpɛlpaskeʒədəmœri:mpɑ:jɛ̃diʃẽ] ‘We [ʒə] picked up, my cousin and I, we [ʒə] picked up some shovels because we didn’t live far from here.’ (J)
(13)
[avɛkno:zamẽ:dəlasɑ̃bjɛ:kɑ̃nusrɑ̃kõtɹnupɑ:lləʒɛ:rjɛj] ‘With our friends from the Assemblée, when we [nu] meet, we speak Jèrriais.’ (J)
(14)
[ʒk̥ rɛkəʃeʃønaɑ̃ ȷp̃ rẽjsøkəʒədire:m] ‘I think that it was that, a press, that we [ʒə] would say.’ (G)
(15)
[kɛ̃matɛ̃tmuriʒələramɑsje:m] ‘When my aunt died, we [ʒə] picked it up.’ (G)
(16)
[isɑjvpɑ:tʃikʃsɛ̃jmɑdi:r] ‘They don’t know what we [ʒe] are saying.’
(17)
[mɑ̃pejrɛmenupɑ:lɛtɛrʒuõdʒɛrnɛzje:] ‘My father and I, we [nu] always spoke in Guernesiais.’ (G)
(G)
¹² In this study, 1 was produced—sporadically—by just six Guernesiais speakers, all of whom were over 80 years of age and all but one of whom were from the same part of Guernsey (the Vale). For these speakers, use of 1 is not governed by any considerations of verb or tense.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
288
.
(18)
[majparɑ̃:puvwɛpɑ:lɐlɑ̃gʎɒjmwɛjʃtɛləpatwɑkəʒəpɑlɛ:miʃẽ] ‘My could speak English but it was patois that we [ʒə] spoke here.’
(19)
[ifɛzɛtɹɒwrydnutɛobʎidʒɪ:datɑ̃dɹ] ‘It was too rough [i.e. the sea], we [nu] had to wait.’ (S)
parents (S)
This extension of meaning of the impersonal pronoun seems to be an emergent change in Norman which, in the Cotentin peninsula, has almost worked through fully and, in the Channel Islands, is at different stages of progress: most advanced in Guernesiais and least advanced in Sercquiais. Taken in isolation, the MN data suggest a straightforward case of pronoun sharing whereby Norman is calquing a syntactic structure of modern spoken French (cf. Heine and Kuteva 2005: 93). However, when considered alongside the IN data, such a parallel development may point to a possible explanation more in keeping with Sapir’s (1921) notion of ‘drift’, defined by Trask (1996: 150) as ‘the curious tendency of a language to keep changing in the same direction’. Given the physical separation of the Channel Islands from one another and from mainland Normandy, it is unlikely that any ‘dialectic interinfluencing’, as Sapir (1979: 172) puts it has occurred in this context. As discussed in §13.1, although speakers of IN may be familiar with standard French, this is usually exclusively as the result of schooling, which will presumably have imparted normative usage (namely first person plural = ) and seems unlikely to have brought about structural change of this kind. An explanation in terms of drift, whereby the economy-driven reduction of contrasting morphosyntactic forms in the verb paradigm of modern spoken French could also be operating in Norman, may therefore at least be worth considering. Interestingly, a similar replacement of 1 is well documented for some North American varieties of French (Péronnet 1989; Neumann-Holzschuh, Brasseur, and Wiesmath 2004: 89; Seutin 1975: 151f.; Dahmen 1995; Lagueux 2005; Thogmartin 1979: 115). 13.4.1.1.3 Conjunctive third person plural subject pronouns a) Feminine pronouns The third person plural subject pronoun of Norman, i [i] (il’ [il] pre-vocalically) is not marked for gender (§13.3 and (20)). This contrasts
Table 13.4. Conjunctive third person plural feminine subject pronouns in modern Norman
mainland Norman Jèrriais Guernesiais Sercquiais
Norman pronoun
French pronoun ()
47 73 81 12
12 (7 speakers) 0 0 0
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
289
with standard French, which has distinct masculine and feminine forms (ils [il], elles [ɛl]; pre-vocalically [i(l)z], [ɛlz]). (20)
[isõbimaʁɑ̃:tlezistweʁ] ‘They [i] are very funny, the stories.’
(21)
[ele:vɑkɛlsõbʁɑ̃ʒi:ɛtu] ‘And the cows, they [ɛl] have red and black patches on too.’ (MN)
(22)
[ʒekɑtfilidmœðtutɑ̃ʒɛ:ri] ‘I have four girls, they [i] all live in Jersey.’
(23)
[le:lɛdʒœ̃ misɑ̃w̃ trofɛt] ‘The vegetables, they [i] are overcooked.’
(24)
[syrlarʏ:jasuvɑ̃dyvɑ̃ejlɛ:fʎœrdɑ̃ləvɑ̃ipiʃɛj] ‘It’s often windy on the road and in the wind the flowers, they [i] were breaking.’ (S)
(MN)
(J) (G)
Evidence of pronoun sharing was present in my data (cf. (21) and Table 13.4) and, as with the third person singular (§13.4.1.1.1), the French pronoun is able to be integrated without any adjustment to the morphosyntactic structure of the MN utterance. Significantly however, pronoun sharing in this context results not merely in a shared pronominal surface form ([ɛl]) but also in the introduction of a hitherto absent opposition of gender. The ALF suggests that this development may already have been present in the Cotentin peninsula to a limited extent at the turn of the twentieth century (map 869), although no evidence of pronoun sharing is recorded in Brasseur (1995: map 6). Pronoun sharing is completely absent from both the synchronic and diachronic IN data (22)–(24), (cf. ALF map 869 and Collas 1931: CLIX).¹³ b) Pre-vocalic pronouns The pre-vocalic Norman third person plural conjunctive pronoun il’ [il] differs from French both in terms of its surface form and, as discussed in §13.4.1.1.3(a)), by the fact that it is not marked for gender (cf. §13.3 and (25)), Spence 1993: 32; Liddicoat 1994: 241). This pronoun is not examined in the ALF nor in Collas (1931), but Brasseur (1995: map 5) records exclusively prevocalic [il] for most of the Cotentin peninsula and the Channel Islands; the ALEN (map 1342), records [il] for much of the Cotentin and for Sark.¹⁴ Although UPNC (1995: 62) describes the use of the shared French pre-vocalic forms [ilz] / [i:z] in Norman as ‘incorrect’ (‘fautif ’), these were produced at least once by most of the speakers interviewed (see (26) and Table 13.5), with one speaker using them to the total exclusion of [il].
¹³ The ALF (map 869) records a form[e] in Jersey but this was not found in my data. ¹⁴ A marked feminine plural form [ɔl] is recorded in a limited area around the base of the Cotentin peninsula and in parts of Calvados and Orne (Brasseur 1995: map 5).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
290
.
Table 13.5. Conjunctive third person plural pre-vocalic subject pronouns in modern Norman masculine pronouns Norman pronoun [il] mainland Norman 118 Jèrriais 192 Guernesiais 226 Sercquiais 27
feminine pronouns
French pronoun [i(l)z]
Norman pronoun [il]
French pronoun [i(l)z]
French pronoun [ɛlz]
64 (12 speakers) 1 0 0
8 1 8 1
3 (2 speakers) 0 0 0
0 0 0 0
(25)
[tutle:sɔsje:tenɔʁmɑ̃:diletɛtutlɑome:dʒwile] ‘All the Norman associations, they [il] were there in July.’ (MN)
(26)
[ilzõʃwɛ:zilapʁɔdyksjɔ̃dlɛ] ‘They [ilz] chose the production of milk.’ (MN)
(27)
[ilõfeʃapuʁləʃtiei:zõvɑ̃dyde:milʒəkʁɛbi] ‘They [il] did that for Ch’ti and they [i:z] sold thousands of them, I think.’ (MN)
(28)
[ebile:walõi:zõœ̃ mjodʃɑ̃silabitpʁɛdbʁysɛl] ‘And the Wallons, they [i:z] are lucky, they [il] live near Brussels.’ (MN)
(29)
[ilõbɑ:tilamejzõ] ‘They [il] have built the house.’
(J)
(30)
[i:zavɛtɛlɒ] ‘They [i:z] had been there.’
(J)
(31)
[ilaprɑ̃dɹõlɑ̃gje:] ‘They [il] will learn English.’
(32)
[ilõmẽɑtʁavɛʁdønmaʃẽpuʁhɑlɐlakʁɛm] ‘They [il] put it into a machine to take out the cream.’ (S)
(G)
The fact that some words in MN utterances such as (26) are effectively homophonous with French may well be motivating the presence of shared pronouns. However, several MN speakers seemed to alternate between Norman pronouns and French pronouns apparently at random (27)–(28), possibly suggesting the presence of personal pattern variation (Dorian 1994, 2010). Apart from one instance (30), produced by a speaker of Jèrriais, IN pronouns remained intact (29), (31)–(32). Unlike the case of the pre-consonantal forms, in those contexts where pre-vocalic [ilz] is shared, it can denote both masculine and feminine referents (33). No evidence was found in my MN or IN data of the French third person plural feminine prevocalic pronoun [ɛlz].
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
(33)
291
[lɛzi:li:zõgaʁdɑlafaʃõdpʁɛ:ʃidyvyltɑ̃] ‘the islands, they [i:z] have kept the way of speaking that belongs to the old days’ (MN)
13.4.1.1.4 Conjunctive impersonal pronouns All the metalinguistic sources consulted record the conjunctive impersonal pronoun of Norman as [no] (liaison form [no: z]) (Cotentin) (34), (36)–(37) and [nu] (liaison form [nu:z]) (Channel Islands) (38)–(40) (ALF maps 407, 651, 1083; Collas (1931: CLI, CLXX; Brasseur 1995: map 3).¹⁵ Once again, in my data (Table 13.6) pronoun sharing is present in MN (35) but absent from IN (38)–(40). As seen with other cases of shared pronouns, French and Norman forms can alternate within the same utterance without any obvious morphological or semantic ‘trigger’ (36)–(37). Table 13.6. Conjunctive impersonal pronouns in modern Norman
mainland Norman Jèrriais Guernesiais Sercquiais
Norman pronoun
French pronoun ()
502 428 614 43
24 (9 speakers) 0 0 0
(34)
[winopœdiʁʃɑ] ‘yes, one [no] could say that.’
(MN)
(35)
[jukɔ̃pœakatede:vylli:vʁ] ‘where one [ɔ̃] can buy old books’
(MN)
(36)
[ɑʃytɑ:̃ lɑno:zamnɛlakolɑsjõepiɔ̃muʒɛdɑ:̃ le:kjɑw] ‘In those times one [no: z] would take a snack and then one [ɔ̃] used to eat in the fields.’ (MN)
(37)
[no:zɑtultɑ̃:diskyteɑ̃patwe:me:ɔ̃savɛpʁe:ʃilø:lɑ̃:gitu] ‘One [no:z] always discussed things in patois but one [ɔ̃] knew their language too.’ (MN)
(38)
[nunpœpɑ:save] ‘One [nu] can’t know.’
(39)
[nupuvevejlamɑjrdiʃɑȷ̃enɑʃi] ‘One [nu] could see the sea from here a long time ago.’ (G)
(40)
[nutravɑʎɛtdʏpɑ̃dɑ̃lagjɛ:r] ‘One [nu] worked hard during the war.’
(J)
(S)
¹⁵ The only exception to this is ALF map 651 which cites [ɔ̃] for Jersey, although this usage is not corroborated by other maps. On the mainland, [ẽ] and [ɑ̃] are noted for data points 358 and 368 respectively.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
292
.
13.4.1.2 Disjunctive pronouns The disjunctive pronouns of Norman reveal similar patterns of usage and may therefore be considered together. On the whole, speakers produced fewer disjunctive than conjunctive forms (see Tables 13.7–13.11). This is unsurprising given that Norman, like French, is a non pro-drop variety, hence conjunctive pronouns are obligatory. Although no disjunctive pronoun sharing is recorded for either MN or IN in any of the metalinguistic sources consulted, a few instances were present in my MN data (42), (47), (52), (57) cf. (41), (46), (51), (56), albeit to a considerably lesser degree than with the conjunctive forms. No sharing of disjunctive forms was found in IN (43)–(45), (48)–(50), (53)–(55), (58)–(60), (61)–(62). Table 13.7. Disjunctive first person singular pronouns in modern Norman (cf. ALF maps 863, 864; Collas 1931: LXXC, LXXIX, CVII, CLXXII, CXCII; Brasseur 1995: map 9)
mainland Norman Jèrriais Guernesiais Sercquiais
Norman pronoun
French pronoun ()
156 86 221 15
5 (3 speakers) 0 0 0
(41)
[tʃœmejavɛpɑ:detɑbjɑvɑk] ‘At my home [me], there was no cow stable.’ (MN)
(42)
[ʒɑneʒymwadzamɛ̃:] ‘I’ve had some friends, me [mwa].’
(43)
[tekummete] ‘You are like me [me], you.’
(J)
(44)
[ʒetrekɑ:me] ‘I have three cats, me [me].’
(G)
(45)
[ilɛvnʏdɒuvmɛ] ‘He came with me [mɛ].’
(S)
Table 13.8. Disjunctive second person singular pronouns in modern Norman (cf. ALF maps 28, 1307; Collas 1931: LXXV, CXII; Brasseur 1995: 333)
mainland Norman Jèrriais Guernesiais Sercquiais
Norman pronoun
French pronoun ()
43 34 51 4
1 (1 speaker) 0 0 0
(MN)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
293
(46)
[etetɑbidzamɛ̃:] ‘And you [te], you have many friends.’
(MN)
(47)
[tʃiktykʁetwa] ‘What do you think, you [twa]?’
(MN)
(48)
[ʃɛbikmɔdpurte] ‘That’s very convenient for you [te].’
(49)
̃ ɑg̃ je] ‘And you [te], you couldn’t speak in English.’ (G) [etetynpuvɛpɑ:dvi:zɑjɑn
(50)
[kjiktʏfɛkɑ̃tʏvɑsjətɛj] ‘What do you do when you go home? [tɛj]’
(J)
(S)
Table 13.9 Disjunctive third person singular feminine pronouns in modern Norman (cf. Brasseur 1995: map 12)
mainland Norman Jèrriais Guernesiais Sercquiais
Norman pronoun
French pronoun ()
12 19 15 3
1 0 0 0
(51)
[ulalɛakatede:pumadi:tʃilɔmɛtdətʃøli] ‘She went to buy apples ten kilometres away from her [li] house.’ (MN)
(52)
[mapəʁmjɛʁfilɔleʒɑ̃viɛl] ‘My first daughter, she is January, her [ɛl].’ (MN)
(53)
[ʒiðekwɔ:rswɔ:treji] ‘I would go and run after her [ji].’
(54)
[mɑ̃pejrpyskəjɛl] ‘My father more than her [jɛl].’¹⁶
(G)
(55)
[dʒəsi:pʏvɪkjɛl] ‘I am older than her [jɛl].’
(S)
(J)
Table 13.10. Disjunctive third person plural masculine pronouns in modern Norman (cf. ALF map 525; Brasseur 1995: map 13)
mainland Norman Jèrriais Guernesiais Sercquiais
Norman pronoun
French pronoun ()
52 71 43 5
5 (3 speakers) 0 0 0
¹⁶ [jɛl] is an indigenous Norman form and not an assimilated French borrowing. It is documented for Guernesiais and for Sercquiais (De Garis 1982: 80; Brasseur 1995: map 12; Tomlinson 1981: 60, 2008: 43) and exists alongside, respectively, [lji] and [ʎɪ].
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
294
.
(56)
[le:gʁɑ̃:kunɛ:smɛ:ly:ʒɑ̃ɑjœ:ijɑ̃natʃikzœ̃ :kilkunɛ:s] ‘Their grandparents know but their parents [jœ:], there are some who know.’ (MN)
(57)
[ʃeøkuladmɑ̃dedaʁpʁɑ̃dʁʃunɑ] ‘It’s them [ø] who she asked to learn that.’ (MN)
(58)
[jajinvulepɑ:pɑ:leɑ̃gjej] ‘Them [jaj], they didn’t want to speak English.’ (J)
(59)
[ʃtɛlpatwakidvi:zesijaw] ‘It was patois that they spoke at their home [jaw].’ (G)
(60)
[ivuli:dralɐsjɛjœ:kɑ̃lagjɛ:rkmɑ̃ʃit] ‘They wanted to go to their home [jœ:] when the war started.’ (S) Table 13.11. Disjunctive third person plural feminine pronouns in modern Norman (cf. Brasseur 1995: 333)
mainland Norman Jèrriais Guernesiais Sercquiais
Norman pronoun
French pronoun ()
0 10 3 0
0 0 0 0
(61)
[le:filtɛjgʁɑ̃:deinõʒamejrejɛlmɑ̃dmœðeɑ̃ʒɛ:rijɑj] ‘The girls were big and they have never really lived in Jersey, them [jɑj].’ (J)
(62)
[ele:filaleɔbjiʒi:dʃẽtɑjkɔmjawdəprɔnõsɑjse:parolkɔmjaw] ‘And the girls, she has to sing like them [jaw], to pronounce her words like them [jaw].’ (G)
13.4.1.3 Reflexive pronouns The reflexive pronouns of Norman are similar to those of French apart from the third person plural which, as noted in §13.3, is lus / leus (hereafter, l-form) (cf. standard French se: hereafter, s-form) (Fleury 1886: 63; Lepelley 1974: 113; UPNC 1995: 67). The l-form is also present with infinitives of reflexive verbs with third person plural reference: e.g. I leus en vyinent et o leus en vount ‘They come and they go’ (MN); Dé de lo, i les veit lus raungui ‘From there, he sees them lining (themselves) up’ (MN) (UPNC 1995: 68). ALF map 869 records the presence of s-forms in the Cotentin peninsula well over a century ago. This is confirmed by Fleury (1886: 63), although he describes them as minority forms. More recently, Brasseur (1995: map 14) records s-forms as predominant in most of Basse Normandie, with l-forms only found in the
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
295
Cotentin peninsula’s northern and western periphery, although UPNC (1995: 68) describes s-forms as generally incorrect because they are influenced by French. For the Channel Islands, the documented evidence differs. For Jèrriais, the ALF (map 869) records l-forms and their more recent presence is confirmed by Brasseur (1995: map 14),¹⁷ although Birt (1985: 79) notes both l-forms and s-forms; for Guernesiais, only s-forms are recorded in Collas (1931: LXXXV), De Garis (1993: 343), and Tomlinson (2008: 105f.), although Jones (2008: 119f.) points to the presence of both l-forms and s-forms; for Sercquiais, l-forms alone are attested (Liddicoat 1994: 247; Brasseur 1995: map 14). The present study reveals a clear divergence between MN, where l-forms are rarely present and where the reflexive pronoun is shared with French (63)–(64) and Table 13.12, and IN, where the l-form is spared to a greater degree, although s-forms are often present, especially in Guernesiais ((65)–(70) and Table 13.12). These data therefore reveal more variation to be present in Jèrriais and Guernesiais than is suggested by previous sources. Utterances such as (66) and (67) indicate that this variation is not governed by the meaning or the morphological form of the verb.
Table 13.12. l-forms and s-forms of the 3 reflexive pronoun in modern Norman
mainland Norman Jèrriais Guernesiais Sercquiais
l-forms
s-forms
2 (1 speaker) 68 (speakers) 16 (speakers) 5 (speakers)
56 (17 speakers) 52 (speakers) 81(speakers) 0
(63)
[ily:zɑ̃võ] (MN) ‘They go away (l-form).’
(MN)
(64)
[insəsõpɑ:ʁakaʃi:tʃøjœ:] ‘They didn’t return home (s-form).’
(MN)
(65)
[nijø̃de:dɑ:ly:maðji:t] ‘Not one of the two got married (l-form).’
(66)
[le:filly:sõlve:tarisəsõlve:ɑdjɛ:zœð] ‘The girls got up (l-form) late, they got up (s-form) at ten o’clock.’ (J)
(67)
[iliðõly:zasjɛðdɑ̃:ləkjowiliðõsasjɛðswɔ:ø̃bwej] ‘They will go and sit down (l-form) in the field, they will go and sit down (s-form) under a tree.’ (J)
(J)
¹⁷ The ALF does not record a reflexive pronoun for Sercquiais and, as stated in §13.2, the ALF data for Guernesiais are not reliable.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
296
.
(68)
[ ʃelɑlətɑ̃:kəle:ʒɑ̃ly:rɛʒwi:] ‘That’s the time when people have a good time (l-form).’ (G)
(69)
[le:ʒɑ̃tʃikfejstruvatrapɑjdədɑ̃:] ‘The people sometimes find themselves caught up in it (s-form).’ (G)
(70)
[ilõkmɑ̃ʃɪɑ̃sɛralɛkɒwlepɪɑhõzɑ̃:ilʏ:zɑ̃nali:dɹɑ̃nɑ̃gʎətɛr] ‘They started school in Sark and then at eleven years of age, they went (l-form) to England.’ (S)
The Guernesiais data are particular noteworthy, given that Collas suggests that l-forms had already disappeared in 1931. Intriguingly, all 16 l-forms recorded were produced by speakers from the north of Guernsey (the Vale), an area often described by speakers as the most ‘French’ part of the island (i.e. where s-forms might therefore be more expected).
13.4.2 Pronoun sharing with English Sustained everyday contact between MN and French has resulted in pronoun sharing. As will be discussed in §13.5, this appears to be facilitated by the typological proximity of the two languages, which enables French pronouns to be incorporated within the morphosyntactic structure of MN without any concomitant modification of that structure. Despite the greater typological distance that exists between IN and its contact language, English, common syntactic slots are occupied by several of their subject pronouns. Unlike in §13.4.1, nothing that might be termed ‘overt’ pronoun sharing with English occurs in the data: indeed, phrases such as **‘I chante’ ‘I sing’ or **‘she voit’ ‘she sees’ are precluded by the Matrix Language Frame model of Myers-Scotton (2002), according to which all function words in a bilingual utterance come from the Matrix language.¹⁸ Contact with English has nevertheless left a more covert mark on the pronominal system of IN, namely a convergence between the abstract lexical structure of IN and that of English (Myers-Scotton 2002: 194; cf. Jones 2018).¹⁹,²⁰ This type of ‘sharing’
¹⁸ The Matrix language is a theoretical construct, defined as the language which sets the morphosyntactic frame for the utterance. This construct, however, is by no means universally accepted: see, among others; MacSwan (2005); Gardner-Chloros (2009); Bhat, Choudhury, and Bali (2016). ¹⁹ The term ‘convergence’ is used here according to Myers-Scotton’s (2002: 101) definition, namely ‘a linguistic configuration with all surface forms from one language but part of its abstract lexical structure from another language’. Unlike the more usual definition (see, for example, Hock and Joseph 1996: 172), for Myers-Scotton (2002: 172), convergence is a largely one-way process. ²⁰ According to Myers-Scotton (2002), all lemmas in the mental lexicon include three levels of abstract lexical structure, namely: (i) lexical conceptual structure (semantic and pragmatic information); (ii) morphological realization patterns (surface realizations of grammatical structure); (iii) predicate-argument structure (the mapping of thematic structure onto syntactic relations).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
297
appears to be produced by what Muysken (2000: 277) describes as the ‘uniformising tendency’ that results from two grammars having to be processed by a single system. Matras (2009: 238) refers to this process as ‘pattern replication’ (as opposed to ‘matter replication’), characterizing it as ‘a change to an inherited structure of the “replica” language, inspired by a structure of the “model” language’.
13.4.2.1 Conjunctive pronouns 13.4.2.1.1 Gender-marking Like French, but unlike English, Norman marks masculine and feminine gender in third person singular pronouns that refer to inanimate objects (§13.3), e.g. The knife is on the table. It is sharp cf. [lkuteejsy:lɑtabjilejejdʒy] (= ‘he’) (J); The apple is in the basket. It is red. cf. [lɑpawmejdɑ̃lpɔɲealejptit] (= ‘she’) (G). All speakers of MN observe the traditional use of masculine and feminine pronouns (71)–(72). Although this is generally also the case in IN (73), (76), (78), some non-traditional usage is apparent and usually involves the generalization of the unmarked masculine pronoun in a context that traditionally requires a feminine form (74), (77), (79)—although cf. also (75). Non-traditional forms are produced without hesitation but they occur so rarely (and each time in the mouth of a different speaker) that they may represent production errors rather than evidence of ‘covert’ pronoun sharing with English (i.e. one pronominal form for all inanimate objects ‘it’). Table 13.13. Gender-marking in the pronouns of modern Norman masculine pronoun + masculine referent mainland Norman 116 Jèrriais 110 Guernesiais 180 Sercquiais 14
feminine pronoun + feminine referent
masculine pronoun + feminine referent
feminine pronoun + masculine referent
31 41 43 4
0 5 7 (7 speakers) 2 (2 speakers)
0 1 0 0
(71)
[labutɛlɔlevjødaʃtœʃefini:] ‘The bottle, it is empty now, it’s finished.’ (MN)
(72)
[ʒɛmbiləpʁɛ:ʃiləpatwe:] ‘I like speaking it, patois.’
(73)
[ulɛtibɑ:slamɛ] lit. ‘How low she is, the sea.’
(74)
[le:ʒɑ̃kisõẽtɛrɛsi:ɑaprɑ̃dlalɑ̃:gʃepɑ:ləjɛrulejkriðmejɑlwipɑ:le] ‘The people who are interested in learning the language, it’s not reading it (masculine) or writing it but hearing it spoken.’ (J)
(MN) (J)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
298
.
(75)
[lẽtɛrenejpɑ:kumutɛtɑ̃:pɑ:sɛ] ‘The interest isn’t what it used to be in past times.’ (J)
(76)
[ʃytvɑkalɛjvjelɔʃtajr] ‘This cow it is old now.’
(77)
[lawtfomilikõtinwisypɑremɑ̃ʔɑ̃twɔrdəlɛdʒi:z] ‘The other family, it continued particularly around the church.’ (G)
(78)
[lɑmwɛʁulesibwɛlɑ̃ɲɛt] ‘The sea, it is so beautiful today.’
(79)
[lɑlwɐdɑ̃:ʃytɑ̃:lɑilɑvwɛtdɒyduzɛnpaʁowl] ‘The law at that time, it had two dozen words.’ (S)
(G)
(S)
13.4.2.1.2 Pronoun calquing Although many Norman and English verbs are semantically equivalent, in contexts where these verbs may trigger a particular category of pronoun, traditional Norman usage does not always ‘match’ that of English. Such ‘non-matching’ verbpairs were examined in the data, although the challenge of finding, in casual speech, comparable data across all four varieties has meant that only three verbs are analysed here: ‘to miss, to lack’, ‘to give’, ˊ ‘to answer’ (Tables 13.14–13.16).²¹ In MN, traditional usage is adhered to by all speakers, with the expected pronoun present in each case (80), (84), (87). In IN, however, the Norman pronoun sometimes calques its English equivalent: specifically, in the case of , the experiencer changes from indirect object to subject (81)–(83); with and ˊ, the recipient changes from indirect to direct object (85)–(86) and (88)–(89) (cf. ALF map 786; Bourdon, Cournée, and Charpentier 1993: 34, 204, 276; Le Maistre 1966: 172, 337, 450; Birt 1985: 75; Tomlinson 1981: 59; Collas 1931: LIX; Liddicoat 1994: 246). However, the number of tokens is insufficient to enable it to be determined whether such usage represents a change in progress. Table 13.14. Pronoun calquing in modern Norman:
mainland Norman Jèrriais Guernesiais Sercquiais
traditional usage (experiencer = indirect object)
non-traditional usage (experiencer = subject)
3 0 0 0
0 3 (3 speakers) 1 1
²¹ No tokens of or ˊ are present in the Sercquiais data.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
299
(80)
[ɑ̃patwe:ily:mɑ̃kɛde:mo:] ‘In patois, they lacked some words.’ (MN) (experiencer = indirect object)
(81)
[imɑ̃kle:sʌndeiroʊst] ‘He misses Sunday roasts.’ (J) (experiencer = subject)
(82)
[imẽtʃɛdʒɛrnəzi] ‘He missed Guernsey.’
(83)
[tɑmɑ̃kjɪkjikʃɒuz] ‘You have lacked something.’ (S) (experiencer = subject)
(G) (experiencer = subject)
Table 13.15. Pronoun calquing in modern Norman: traditional usage (recipient = indirect object) mainland Norman 11 Jèrriais 24 Guernesiais 1
non-traditional usage (recipient = direct object) 0 3 (3 speakers) 1
(84)
[ʃɛtɛde:nɔ̃:knoly:dɔnɛ] ‘They were names that we gave them.’ (MN) (recipient = indirect object)
(85)
[ø̃livrkeʒledune] ‘A book that I have given him.’ (J) (recipient = direct object)
(86)
̃ æpkmnãw̃mledõwni] ‘It’s a map that my husband gave him.’ [ʃejɑ̃j m (G) (recipient = direct object)
Table 13.16. Pronoun calquing in modern Norman: traditional usage (recipient = indirect object) mainland Norman 1 Jèrriais 5 Guernesiais 0
non-traditional usage (recipient = direct object) 0 2 (2 speakers) 1
(87)
[ipʁɛ:ʃpatwe:eily:ʁepɔ̃tʃikku:ɑ̃fʁɑ̃se:] ‘he speaks patois, he sometimes answers them in French.’ (MN) (recipient = indirect object)
(88)
[ʒəlejrepõnyɑ̃ʒɛ:rjɛj] ‘I answered him in Jèrriais’ (J) (recipient = direct object)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
300
.
(89)
[ʒəlrɛpãw̃ɑ̃frɑ̃se:] ‘I answer him in Guernesiais’²² (G) (recipient = direct object)
13.4.2.2 Pronominal reflexive verbs Certain Norman verbs can be reflexive or non-reflexive in both their form and meaning (Birt 1985: 79–82; Liddicoat 1994: 248; UPNC 1995; Tomlinson 2008: 105). Lever, for example, means ‘to raise [something]’ in its transitive form but ‘to get up’ in its reflexive form (s’lever) (ALF maps 763, 764). When not being used reflexively, lever can only be transitive: in other words, **il leve (with the meaning ‘he gets up’) is an impossible structure. In this context, the clitic pronoun therefore forms part of the lexical specifications of the verb. All speakers of MN produced forms that are consistent with the usage described in the above metalinguistic works (90), (93), (98), (103). However, although traditional usage is often adhered to in IN, (91), (94), (96), (99), (101) some verbs are found to lose their traditional reflexive pronoun (92), (95), (97), (100), (102), (104), (105). The four verbs analysed in Tables 13.17–13.20 (’, ’, , and ’ˆ are those which occur most frequently across the different varieties.²³ Table 13.17. Reflexive pronouns in modern Norman: ’ (cf. ALF map 62; ALEN map 1184; Tomlinson 1981: 115; Brasseur 1995: map 14) reflexive pronoun present mainland Norman 3 Jèrriais 15 Guernesiais 0
no reflexive pronoun 0 0 3 (3 speakers)
(90)
[ivœsasjɛʁsysɑtʃajʁ] ‘He wants to sit on his chair.’ (MN) (reflexive pronoun present)
(91)
[typœtasjɛðlɒ] ‘You can sit there.’
(92)
[vjɑ̃ʔasjejsylaskabɛ] ‘Come and sit on the stool.’ (G) (no reflexive pronoun)
(J) (reflexive pronoun present)
²² Guernesiais is usually referred to by native speakers as [frɛ̃se] ‘French’. ²³ Respectively, reflexive: ‘to sit down’; non-reflexive: ‘to sit’; reflexive: ‘to be called’; non-reflexive: ‘to call’; reflexive: ‘to get up’; non-reflexive: ‘to lift, raise’; reflexive: ‘to come to a stop’; non-reflexive: ‘to stop’ (Le Maistre 1966: 27, 22, 322, 25; Société Jersiaise 2008: 24, 21, 123, 198; Liddicoat 1994: 248). No tokens of these verbs are present in the Sercquiais data.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
301
Table 13.18. Reflexive pronouns in modern Norman: ’ (cf. Brasseur 1995: map 14) reflexive pronoun present mainland Norman 3 Jèrriais 6 Guernesiais 7
no reflexive pronoun 0 4 (4 speakers) 3 (2 speakers)
(93)
[ʃɑsapølø̃kabeʁjolɛ] ‘That’s called a cabriolet.’ (MN) (reflexive pronoun present)
(94)
[ilpɑ:labsɔlymɑ̃bɛ̃purɛ:tø̃f iʃyɑ̃gjejkumilsapɛl] ‘He speaks very well for a dratted Englishman, as he’s called.’ (J) (reflexive pronoun present)
(95)
[dykoutɛderouzeilapɛlpõdejvejprilapɛldezɛðaɲi:] ‘In the area around Rozel, they are not called vêpres they are called ithangnies.’ (J) (no reflexive pronoun)
(96)
[larutiʃɑjbɑ:sapœlle:sɔ:ltpænz] ‘That road down there is called the Salt Pans.’ (G) (reflexive pronoun present)
(97)
[ilapœltuma] ‘He is called Thomas.’
(G) (no reflexive pronoun).²⁴
Table 13.19. Reflexive pronouns in modern Norman: (cf. ALF maps 763, 764; Collas 1931: CCXVIII; Brasseur 1995: map 14) reflexive pronoun present mainland Norman 6 Jèrriais 14 Guernesiais 4
no reflexive pronoun 0 7 (6 speakers) 7 (4 speakers)
(98)
[ʒəmələvɛtɛʁʒuɑʃẽkœ:ʁ] ‘I always got up a five o’clock.’ (MN) (reflexive pronoun present)
(99)
[ily:lvi:teisɑ̃fy:t] ‘They got up and they left.’ (J) (reflexive pronoun present)
²⁴ Note that the context makes it completely clear that the intended meaning is ‘He is called Thomas’ rather than ‘He calls Thomas’.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
302
.
(100)
[le:filõlveaø̃kardedʒi:] ‘The girls got up at a quarter to ten.’ (J) (no reflexive pronoun)
(101)
[ʒəmɑ̃veməlvɑjpuravemõdejʒœnɑj] ‘I’m going to get up to have my breakfast.’ (G) (reflexive pronoun present)
(102)
[sivunlvɑjpɑ:ʒəvudãw̃rɛpɑ:dykolɛt] ‘If you don’t get up I won’t give you any warm milk.’ (G) (no reflexive pronoun) Table 13.20. Reflexive pronouns in modern Norman: ’ˆ (cf. Brasseur 1995: map 14) reflexive pronoun present mainland Norman 4 Jèrriais 0 Guernesiais 0
no reflexive pronoun 0 8 (7 speakers) 5 (3 speakers)
(103)
[isarɛtepuʁmɛ] ‘He waited for me.’
(104)
[ʒarejti:mpurlɛbʊs] ‘We waited for the bus.’ (J) (no reflexive pronoun)²⁵
(105)
[ʒarejtiɛndɛmiɑjr] ‘I waited for half an hour’ (G) (no reflexive pronoun)
(MN) (reflexive pronoun present)
IN usage varies from speaker to speaker and from verb to verb, with no clear pattern discernible across the different varieties. Since, in their non-reflexive form, these verbs can traditionally only be transitive, and therefore require an expressed object, non-traditional usage reflects a change in the predicate-argument structure of IN. As Tomlinson (1981: 113) acknowledges, this may be due to contact with English, which does not always mark reflexivity as overtly as some other Germanic languages (cf. McWhorter 2002: 220f.).²⁶ A further piece of evidence pointing to English influence in this context is that, in IN, the infinitive of reflexive verbs, which traditionally change their reflexive pronoun according to the subject of the utterance (Birt 1985: 81; cf. §13.4.1.3) are occasionally found in the data with what appears to be a generalized se (traditionally the third person singular reflexive pronoun—cf. §13.3) (106–113). In such cases, it seems that the reflexive pronoun may—on the basis of English, which
²⁵ Note also the calque of the English preposition here. ²⁶ Cf. for example, the verbs ‘to shave’, ‘to hurry’ and ‘to remember’, which are overtly reflexive in German (sich rasieren, sich beeilen, sich erinnern) but not in English.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
303
does not have a reflexive pronoun in these contexts—have been reanalysed as part of an ‘invariable’ infinitive. In turn, this can sometimes motivate a doubling of the reflexive pronoun (108–109). This invariability of the reflexive pronoun se is particularly noteworthy in the three persons of the plural of modern Guernesiais (cf. Tomlinson 1981: 61f.; 2008: 40). (106)
[ʒavjõabityddəsnaleomarʃi] ‘We used to go to the market.’
(J) [s]
(107)
[ʒavõɑ̃vi:dsnale] ‘We want to go.’
(J) [s]
(108)
[ʒənusɑsivi] ‘We sat down.’ (G) (Tomlinson 1981: 62) (doubling:[nu] and [s])
(109)
[dʒənwɔsnalõowkʎɒw] ‘We go to the field.’ (S) (doubling:[nwɔ] and [s])
(110)
[evusnalɑjɔɲɛ] ‘And you are going today.’²⁷
(111)
[vusɑ̃vne:tiʃɑ̃jvunavɑjpɑ:jɑ̃ɑalɑj] ‘You were coming here, you don’t have far to go.’ (G) [s]
(112)
[eʒdiʃɛmy:purnãdskwɔʃjeɑsese] ‘And I said “It’s better for us to go to bed tonight”.’ (G) [s]
(113)
[ɑmwõkəvunsləvɑjvunerɑjpɑ:dədeʒœnɑj] ‘Unless you get up, you won’t have any breakfast.’ (G) [s]
(G) [s]
13.5 Conclusion Although, as Winford (2003: 5) states, ‘[t]here are in principle no limits [ . . . ] to what speakers of different languages will adopt and adapt from one another’, it seems generally accepted that open-class items are more easily and frequently borrowed than closed-class items such as pronouns, which are considered to be embedded at a ‘deeper’ level of the linguistic system. This study has demonstrated, however, that intensive daily contact with their different superstrates is clearly affecting pronoun usage in MN and IN. In his examination of pronoun borrowing in the Mayan language family, Law (2014: 106) emphasizes the importance of taking into account the structure of both the source and the recipient language. This study’s findings underline Law’s ²⁷ Cf. Collas (1931: CCXVIII), which records the traditional reflexive pronoun [vu] in this context, suggesting that the generalization of [sə] is relatively recent; also (111).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
304
.
point by demonstrating that the fact of being in contact with two typologically different superstrates is causing the pronouns of MN and IN to diverge (cf. Jones 2015). Law continues: Pronominal borrowing is facilitated because the donor and recipient languages have virtually identical syntactic structures in several areas specifically involving the position and use of pronominal markers. In these languages [ . . . ] one can simply insert a borrowed pronoun into the recipient language without additional structural accommodation because of their highly convergent typological and structural profiles.
Such ‘overt’ pronoun sharing was found only in MN, with French pronouns inserted into a MN syntactic frame in a similar way to a lexical borrowing. The fact that this occurred relatively frequently with conjunctive pronouns (§13.4.1.1) but to a far lesser degree with disjunctive pronouns (§13.4.1.2) is interesting, given that ‘free’ pronouns are generally considered to be more ‘borrowable’ than clitic pronouns (cf. Jake 1994)—however, as Muysken (2000: 177) points out, the fact that MN and French are closely related varieties is not without significance here. Alongside such ‘overt’ pronoun sharing, evidence was found in MN of more ‘covert’ influence, whereby the first person plural conjunctive pronoun had virtually disappeared from everyday usage, becoming replaced with a syntactic calque of colloquial French + third person singular (§13.4.1.1.2). That this structure was also found in IN (particularly in Guernesiais), which is not in everyday contact with French, is intriguing and, as discussed in Jones (2015, 2017), may be attributable to syntactic drift. The greater typological distance between IN and English means that IN pronouns have been being largely ‘spared’ by ‘overt’ contact: no examples were found of the surface forms of English pronouns being used with Norman verbs. However, evidence was found of ‘covert’ pronoun sharing in the context of third person singular gender marking, pronoun calquing and certain reflexive verbs (§13.4.2), with the data suggesting that some convergence has occurred between IN and the abstract lexical structure of English. This seems to correspond with Matras’s account of ‘pattern replication’, which he describes as arising from ‘an individual speaker’s scan for an optimal construction through which to communicate local meanings’ (2009: 243) and thus considers to be more erratic that what he terms ‘matter replication’. Viewed from this perspective, the fact that, in the varieties of modern Norman examined, ‘covert’ pronoun sharing with English occurs less frequently than ‘overt’ pronoun sharing with French is therefore not at all unexpected.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
PART III
WORD STRUCTURE
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
14 On the origins of French and Occitan Clive R. Sneddon
14.1 Introduction 14.1.1 The problem The proverbial alien from outer space wanting to know how French and Occitan came to be distinct languages might begin by looking at the earliest written evidence. For French, we have two ninth-century texts, the Strasbourg Oaths and the Eulalia sequence, but the earliest Occitan is not earlier than c.1000 and is predominantly eleventh century. In between we have the late tenth-century Clermont-Ferrand manuscript, whose mostly Latin additions include two texts, the Passion and St Leger, which are normally said to be in French, but with marked Occitan influence in the Passion.¹ Can it be said that this reveals an early stage in the emergence of French and Occitan from their common Gallo-Romance source? The difficulty with this claim is that, if French has already emerged in the ninth century, it is hard to see why a text reflecting common Gallo-Romance should be written in the tenth. Could the Occitan elements in the Clermont-Ferrand MS 240 be a sign of conscious vernacular bilingualism in writing the Passion, or simply reflect its composition by a native of the Auvergne? If the latter, why was the author writing in French? The usual alternative explanation, that a French text was being copied by a local scribe, leaves us little the wiser, as the Passion shows its scribe capable of copying accurate French, and yet from time to time, consciously or otherwise, writing his own language and thus deviating from the source text. Only a study of the language of the earliest vernacular texts can contextualize the Clermont-Ferrand texts, and help us understand their choices. The data to be used here are taken from a concordance of fifteen early texts, with a reverse concordance to identify morphology. Full primary source details are in the list of references and a summary is in Table 14.1, but the witnesses range in date from s. ixex to s. xii¹ (Sequence or Prose² of St Eulalia to Epître farcie de St Etienne), and ¹ My thanks to Mr Michael Guggenbuhl, Directeur de la Bibliothèque du Patrimoine, Clermont Auvergne Métropole, Clermont-Ferrand, for sending me a complete set of images of MS 240 together with the library’s current descriptions. ² The sequence follows the Alleluia in the mass by setting words in rhythmic prose to its last syllable, and by extension designates a hymn with rhyme in an accentual meter. The genre was popularized in the ninth century by Notker Balbulus in his Liber Hymnorum. Variation and Change in Gallo-Romance Grammar. First edition. Sam Wolfe and Martin Maiden (eds). This chapter © Clive R. Sneddon 2020. First published in 2020 by Oxford University Press.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
308
.
comprise seven French texts, seven Occitan, and one Francoprovençal. In addition, a French text first identified about forty years ago, the Augsburg Passion, will be compared with what has emerged from the others.³
14.1.2 Identifying a vernacular language Before looking more closely at the data, there is an initial question to consider. What confidence can we have that the traditional identification of the language of these early texts is correct? The mostly nineteenth-century philologists who first identified them worked by placing their spellings and morphology in a line of development from Latin to a modern dialect or standard Romance language. Most early texts are in verse, so authorial language could be deduced from the assonances or rhymes, independently of the spellings in the preserved manuscript. Forms found that did not fit into this history of the modern language were characterized as scribal, reflecting the language of the person who wrote the copy as distinct from that of the author. The existence of such non-compliant forms makes it possible to disagree with the traditional identification of a text’s language. For example, both Koschwitz (1913) and De Poerck (1963: 144, 147) have rejected the idea that the bilingual Sponsus is Occitan, assigning it to French. More recently, Hilty (1994, 1995) has denied that the newly discovered Augsburg Passion is French, assigning it to Occitan, but printing an ‘original’ Occitan text (1995: 32) to correct scribal errors or French transmission. For this chapter I have accepted each text’s traditional assignment.
14.1.3 The nature of the evidence For historical linguists, the history of a language can be reconstructed from its use today including from dialects, or by deducing speech from the spellings in our documents. Only when the two methods give consistent results can the results be considered reliable. However, Remacle (1948) studying Wallon came to an unexpected conclusion. He could say that the Eulalia provided the earliest attestation of forms known to be Wallon from later texts and the modern dialect. However, the oldest Liège document from 1236 had a majority of forms belonging to what would become standard French, together with some Wallon forms. It is not a transcript of speech, neither Wallon nor Francien, nor does it record any written old French standard. It is this mixture of forms that led Gossen (1967) to create the term ‘scripta’ for medieval documentary French. In identifying four ‘scriptae’ as Central, Western, Northern and Eastern, he is attributing linguistic and regional identity to ³ My thanks to the staff of the Augsburg Stadtarchiv for a digital image of each side of their document.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
309
the scribes; each has a common fund of French spellings, while permitting more localized forms. Two atlases, by Dees (1980) mapping forms from French documents and Dees (1987) mapping forms from French literary manuscripts, show fewer local features in the latter, implying the existence of spellings which are more widespread. The most likely reason for this is that documents and books have different audiences. Documents are for local use and need to be understood locally. A book has a value and can be sold on, making it undesirable to use less widelyknown spellings. Both documents and books are produced by educated people whose normal work involves reading and writing Latin. The working hypothesis for the present study is that scribes are familiar with Latin conventions, so that writing the vernacular is an exercise in making material more widely available at minimal effort, by changing the conventions learned for Latin as little as possible. Whereas a linguist studying the history of a language and its dialects will use the more numerous local forms found in documents, one seeking any common fund of spellings will need to study the written forms of books.
14.1.4 The books we have for early texts It is not unusual to consider the earliest texts as ‘flyleaf literature’. A flyleaf does not constitute a book, whereas the word literature implies a culture with conventions of, for example, genre, rhetoric or versification. Zumthor (1963: 71–121), in his account of vernacular poetic language, studies what he calls ‘l’écart rhétorique’, including typical literary elements found in 12% of the St Léger (Zumthor 1963: 62–6). In comparing the French Eulalia with its Latin counterpart, Zumthor (1963: 53–5) does not mention music, no doubt because of the absence of neums, the signs used to represent plainchant. In a broader study of the progress of vernacularization, from initial practical use, to use in literature, to being the basis of a learned culture with associated book production and storage, Mortensen (2018) assumes a book is a bound volume containing one or more substantial works.⁴ He is not interested in flyleaves, saying that ‘marginal afterthoughts in empty spaces like the Hildebrandslied should count differently than whole books or booklets made for vernacular texts’ (Mortensen 2018: 72). The early texts considered here are all short, and survive in copies, not the author’s original. We can list them with their sigla by language (numbers 1–7 are French, 8–14 Occitan, 15 Francoprovençal, and 16 the additional French text), add their paleographic date, indicate neums if present, and classify their mode of survival, as flyleaves, on blank spaces, or in books or booklets. ⁴ Mortensen (2018: 80–3) complicates the shift to literary use of the vernacular by noting in tenth century southern Italy what he calls a vernacular use of Latin, in which cultural references are removed and the language simplified to be better understood if read aloud.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
310
.
Table 14.1. Early vernacular texts from formerly Roman Gaul text, date and any music 1. Strasbourg Oaths (SO), s. xex
flyleaf or in empty space No
2. Sequence or Prose of St Eulalia (E), s. ixex (after 3 August 881); fol. 141 was blank after the main text but has the quire signature xviij 3. Jonas sermon (Jr&Jv), s. x¹
No
4. Passion de Jésus-Christ (P), s. x–xi, with neums above fol. 109v1 lines 1–3 5. Life of St Léger (L), s. x–xi, with neums above four different columns 6. Song of Songs poem ‘Quant li solleiz’ (CA), s. xi-xii
7. Epître farcie de St Etienne (EP), s. xii¹
8. Aube bilingue (A), s. x–xi, with neums throughout 9. Chanson de Sainte Foi (F), ximid
Yes, a binding fragment now a flyleaf; it was a single sheet in Latin except for vernacular aidemémoire purposes Yes, on blank space after letter E of the main text ends on fol. 109r Yes, on blank space after Letter In of the main text ends on fol. 159v Yes, on blank space after the main text, this being the fifth added after four Latin texts Yes, on blank space at the back of a miniature, using about two-thirds of that space Yes, using the full width of the blank space, but not the rest of the column No
10. Trope in Occitan on Tu autem (PLI), s. xiex, with neums throughout
No
11. Poème farci In hoc anni circulo alternating Latin and Occitan (PLII), s. xiex, with neums throughout
No
12. Versus Sancte Marie in Occitan (PLIII), s. xiex, with neums throughout 13. Sponsus, Latin play with some roles in Occitan (SP), s. xiex, with neums throughout
No
No
booklet or book Yes, a book, which cited 842 original as focal point of its Book III by 14 June 844 Yes, a booklet having been created by adding fols 142–3 for the French Eulalia and the German Ludwigslied on fols 141v–143r No
No, but at 516 lines it could have been copied from a booklet No, but at 240 lines it could have been copied from a booklet No
No
No
Yes, a booklet of two quires fols 14–29, with blank outer pages, fols 23v–28r having a Latin text Yes, a booklet of one quire fols 40–7 intercalated into this manuscript’s original five quires fols 32–9, 48–79 Yes, a booklet of five quires fols 32–9, 48–79, the original part of fols 32–118, this manuscript’s oldest section Yes, a booklet of five quires fols 32–9, 48–79, the original part of fols 32–118 Yes, a booklet of five quires fols 32–9, 48–79, the original part of fols 32–118
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
14. Boeci (B), s. xi–xii
15. Alexander text by Albéric (AL), s. xiiin
16. Augsburg Passion (AP), s. xex, with neums at first refrain
Yes, copied on blank space at the end of the manuscript, which is not enough to finish Yes, on a blank space using the ruled two columns on fol. 115v, but writing over them in long lines on fol. 116r, the rest of 116r–v staying blank Yes, on a book’s flyleaf later used for a 1067 Augsburg document, after the flyleaf was separated from its book
311
No, but at 258 lines when interrupted it could have been copied from a booklet No
No
Table 14.1 shows that the categories flyleaf, booklet, and book are permeable. A single sheet has become a flyleaf from being used for binding and has survived in a book (Jr&Jv);⁵ an original flyleaf has separated from its book and is now a document (AP). A booklet has been constructed from an unused final leaf and perhaps a bifolium, now all singletons (E). What will have been a single leaf has been copied into a Latin work (SO). A book may have spaces built into its organization (P and L), while spaces invite their use (CA, EP, A); the space used is usually sufficient, mostly with room to spare (AL), but can overrun or have lost its final leaves (B). A booklet can survive independently until bound with others (F), or it can become the core around which others form a book (PLII, PLIII, SP, and PLI). The format in which a text has survived, seven in books or booklets, seven on blank spaces, two on flyleaves, does not tell us about the format of previous copies. Mortensen’s three vernacularization stages are represented, the first of practical use applies to SO and Jr&Jv, the second of literary use, including SO, applies to fifteen (eight with neums and a ninth, E, requiring music), and all sixteen with the possible exception of F involve, earlier than Mortensen expects, a learnèd culture with book production and storage in ecclesiastical centres. Excluding F and AP, fourteen texts are preserved today in ten books named in the list of references; three are history books, two patristic books, two liturgical books, one a reference work, one a glossed Bible volume, one a paraliturgical collection. It is unlikely that the chance availability of space explains the copying of texts on blank spaces. Rather the implication is that these copies were expected to survive in an institutional library, and to be read by those who shared the learnèd culture which produced them. ⁵ For the abbreviations used to identify the texts hereafter, see Table 14.1.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
312
.
14.2 Comparing the vernacular forms 14.2.1 Writing the vernacular alongside Latin One practical and one literary reason for writing the vernacular involve Latin. The Jonas sermon is in Latin as the language in which sermons were written, but delivery after the 813 Council of Tours required the audience to understand; the Jonas preacher wrote in French those phrases he was unsure of getting right when speaking from his Latin script. The literary effects to be achieved by contrasting Latin with the vernacular explain our bilingual texts, for French Nithard quoting SO and the farciture of EP, for Occitan A, PLI, PLII, and SP. Concording these seven texts involves the vernaculars appearing alongside Latin. For an overview of how educated people present vernaculars beside the language of writing, the initial area of interest will be morphological, looking for examples across as wide a range of texts as possible, since the longer texts will occur disproportionately often.
14.2.2 Definite articles in the earliest texts Both Old French and Old Occitan have a two-case system, nominative and accusative. The data given by Kukenheim (1967: 16) and Anglade (1921: 211) for the expected definite article forms in the twelfth and thirteenth centuries provide a basis for comparing the definite articles found in the earliest texts (Table 14.2). Definite articles are used neither in the Latin portions of the concorded texts nor in SO, limiting the maximum number of texts which can attest a form in the concordance to fourteen. Setting aside the elided forms because by definition they lack distinctive vowels, each form attested by Kukenheim and Anglade will be verified in the concordance. For the singular forms, the concordance shows: first, the nominative li is not attested as a feminine singular, but as a masculine singular in E, Jv, P, L, CA, EP, and AL, so in the remaining six French texts and the one Table 14.2. Expected definite article forms in Old French and Old Occitan
Old French definite articles
Old Occitan definite articles
singular
singular
masculine
feminine
masculine
feminine
li, l’ lo, le plural li les
la, l’ la, l’
lo (or le) lo plural li (or lhi, rarely los) los
la (rarely li) la
les les
las las
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
313
Francoprovençal text, but in no Occitan text. Second, the nominative lo is attested in P, L, EP, F, PLI, PLII, B, and AL, three French texts, four Occitan and one Francoprovençal, and the accusative ‘lo’ in E, P, L, EP, F, PLI, PLIII, SP, B, and AL, four French texts, five Occitan and one Francoprovençal. Third, the nominative la is attested in E, Jr, P, CA, F, PLII, B, and AL, four French texts, three Occitan and one Francoprovençal, and the accusative la in E, Jv, P, L, CA, EP, F, PLIII, SP, B, and AL, six French texts, four Occitan and one Francoprovençal. Fourth, the nominative ‘le’ is attested solely in P, whereas the accusative le is in Jv, P, CA, EP, which could mean that P regards le as applicable to both cases, just like lo; uniquely one EP example of le is feminine, a northern form in that Central text. For the early texts, the forms lo and la are thus the normal forms for masculine and feminine respectively, in both nominative and accusative cases, across all three languages; since they represent the Late Latin development of the unstressed vowel, it is likely ‘lo’ is earlier than le, and it may be earlier than li. The form li first appears in the ninth century in the Wallon E, and le in the tenth-century Jv; these two distinctively non-Occitan forms compete with lo and la through to the early twelfth century. The plurals occur less often in the concordance. The masculine nominative li is attested in E, Jv, P, L, CA, EP, F, B, six French texts, and two Occitan, with Anglade’s lhi and los unattested. The masculine accusative les is attested in E, Jv, L, CA, the feminine nominative ‘les’ in CA, and the feminine accusative in P, CA, EP, a solely French form in six texts overall. As a masculine nominative, los is unattested in the concordance, but as an accusative it occurs in P, L, F, PLI, SP, B, two French and four Occitan texts. The feminine las is attested as a nominative in P, F, B, and an accusative in P, L, F, B, in overall two French and two Occitan texts. The distribution of these plural forms tells a similar story to that of the singulars, in that both French and Occitan attest the same forms, but with fewer examples there is greater reliance on the longer texts, P, L, F and B; of the forms los and las, only los is attested by other Occitan texts, and neither is attested by other French texts. Vowels are present in contracted article forms following some of the prepositions listed by Kukenheim (1967: 16), a, de, and en, and Anglade (1921: 212), a, de, per, sus and vers. The early texts have no contracted forms involving per, sus or vers, but do after a, de, and en. The singular al occurs in Jr, P, L, CA, F, PLII, PLIII, B, and AL, in four French texts, four Occitan, and one Francoprovençal; the vocalized au occurs in EP, F, PLII, one French text, and two Occitan. The plural ‘als’ occurs in P, L, F, two French texts, and one Occitan, and the variant alz in the Occitan F alone. The singular del occurs in P, L, CA, F, B, AL, three French texts, two Occitan, and one Francoprovençal. The plural ‘dels’ occurs in P, L, F, two French texts and one Occitan, and the variant delz in the Occitan F alone. The singular el occurs in Jr, P, L, EP, F, PLIII, B, AL, four French texts, three Occitan, and one Francoprovençal, the vocalized eu occurs in SP, one Occitan text, and ou
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
314
.
is not attested; neither is any potential plural such as els, elz, es, or ez. After a, de and en, it is the singular forms which occur most often, and across all three vernaculars; for the plurals it is only the forms ending in -z which are confined to Occitan, and they are peculiar to F alone. The overall impression given by the attested forms of the definite article is that a core set of forms for each case and gender is known everywhere, and that the differentiation by language reflected by Kukenheim and Anglade begins with French but is in its early stages.
14.2.3 Unstressed pronouns in the earliest texts Disregarding the nominative pronouns, essentially the same forms as for the definite articles also appear as third person unstressed pronouns, though less frequently. Again, the data given by Kukenheim (1967: 34) and Anglade (1921: 247–8) in Table 14.3 for the expected unstressed pronoun forms provides a basis for comparing the forms found in the earliest texts. This table shows fewer differences between the two languages than Table 14.2 above. Beginning with the dative forms, the singular li is attested for masculine and feminine in SO, E, Jv, P, L, CA, EP, F, PLIII, B, seven French, three Occitan, and no Francoprovençal texts. Similarly, the plural form lor is attested for masculine and feminine in Jv, P, L, EP, B, F, four French, two Occitan, and no Francoprovençal texts; Anglade’s rare form lur is found only in F, so in no French texts, and one atypical Occitan text. The more varied accusative forms attest la for the feminine singular in E, CA, EP, F, PLIII, B, three texts each from French and Occitan, but none from Francoprovençal. The masculine singular lo is attested in Jv, P, L, EP, F, PLII, PLIII, SP, B and AL, so in four French texts, five Occitan, and one Table 14.3. Expected unstressed third person pronouns in Old French and Old Occitan
Old French third person unstressed pronouns
Old Occitan third person unstressed pronouns
singular
singular
masculine
feminine
masculine
feminine
li lo, le plural lor les
li la
li lo plural lor (rarely lur) los
li la
lor les
lor (rarely lur) las
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
315
Francoprovençal. The masculine singular le is attested in CA, EP, so in two French texts, one of which also has lo. The masculine plural los is attested in P, L, B, F, two French, and two Occitan texts. The feminine plural las is attested in P, F, one French, one Occitan. The plural les is attested in Jv, L, two French texts. This is consistent with the idea that los is the oldest form, and that like lo it is used throughout Gallo-Romance before it begins to be replaced in the French areas. Their lack of vowels makes the cliticized forms mentioned by Kukenheim (1967: 34) and Anglade (1921: 247f.) difficult to assess. Suffice it to say for Occitan that no word ends in -lh, that -ll seems to be reduplicating a final consonant, and that the -ls examples behave as for the articles. Overall, although there is less differentiation and there are fewer examples than for the definite articles, these unstressed pronouns confirm that a core set of forms is known everywhere, and that the French le and les are not often used.
14.2.4 The preposition ab in the earliest texts The existence of ab in the French Augsburg Passion is one of Hilty’s (1994, 1995) reasons for regarding that text as Occitan. It is also a form that occurs in the two Clermont-Ferrand texts. It means ‘from’ in Latin, and, as a derivative of , ‘with’ in Romance. The concordance shows only two occurrences meaning ‘from’, one in a Latin line of A, the other in the French CA. In the sense ‘with’, it occurs as ap once in the Occitan PLIII, as ob four times in the French L, not at all as ou, and as o once in the Occitan F, once in the French P and twice in the French EP; it occurs as ab 57 times across seven texts, three French, (SO 1, P 9, L 5), three Occitan (F 25, PLII 1, B 8), and one Francoprovençal (AL 8). As with the articles, there is one widespread form from the beginning, with alternatives in o- predominantly French. Berschin and Berschin (2011: 212) point out that, according to Französisches etymologisches Wörterbuch 1970–2002, vol. 25, Latin develops phonetically to ab in the earliest French texts, before evolving to od.
14.2.5 Third person plural endings from unstressed Latin vowels The third person plural endings found in the concordance can be compared with expected and attested outcomes from Meyer’s study (Meyer 1880) of Occitan material, both documents and texts. He distinguishes three principal outcomes. The first is for (also involving the future), and , where French has ont, font and vont, and Occitan has aun / au / an / ant, faun / fau / fan / fant, and vaun / vau / van / vant. The second is for Latin -, involving the Latin conjugations I for the present indicative, II–IV (Meyer groups III and V as III) for the present subjunctive, all the imperfects, the conditional from the Latin
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
316
.
pluperfect indicative and the -essa imperfect subjunctives, all as detailed in Meyer 1880: 200–1; he has found forms in -avan, -ian and -eran, with later forms in -on or -ien, -ant, -unt or -ont. The third is for Latin - and -, involving the Latin conjugations II–IV for the present indicative, I (-) for the present subjunctive, and all (-) the preterites, with the same outcomes as for the second. The concordance results are shown for the first outcome, and then the second and third in Tables 14.4 and 14.5 respectively. Table 14.4. Third person plural endings from Meyer’s first group of unstressed Latin vowels outcomes of
outcomes of
outcomes of
ont P, CA
font not attested
vont P
aun F au F
future in -ront P
an P, F, B ant P, F, B future in -raun not attested future in -rau not attested future in -ran P, F future in -rant P
faun not attested fau not attested fan P, F, B fant F
vaun not attested vau not attested van P, F vant P Cf. outcome of stressed sont CA
Table 14.5. Third person plural endings from Meyer’s second and third groups of unstressed Latin vowels from Latin forms in -ent or -en pluperfects
forms in -eient or forms in forms in forms in -event or -oent or -an or -au -on or -ant, -unt -oient -ien or -ont
auran P
-eient Jv, EP
aurent L aurien F ( auret E, L, SP) ( pouret E)
eren P
-ant not attested erent Jv, P -event not attested -an P, L, -ien P, B -unt P, F, PLIII, B EP, AL ( iert CA) -oent not attested au F, PLII -ont not attested ( eret E, Jv, P, CA) -oient not attested -au F, B -ent Jr, Jv, P, L, CA, EP, F, SP, B, AL -en P, L, F, PLI, B
eran F
-on L, F
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
317
For this first outcome, there are not many examples, with a resultant bias to the longer texts. Nonetheless, P shows the full range of forms, whereas F and B are confined to Meyer’s Occitan forms and CA only has French forms, one of them stressed. The second and third outcomes have more examples. The best attested ending is -ent, in five French texts, three Occitan and one Francoprovençal, with -an attested in two French texts, three Occitan, including all the longest. The only French imperfect found is -eient, but both the Occitan forms that Meyer thought late, and the regional French imperfect endings, are unattested. These examples confirm that the principal endings are common to all vernaculars.
14.2.6 Some inconsistent usages The early texts show two survivals of the reflex of the Latin nominative singular , forcing a choice of how to represent the support vowel, and a similar choice arises for and ; the choices made in other cases are shown for comparison. To show consensus is possible, the forms of Latin are given in Table 14.6. Table 14.6. Outcomes from Latin , , , and
.
.
.
qant F=1
sendra SO=1 seindrae P=1 seinner F=2
nostra F=2, P=1 nostrae P=1 nostre P=1, EP=1, F=2
vostre F=1
quan F=9, B=6 quand F=1 quant SO=1, Jr=1, P=1, L=3, CA=1, EP=2, F=4, PLI=1, B=7, AL=2
senior L=1 seignor EP=1 . senior P=3, L=1
nostres P=1 noster (Latin) Jv=1 . nostro SO=1
sennior P=3 sennor F=1
nostra F=1, PLI=1 nostrae P=1
senor B=2
nostre P=2, CA=2, F=1, PLIII=1, B=1 . cf nostrorum (Latin) Jv=1, SP=1
. seniors L=1
. vostr’oleo SP=1 vostra F=1 vostre F=1, P=1 . vostras SP=1 vostres P=2, SP =1
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
318
.
For , SO and P choose a or ae for the support vowel, add the glide d, but only P represents any palatal with i; other support options are e in F, Latin o in L, EP, with the palatal inn F, ign EP, or Latin ni L. The accusative keeps the Latin P, L, adding s in L, doubling the n in P, F, removing the i in F, B. With five nominative solutions, four French one Occitan, and five accusative, three French two Occitan, a palatal is mostly represented, and the Latin spelling can be kept. For and , ignoring the Latin forms in Jv, SP, the nominative support vowel becomes a P, F, ae P, e P, EP, F (cf. vostre F), with nominative s in P. The accusative singular uses the nominative choices a F, PLI (cf. vostra F and plural vostras SP), ae P, e P, CA, F, PLIII, B, (cf. vostre P, F, and plural vostres in P, SP), but evades one choice by elision, SP. On the evidence of and , the ae ending is peculiar to P, whereas a is found in two French texts, two Occitan. For and , SO alone uses the etymological accusative o, whereas a is shared by one French text, three Occitan, and e by three French texts, three Occitan. As for the articles, the same forms are a shared resource, found in both languages. For , two options appear for the initial and three for the final consonant. All attested forms appear in F and two in B, both Occitan. Ten texts, six French, three Occitan including these two, and one Francoprovençal agree on a single solution.
14.2.7 Possible inferences It is too early to draw firm conclusions from this material. In 1963, De Poerck localized the manuscripts he studied with a view to understanding early dialects. While it is true that scribes writing books are exposed to the local dialect, the question is whether that dialect affected their scripta. The data presented here shows that the earliest vernacular texts shared some written forms across all three languages, and that it was French which began the process of individuation. The culture of tenth-century Clermont cathedral was very Latin, on the evidence of today’s MS 240. The differences between its two French texts seem less marked than has been claimed; the most that can be said is that the longer P has a wider range of forms than L. Chambon and Olivier (2000), the latter for the late medieval section, have provided an excellent overview of the materials available for writing the language history of the area, in particular its early history. Chambon’s account (in Chambon and Olivier 2000: 93–105) stresses the long lasting Romanization of the area, the deliberate conservatism of the region Aquitaine Provence, the local resistance to innovations from Lyon, and the two-case simplified ‘scripta latina rustica’ of sixth to eleventh century charters, written alongside proper Latin. With this linguistic background, the fact that the two vernacular texts in Clermont 240
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
319
were copied by different scribes should not alter the picture of educated writers who knew their Latin and were resistant to innovation. Pace Chambon (Chambon and Olivier 2000: 114)—who rightly rejects the view that P is more Occitan than L—it does not follow that ‘ces deux œuvres ont, au minimum, fait l’objet d’une très forte occitanisation’ (‘these two works have, at the very least, been the subject of very considerable occitanization’), because the scribes could simply be using the common fund of spellings they had. Before any conclusion is attempted, the Augsburg Passion will be examined to see what difference it may make to what has emerged from the concordance of the earliest texts.
14.3 Writing the vernacular 14.3.1 The Augsburg Passion This text was first published by Berschin et al. (1981: 252 and plates VI–VIII), after which Berschin and Berschin (2011) responded to subsequent work on it. The text is copied above a document of 1067, in a minute hand which gives a sixline poem complete in a single line; Berschin et al. (1981: plate VII) shows how faint it is, and plate VIII magnifies it considerably. The most natural explanation of the minute hand is that it was to fit the space above the existing document, which would place its transcription in or after 1067. However, an apparently contemporary hand appears on the document’s verso, recording in a Latin hexameter a donation by Bishop Erkanbald: Erkenbald praesul sanctae dat dona mariae (‘Bishop Erkenbald gives gifts to [the church of] St Mary’) (1981: 251 and plate VI). The only known Bishop Erkanbald was Bishop of Strasbourg from 965 to 991, significantly earlier than 1067. His political importance is indicated by Berschin et al. (1981: 270), and stated more explicitly by MacLean (2018: 242–6) on his relationship with the first two Ottos. Berschin et al. (1981: 254, 255) place the gift to his cathedral church in Erkanbald’s lifetime and the Passion, with neums over the first occurrence of the refrain, in the tenth century; Berschin and Berschin (2011: 209) date the text on palaeographic and transmission grounds to the late tenth century. The reconstructed transmission, Berschin et al. (1981: 251 and note 2, 269f., 256, 259), has Erkanbald’s donation note to Strasbourg Cathedral, found also in two other books, inscribed on the recto of the donated book’s first leaf, with the Romance text added on the blank verso in Erkanbald’s lifetime, and this book later transferred, perhaps via their shared Archdiocese of Mainz, to Augsburg cathedral, where the donation leaf was separated and used, a not uncommon practice at Augsburg, for a draft charter, whence the 1067 draft of Bishop Embriko of Augsburg eventually made its way to the Stadtarchiv. This transmission does not explain the extremely small hand in which the Passion is written, but if Bernhard Bischoff is correct in identifying the
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
320
.
work’s genre as a Sybilline prophecy (Berschin and Berschin 2011: 216), it may have been written this way to ensure its difficulty. The text is mostly clearly written, with word division respected by its editors. However, some different readings are possible, though not materially affecting its meaning as established by Berschin et al. from its content (Berschin et al. 1981: 261–3). The editors have published the text in six lines, two stanzas ending in a refrain, with four transcription points annotated before the edition (Berschin et al. 1981: 252). What I see from the photographs kindly supplied to me by the Augsburg Stadtarchiv would read, when divided into the six lines, as set out in Table 14.7. This revised diplomatic transcript needs interpreting to understand some of its forms. For clarity, prepositions should be separated as per modern word divisions. The morphology of caus is left as per Berschin and Berschin (2011: 212), and si in line 5 is understood as predicative, which makes this line interrupt the syntax of the stanza. Neither staudir nor haudir are attested in the concordance, though star and audir are. From the Passion narrative, the meaning is ‘they will pierce his side’, but if audir is accepted as the closest form, the line could personify lad expressing pain, ‘they will hear his side’. In line 5, the Berschin argument (Berschin et al. 1981: 263) that this line is from Psalms 68.22 seems well founded, and would dictate reading a c, whether the scribe wrote c or e, but l’acid is unattested in the concordance. For its part, greu is attested in the concordance but not the verb grever. Table 14.7. The Berschin et al. text as read and annotated from the manuscript Berschin et al. text Sneddon reading annotation 1. alespins batraunt sos caus 2. et abes lan staudiraunt sos lad 3. et en la crux lapenderaunt 4. et oblaeid lopotaraunt
abespins batraunt sos caus et abes lan haudira\u/nt sos lad & en la crux lapenderaunt et oblaeid lopotaraunt
5. si greu est a paerlaer
si greuent a pa\e/rlaer
6. et en la crux lapenderat
et en la crux lapenderat
Second letter has a faint bulb, and is b not l. No letter is missing after lan, but the first letter of the next word is not an st ligature but an h; its ending has a u placed above a. All letters are certain. The e in the second word is a two-compartment letter, neither e nor c; the i is like the first superscript minim on haudira\u/nt. The st ligature here differs from their line 2, and from those in A, SO and E. Rather it looks like a decorated second minim before t. The second ae has the letters back to back, whereas the first is closer to the two-compartment letter of line 4. The letters er are very small and have a slight space before at, with nothing after the final t, making the loss of un a scribal error.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
321
1. Ab espins batraunt sos caus 2. Et ab es lan haudira\u/nt sos lad 3. Et en la crux l’apenderaunt
With thorns they will beat his head And with that lance they will hear his side And on the cross they will hang him
4. Et ob l’acid lo potaraunt 5. Si greuent a pa\e/rlaer 6. Et en la crux l’apendera[un]t
And with the vinegar they will slake his thirst They grieve to speak of it And on the cross they will hang him
In terms of the forms discussed from the concordance, its use of ab fits into the pattern found there, as does the form ob. However, the striking thing about its morphology is that no -aunt ending is found in the concordance. Meyer (1880: 192) expected for the initial form *aunt, of which he had no example. The futures of the Augsburg Passion have supplied the necessary attestations, and Meyer’s asterisk can now be removed. This is another early form, still usable in late tenth-century Strasbourg cathedral for French, which was potentially available for all three vernaculars.
14.3.2 The circumstances of writing The evidence of the copies we have suggests that the majority of our early texts were written in a church environment. This is consistent with the idea that the first vernacular literary works could only be created by the educated, who could read and write Latin. As long as education was monastic, the genre of text would be broadly religious, and the writing would be as taught in the religious house concerned. Religious houses were in touch with each other, at least in their own area, enabling ideas and practices to be exchanged. However, a separate training to read and write in each house is more likely to result in a house style than a common fund of writing practices. A historian’s account of French document production in the thirteenth to sixteenth centuries in England, France, and the French-speaking parts of the Holy Roman Empire may help understand the role of training for writing. Lusignan (2004) comes to the conclusion that scribes had to be versatile in their use of language, and would adapt to the expectations of their employer. Their basic education would have made them literate in the medieval sense, able to read and write Latin. Writing in the vernacular was then a matter of adapting what they had learned in writing Latin, but avoiding reinventing the wheel by adopting the house style of their workshop. Knowing the vernacular was not straightforward because of the variety of ways of writing French, and a scribe who changed employer could find himself confronted with another variety of French. Lusignan (2004: 67) cites the example of La Rochelle, whose chancellery had a house style which all scribes must apply, and some 100 scribes have been identified there, not all of them local.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
322
.
If a chancery felt the need to have a house style, a workshop producing books for the lay market would find consistency across its production a help in securing commissions and finding customers. This commercial pressure will not apply before the commercial book trade begins in Paris under the impulse of the University in the late twelfth century, and then slowly but surely the lay market for vernacular books in the thirteenth century. Is it reasonable to look for equivalent pressures on those who wrote our earliest texts? And if so, what would constitute evidence of such pressures? Lusignan’s later evidence suggests that the first pressure on scribes, even at the beginning of vernacular literature, will have been to keep things simple. This would give rise to rules such as: ‘Do not diverge more than you have to from the Latin you know. Do not attempt to come closer to speech than you have to. Do make use of the practices of the scriptorium you work in’. Such rules do not claim that any standardization has occurred, the variety of forms found makes plain that is not the case, nor as Lodge (1993) has shown in his history of French are the sociolinguistic prerequisites for standardization met before the end of the Middle Ages. Since any such rules were transmitted verbally, there can be no direct evidence of their existence today. The indirect evidence of common forms derived from the concordance of early texts will be as close as it ever comes to proof.
14.4 Conclusion The underlying theme of this chapter has been that the earliest texts are both experimental in writing the vernacular as literature, and conservative in working within the Latin tradition. This tradition applies to the rules of writing, both as regards representing language on parchment and as regards adapting the chosen material to its intended audience, as per the rules of rhetoric. This blend of conservatism and innovation is what has appeared in the concordance of the earliest texts. A commonality of form underlies the numerous differences of detail. The morphological conservatism of our texts makes it easier for them, wherever they were composed, to be copied and read anywhere across Roman Gaul. If we look at the texts composed and copied before c.1000, SO and AP come from Strasbourg, a German-speaking area with a bilingual French-German ruling class well into the ninth century, but not necessarily as late as the end of the tenth century, which might put the composition of AP further back into the tenth century than its surviving copy. By contrast E and Jr/Jv come from a definitely bilingual French-German area, where distinctively French forms are more likely to be created. The last three texts copied before c.1000 are P, L and A, from respectively conservative Auvergne and the French/Occitan border of the Loire valley. The five eleventh-century texts come also from the Loire Valley (F), and from Limoges (PLI, PLII, PLIII, and SP), in what Chambon has identified (Chambon and Olivier 2000) as conservative Aquitaine. From about 1100, B too is from the
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
323
Loire Valley but CA is from non-francophone Brittany, while from the twelfth century EP is from francophone Tours; AL is unlocalized, but its Francoprovençal is thought of as originating in Lyon as discussed by Chambon through innovations not found outside its area. The conclusion to draw seems to be that our texts come mostly from conservative areas of Latinity or from French frontier areas, the latter being more likely to encourage experimentation. It is the conservatism which keeps a common stock of early vernacular morphology in use; the distinctive spoken morphology of French will be used, but not to the exclusion of the common stock of forms until it is no longer so daringly experimental to write in the vernacular at all, which on the evidence of our texts is the first half of the twelfth century, by which time French is already being written in bilingual England, with a new set of spelling conventions. Seeing Occitanisms in the Clermont manuscript is to miss the conservatism of the Auvergne and its preference for the common stock of forms.
List of primary sources Manuscripts, where accessible online accessed 10 April 2019; editions used for the concordance specified in full, or abbreviated if in secondary sources below. 1. Strasbourg Oaths (SO) Paris, BnF, MS latin 9768, Nithardus, De dissensionibus filiorum Hludovici Pii libri quatuor fols 1–18r, s. xex, at fol. 13r–v, from Abbey of Saint-Riquier (Somme) or Saint-Medard de Soissons (Aisne), a copy of a chronicle composed before 14 June 844 quoting the 842 vernacular oaths; link to digital copy https://archivesetmanuscrits.bnf.fr/ark:/12148/ cc572968 Lauer, P. (ed.), (1964). Nithard: Histoire des fils de Louis le Pieux. Paris: Les Belles Lettres, 104–8, two plates after XX 2. Sequence or Prose of St Eulalia (E) Valenciennes, Bibliothèque de la Ville, MS 150, Gregory of Nazianzus fols 1–140, s. ixmid, at fol. 141v, s. ixex (after Louis III’s victory at Saucourt celebrated in the Ludwigslied), from Abbey of Saint-Amand (Nord), in which the blank space fols 140v–141r and 143r–v around the French and German poems was used for Latin texts, including the Latin Eulalia on fol. 141r; link to digital copy from https://archivesetmanuscrits.bnf.fr/ark:/12148/cc91009c Koschwitz (1913: 7) 3. Jonas sermon (Jr&Jv) Valenciennes, Bibliothèque de la Ville, MS 521, Heraclides of Cyprus’s Paradysus etc. fols 1– 106, s. xin, leaf eroded on recto bound before fol. 1, s. x¹, from Abbey of Saint-Amand (Nord); link to digital copy from https://archivesetmanuscrits.bnf.fr/ark:/12148/cc90982r Koschwitz (1913: 9–14) 4. Passion de Jésus-Christ (P) Clermont-Ferrand, Bibl. du Patrimoine, MS 240, Liber glossarum fols 1–253 with end of quire spaces after each letter used mostly for Latin texts, s. x, fols 109v–111r, s. x–xi, from Clermont cathedral (Puy-de-Dôme), with neums above fol. 109v1 lines 1–3
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
324
.
Avalle, D’A. S. (1962). Cultura e lingua francese delle origini nella “Passion” di ClermontFerrand. Milan: Ricciardi, 95–126 5. Life of St Léger (L) Clermont-Ferrand, Bibl. du Patrimoine, MS 240, Liber glossarum fols 1–253 with end of quire spaces after each letter used mostly for Latin texts, s. x, fols 159v–160v, s. x–xi, from Clermont cathedral (Puy-de-Dôme), with neums above fols 159v3 line 1 and 160r3 line 1, and above 160v1 lines 1–4 and 160v2 line 1 Linskill, J. (1937). Saint-Léger: Étude de la langue du manuscrit de Clermont-Ferrand suivie d’une édition critique du texte. Paris: Droz, 151–77 6. Song of Songs poem ‘Quant li solleiz’ (CA) Paris, BnF, lat. 2297, Sacramentary fols 1–90, s. xex, the blank fols 91–2 being used for four Latin texts the last unfinished but for which the scribe left room, so that the fifth text fol. 92v, s. xi–xii, only just has room to finish, from Brittany perhaps Dol (Ille-et-Vilaine); link to digital copy from https://archivesetmanuscrits.bnf.fr/ark:/12148/cc601249 fol. 92v Meyer, P. (1877). Recueil d’anciens textes bas-latins, provençaux et français. 2e partie: Ancien français. Paris: Vieweg, 206–9 7. Epître farcie de St Etienne (EP) Tours, Bibl. municipale, Diocèse 1 (olim Petit Seminaire 583), Missal fols 1–332 in two parts between which is EP and at fols 209r–v four Latin items beginning with a settlement between priest and parishioners of Notre Dame at Avon-les-Roches, s. xi, fol. 208r, s. xii¹, from collégiale St. Gatien de Tours (Indre-et-Loire) Foerster, W. (1879). ‘Épître farcie de la Saint-Étienne, En vieux français du XIIe siècle’, Revue des langues romanes 16:5–15 (6–8), preceded by a facsimile 8. Aube bilingue (A) Vatican, Bibl. apostolica, Reg. lat. 1462, Fulgenti De Fabulis gentilium etc fols 1–50v1, s. viii–ix, fol. 50v2, s. x–xi, perhaps from Abbaye de Fleury in Saint-Benoît-sur-Loire (Loiret), with neums lines 1–7; link to digital copy from https://digi.vatlib.it/view/MSS_ Reg.lat.1462 Foerster, W. and Koschwitz, E. (1921). Altfranzösisches Übungsbuch. Leipzig: Reisland, cols 265–70 (267–8) 9. Chanson de Sainte Foi (F) Leiden, University Library, Codex Latinus oct. no. 60, a booklet Liber sancti Benedicti fols 1–13 and an unrelated booklet Occitan Sainte Foi and Latin text on the transfer of Madeleine relics to Vezelay fols 14–29, both s. ximid, fols 14v–23r, perhaps from Abbaye de Fleury in Saint-Benoît-sur-Loire (Loiret); description based on Thomas (1925: XVI–XVIII) Thomas, A. (1925). La Chanson de sainte Foi d’Agen: poème provençal du XIe siècle. Paris: Champion, 2–37 with facing facsimile of the poem’s 593 lines 10. Trope on Tu autem (PLI) Paris, BnF, lat. 1139, collection of paraliturgical pieces etc fols 1–236, s. xi–xiii, fol. 44r, s. xiex in an intercalated quire described by De Poerck (1969: 307–8), from Abbaye de Saint-Martial de Limoges (Haute-Vienne), with neums throughout; link to digital copy from https://archivesetmanuscrits.bnf.fr/ark:/12148/cc59066k
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
325
Thomas, L.-P. (1951). Le ‘Sponsus’ (Mystère des Vierges sages et des Vierges folles) suivi des trois poèmes limousins et farcis du même manuscrit: Étude critique, textes, musique, notes et glossaire. Paris: Presses universitaires de France, 195 11. Poème farci In hoc anni circulo (PLII) Paris, BnF, lat. 1139, collection of paraliturgical pieces etc fols 1–236, s. xi–xiii, fols 48r–49r, s. xiex in five quires described by De Poerck (1969: 303–7), from Abbaye de Saint-Martial de Limoges (Haute-Vienne), with neums throughout; link to digital copy from https://archivesetmanuscrits.bnf.fr/ark:/12148/cc59066k Thomas, L.-P. (1951). Le ‘Sponsus’ (Mystère des Vierges sages et des Vierges folles) suivi des trois poèmes limousins et farcis du même manuscrit: Étude critique, textes, musique, notes et glossaire. Paris: Presses universitaires de France, 196–8 12. Versus Sancte Marie (PLIII) Paris, BnF, lat. 1139, collection of paraliturgical pieces etc fols 1–236 s. xi–xiii, fols 49r–50r, s. xiex in five quires described by De Poerck (1969: 303–7), from Abbaye de Saint-Martial de Limoges (Haute-Vienne), with neums throughout; link to digital copy from https://archivesetmanuscrits.bnf.fr/ark:/12148/cc59066k Thomas, L.-P. (1951). Le ‘Sponsus’ (Mystère des Vierges sages et des Vierges folles) suivi des trois poèmes limousins et farcis du même manuscrit: Étude critique, textes, musique, notes et glossaire. Paris: Presses universitaires de France, 199–200 13. Sponsus (SP) Paris, BnF, lat. 1139, collection of paraliturgical pieces etc fols 1–236 s. xi–xiii, fols 53r–55v, s. xiex in five quires described by De Poerck (1969: 303–07), from Abbaye de SaintMartial de Limoges (Haute-Vienne), with neums throughout; link to digital copy from https://archivesetmanuscrits.bnf.fr/ark:/12148/cc59066k Thomas, L.-P. (1951). Le ‘Sponsus’ (Mystère des Vierges sages et des Vierges folles) suivi des trois poèmes limousins et farcis du même manuscrit: Étude critique, textes, musique, notes et glossaire. Paris: Presses universitaires de France, 174–87 14. Boeci (B) Orléans, Médiathèque, MS 444, glossed Jérémie and Ezéchiel etc. fols 1–269 s. xi, fols 269v–272v, s. xi–xii, from Abbaye de Fleury (Loiret); link to digital copy from http://aurelia. orleans.fr/archives-et-manuscrit by looking for 444 or 444 Jérémie Schwarze, C. (1963). Der altprovenzalische ‘Boeci’. Münster: Aschendorff, 18–161 15. Alexander text by Albéric (AL) Florence, Bibl. Laur., Plutei 64.35 Quinti Curtii Rufi Historiarum Alexandri Magni Libri fols 1–124 s. xiin, Francoprovençal text on Alexander’s education fols 115v–116r, s. xiiin; link to digital copy from http://mss.bmlonline.it/s.aspx?Id=AWOItZ95I1A4r7GxMMFZ#/ book Meyer, P. (1886). Alexandre le Grand dans la littérature française du moyen âge. Bibliothèque française du moyen âge, 4–5. Paris: Vieweg, Tome premier: Textes, 1–9 16. Augsburg Passion (AP) Augsburg, Stadtarchiv, UrkS Nr 5 document dated 1067, French text s. xex, apparently from Strasbourg cathedral; text not included in the concordance Berschin, Berschin, and Schmidt (1981:252 and plates VI–VIII)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
15 Appositive compounds in dialectal and sociolinguistic varieties of French Brigitte L. M. Bauer
15.1 Introduction Well-attested in contemporary French, appositive compounds—compounds which contain two nouns in apposition—have undergone a complex development in Indo-European. Productive from the early languages onwards, they have undergone a sharp increase over the past hundred years in French and elsewhere in Romance and they are now among the most widespread and productive types of compound, cf. e.g. Fr. train-éclair ‘train-lightning’ > ‘bullet train’, médecinjuriste ‘physician-lawyer’, hôtel-restaurant ‘hotel restaurant’, bébé-phoque ‘baby-seal’ > ‘young seal’, homme-grenouille ‘man-frog’ > ‘diver’, and so forth. Recent diachronic research has determined that compounds based on apposition in very early Indo-European languages were formations in which the component nouns were fully equivalent elements. As a result of grammatical change, French and Romance appositive compounds are somewhat different in that one of the component nouns may prevail semantically and / or grammatically. Moreover, it has been established that these formations have distinct characteristics and that they have undergone a change in word order. While detailed historical research into this phenomenon to date has focused on (standard) French, this article aims to further complement the picture with dialectal and sociolinguistic data. Although the different types of appositive formation and their evolution have now been established (Bauer 2017), further analysis is needed to trace and account for their dramatic spread in French. Little is known, for example, about their incidence and usage in non-standard French, especially elsewhere in GalloRomance. Yet these data may be quite useful because they often reflect spoken language and provide information as to the geographical spreading of types of appositive compound, related grammatical characteristics, and potential archaic features. This article will therefore analyse appositive compounds in a number of Gallo-Romance dialects, taking into account the potential role of metonymy, metaphors, word order and early attestations of later trends. These findings will be evaluated against data from poilu, the sociolect spoken by the French soldiers
Variation and Change in Gallo-Romance Grammar. First edition. Sam Wolfe and Martin Maiden (eds). This chapter © Brigitte L. M. Bauer 2020. First published in 2020 by Oxford University Press.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
327
during the First World War. Poilu is of special interest in this context because it was a rather expressive, primarily spoken, sociolect, with its speakers coming from all over France, often with a dialectal background. Analysis will help identify a potential correlation between certain subtypes and dialectal and social stratification. First, this article briefly assesses appositive compounds and their development in Indo-European and French, identifying the different types and their characteristics (§15.2). Subsequently, §15.3 investigates data drawn from a number of Gallo-Romance varieties, whereas §15.4 focuses on appositive compounds in poilu. The aim of both sections is to identify and evaluate patterns in variation and change. Section 15.5 discusses the findings and formulates conclusions.
15.2 Appositive compounds 15.2.1 Introduction Composition—the combining of two autonomous parts of speech into a new lexeme—reflects an underlying grammatical relation (e.g. Benveniste 1974), e.g. that of an adjective and a noun (e.g. Fr. vinaigre ‘vinegar’, < vin ‘wine’ + aigre ‘sour’), a verb and a direct object (e.g. Fr. tire-bouchon ‘corkscrew’, < tire ‘pull’ + bouchon ‘cork’), a preposition, a noun and a preposition (e.g. Fr. au sommet de ‘on top of ’), and numerous others. In the handbooks (e.g. Brugmann 1906) these compounds commonly are referred to as determinatives, in which one element explicitly determines the other, in a relation of government or attribute (adjective + noun); for an overview of the classification of compounds since Bloomfield and its inconsistencies over the past thirty years, see Scalise and Bisetto 2009: 35–44). A separate category of compounds is widespread in contemporary French and is formed by compounds that include two identical parts of speech, e.g. nouns in apposition, as in homme-grenouille ‘diver’ (< homme ‘man’ + grenouille ‘frog’). They are the topic of this article (for the position of appositive compounds in the classification of compounds, see Bauer 2017: 290–307).
15.2.2 Main characteristics Appositive compounds are the result of the combining of two nouns in apposition to form a new lexeme, as in French visite-éclair ‘visit + lightning’ > ‘a lightning visit’, coq-faisan ‘cock + pheasant’ > ‘male pheasant’, or hôtel-restaurant ‘a public place that offers hotel and restaurant services’. In languages that feature case, the equivalence of the nouns translates in case agreement: in similar instances in Latin, for example, the two nouns feature the same case, cf. Lat. arbore
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
328
. .
olea (Varro¹) ‘tree-ABL olive-ABL’> ‘(from an) olive tree’ or porcum feminam (Cato²) ‘pig-ACC female.animal-ACC’ > ‘sow’. Appositive compounds have been attested from early Indo-European. Yet the relation between the component elements has shifted over time. Among the earliest attestations were so-called dvandvas (lit. ‘pairs’), which combine two ́ ‘father.mother’. In this complementary nouns in correlation, e.g. Skt mātárārā compound denoting ‘parents’, ‘father’ is inherently complementary to ‘mother’ and vice versa and their complementarity is exclusive (see Benveniste 1974). These formations co-existed in early times with additional compounds and hybrids, which were also appositive compounds and based on a relation of coordination. In additional compounds, the component nouns form an entity, without being complementary (e.g. Skt ajāvāyas ‘goats+sheep’ > ‘small cattle’, but also Gk. χορτάρακος (< χορτ- ‘grass’ + ρακ- ‘strip of cloth’ > ‘fodder’). Hybrids, on the other hand, combine two nouns that refer to two qualities in a person or object, e.g. OHG arzatgot ‘physician+god’ (>Asclepius), OE werewulf (‘man+wolf ’), which denotes a creature who is a man and a wolf at the same time, Ger. Dichter-Komponist is both a poet and a composer, or Eng. Dallas-Fort Worth, where reference includes both Dallas and Fort Worth (see L. Bauer 2010 for a discussion about these last very productive formations in Germanic). Finally, appositive compounds include so-called ‘type-of compounds’, which are based on hypotaxis because one of the two nouns specifies the other, cf. OE lēonhwelp ‘lion+young.animal’ > ‘lion cub’, Ger. Walfisch ‘whale+fish’ > ‘whale’, or Fr. bébéphoque ‘baby-seal’ > ‘young seal’, and so forth (for an overview of these compounds in early Indo-European languages, see Bauer 2017: 307–55). Over time, an important change has taken place cross-linguistically: the earliest Indo-European languages typically include dvandvas, additional compounds, and hybrids—with a very limited number of type-of compounds. The compounds of modern times as reflected in Romance (and Germanic), on the other hand, are typically hybrids and type-of compounds, with a few attested instances of additional compounds (for details, see Bauer 2017: 288–371). In Romance, this shift translates into a dramatic increase in recent times, especially of type-of compounds: their incidence and productivity has gone up sharply since the early twentieth century.
15.2.3 Appositive compounds in French As in the other Romance languages, appositive compounds in French may be divided into two groups: (a) ‘hybrids’—e.g. hôtel-restaurant—and (b) ‘type-of
¹ R.R.: De Re Rustica.
² Agr.: De agricultura.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
329
compounds’, such as train-éclair, which refers to a type of train. Dvandvas and additional compounds are almost non-existent. While hybrids have always been widespread and productive, with ample attestations in early as well as contemporary Indo-European languages, type-of compounds had a slow start and came into full development in later times, often with their own characteristics in the individual languages. While type-of compounds in Germanic, for example, are often metonymic (e.g. Ger. Walfisch above, ‘whale+ fish’ > ‘whale’, including a generic noun and a noun denoting a subspecies), the components of those in Romance are often in a metaphorical relation, for example, as illustrated, e.g. in bateau-mouche ‘boat-fly’ > ‘tourist boat’ (in Paris), in which the second noun denotes a concept with which the first noun is tacitly compared: the Paris tourist boat reminds one of a fly. Similarly, in homme-grenouille ‘diver’ the man is compared to a frog, and in roman fleuve ‘novel-river’ > ‘novel sequence’ the novel (roman) reminds one of a river (fleuve).³ With the most common types of appositive compound in French being hybrids and type-of compounds, numerous instances can be quoted. Hybrids: e.g. hôtel-restaurant; médecin-juriste, écrivain-cinéaste, maître-maçon Type-of compounds: e.g. chef-lieu ‘head-location’> ‘capital’; coq-faisan ‘cockpheasant’ > ‘male pheasant’; mère-patrie ‘mother-fatherland’ > ‘motherland’; langue-mère ‘language-mother’ > ‘mother language’; ferme-école ‘farm-school’ > ‘training farm’; homme-grenouille ‘man-frog’ > ‘diver’; bateau-mouche ‘boat-fly’ > ‘tourist boat’; train-éclair ‘train-lightning’ > ‘bullet train’. Although these compounds combine equivalent nouns, there may be a prevalence of one of the nouns, in grammar and / or meaning: bateau-mouche, for example, denotes a type of ship while train-éclair denotes a type of train. Similarly, a ferme-école is a farm where agriculture in taught and a navire-école is a training ship (navire). In other words, one of the two nouns determines the overall meaning of the compound. For hybrids, this semantic prevalence may be absent at face value, because the nouns refer to two qualities in the same referent. Yet there may be a subtle but pertinent difference. In an article discussing a trend among contemporary French authors to make films, Beigbeder (2012), refers to them in terms of écrivainscinéastes (lit. ‘writers-film makers’): they are primarily writers who expand their professional activities into cinema. Similarly, a médecin-juriste in all likelihood is a practising physician with an additional degree in law, whereas the individual who practices law, but also has a medical degree, will be referred to as juriste-médecin. ³ A roman fleuve ‘sequence novel’ refers to a series of novels—each with its own title and storyline— that share characters and/or settings, e.g. Les Thibauts by Roger Martin du Gard (1922–1940) narrating the life of two brothers.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
330
. .
Similarly canne-parapluie ‘cane-F umbrella-M’ is a cane that may also be used as an umbrella, not the other way around. In this instance, the prevalence of canne is also reflected in gender assignment: feminine canne—rather than masculine parapluie— determines the gender of the compound. In type-of compounds as well, the prevalence of one of the two nouns may be manifest in gender use, when the compound includes nouns with different gender, as in bateau-mouche, that combines masculine bateau and feminine mouche, the compound is masculine. Similarly it is un navire-école ‘training ship’, une visite-éclair ‘lightning visit’, and so forth. The noun that determines the gender of the compound systematically is the noun that determines its prevalent meaning as well. Moreover, in the vast majority of the examples above the element that has semantic and grammatical prevalence occurs in initial position, following the patterns of an SVO or right-branching language, which modern French is. In a right-branching language the head element comes first, followed by its complement, be it a direct object, a genitive, or an adjective or appositive noun, cf. il rédige un livre ‘he edits a book’; la vie de son père ‘his father’s life’ (lit. ‘the life of his father’), un livre anglais ‘an English book’ (lit. ‘a book English’), rue Bonaparte ‘Bonaparte street’, l’homme-grenouille ‘diver’ (lit. ‘man-frog’), and so forth (see Greenberg 1963; Bauer 1995, 2017). This observation brings us to a group of formations in French that are different; see Table 15.1. In these, the second element is the head of the structure, determining its semantic reading and grammatical gender. Diachronic analysis has shown that these are early formations, going back to the Middle Ages and reflecting the archaic left-branching structure. It is noteworthy that mère-branche has over time become branche-mère (first attested in 1721). In this light, appositive compounds referring to contemporary phenomena are of special interest: mère typically comes second, as recent databases such as
Table 15.1. Compounds with second element as head le chef-lieu le coq-faisan la mère-patrie la mère-branche⁴ la bouc-chèvre
head-location cock-pheasant mother-fatherland mother-branch buck-goat
> > > > >
‘capital’ ‘male pheasant’ ‘country of origin’ ‘main branch’ (of a tree) ‘sterile goat’
type of location type of pheasant type of fatherland type of branch type of goat
⁴ A mère-branche is a branch that emerges directly from the trunk of the tree. Mère-patrie is the country that has colonies, the country where one is born, and the country where one’s ancestors were born (Petit Robert 1993: 1388).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
331
Wikipédia show,⁵ cf. société mère ‘company mother’ > ‘parent company’, fichier-mère ‘mother file’, idée mère ‘main idea of a theory or book’ langue mère ‘language-mother’ > ‘language from which other languages descend’, maison-mère ‘headquarters of an organization’, roche mère ‘rock at the base of certain terrains’, solution mère ‘core solution’, liqueur mère ‘salted water after crystallization’, microfiche-mère, and others. This example further illustrates that appositive compounds play an essential role in serial compounding: many nouns recur in appositive compounds conveying a central concept, for example, école in navire-école, auto-école, ferme-école, with the concept of training, industrie-pilote, projet-pilote, with the concept of ‘experiment’, attaque-surprise, grève-surprise (lit. ‘strike-surprise’), fête-surprise (lit. ‘party-surprise’), conveying the concept of surprise, and so forth. These processes are complemented by instances denoting feminine occupational terms (e.g. femme-professeur ‘woman-professor’) and diminutives (e.g. bébé-Peugeot ‘small Peugeot’; see further §15.5). In sum, appositive compounds in Indo-European have undergone a shift from fully equivalent elements (dvandvas), based on correlation, to structures in which one of the component elements determines the gender of the compound and has semantic prevalence. In modern French that element is in initial position, with a few left-branching residues (for a full analysis and discussion, see Bauer 2017). The following sections will examine the early days of the spread of these compounds, focusing on the occurrence of appositive compounds in twentiethcentury Gallo-Romance varieties and poilu.
15.3 Appositive compounds in Gallo-Romance dialects 15.3.1 Introduction The varieties that have been analysed for this study include the northern GalloRomance dialects of Lille (Picard) and La Gleize (Walloon), Gallo-Romance dialects in southern France, such as Gascon, Occitan, and Limousin, and Francoprovençal in Provence, and the dialect of Bagnes (Switzerland). The related societies vary from agricultural communities to regions with agriculture, forestry and industry (e.g. Limousin, in Robert et al. 1968: xxxii–xxxiii), to urban societies, as in Lille. Dialects of mountainous areas are rather prominent in this corpus, reflecting the focus of certain publications used for this study. ⁵ I am grateful to Prof. A. Meyer and Mr J. van Paridon (Max Planck Institute for Psycholinguistics) for providing me access to a major French text database, a snapshot of the French-language Wikipédia comprising 543 million words (https://dumps.wikimedia.org/frwiki/latest; accessed 1 December 2018). The tools necessary to access the data can be found in Van Paridon, J. and Thompson, B. (2019). subs2vec: Word embeddings from subtitles in 55 languages. (DOI: 10.31234/osf.io/fcrmy).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
332
. .
Rohlfs (1970), for example, found that mountain dialects are best preserved and least affected by standard French (Rohlfs 1970: 2). Similarly, the dialect of Bagnes is spoken in a mountainous tributary valley of the Rhône and was very much alive in the mid-twentieth century because of its geographical isolation, its economic isolation as defined by its self-sustaining agriculture, and its cultural isolation as a Catholic community in an otherwise protestant region (for details, see Bjerrome 1957: 9–12). The dialect of La Gleize is a rural dialect as well, spoken in a group of villages not too far from the town of Liège.
15.3.2 Sources and general trends Data have been drawn from a variety of corpora, including several dictionaries of Gallo-Romance dialects, such as the dictionary of the Limousin dialect (Robert et al. 1968), Gascon (Palay 1974), and the dialect of Haut-de-France as spoken in Lille (Vermesse 1861). While these dictionaries were systematically scrutinized, others were consulted as well, such as Alibert (1965), offering dialectal varieties over a fairly wide area. In addition, grammars and linguistic descriptions have been investigated, since they often provide extensive vocabulary data, such as Ronjat’s grammar of Provençal (Ronjat 1930, 1937) and Bjerrome’s analysis (Bjerrome 1957) of the dialect of Bagnes, featuring an extensive lexicon. Finally, detailed lexical studies of given varieties have been scrutinized, such as that by Rohlfs (1970) on Gascon or the syntactic analysis by Remacle (1952) of the dialect of La Gleize. Instances of appositive compounds found in any given source have been cross-checked in the other publications. The sources are based on studies by the authors in situ—trained linguists— most of them covering a long period of research (e.g. Remacle 1952: 15–39; Bjerrome 1957: 12f.; Rohlfs 1970: 2; Palay 1974); one of the authors was also a native speaker of the given dialect (Remacle 1952: 16f.). Most authors specify that their data reflect ‘spontaneous speech’ (e.g. Bjerrome 1957: 12–14; Remacle 1952: 17f., 19, 23, passim) and come from extensive exchanges with the local population reflecting all social groups (e.g. Rohlfs 1970: 2; Remacle 1952: 17f.). Where possible, written sources have been included as well (e.g. Ronjat 1930: 55). On the whole, there is no specification as to social groups in the actual descriptions in the lemmata. A few additional comments about the sources are called for. First, none of them is exhaustive or tries to be so, not even the dictionaries. Many appositive compounds may therefore simply have been left out. Moreover, appositive formations may be easily interpreted by some as noun phrases rather than compounds and may therefore not have been included. Finally, the sources do not provide the detailed information about the entries that is found, for example, in the Robert: generally there is no information about the plural, context, register or date of attestation, and so forth.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
333
In terms of general trends, the data show—first of all—that appositive compounds are not overwhelmingly widespread in the dialects investigated. Their systematic analysis brought up around eighty instances. In all, compounds other than appositive compounds are distinctly more common, taking the form of verb + noun (e.g. Lim. s(o)ufro-doulour ‘suffer-pain’> ‘victim’, noun + adjective (e.g. Lim. mai-grando ‘mother-grand’> ‘grand-mother’), adjective + noun (e.g. Lim. gran-pai ‘grand-father’ > ‘grand-father’; fāū ple ‘wrong-fold’ > ‘bad habit’), noun + preposition + noun (e.g. Lim. frui de gardo ‘fruit of conservation’ > ‘fruit that keeps during winter’), and so forth: these are determinative compounds, including two or more different parts of speech and based on an explicit hierarchical relation. While appositive compounds form a relatively small group in Gallo-Romance, they seem to be rather diverse at face value. Yet there are clear patterns. The following paragraphs will assess the most common types of appositive compound, evaluating their structural and semantic characteristics.
15.3.3 Appositive compounds: hybrids Hybrids—conveying two qualities in the same referent—are relatively widespread, cf. for example, Lim. hōme-Dī and Gsc. omi-Dioü (Fox 2003: 245) ‘man-God’ > ‘Christ’, an important religious concept, but not attested in all consulted publications. Hybrid formations in reference to human-beings in a professional capacity are commonly attested in all dialects, as in Prv. baile-pastre ‘chief-shepherd’, capmèstre ‘chief-master’ > ‘grand master’ (Ronjat 1937: 476). The Gascon variety capmèstę or cap-mèstre, denotes ‘foreman’ (e.g. of a construction crew). Gascon also features mèstę-baylèt ‘master-valet’ > ‘foreman’, but also ‘managing agent’. In Limousin, meitre occurs in contexts of this type as well, e.g. meitre-oubriei ‘master-worker’ > ‘master’, in addition to its feminine variety meitresse, as in Lim. meitresso fenno ‘master-F woman’; its precise meaning remains elusive. The French translation runs ‘maîtresse femme’ (Robert et al. 1969: 240), and it could refer to a woman who is ‘wife and lady of the house’ at the same time. Yet in isolation, meitrisso denotes ‘fiancée’, a woman who has a ‘suitor’ (Robert et al. 1969: 240). Similarly, Gsc. mèstro-bayléto ‘mistress-valet-F’ > ‘ female tenant’ or ‘mèstę-baylèt’s wife’. In rare instances, the compound may include more than two elements, as in Lim. reigranpai and feminine reino grando-mai, referring to great-grandparents. The common adjective + noun compound conveying ‘grandparents’ has been extended with an additional noun, rei ‘king’ and reino ‘queen’ respectively, cf. Lim. reigranpai lit. ‘king.grand.father’ ‘great-grand-father’; reino grando-mai lit. ‘queen grand-mother’ ‘great-grand-mother’. The formation may be shortened, suggesting it is commonly used, e.g. Lim. reigrando-mai lit. ‘king.grand-mother’ ‘great-grandmother’. An important feature of these formations is the alternation rei vs reino,
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
334
. .
suggesting that they are genuine and productive appositive compounds. Dictionaries of other dialects do not provide this formation. Instead they feature equivalents of Fr. arrière-grand-père, etc. Moreover, hybrid formations in the dialects also denote concrete objects and parallel forms may be found in several dialects, such as LG banc-cofe ‘bench-chest’ > ‘an elongated chest that also functions as a bench’ (Remacle 1952: 80), which includes banc (< Gmc bank ‘bench’). Cofe goes back to Lat. ‘basket’. The dialect of Bagnes and Gascon offer similar examples, but with the reverse sequence: Ba. artse-bã ‘chest-bench’, which is defined as a ‘chest that also functions as a bench, seat bench’ (Bjerrome 1957: 154) as well as Gsc. arquebanc and Prv. archibanc, with the same meaning. They go back to Lat. ‘chest’, which is rare in contemporary literary French (Bloch & von Wartburg 1996 [1932]: 34) and Gmc bank ‘bench’). The question is whether these all denote the same concept. As indicated earlier (§15.2.3), compounds vary in meaning according to the prevalent noun and the question is whether in LG banc-cofe, Ba. artse-bã, and Gsc. arquebanc emphasis is on the concept ‘bench’ (seating) or on the concept ‘chest’ (storage). In the translations by Bjerrome (1957) and Ronjat (1937) artse ‘chest’ is semantically prevalent. Yet, archibanc, Ba. artse-bã, and Gsc. arquebanc—including masculine banc / bã and feminine artse / arque—are masculine compounds. This distribution of gender suggests that banc / bã are head nouns and that the compound therefore denotes a type of bench, rather than a chest. The compound therefore is leftbranching. This observation is relevant in light of the Walloon example, with the specifying noun preceding the head noun as well (see §15.3.7).
15.3.4 Appositive compounds: additional compounds The data provide several instances of additional compounds, which as noted above are very rare in standard French. The well-known Mediterranean sauce made with egg, garlic and oil, for example, is referred to by an additional compound, such as Prv. aiòli ‘garlic-oil’, which combines two ingredients that together form an entity, but are not complementary. Similarly, the concept ‘all seasons’ is conveyed by two nouns in apposition in Prv. ivèr-estiéu ‘wintersummer’. But note that time indications as a rule are not additional compounds. The concept ‘day and night’, for example, is conveyed by Prv. nieuch-e jour ‘nightand day’, with a connecting element. Additional compounds are also found in formations of the type Prv. aigo-nèu ‘water-snow’ referring to ‘rain mixed with snow, sleet’ (Gsc. aygue néu ‘water-snow’) as well as Gsc. aygue-sàu ‘water-salt’ > ‘brine’ (Prv. aigo-sau; examples from Ronjat 1937). Moreover, additional appositive compounds may also have an adjectival variety, as found, e.g. in nominalized Gsc. douçe-amare and douçamèro ‘sweet-bitter’ > ‘licorice’, in which sweet and bitter are components of the flavour. In Provençal,
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
335
the adjectival compound douço-amaro—including the adjectives sweet and bitter—refers to an unspecified plant (Ronjat 1937: 465), which means that it is impossible tell whether the formation is a hybrid or an additional compound.
15.3.5 Appositive compounds: type-of compounds Type-of compounds refer to (everyday) objects, plants and animals. Ba. söpa palūva and mësa palūva, for example, denote a drink and a liturgical mass respectively. In these compounds palūva denotes a ‘mother who has just given birth’ and for whom a special drink is prepared (söpa palūva) and a special religious service is held upon her return to normal life (mësa palūva). The noun palũva is a derivative of pàle ‘straw’, which is connected to child birth, cf. the expression in the same dialect a fïna p’i pàle ‘his wife (fïna) is in child birth’ (Bjerrome 1957: 214). In other dialects as well, the link between child birth and straw is attested (e.g. Occ. palholada ‘care provided to a woman shortly after giving birth’ (< palha ‘straw’), even if no explicit reference is made to traditions surrounding the event of childbirth. Type-of compounds are more commonly attested in the names of plants, such as Ba. tsū-r˙á ‘cole-rape’, denoting a type of cole or Prv. bledo-rabo ‘beet-rape’ > ‘beet’ (Ronjat 1937: 465). In addition, compounds that include a generic noun are rather exceptional in standard French, but there are several instances in the dialects, such as Prv. sèco magardo in which sèco is a generic plant name (‘rose’) and magardo denotes the subspecies (‘eglantine’). Similarly, in Ba. pai basë, generic pai ‘pea, bean’ combines with the subspecies basë ‘dwarf bean’. The pai basë differs from the pai a pèrtse ‘bean at pole’ > ‘pole bean’. These types of formation are based on the metonymical relation between a generic noun and a noun conveying its subspecies. Within the category of animal nouns as well, there are several formations that combine a generic noun with a noun conveying a subspecies, such as Lim. chi dōgue ‘dog watchdog’ > ‘watchdog’ and chi lebriei ‘dog greyhound’ > ‘greyhound’, in which chi functions as a generic noun and dōgue and lebriei—commonly used as an independent noun as well—refer to a subspecies. While chi was obviously inherited from Latin (< ), dōgue was a borrowing from Germanic, possibly from Eng. dog. Chi may originally have had an explanatory function. These formations—combining a generic noun with a noun denoting a subspecies— were and still are widespread in Germanic languages, especially in animal names (e.g. OHG tigertier ‘tiger+animal’ > ‘tiger’, MHG walfisch ‘whale+fish’ > ‘whale’, and so forth (for further data, see Carr 1939; for additional analysis, see Bauer 2017: 334–6). Similarly, the sex or the age of animals is often conveyed by appositive compounds in Germanic from early times on, as well, e.g. ON karlfugl ‘male+bird’, ON kvennfugl ‘female+bird’ > ‘henbird’, OHG windzôha
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
336
. .
‘greyhound+bitch’, OE gātbucca ‘goat-buck’ > ‘he-goat’, and Germanic age indications (e.g. OE hindcealf ‘deer+young.animal’ > ‘fawn’, OE lēonhwelp ‘lion +cub’ > ‘whelp’). Several of these formations continue to exist in contemporary Germanic, cf. e.g. Eng. turkeyhen, buckgoat, peacock, or Ger. Hirschkuh ‘hind’, Ziegenbock ‘billy goat’, Schafbock ‘ram’ (see further Carr 1939: 326–8 and passim). Even if Gallo-Romance dialects feature these formations, they do so to a much lesser extent than Germanic languages. One explanation of this discrepancy between French (Romance) and Germanic may be that the lexicon referring to animals in Romance is rather detailed. Standard French as well as the dialects investigated here indeed provide many instances of precise animal reference, often on the basis of a productive system of derivation, cf. Lim. merle ‘male blackbird’, but merlāto ‘female blackbird’. Details further pertain, for example, to age, moment of birth or general quality, cf. Gsc. abeliot (Aure) ‘ovine that is raised outside the herd’ (Palay 1974: 4), Lim. cougulen ‘late-born lamb’ (often a weak animal, Robert et al. 1968: 78), Lim. rōsso ‘bad horse’ (early Fr. rosse), Ba. kolasõ ‘young male sheep’, Ba. kwàse ‘mother hen’, Ba. mœutõ ‘young sheep’, Ba. modzõ ‘two-year-old heifer’, Ba. nẅḗ ‘pig born in the spring’, and so forth. The dialect of Bagnes also features verbs that denote specific animal activities, e.g. Ba. ébàtre ‘play about while waiting to be milked (of cows)’. Further research is needed to account for this important discrepancy, which should include not only additional Gallo-Romance dialects, but also Germanic data—including the farmers’ sociolects—to see whether they indeed lack non-compound terminology for animals. Finally, a special instance is provided in Lim. lou-gorou / lougorou ‘imaginary person who runs around in a wolf skin during the night’, which Robert et al. (1969: 189) interpret as a noun + adjective compound. Yet its etymology is more complex and is nominal. Lim. lou-gorou, Gsc. loup-garoû, etc. are cognates of Fr. loup-garou and OF leu-garoul, and go back to Lat. ‘wolf’ and late Lat. garulfus ‘werewolf’. The late Latin noun garulfus has its origins in Gmc. *wariwulf (*wari ‘man’ + wulf ‘wolf ’). This compound (cf. OE werewulf ) was a hybrid denoting a creature that is a human being and a wolf at the same time. When the etymology of the borrowed compound in French had become opaque, lou(p) was added for explanatory purposes. Only the wolf- part is clarified. For that reason, the formation is not a tautological hybrid, but rather a type-of compound, because loup features as a generic noun (see Bauer 2017: 334f.).
15.3.6 Appositive compounds: the role of metaphors Appositive compounds in which the component nouns are in a metaphorical relation, are attested in all dialects, as in Provençal, cf. terro-maire ‘earth-mother’ > ‘mother earth’ and Gsc. terre-may ‘earth-mother’, in reference to the earth as a source of food—like a mother; see also Occ. tèrramaire ‘earth.mother’ as a
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
337
one-word formation (Alibert 1965: 43). The concept of ‘mother as source’, is also manifest in Lim. mai-sucho ‘mother-root’, which refers to that part of the tree (certain roots) from which new trees will grow. While ‘master’ and ‘chief ’ are productive in hybrids (see §15.3.3), they are also well represented in metaphorical usage, cf. Prv. mèstre-autar / Lim. meitre-ōtar ‘master-altar’ > ‘main altar’ and with a feminine form Lim. meitresso-brancho ‘master branch’. For cap ‘head, chief ’ the following compounds have been identified: Gsc. cap-counte, ‘master beam’ in a construction, Prv. cap-liò ‘chieflocation’ > ‘capital’, Occ. cap-lòc. It is noteworthy that in these ‘southern’ instances cap systematically occurs in initial position, suggesting that standard French chef-lieu—an early formation—is fully appositive and not a Germanic borrowing or calque (see further §15.3.7; for a discussion of the origins of chef-lieu, see Tobler-Lommatsch 1936: 384). Metaphorical relations also underlie appositive formations, such as Gsc. ferbaylét ‘iron-valet’ > ‘additional iron bar that keeps doors closed’ or the botanical term Ba. perĩ botële ‘pear-bottle’, denoting a type of pear that has the shape of a bottle (Bjerrome 1957: 216). While few instances of animal name compounds are based on metonymy (see above), many more are based on a metaphoric, relation between the composing elements, as in Gsc. gat-loup ‘cat-wolf ’, which denotes a wild cat, a lynx. Similarly, Gsc. bernàt-pasquè and Gsc. bernàt-pescàyrẹ (Palay 1974: 122) ‘martin-angler’, denoting a ‘heron’. A metaphorical relation is also manifest, for example, in Lim. chi mōūtou ‘dog-sheep’: the dog’s coat reminds one of a sheep’s wool (Robert et al. 1969: 257), but also in Ba. tsa-fweina ‘cat-stone.marten’ > ‘angora cat’, to mention just a few instances. In agricultural communities, fertility of the livestock is of vital importance, which accounts for the productivity of the following formations that denote infertile animals, cf.: Prv. ouélho muļo ‘sheep-mule’ > ‘sterile sheep’ and Prv. vaco mulo ‘cow-mule’ > ‘sterile cow’ (examples from Rohlfs 1970: 49). Similar examples are reported for Limousin, cf. vacho mulo ‘cow-mule’ > ‘sterile cow’. These formations refer to a type of sheep and cow respectively and the implicit comparison with a mule is obvious. Formations may also be based on an opposition, rather than the concept ‘mule’ as a sterile animal, as in the following examples that include both the feminine and male variety of the same species, cf. Prv. chabro-bou ‘goat-buck’ refers to a sterile goat as well, but the specifying noun comes second, in contrast to the standard French equivalent, bouc-chèvre. In both types of formation, chabro and chèvre respectively determine the basic meaning of the compound (a type of goat) as well as its grammatical gender. The sequence in both formations is fundamentally different and it is striking that the sequence ‘goat-buck’ is attested in a southern dialect. Within this type, Limousin offers a remarkable instance which at face value may be interpreted as a type-of compound specifying the sex of an animal, but in
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
338
. .
reality is a metaphorical compound: Lim. cherbo-femēlo. Cherbo (F) ‘hemp’ denotes a plant, but the noun femēlo ‘female’ does not refer to it being a female plant, but rather to parts of the plant that carry the seeds. Finally, the Limousin masculine compound ōli viergeo (< oil-virgin), with many attestations in other dialects, denotes a type of unrefined olive oil, hence the metaphor of virgin. Similarly, vigno viergeo refers to a plant (i.e. vigno ‘vine’) that will not carry any fruit and therefore is purely ornamental.
15.3.7 The sequence of the component nouns The examples provided in the previous sections, drawn from the dialect of Limousin, Bagnes, and Gascon with a few rare exceptions typically feature the sequence [head noun + specifying noun]. In this light, the data from the dialect of La Gleize are of interest, because the reported appositive type-of formations typically show a left-branching sequence, cf. LG fièsse-djoûrs ‘feast-days’ > ‘holidays’, LG wèyin-tins ‘win-time’ > ‘autumn’, and LG homme-jour ‘man-day’ > ‘business day’⁶ (Remacle 1952: 8). The Wallon instances follow Germanic ordering pattern, as in Fl. feestdagen ‘holidays’ and man-dagen ‘business days’ respectively, denoting certain types of day. LG wèyin-tins includes Romance tins (Fr. temps ‘time’) and denotes ‘temps du regain’, ‘the period of the second harvest’ (Remacle 1952: 93).⁷ In this formation, win in all likelihood is a borrowing from Flemish, where the verb winnen had the meaning ‘to acquire’ in early attestations. More specifically, in the context of agriculture its meaning was ‘cultivate (the land), grow crops, and harvest’. So wèyin-tins may indeed have referred to the time of harvesting. Similarly, LG pî passê, later pî-pazê (Remacle 1952: 88), denotes a type of pathway, with the specifying noun preceding the head noun. It may be a calque from Flemish or Germanic in general, which includes compounds of the type footpath; in them the noun that identifies the generic quality of the referent comes last, e.g. Du. voetpad. A formation similar in sequence and choice of words is attested in the dialect of Lille: piedsente ‘footpath’ < pied ‘foot’ + sente ‘path’ (standard French sentier) (Vermesse 1861: 131). While left-branching compounds are found in the dialects of La Gleize and Lille, the linguistic context is different because left-branching appositive structures are not an isolated phenomenon in
⁶ Homme-jour and jour-homme refer to the amount of work a human being carries out during one day. ⁷ Fr. regain denotes the second hay harvest, when hay is cut for a second time in late summer (early September).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
339
the dialect of La Gleize, as street names illustrate, for example, cf. potî-rowe ‘potter-street’ (Remacle 1952: 90), where the sequence elsewhere would be rue des potiers. By contrast left-branching is distinctly less strong in the more southern dialect of Lille, as manifest in Li. pain-perbole referring to a cake (pain) made with rye, aniseed and honey. Pain-perbole must originally be a borrowing from Flemish, that is peperbol, a type of roll with sweet spices (< Fl. peper ‘pepper’ + bol ‘roll’).⁸ In all likelihood, the opaque Fl. peperbol has been reanalysed as pain-perbole—a compound referring to a type of bread—in which the generic noun comes first (pain) followed by the specifying element according to French right-branching grammar. The left-branching Flemish formation has therefore been re-analysed as a right-branching structure in the dialect of Lille. The meaning of perbole remains unclear. It may be a non-word, resulting from the reanalysis, which does not affect the identification of the basic rightbranching structure. Finally, while GL banc-cofe ‘bench-chest’ is structurally in line with the compounds discussed above, the situation pertaining to southern instances of the type Gsc. arquebanc remains elusive. More historical data are needed in terms of context and timing of the borrowing before these formations can be fully interpreted.
15.3.8 Preliminary conclusions While appositive compounds are not widespread in the dialects, they show a number of patterns. These are important because the data show that the formations are not simply calques or interdialectal borrowings, but original dialectal features. Hybrids and type-of compounds are well-represented. And in this last category metonymic relations are relatively prominent—especially in reference to animals and plants—a feature rarely found in contemporary French, if at all. Another distinguishing feature of dialectal compounds is the occurrence of additional compounds. By contrast, the strong presence of appositive compounds based on a metaphorical relation between the composing elements is found in both standard French and the Gallo-Romance dialects investigated here. In terms of difference between the dialects, it is noteworthy that northern dialects are more prone to left-branching compounds, where especially in southern dialects right-branching prevails. Also in formations where standard French still continues the early structure, as in Prv. chabro-bou ‘goat-buck’ vs standard Fr. bouc-chèvre.
⁸ Pepper in this context refers to spices in general, but recipes for this cake typically do not include pepper.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
340
. .
15.4 Le poilu 15.4.1 Introduction One of the first to draw attention to the importance of appositive compounds in French—without using that term—was Frei (1929), who identified them as part of a more general trend that entails the replacement of the adjective by a noun, a widespread device in expressive language, according to Frei (Frei 1971: 243). He identified this trend in spoken as well as written French, illustrating it with instances of predicative as well as attributive usage, e.g. c’est pas sorcier ‘that is not very difficult’ (lit. ‘it.is not wizard’), but also un succès bœuf ‘great success’ (lit. ‘a success bullock’), un diner monstre ‘a colossal dinner’ (lit. ‘a dinner mammoth’), or un style nature ‘an unaffected style’ (lit. ‘a style nature’) (Frei 1971: 243f.). A substantial part of his data came from the Agence des Prisonniers de Guerre, a repository of letters written during the First World War. The authors were predominantly women from all over France—‘généralement des femmes du peuple’ (Frei 1971: 37)—who wrote to their loved ones, French prisoners of war, offering a rather complete picture of the state of the French ‘langue courante et populaire’⁹ at that time (Frei 1929: 37). Similarly le poilu,¹⁰ the sociolect used by the French soldiers in the trenches during the First World War, was spoken by individuals who came from all over France, many of whom had a dialectal background. Given the increase of appositive compounds from the beginning of the twentieth century, poilu may provide important data to the study at hand. In the following pages the findings will be discussed.
15.4.2 Appositive compounds in poilu The data, which have been drawn primarily from Renault’s (1919) dictionary, reveal first of all a significantly higher incidence of appositive compounds in comparison to the dialects investigated. Yet, in accordance with the dialects, appositive compounds represent a relatively small group within a much larger collection of compounds, which are determinatives in majority, such as verb + noun-obj (e.g. fauche-pattes ‘mow-feet’ > ‘a shell’,¹¹ rase-terre ‘brush-ground’ > ‘short man’), verb + preposition + noun (e.g. roule-par-terre ‘roll-over-ground’ > ‘liquor’), noun + de + noun (e.g. parc des refroidis ‘park of.the cooled’ > ‘cemetery’, fil de fer ‘string of
⁹ ‘the everyday language and colloquial French’. ¹⁰ Le poilu refers to the soldiers as well as their language. ¹¹ At explosion its contents remain close to the ground, mowing away unprotected feet.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
341
wire’ > ‘thin man’), noun + à + noun (e.g. panier à salade ‘basket for salad’ > ‘helmet’,¹² parc aux os ‘park for bones’ > ‘cemetery’), and so forth. In addition to these widespread formations, poilu features numerous appositive compounds as well. In Renault (1919), for example, over forty instances were identified. Several of them are part of the (semi-)official military language, which includes many appositive compounds in its core terminology, including ranking terminology, e.g. lieutenant-colonel. Several of them refer to two qualities in the same person, such as: officier tankeur, an officer and driver of a tank, médecin chef (de service) denoting a doctor who is chief of his department, chauffeurmécanicien, who is a driver and mechanic at the same time, and médecin-major who is in all likelihood a doctor with the rank of major. There are quite a few formations that include the name of a military rank and a noun referring to the individual’s job. Caporal-mitrailleur and sergent-mitrailleur, for example, denote the soldiers who operate machine guns, with mitrailleur, referring to the soldier in charge of the mitrailleuse, the actual gun. Similarly caporal-fourrier (also found as cab-four) and brigadier-fourrier refer to individuals who are fourrier in a given rank. A fourrier was a low-ranking officer in charge of the distribution of food, equipment, and accommodation of the troops. Similarly, a caporal-cuisto—with the français familier term cuisto ‘cook’—was a corporal in charge of cooking.¹³ In addition to tanks and machine guns, aircraft were in an innovation in 1914–1918 warfare as well. Their existence is reflected in the lexicon, e.g. the hybrids soldat aviateur ‘soldier aviator’ and pilote-aviateur ‘pilot-aviator’, which recalls the nautical origins of the noun pilote. A pilote originally was in charge of steering ships safely through hazardous waters of which he had expert knowledge. Via pilote-aviateur, the concept ‘airman’ eventually came to be referred to by pilote, without further semantic specification. In addition, hybrids in poilu may anticipate structures that were to become widespread, such as apprenti-cadavre ‘apprentice-corpse’ and—to a lesser extent—élève-mort ‘student-dead.man’. Especially apprenti- ‘apprentice’ is a productive formation in today’s French (e.g. apprenti-maçon ‘apprentice mason’, apprenti-pilote ‘apprentice-pilot’, apprenti-boucher ‘apprentice butcher’, and so forth; see §15.5). These hybrids denote a person who is a trainee and a workman at the same time. Semantically, the compound apprenti-cadavre denotes an ambitious person who wants to advance in ranks risking his life; un élève-mort is a wounded soldier. These examples illustrate how the interpretation of formations that are similar in form and meaning, may in fact have a very different
¹² In today’s français familier, panier à salade refers to the police van used by law enforcement to transport prisoners and/or people under arrest. ¹³ Fam. cuisto < cuistancier ‘cook’; cuistancier < cuistance ‘kitchen’.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
342
. .
interpretation, showing that if compounds are linguistically motivated instances for the speaker, they may easily be opaque to the listener. Among the high number of hybrids that denote human beings, Poi. niche-abri (F) is rather exceptional in that it conveys an object: it refers to housing or a place to sleep (niche [F]) that at the same time provides protection from military attacks and severe weather (abri [M]) (< ‘digs-shelter’). In this reading the compound is a hybrid. A second large group of formations includes appositive type-of compounds that include an animate noun (often a rank) and a second noun—in apposition— specifying the person’s function in metonymical usage, e.g. Poi. homme-perco, for example, was detailed to get coffee. Perco is a truncated form of percolateur à café ‘coffee percolator’, which was not part of military equipment in the trenches. The homme-lettres (> omelette), a popular version of vaguemestre, was in charge of the mail. The sergent-ballot was the sergeant in charge of transporting parcels (< Fr. ballot a small bundle of parcels, typically carried by donkeys). The brigadiermarcassin denoted the brigadier in charge of pigs: Fr. marcassin is a young boar. Several instances include caporal ‘corporal’ followed a specifying noun that expresses the troops’ appreciation, cf. pejorative caporal patates ‘weak corporal, imbecile’, in which patates ‘potatoes’ is used in reference to the assumed predominant food of provincials and farmers (Renault 1929). Finally poilu has a substantial number of instances of appositive compounds in which the composing elements are in a metaphorical relation, such as Poi. avionmoineau ‘plane-sparrow’ > ‘one-engine plane’, in which the smallness of the plane is tacitly compared to that of a sparrow. It is noteworthy that the bird’s minuteness is also conveyed in the French expression avoir un appétit de moineau ‘to have very little appetite’. Similarly, in Poi. une équipe-chameau—team (F) – camel (M) ‘group of carriers’—the soldiers in charge of transporting and carrying beams and timber during the construction of a (provisional) bridge are compared to camels. In Poi. feminine blessure filon (< wound [F] – luck [M]) ‘wound that allows one to return behind the front’ the wound is tacitly compared to a stroke of of luck. All metaphorical appositive compounds were not innovations. Several existed before 1914, but came to be commonly used during the war, often in cynical context, such as papier buvard ‘the type of paper used as desk blotter’: it occurred in expressions such as Poi. ramasser un aviateur avec la cuiller à café et le papier buvard ‘to collect the remains of a pilot (after a crash) with a coffee spoon and blotting paper’ (example from Renault 1919: 447). The following instances are appositive compounds based on metaphors as well, but more than one implicit comparison is at play, at various levels. Poi. mère pingouin [F] ‘mother penguin’, for example—with masculine pingouin—denotes a teaching plane that could not fly. This incapacity accounts for the penguin metaphor: despite their wings, penguins cannot fly. Part of the formation may be a borrowing as penguin ‘non-flying training airplane’ was also an English term
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
343
and this type of plane was manufactured in the USA.¹⁴ The mère metaphor in all likelihood is French and may underscore the teaching part of the concept (Renault 1919: 422). In §15.2, the occurrence of mère in various appositive compounds was pointed out, its metaphorical use emphasizing the concept of ‘source’. In mère pingouin, mère may indeed highlight reference to the source of knowledge because mère pingouin was an instrument used in the initial stages of aviation training. Yet the sequence of the component nouns and the feminine gender of the compound do not fully support that interpretation. Metaphorical mère pingouin may also be a term of endearment, accounting for the occurrence of mère in initial position (comparable to today’s [la] maman-pingouin in children’s stories). Instances that illustrate the usage of the compound have not been provided by Renault and without context it is impossible to assess the precise function of mère in these formations. Another innovative feature of warfare during the First World War was the observation balloon, which became especially important when the troops became literally entrenched. The metaphorical nouns for observation balloons in poilu were saucisse and boudin ‘sausage’ in reference to their shape. In that light, Poi. saucissemann is remarkable in that it includes a borrowing (-mann) and that the specifying noun comes first (‘sausage-man’). The compound denotes the observer—military or meteorological—who was positioned in an observation balloon: Ger. -mann obviously refers to the man in question, alluding to the alleged German liking for sausages (Renault 1929: 481). The noun, as well as the left-branching structure, are Germanic. By contrast, in boudin-cavaleur—co-existing with non-compound boudin ‘observation balloon’—the specifying noun follows the head noun and establishes the metaphor of running or moving. The formation follows regular French patterns, both in lexical elements and structure. The formation choucrouteman (‘sauerkraut-man’) finally needs some clarification as well. Nationalities were often referred to in the French armed forces by the nation’s favourite dishes. The English, for example, were called les rosbifs, the Italians were les macaronis, and the Germans commonly were referred with the term choucroutes ‘sauerkraut-PL’. Choucrouteman obviously combines choucroute and Ger. Mann. In English as well, nationalities were referred to in terms of foods, but to a lesser extent, with kraut ‘German’ (< sauerkraut; e.g. Dickson 2007: 75, with derivatives) and frog ‘Frenchman’ being well known.¹⁵
¹⁴ Penguins were training airplanes with short wings and a small engine that could not fly. They were used in early aviation training to acquire ‘the feel of airplane controls at near-flying speeds, without the danger of actual flying’ (https://www.cradleofaviation.org/history/exhibits/exhibit-galleries/world_ war_i/breese_penguin.html; accessed April 2019). ¹⁵ Eng. frog ‘Frenchman’ in reference to a French delicacy was first used in the 1780s and revived by American soldiers during the First World War (Dickson 2007: 62).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
344
. .
15.4.3 Summary of findings Appositive compounds have a remarkably higher incidence in poilu than in the dialects. While they reflect in principle the same type of formation as in the dialects, they seem to reflect more often instances of expressive language. Hybrids are very productive, often combining the name of a rank with an occupational term—to the extent that they become regular, semi-serial formation, e.g. caporal-mitrailleur, caporal-fourrier, soldat aviateur. Type-of formation may also take a repetitive pattern, as in homme-lettres, sergent-ballot, in which the specifying element refers to a distinguishing object of the individual’s task. The role of metaphors is exceptionally strong in the poilu compounds, where they are often in play at various levels, as in mère-pingouin.
15.5 Discussion of the findings and conclusions This section brings together the findings from the dialects and poilu, relating them to standard French and—occasionally—to argot, which has been profoundly influenced by poilu and exists to date. The data show that appositive composition is a productive process in the dialects: appositive compounds are well attested—without being widespread. They include hybrids, additional compounds, type-of compounds, and compounds based on a metaphorical relation. Historically, hybrids have been very productive in all Indo-European languages. Their relatively high incidence in the Gallo-Romance dialects—denoting objects and especially professionals—is in line with that trend and with data from standard French. In poilu, however, hybrids are distinctly more productive than in the dialects, especially in reference to professional jobs of the military. The existence of additional compounds in the Gallo-Romance dialects is remarkable in light of their near absence in standard French. The additional compounds found in the dialects (e.g. Prv. aioli, Gsc. aygue-sàu ‘brine’), parallel instances in (early) Indo-European languages, where they occur, for example, in reference to substances and dishes that involve several ingredients (see Bauer 2017: 315–53). That additional compounds are not merely a dialect phenomenon is shown in their occurrence in argot, where a few additional compounds are found as well, cf. arg. bœuf-carotte ‘beef-carrot’, in reference to a person who makes a suspect ‘simmer’ during interrogations (Caradec and Pouy 2009). Similarly, the argot compound un couple couscous-pommes frites (lit. ‘couple couscous-fried potatoes’), including two nouns in reference to dishes, denotes a mixed Maghreb–French couple (Caradec and Pouy 2009).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
345
Another difference with standard French is the occurrence of generic and metonymic type-of compounds in the dialects denoting plants, but especially animals (e.g. Lim. chi dōgue ‘dog-watch.dog’). As pointed out earlier, these formations have always been widespread in Germanic. In that light, it is noteworthy that they are attested in Gallo-Romance dialects that are not in regular contact with Germanic languages. Yet these non-contact-induced formations remained exceptional in the dialects investigated, possibly because the detailed lexicon in reference to animals does not require them. In terms of shared features, the prominence of metaphors stands out: type-of compounds based on a metaphorical relation between the composing nouns are remarkably productive—conveying a wide variety of referents—in the dialects, poilu, argot, and standard French. In the dialects, metaphorical compounds often refer to animals and plants, whereas in poilu they denote a much wider range of referents and the metaphorical usage plays at various levels. As to the sequence of the component nouns, differences may be pointed out along geographical lines. While in the vast majority of appositive compounds in the dialects the head noun comes first, there are a few exceptions, in which the initial position is taken by the specifying noun. These are typically found in the dialect of La Gleize, which is in close contact with Germanic, and to a much lesser extent the dialect of Lille, in the north of France. Yet left-branching instances are also attested in dialects without Germanic contact languages, e.g. Lim. meitre-ōtar (lit. ‘masteraltar’). These instances are residues from an earlier stage, when left-branching attributes were more prominent than in the contemporary language. It is noteworthy that Provençal features right-branching appositive compounds, where standard French continues the early formation, e.g. Prv. chabrou-bou (vs Fr. boucchèvre ‘sterile goat’), and Prv. terro-maire. In this context, it is noteworthy that in poilu the appositive compounds typically are right-branching, with a few exceptions that involve borrowings. This pattern is further confirmed by data from argot. There is an important difference between the findings from the dialects and those from poilu in terms of repetitive patterns. In the dialects, most formations are rather exclusive and isolated, with only an occasional noun occurring in more than one compound (e.g. mèstre ‘master’). The situation is different in poilu, where quite a few instances are based on repetition, e.g. caporal-fourrier, caporal-cuisto, caporal-mitrailleur, with similar repetitive patterns for sergent, and so forth. This repetitive trend is also manifest in instances of expressive language as in diner-monstre, procès-monstre, concert monstre, and others that Frei (1929: 243) found in letters from the First World War. In argot, this pattern is even stronger and typically involves metaphorical usage, especially in expressive language, e.g. effet bœuf ‘impact-bullock’ > ‘major impact’ or prix choc ‘price-shock’ > ‘shock price’. But the phenomenon is also attested in non-emphatic language: bidon ‘bluff ’, for example, recurs in reference to the
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
346
. .
concept of ‘fake’, arg. grippe bidon ‘flu-bluff ’ > ‘fake flu’, arg. attentat bidon ‘attack-bluff ’ > ‘fake attack’, and so forth. Similarly, minute, for example, conveying the concept of speed, appears in appositives such as arg. entrecôte minute (data from Caradec and Pouy 2009). The formations have been very productive in advertising ever since the 1960s (e.g. magasin pilote [lit. ‘store pilot’ > ‘experimental store’], soupe minute [lit. ‘soup minute’ > ‘instant soup’]), which was condemned by certain scholars as a symptom of linguistic ‘disintegration’ (e.g. Etiemble 1965: 107). These trends have become established processes in contemporary French, where they are very productive. They are a vital source of new lexical formations. In addition, serial compounding has come to play a semi-grammatical role as well, to the extent that femme and bébé have taken on grammatical function conveying feminine gender and diminutive value respectively (Bauer 2017: 364–7). Large contemporary databases such as Wikipédia (see note 5) show that when conveying grammatical value, these nouns typically occur in initial position (e.g. femmeprofesseur ‘woman professor/teacher’; bébé-piscine ‘baby pool’). It is noteworthy that in French synthetic diminutive formations are relatively limited (see Hasselrot 1972 for French, Bauer 2011 for Romance). Appositive feminine formations are productive and widespread as text corpora show. It is therefore remarkable that the Académie Française (2019) in its overview of linguistic processes underlying feminine occupational terms, identifies appositive compounds as a device that is and should be used only when morphological endings would create ambiguous or unfortunate formations (e.g. femme médecin vs **médicine ‘female doctor’ or femme marin vs **marine ‘female sailor’): the number of formations is assumed to remain limited (Académie Française 2019: 7). In sum, data show that appositive compounding was a productive formation process in the dialects, that features many parallels with standard French itself. Appositive compounds therefore are an integrated part of grammar in Gallo-Romance dialects, with emphasis on hybrids and type-of formations and a major role for metaphors. Data further show that the patterns observed for the dialects are even stronger in poilu and argot. Their innovative features include (a) a substantial increase in incidence; (b) early attestation of serial compounding in poilu; and (c) ongoing productivity of metaphorical language use, which is a rich source of language innovation. These trends account for the increase in appositive compounds over the last hundred years.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
16 Complex versus compound prepositions Evidence from Gallo-Romance Nigel Vincent
16.1 Introduction Viewed from a diachronic perspective prepositions in Gallo-Romance, and indeed in Romance more generally, fall into four broad classes: a) those like Fr. à, de and en (with cognates across the whole of Romance) which derive from a simple clause-level particle and which can be traced back to proto-Indo-European (Vincent 1999, 2017); b) those such as Fr. pendant ‘during’, Occ. cap ‘towards’, Lauragais costo ‘beside’, which derive from an item belonging to another grammatical category, whether nominal, verbal or adjectival; c) those such as Fr. autour ‘around’, Occ. a dich de ‘by virtue of ’, Lgd. demest ‘among’, which derive from a prepositional phrase and which, following Fagard and De Mulder (2007) and Roy and Svenonius (2009), we will call complex prepositions; d) those like Fr. avant ‘before’, Frp. dézò ‘under’, Occ. dins ‘in’, which derive from a structure involving an original particle (Lat. , , or ) prefixed to an existing prepositional or adverbial item and which, following Vincent (2017), we will call compound prepositions. The goal of the chapter paper is then to compare and contrast the synchronic structures and diachronic trajectories of these last two classes of items.¹ To this end we begin with a brief review of some general issues posed by prepositions as a category (§16.2). We move on to examine the origins of Gallo-Romance complex and compound prepositions (§16.3), the diachronic stages by which they have ¹ It is a pleasure to dedicate this chapter to J. C. Smith. I am grateful to Louise Esher for her help with the Occitan material and to those who responded to my request on the ALT list for information about complex adpositions around the world: Alexandre Coupe, Anders Holmberg, Gerd Jendraschek, Jana Kellersmann, Christian Lehmann, Pieter Muysken, Doris Payne, Mark Post. Thanks too to Delia Bentley, Adam Ledgeway, Christian Lehmann, and the editors of this volume for their comments on an earlier draft. Errors of interpretation or attribution are entirely my own responsibility. Variation and Change in Gallo-Romance Grammar. First edition. Sam Wolfe and Martin Maiden (eds). This chapter © Nigel Vincent 2020. First published in 2020 by Oxford University Press.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
348
developed (§16.4) and their place in the broader typological context (§16.5), before considering the light such historical evidence can shed on the more general question of the syntactic status of prepositions, prepositional phrases and prefixes (§16.6, §16.7).
16.2 Prepositions: some preliminary considerations It would take us too far afield to attempt to review the substantial literature on the concept of adpositions in general or prepositions in particular (see Hagège 2010 and Libert 2013 for recent overviews and references). For our purposes, it will suffice to say that a preposition is an independent word which takes an argument and serves to link that argument to a governing category whether nominal, adjectival, verbal or phrasal. In that sense, it serves as what Lehmann (2019) calls a relator. Such relations can be purely functional as with fier de toi ‘proud of you’ or je me souviens de ces moments ‘I remember those times’, or express independent semantic content as with pairs like avant ‘before’ vs après ‘after’ and avec ‘with’ vs sans ‘without’ or oscillate between the two as with the directional or locational meaning of à in à Paris ‘to, in Paris’ beside its role as the marker of the demoted subject of the infinitive in j’ai fait chanter la chanson à Pierre ‘I made Pierre sing the song’. We return to the issue of the place of prepositions within a larger account of syntactic categories in §16.6. Both functional and semantic definitions of the kind just outlined are neutral as to the form of the prepositional element and it is not uncommon in languages to find simple and complex items with similar meanings—as for example French pendant and au cours de (lit. ‘in the course of ’) ‘during’—or pairs of antonyms where one member is (synchronically at least) monomorphemic and the other is polymorphemic, as with English behind vs in front of. In addition, as we will see, the mechanisms of language change may bring items with different origins into a synchronic relation. Thus, avec derives from a phrasal expression + *hoque ‘at this’, while its antonym sans is a direct descendant of the Latin preposition ‘without’. A further distinction that will turn out to be particularly relevant in the diachronic context is that between prepositions and adverbs. Although there are marginal cases in which an adverb can be said to take an argument—for example malheureusement pour toi ‘unfortunately for you’—in general, adverbs can be distinguished from prepositions by the absence of an argument. By contrast in some instances items that usually require an argument may occur without as in il a pris la clé, il est parti avec ‘he took the key, he has left with (it)’, a usage variously described in dictionaries as colloquial, Belgian or Canadian! And indeed one of the common ways for an adverb to develop into a preposition is precisely through a process of argument capture. This is the mechanism which yields the
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
349
core Latin prepositions such as , , from original proto-Indo-European sentential particles (Vincent 1999) and the same process is repeated in the development from Latin into Romance. Thus, modern French sous ‘under’ and OFr. sus ‘on, over’ originate in the Latin adverbs and and over time come to take arguments and thus acquire prepositional status. It is for this reason that in synchronic theories of syntactic categories, prepositions and adverbs together with subordinating conjunctions can be seen as members of a larger grammatical class, an analysis which goes back at least to Jespersen (1924 [1992]: 87–90) and which was revived in the context of the X-bar theory of syntax by Jackendoff (1977) and Emonds (1985). Conversely, adverbial marking may become attached to a preposition. Thus, old French has an adverbial suffix ‑s, etymologically derived from Latin words ending in that consonant such as > plus, > moins, and thence generalized to longues ‘for a long time’, sempres ‘always’ and from there to the modern preposition sans ‘without’ (< +s) (Buridant 2000: §407). Both processes are combined in the history of Fr. derrière ‘behind’. Its origin is the Latin adverb reinforced with the prefix according to the mechanisms which we will consider in more detail below. Suffixation of -s yields the old French adverb derriers which is then extended to a prepositional function as in deriers son dos mist son tinel poser ‘he placed his rod behind his back’ (Aliscans 4448)² before the final ‑s is dropped again in the modern language (Buridant 2000: §369). Diachrony is important here in another respect. Tremblay et al. (2005) note that in old French both the simple form sus and the complex dessus could be used with or without an object, in traditional terms as either a preposition or an adverb, as the examples in (1) (= their (4) and (5)) demonstrate: (1)
a. Et fetes escrivre sus la tombe . . . ‘And have it written on the tomb . . . ’
(Artu, 220)
b. Lors pensent des plaies Lancelot et metent sus ce qu’il cuident . . . (Artu, 203)3 ‘Then they dress Lancelot’s wounds and put on what they consider . . . ’ c. Desus la tombe . . . avoit letres qui disoient . . . ‘On the tomb were letters which said . . . ’
(Artu, 251)
d. après firent metre desus letres qui disoient . . . ‘after they had placed thereon letters which said . . . ’
(Artu, 262)
Over time this freedom has been lost and modern standard French has developed a twofold system with the items sur and sous taking an overt object while dessus and dessous do not. ² See Régnier (1990–1).
³ See Frappier (1954).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
350
An extra dimension of complexity is raised by the structures au-dessous (de) ‘below’ and au-dessus (de) ‘above’, which in traditional French grammar are dubbed ‘locutions prépositionelles’. These issues are addressed by Adler (2008), who investigates a range of expressions made up of the sequence ‘preposition + definite article + noun + preposition’, to which she applies a series of tests such as omission or substitution of the article and modification of the noun. Thus, an expression like dans le coin de ‘in the corner’ allows le to be substituted by ce ‘this’ or un ‘a, one’, coin to be modified (dans l’autre coin ‘in the other corner’) and the dependent de-phrase can be omitted altogether, whereas none of these changes are possible with a fixed expression like à l’instar de ‘after the fashion of, like’. On the basis of a range of such tests, Adler concludes that it is necessary to draw a threeway distinction as follows. i. prepositional phrases constructed according to the productive rules of the contemporary language: for example sur le dessous de ‘on the top of’ exactly parallel to dans le coin de ‘in the corner of ’ or au commencement de ‘at the beginning of; ii. items for which there is ‘lexicalization in progress’ such as au bord de ‘at the edge of’, au milieu de ‘in the middle of ’; iii. complex prepositions which have already undergone lexicalization such as à côté de ‘beside’, au-dessus de ‘above’, and au fur et à mésure de ‘as and when’. What these tests show, then, is a scale from full syntactic independence through to full incorporation as distinct lexical entries, something which, as we will see, is entirely consistent with the expected outcome of the mutually supportive and gradual diachronic processes of grammaticalization and lexicalization.
16.3 Origins A core class of Romance prepositions are continuers of words that were already prepositional in Latin. Thus, across the whole of the family we find reflexes of Latin ‘to’, ‘down, from’, ‘in’, ‘for’, albeit in some instances with local differences of meaning and function. By contrast, there is a considerable degree of regional diversity in those items which come from different structures in Latin. On the one hand, there are those which derive from members of other word classes such as OFr. lez ‘beside’ < Lat. ‘side’, Occ. tenent ‘next to’ < Lat. ‘holding’, Fr. vers ‘towards’ < Lat. ‘turned’, about which, however, I will have little to say in the present context. On the other, there are the items which will be my main focus, those which originate in multi-words expressions in Latin. In Vincent (2017) these are divided into two sub-groups labelled ‘compound’
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
351
and ‘complex’, and I will maintain the same terminological and conceptual distinction here.⁴ Complex prepositions arise when formerly transparent and productive syntactic structures become fixed and fused, as with the examples drawn from Adler and discussed above. Compound prepositions, by contrast, are the result of a particle being prefixed to an item that already has, or later acquires, prepositional function just as compound verbs derive by prefixing a particle to an existing verb stem. On this account, then, a preposition like depuis ‘since’ < is parallel in its formation to a verb like déduire ‘deduce, infer’ < ‘lead down from’. Examples of both kinds are given in (2) and (3).⁵ (2)
(3)
Compound prepositions form etymon meaning devant in front of drjɛr behind dintre inside denfra below avant before
variety French Jèrriais Languedocien Occitan French
Complex prepositions form etymon pa l ho: d a biais de a cap de a dich de de bal (de) () a cause de entrémé di
meaning up in the guise of at the end of, after in virtue of down because among
variety Jèrriais Languedocien Languedocien Languedocien Languedocien French Francoprovençal
Regional differences emerge clearly from examples such as these. There are formations which are unique to a particular region or dialect as well as ones that are shared between some parts of Gallo-Romance and other adjacent regions. For instance, the Lgd. a cap de ‘after’ shows the regional noun cap ‘head’ entering into prepositional expressions as also cap a ‘towards’ that are matched by similar ⁴ Our use of the term ‘complex preposition’ corresponds to what is found in much of the relevant literature (see for example Roy and Svenonius 2009 and Lehmann 2019), while our choice of ‘compound’ is based on its use particularly in accounts of the mechanisms at work in the derivational morphology of verbs. However, readers should note that the two classes of elements are sometimes collapsed under the single label ‘compound’ as in the abovementioned discussion by Adler (2008) or in some descriptive works such as for instance the sections devoted to prepositional structures in Liddicoat (2011). ⁵ In these tables the Jèrriais examples are taken from Liddicoat (2011) and the Languedocian ones from Ronjat (1937), to which the reader is referred for more detail of usage and etymology, as well as further instances of the patterns exemplified here.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
352
developments in Catalan. In this sense, the distribution of these items is, not surprisingly, like that of other lexemes rather than the patterns of grammatical and phonological structure which tend to define subbranches of Romance. Lehmann (2019) draws a distinction similar to the one we have made between compound and complex when he contrasts the mechanism of word formation with the diachronic process of univerbation, well known as a frequent concomitant of grammaticalization. That said, his account differs from ours insofar as he attributes the origin of prepositions of both the depuis and the à côté de type to the effects of univerbation. Writing about Latin sequences such as —source of Fr. avant ‘before’, Sp. abante, It. avanti ‘forward’ and cognates across Romance as a whole— , , and the like, Lehmann (2019, §6.1.2) argues as follows. After univerbation, the structural appearance of the erstwhile superordinate preposition is that of a prefix. [ . . . ] It is, however, important to put this into diachronic perspective [ . . . ] The genesis of the construction is no process of prefixation—thus, no process of word formation. The historical evidence of the original constructions leaves no doubt that these complex prepositions originate in univerbation of adjacent sequences in erstwhile syntactic constructions. This does not, of course, exclude the possibility that, at a subsequent point in diachrony, a set of univerbations of the same structure constituted a model for the formation of further complex prepositions. It is, however, not clear that such an assumption is actually needed for the complex prepositions under review.
There is, however, no need to rule out in this way compounding as a mechanism for prepositions given that it is freely available for other categories such as verbs, adjectives, and nouns as the examples in (4) demonstrate (and compare here the discussion of formations in in Vincent 2017: 296): (4)
a. Verbs compounded with : ‘be present’, ‘go to’, ‘swear’ b. Adjectives compounded with ‘turned towards’
:
‘pursuing,
hostile’
c. Nouns compounded with : ‘adage’, ‘adverb’ d. Verbs compounded with : ‘call forth’, ‘run forward’, ‘flatten out, explain’ e. Adjectives compounded with : ‘small’, ‘banished from, deprived of ’ f Nouns compounded with : ‘quarryman’, ‘conclusion’
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
g. Verbs compounded with : ‘copy’, ‘cross’
‘bring
353 over’,
h. Nouns compounded with : ‘transgressor’, ‘transport, translation’ i. Adjectives compounded with : ‘on the other side of the Danube’, ‘changed in shape’ Moreover many of these examples show the same shading from straightforward semantic transparency to gradual loss of compositionality, as we have seen with prepositional formations. Compare for example built on the stem - ‘flat’ but with the derived meaning ‘explain’ or which contains the stem - ‘step’ but comes to refer to an individual who has stepped across a particular moral or legal boundary. More generally, univerbation often goes hand in hand with the loss of compositionality (Bauman and Torres Cacoullos 2016). In addition, over time the categorial range of such prefixation mechanisms changes as does their frequency. Buchi (2009) demonstrates in some detail, for example, how the Latin prefixes - and - become restricted to a preverbal function in Romance and how, even then, considerable care is needed to avoid overproliferation of such forms in reconstructed proto-Romance.
16.4 Diachronic stages The examples and the patterns we have discussed so far correspond closely with the diachronic trajectory first proposed in Lehmann (2002a) and taken up by Fagard and De Mulder (2007) and Lehmann (2019): adpositional locution > complex adposition > simple adposition > functional adposition
This historical sequence is in turn but a special case of the general directionality exhibited by changes which fall under the heading of grammaticalization. Two caveats are, however, in order. First, as Fagard and De Mulder (2007) argue, the initial phases of this kind of development create items that constitute new lexical rather than grammatical elements; otherwise put, the process is one of lexicalization rather than grammaticalization. This is clear from the range and variety of meanings such items may express, the many language and dialect particular idiosyncrasies of form and content, and the fact that they do not form closed classes. It is only in the latter stages of this trajectory that it is reasonable to identify new grammatical markers, that is to say members of small closed classes of items that are the exponents of a syntactico-semantic functions which are readily identifiable cross-linguistically. At the same time, it should be clear that
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
354
what we see here are not instances of degrammaticalization, a change whereby in the words of Willis (2017: 28) ‘formerly functional items unexpectedly acquire lexical status and bound items unexpectedly gain greater freedom’. The latter are often taken to counter-exemplify the claimed directionality of grammaticalization processes (on which see Börjars and Vincent 2011), whereas the developments under consideration do, as we have said, support the general claim for directionality of change but with the emergence of new lexical items as an intermediate stage between the original phrasal structure and items which may, but do not have to, develop into markers of grammatical function.⁶ A second caveat is that the above trajectory conflates form and function in ways that are frequently but not necessarily linked. Although functional adpositions are commonly simple in form they not have to be. For example, the Italian complex preposition da parte di ‘on the part of’ is frequently used to mark source as in saluti da parte di mio padre ‘greetings from my father’ or agent as in una proposta da parte di Giorgio ‘a proposal by George’ but shows no sign of incipient formal simplification. Finally, we should note that although the mechanisms of grammaticalization are involved in the development of complex prepositions, and in some cases the passage from complex to simple prepositions, they do not provide support for the view that has been advanced that French is more grammaticalized than other Romance languages (Fagard and Mardale 2012: 335) or more generally that there is a scale of grammaticalization within Romance at the level of languages as a whole rather than at the level of individual forms and constructions (Lamiroy and De Mulder 2011). As we have seen, different structures fuse into complex prepositions in different parts of the Gallo-Romance, and a fortiori the pan-Romance, domain, but there is no evidence that we are aware of that some parts of the Romance-speaking area are more disposed to develop such items than others.
16.5 The typological context Our concern in the present chapter is with prepositions, but these in turn are part of a larger class of adpositions for the simple reason that the relational properties that characterise such items are independent of linear order. There is ample crosslinguistic evidence of postpositional items displaying the same range of formal variety and undergoing the same diachronic processes as we have been considering here. As one example, we can compare the forms found in the Chibchan ⁶ This is not to say that an adposition may not develop through degrammaticalization. A commonly cited instance is North Saami haga ‘without’, which derives from an abessive case suffix (Ylikoski 2016). The absence of any such formations in Romance is to be attributed to the more fusional than agglutinative nature of Latin case morphology, rather than to any general properties of universal grammar.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
355
language Cabecar spoken in Costa Rica as described in González Campos and Lehmann (forthcoming, ch. 8).⁷ As the authors note, this language is systematically postpositional just as French is systematically prepositional and we find there the same variety of forms and functions (glossing terms due to González Campos and Lehmann): (5)
a. Carlos da Carlos ‘with Carlos’ b. ják wa stone ‘with a stone’ c. ju wäś a ka
irä ju house front and house ‘in front of and behind the house’
shäbë ́ kjane back
ka
d. yikí ka=na yesterday = ‘since yesterday’ Although Cabecar is a language with a system of case marking, the noun that accompanies a postposition is not itself case-marked. In that respect there is no overt equivalent to the French de in à côté de. By contrast, the Hindi formations in (6), kindly provided by Jana Kellersmann, show not only postpositional items derived from nominal expressions but also the use of a linking word ke, the oblique case of the possessive marker kaa/kii, parallel to the use in Romance of de/di: (6)
kii taraf ‘towards’, ke andar ‘inside’, ke aage ‘in front of, ahead of ’ ke uupar ‘on top of ’, ke niice ‘beneath’, ke/se pahle ‘before’, ke paas ‘near’, ke piiche ‘behind’, ke baad ‘after’, ke baare mem ‘about, concerning’, ke baahar ‘outside’, ke lie ‘for’, ke saamne ‘opposite’
For some languages, complex elements appear to be the norm. Holmberg (2002: 162) writing about the Chadic language Zina Kotoko observes that ‘most adpositions in Zina Kitoko have two parts, both of which precede their complement, forming a complex preposition’, as in (7):
⁷ I am grateful to Christian Lehmann for sharing draft material from this grammar with me, though he is not to be held responsible for any errors of interpretation on my part.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
356
(7)
Tà kwìcə àsú dé má gmá they cut meat on ‘they cut the meat on the table’
tábəl table
Like our Romance examples and the Hindi ones in (6), there is a lexically derived part gmá most plausibly a reflex of gəmáyá ‘head’ and what he terms a relator má, which elsewhere mark source or possession but here serves to link the semantic core of the preposition to its argument, very much as does Latin and its reflexes in formations like Occ. a dich de. A further argument in support of the equivalence of pre- and postpositions comes from borrowing. Turkish is a language with a variety of simple and complex items, all of which are postpositional (Jendraschek, forthcoming). Some of these have been borrowed into the largely prepositional Indo-European languages of the Balkans where for the most part they now precede rather than follow their arguments (Friedman and Joseph, forthcoming, §4.3.3.2).⁸ Thus, in both Macedonian and Bulgarian there is a preposition karši, as also Albanian karshi, deriving from Turkish karşı, although the original Turkish order is to be seen for example in Aromanian doĭ oamenĭ carşí ‘opposite two men’. By contrast, Turkish has a complex postposition kanal-ı-yla channel.. ‘through (the mediation of)’ built around the French noun canal and calquing the French prenominal sequence par le canal de. In sum, the necessarily restricted range of languages I have cited in this section serves to demonstrate, if demonstration were needed, that the patterns from Gallo-Romance under discussion here are far from unrepresentative of the way many, perhaps most, languages deploy and develop complex adpositional forms. For more discussion and exemplification of these patterns in the languages of Europe, see Fagard et al. (forthcoming).
16.6 Modelling P and PP What we have seen so far is a substantial body of evidence, from both within and beyond Gallo-Romance, for recurrent patterns of development over time of prepositional expressions of different kinds. This in turn raises the important question of the light such changes can shed on the way these expressions are to be modelled within synchronic grammatical frameworks. Broadly speaking, we can divide the available approaches into two classes: those which see the units of grammar as constructions of various sizes and types and those which seek instead
⁸ I am grateful to Gerd Jendraschek and to Brian Joseph for supplying me with pre-publication drafts of their work.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
357
to encompass these patterns within a more richly articulated, universally available, model of syntax. The constructional approach is, in a sense, the default since if no alternative general way of dealing with the data is found then each single item will have to be described in its own terms. That said, both types of approach of necessity subsume theories of category and phrase structure. In this context, it is immediately clear that the existence of a lexicalization stage in the formation and historical development of complex prepositions serves to disconfirm one common view to be found in the literature, namely that prepositions constitute a purely functional class, as in different ways is argued by Baker (2003: 303–25) and Cinque (2010). Indeed, Déchaine (2005: 8) declares this to be ‘far and away the majority view’ before going on to adduce a number of arguments against it and in favour of considering prepositions to be a lexical class. She points in particular to the sheer range of meanings prepositions convey, to the way they construct patterns of semantic contrast in expressing direction, distance, time, location and the like, and to the fact that any attempt to consider them a closed functional class runs into problems about where to draw the boundaries. At one point she lists eighty items as belonging to the family of prepositions in English, while the discussion by Pullum and Huddleston (2002) raises that number considerably, though at the expense of some categorial assignments that would probably not find general assent such as all items ending in -ward(s) including, in addition to natural candidates such as toward(s), words like homeward(s), heavenward(s), and the like. But such examples serve well to make the point which we have already seen emerge from Adler’s (2010) study, namely that the boundary between fully fledged prepositions and other expressions is at best fuzzy if not downright indiscernible. In addition, what the examples discussed in this chapter demonstrate is that the fuzziness of this boundary is intrinsic since new items are constantly entering on the scene via the mechanisms of linguistic change, which in turn build on the productive rules of word formation. One of the arguments behind the decision by Baker (2003) to relegate the category P to a purely functional status is, however, worth considering further, namely the difficulty of finding a plausible featural decomposition of the category ‘preposition’. In the original X-bar theory of categories, P is fitted in as something of an afterthought with the doubly negative feature composition [–N, –V]. This, however, is not consistent with the fact, as pointed out by Jackendoff (1977: 78f.), that prepositions share with verbs the property of taking arguments and in many languages are like verbs in determining the case of such arguments. Hence, his alternative proposal that P should be defined as [– Subj, + Obj]. The problem here, noted by Baker (2003: 11f.), is that these features are not themselves very easily defined and lead to some strange groupings as when both nouns and verbs are considered to be [+ Subj]. On the other hand, Baker’s semantically based alternative—namely to treat nouns as definable in terms of reference while verbs are defined in terms of predication—has much to recommend it provided one does
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
358
not, as he does, then try to generalize it across all word classes and thereby end up, as we have seen, being forced to deny the existence of lexical prepositions. An alternative approach is to abandon the search for one set of features that can cross-classify to identify categorial groupings and to look instead to break individual items down into their core semantic ingredients within a universally defined syntactic architecture. This is the tack taken within the framework of nanosyntax (see Baunaz et al. 2018 for discussion, exemplification, and references). Its application in the domain of prepositions is to be seen in Roy and Svenonius (2009), whose proposed structure for a côté de la maison ‘beside the house’ is shown in (8) (we use NP as the head for what they label as DP, but nothing hangs on that difference for present purposes):
PlaceP
(8)
à
AxialPart côté
KP K
NP
de la maison It is immediately obvious that there is no single head labelled preposition here. Rather there are three components. Place identifies the overall semantic function of the constituent within the phrase or clause to which it belongs while K is the functional head governing the element which expresses the point of reference for the location, what they call the ‘ground’. This is the natural extension of the original use of K to indicate an item’s abstract Case role as in Bittner and Hale (1996) and is supported by the fact that where Romance languages have reflexes of earlier Latin used the genitive. Compare too the origin of the Hindi item ke in the examples in (6), which on this account would also occupy K. One difference, however, is that K in the nanosyntactic version also has the semantic function of linking the ground item to the relevant spatial dimension, here the side, and it is the head dominating the latter, the Ax(ial)Part, which is the most original ingredient of this way of conceiving things. The AxPart is a function from the ground to a specific domain, which is in turn linked by the Place head to the larger space. The example here works in terms of spatial orientation and location, as is to be to expected, given the overall Place head but the same structures can also be deployed in the analysis of temporal (TimeP), causal (CauseP) expressions and the like. Crucially, however, the element occupying the AxPart slot, although it may bear phonological resemblance to a full lexical item does not have an independent syntactic category. In the diachronic terms which are our interest here, that means that côté has undergone a shift from N to AxPart, an instance in the conception of
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
359
Roberts and Roussou (2003) of grammaticalization as an upwards shift from a lexical to a functional head. Syntactic structures like (8) are constructed according to universal principles, two of which are relevant here: a) all trees are binary branching so a single node dominating the three etymological ingredients of complex expressions such as à côté de or Occ. fin-qu’à ‘until’ would a priori not be an option; b) decompositions such as these must obey the ‘one feature one head’ maxim (OFOH), which forbids clusters of features under one end and enforces a strict syntax-semantics mapping where each identified semantic feature occupies its own head. This applies even when the overt morphological form might suggest the opposite. Thus, Roy and Svenonius (2009: 106) argue: Since e.g. behind the house means exactly the same thing as in back of the house, and we have assigned meanings to the Place and AxPart components (plus a licensing component K) it stands to reason that a simple preposition like behind consists of the same three parts abstractly.
This in turn requires the English word behind to give realization to the three heads Place + AxPart + K, which do not at first sight constitute a well-formed subtree. The mechanism to achieve this is spanning, by means of which a sequence of adjacent heads can be spelled out together by one phonological form. And it follows that since the structures themselves are universally defined, any changes will also have to be stated in terms of spell-out. In particular, on this approach change does not involve loss of compositionality since the structures which are input to the semantics do not alter over time. Rather it is what Langacker (1987: 448) calls ‘analyzability’, that is to say ‘the ability of speakers to recognize the contribution that each component makes to the composite whole’, which is obscured. Two criticisms can be levelled against this approach, one synchronic and one diachronic. On the synchronic, or perhaps better metatheoretical, side there must be a concern that the OFOH maxim or hypothesis contains an intrinsic redundancy. If features map one-to-one onto heads, the question arises as to why we need both. And since the crucial ingredients here are the features, which are the semantic building blocks out of which the interpretations of sentences and utterances are constructed, mapping each one onto a single syntactic head can reasonably be dispensed with except when the overt surface properties of a language require it. The second issue lies in the diachronic process by which complex prepositions develop, namely through the incorporation of a nominal head such as côté ‘side’,
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
360
force ‘strength’, dich ‘saying’. Although there are recurrent patterns across languages, particularly as regards the use of spatial terms and names of body parts like ‘head’ and ‘hand’, there do not appear to be any absolute limits on what may, at a given time and place, come to form part of a complex preposition, just as more generally there are no limits on what new lexical items a language may develop through the processes of semantic change. And since the mechanism of grammaticalization is a gradual one with intermediate stages for example between a full verb and its various auxiliary or auxiliary-like uses, this in turn means that the number of intermediate heads required is very large and indeed potentially infinite. It is not plausible therefore to assume in advance a universal set of heads/ features which can enter into the nanosyntactic computations. Both these criticisms reflect in different ways concerns which have been expressed over the years about the cartographic approach to the mapping out of linguistic structure, of which nanosyntax is ‘a direct descendant’ (Baunaz and Lander 2018: 3). In this connexion, a valuable via media is opened up by Wiltschko’s (2014) concept of a ‘universal spine’. Her starting point is a distinction between what she labels the ‘no-base hypothesis’ (NBH) and the ‘universal base hypothesis’ (UBH). The former, as the name implies, assumes there are no a priori givens in universal grammar but rather new items emerge and become routinized precisely through the mechanisms of linguistic change which we have been discussing. This way of viewing things is standard with those who adopt a constructionist or cognitivist view of natural language, but, Wiltschko argues, is too weak to explain the many recurrent typological patterns attested across the languages of the world. On the other hand, the UBH associated with work within the various Chomskyan frameworks over the years, but particularly with the cartographic and nanosyntactic approaches, is too strong since the range of cross-linguistic categorial and subcategorial diversity is too great to be assumed in advance as part of the child language learner’s innate grammar. She suggests instead that what is uniformly and universally available to the language learner is a mechanism for building new categories based on whatever input is available in the external language community, the so-called universal spine hypothesis (USH). In her words (Wiltschko 2014: 24): The USH differs from the UBH in the assumption that the spine does not serve as the repository for language-specific categories. Instead it serves as the basis for the construction of categories, as a universal categorizer [italics in original] . . . Language specific categories are constructed from a small set of universal categories K and language specific units. . . . The set of universal categories K is hierarchically organized
Space does not permit us to enter here into more detail but it should be clear that a framework such as this creates the necessary conceptual space for the observable
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
361
cross-linguistic variety of prepositions and preposition-related structures without requiring that all the possible options are built in ahead of time. In particular, there is now a natural way to capture Déchaine’s (2005) insight that what is special about prepositions is not that they are not a lexical category, which as we have seen they manifestly are, but that they constitute a ‘borderline’ or ‘elsewhere’ category, sharing properties with both full lexical categories such as noun and verb but also available for use in a range of grammatical functions, the precise details of which vary from language to language. And, it is precisely this border which is crossed in the course of change. How exactly the class of adpositions plays out in different languages is then the joint product of the mechanisms provided by the universal spine and the data that it is available to feed into it in particular instances.
16.7 Prefixes and PPs Given the fact that prefixes and prepositions derive from the same etymological sources (Vincent 1999, 2017), it is reasonable to ask how they are connected in synchronic grammars. In Vincent (2017: 305f.) we argued for an account whereby the original non-projecting particles fuse with a lexical stem to yield a compound which is in the first instance semantically transparent, though such transparency may be obscured with the passage of time as we saw with some of the examples in (4) above. This may in turn lead to a change in the valency of the prefixed item as when intransitive Latin ‘go’ combines with the particle ‘across’ to give the transitive verb ‘cross’. However, it is also possible to find instances in which the prefix co-occurs with its cognate adposition as in the Latin examples in (9) (the relevant parts are in bold): (9)
a. plebem de sacro monte deduxit people. sacred. mountain. lead..3 ‘he led the people down from the sacred mountain’ (CIL xi.1826) b. Caesar . . . militesque ex oppido exire iussit Caesar. soldier.. city. go out. order..3 ‘Caesar ordered the soldiers to leave the city’ (Caesar De bello Gallico 2.33) c. ne quam multitudinem hominum amplius . crowd. man.. more trans Rhenum in Galliam traduceret Rhine. in Gaul. bring across...3 ‘that he should not bring across the Rhine into Gaul any more groups of men’ (Caesar De bello Gallico 1.35)
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
362
On our view, which is no more than what has been assumed in traditional grammar, examples like these represent different ways in which the semantic content of the prefix/preposition may be realised and these can vary from language to language. One can compare for example English he entered the room with its French equivalent il est entré dans la chambre. However, an alternative nanosyntactic account has been developed by Tolskaya (2018) in work on Russian aspectual prefixes. In her Russian example given here as (10), the verb with the general goal prefix za- co-occurs with a PP expressing the precise destination, while in her example given here as (11) we find a pattern corresponding to the Latin examples in (9) with the same element both prefixed to the verb and as part of the associated prepositional phrase: (10) my za-šli v dom we -go. in house. ‘we went into the house’ (11) oto-jti ot doma from-walk from house ‘to walk from the house’ She proposes to account for this by assuming, following Svenonius (2004), that the PP which serves as an argument of the verb is linked to the choice of prefix not only semantically but also syntactically in that the prefix starts life as a subconstituent of the PP. For an example like (9a) the starting point in the syntactic derivation would be a configuration as in (12) with the Pref(ix)P then raising to adjoin to the verb before lexical spell-out.
PP
(12)
PrefP Path de
PlaceP Place GROUND
PathP Path de
PlaceP Place
NP sacro monte
Once again we are faced with an analytical choice between an account which states the semantics independently of the syntax and one which mediates the former through an elaborated version of the latter. And once again the evidence of linguistic change—at least in the Romance case, since I cannot speak for the diachronic possibilities in Russian—would seem to argue in favour of representing
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
363
the semantic structures and their historical shifts independently of their syntactic encoding. A similar distinction between the representation of semantic content and syntactic structure is also supported by the typological differences between the way languages grammaticalize such content, as is well demonstrated, for instance, in Talmy (2007).⁹
16.8 Conclusions The time had come to sum up the conclusions of the present chapter. In so doing I will draw some links between the material discussed here and in Vincent (2017). The two chapters are complementary in that the former considered the role of a single etymological item, Latin , across a range of Romance languages, while here I have narrowed the geohistorical focus within Romance while widening the inventory of etymological sources and drawing explicit comparison with languages beyond Romance. From both it emerges that there is considerable variation in both the complex and the compound forms which have developed in different regions. This in turn is consistent and on a par with the larger patterns of lexical variation within the family, and is in sharp contrast with the cross-linguistic unformity in the development of the core functional prepositions. It also serves to reinforce from a diachronic perspective the view argued for in Déchaine (2005) that prepositions, as distinct from nouns, verbs and adjectives, constitute a borderline but nonetheless lexical category. Part of her argument consisted in the availability of prepositions in the formation of compounds, both with other lexical classes and with each other, and here too the Romance data provides supporting evidence. In particular, and pace Lehmann (2019), there is reason to suppose that some Romance items derive by direct prefixation without prior univerbation, and that is particularly the case in those items which develop early in the passage from Latin to Romance (Buchi 2009). Finally, I have argued, on both diachronic and synchronic grounds, that an account which treats morphosyntactic and semantic patterns as in principle independent is preferable to one like the nanosyntactic approach in which the two are intrinsically linked via the OFOH hypothesis. To the extent that they are connected or connectible in any given languages, it is better to have recourse to category-building mechanisms of the kind envisaged within Wiltschko’s Universal Spine Hypothesis. More generally, the conclusion—which will come as no surprise to JC, given his own work over the years—is that in the pursuit of understanding language as a general phenomenon diachronic and synchronic considerations are inseparable.
⁹ I am grateful to Delia Bentley for reminding me of the relevance of Talmy’s work in this context.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
17 Syncretism and metamorphomes in northern Occitan (Lemosin) varieties Louise Esher
17.1 Introduction This study explores the origin and behaviour of syncretism patterns (i.e. patterns in which ‘two or more distinct morphosyntactic values are collapsed in a single inflected word form’, Baerman 2007: 539) in the verb paradigm of northern Occitan varieties spoken in the Limousin region of central France (henceforth ‘Lemosin varieties’).¹,² In Lemosin varieties, syncretism is overall more prevalent than in southern Occitan varieties, and can be found between person/number forms within a single TAM (tense, aspect, mood) category, as well as between forms realizing different TAM categories (§17.2). On the basis of a comparative study of data from modern Lemosin varieties and historical grammars, the majority of syncretism patterns in these varieties are shown to result from regular sound change causing mergers of previously distinct inflexional forms (§17.3). The syncretism patterns are also observed to share key characteristics with the structures termed ‘metamorphomes’ (Round 2015; for Romance examples and discussion, see e.g. Maiden 2009a, 2016a, 2016b, 2018a).³ Metamorphomes are a phenomenon of ‘autonomous morphology’ (Aronoff 1994); they can be characterized as groupings of paradigm cells which share inflexional exponents, as bundles of implicational relationships between paradigm cells, and as recurrent patterns of paradigmatic distribution of inflexional exponents. Such patterns may ¹ The research reported here was begun during a Junior Research Fellowship funded by St John’s College, Oxford (2013–16) and continued under the auspices of two projects overseen by the French National Research Agency: ANR-17-CE27-0001-01 (Project ’The Linguistic Crescent: A Multidisciplinary Approach to a Contact Area between Oc and Oïl Varieties’) and ANR-10-LABX-0083 (Programme ‘Investissements d’Avenir’, Labex EFL, Strand 3, Workpackage LC4—‘Les parlers du Croissant: une aire de contact entre oc et oïl’). A version of the study was presented at the 18th International Morphology Meeting (Budapest, 10–13 May 2018) and I thank the audience there for their helpful comments. I am grateful to Xavièr Bach, Pierre-Joan Bernard and the CIRDOC in Béziers for assistance with access to a number of publications not widely available; and to Hans-Olav Enger, Martin Maiden and Nicolas Quint for their help in untangling various of the data and issues discussed here. ² Lemosin /lemuˈzi/ is the Occitan term for the area and by extension its speech varieties. ³ Note that Maiden refers to these patterns as ‘morphomes’ throughout.
Variation and Change in Gallo-Romance Grammar. First edition. Sam Wolfe and Martin Maiden (eds). This chapter © Louise Esher 2020. First published in 2020 by Oxford University Press.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
365
align partially, entirely or not at all with phonological or syntactic/semantic natural classes (Smith 2013). Two significant properties of metamorphomes are their systematicity (the same distributional pattern is found across multiple lexemes), and their productivity as templates for morphological analogy affecting inflexional exponents. As both these properties are shared by syncretism patterns in Lemosin varieties, and the interaction of the syncretism patterns with established metamorphomes parallels interaction between established metamorphomes, this study proposes that the behaviour of the syncretism patterns is most accurately captured by treating them as metamorphomes (§17.4).
17.2 Syncretism patterns in the north Lemosin variety of Gartempe (Creuse) Some impression of the patterns observed in Lemosin may be gained from the examples reproduced below, taken from Quint’s (1996) descriptive grammar of the variety of Gartempe (Creuse), a village situated in the northern part of the Lemosin dialect area, within what is called the ‘Croissant linguistique’, a transitional area between oc and oïl varieties (Brun-Trigaud 1990). The most pervasive and systematic type of syncretism in this variety concerns person/number values, for which three patterns are found (1=3, 1=3, and 2=2). The variety of Gartempe also presents TAM syncretism: between the present indicative and present subjunctive for all person / number combinations in the first conjugation; between the present indicative and present subjunctive for all person / number combinations except the third person singular in the second and third conjugations, including many (though not all) irregulars; and between the imperfect indicative and conditional in a subset of third-conjugation verbs. Table 17.1 shows the synthetic forms of the first-conjugation verb chantar ‘sing’ (Quint 1996: 115f.).⁴ In five of the seven synthetic paradigm categories (present indicative, present subjunctive, imperfect indicative, imperfect subjunctive, conditional), all three patterns of person syncretism apply: 1=3, 1=3, and 2=2. In the remaining two categories, the preterite and future, only one of these patterns is found (1=3), while the other four forms are each distinct. The present indicative and present subjunctive are syncretic with each other for each person / number combination. In first-conjugation verbs which display root allomorphy, typified by sauvar ‘save’ (Table 17.2) and gaitar ‘look (at)’ (Table 17.3), such allomorphy is
⁴ Note that stress is not systematically indicated in the source paradigms; for the purposes of this study, stress placement is inferred based on Quint’s (1996) description of the phonological system in the variety of Gartempe.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
366
Table 17.1. chantar [tsãˈta] ‘sing’, Gartempe (Quint 1996: 115f.)
1 2 3 1 2 3
.
.
.
.
ˈtsãtə tsãˈta: ˈtsãtə tsãˈtã tsãˈta: tsãˈtã
ˈtsãtə tsãˈta: ˈtsãtə tsãˈtã tsãˈta: tsãˈtã
tsãˈtavə tsãtaˈva: tsãˈtavə tsãtaˈvã tsãtaˈva: tsãtaˈvã
tsãˈti tsãtəˈta: tsãˈte tsãtəˈtã tsãtəˈte tsãtəˈtã
tsãˈtœsə tsãtəˈsa: tsãˈtœsə tsãtəˈsã tsãtəˈsa: tsãtəˈsã
tsãtəˈraj tsãtəˈra: tsãtəˈrø tsãtəˈrã tsãtəˈre tsãtəˈrã
tsãtəˈjø tsãtəˈja: tsãtəˈjø tsãtəˈjã tsãtəˈja: tsãtəˈjã
Table 17.2. sauvar [soˈva] ‘save’, Gartempe (Quint 1996: 118)
1 2 3 1 2 3
.
.
ˈsawvə soˈva: ˈsawvə soˈvã soˈva: soˈvã
ˈsawvə soˈva: ˈsawvə soˈvã soˈva: soˈvã
Table 17.3. gaitar [ɡeˈta] ‘look (at)’, Gartempe (Quint 1996: 118)
1 2 3 1 2 3
.
.
ˈɡajtə ɡeˈta: ˈɡajtə ɡeˈtã ɡeˈta: ɡeˈtã
ˈɡajtə ɡeˈta: ˈɡajtə ɡeˈtã ɡeˈta: ɡeˈtã
distributed in line with the patterns of syncretism observed for chantar: the alternants [aw] and [aj] occur only in the first person singular and third person singular forms of the present indicative and present subjunctive.⁵ Observation of sauvar and gaitar demonstrates that the pairings found in Lemosin are not simply cases of identity between desinences: instead, each pair of syncretic forms involves identity between entire wordforms.
⁵ The forms of gaitar and sauvar given as present subjunctive here are labelled ‘imperfait / imperfach’ (i.e. imperfect indicative) in the original source. I assume this to be a misprint: the desinences of the forms reproduced are characteristic of the present subjunctive in this variety, whereas [soˈvavə], [ɡeˈtavə], etc. would be expected in the imperfect indicative (compare Table 17.1).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
367
Table 17.4. partir [parˈtir] ‘leave’, Gartempe (Quint 1996: 121)
1 2 3 1 2 3
.
.
.
.
parˈtisə partiˈse parˈti partiˈsã partiˈse partiˈsã
parˈtisə partiˈsa: parˈtisə partiˈsã partiˈse partiˈsã
parˈtiʃə partiˈʃa: parˈtiʃə partiˈʃã partiˈʃa: partiˈʃã
partiˈsi partisəˈte partiˈse partisəˈtã partisəˈte partisəˈtã
partiˈsœsə partisəˈsa: partiˈsœsə partisəˈsã partisəˈsa: partisəˈsã
partiˈraj partiˈra: partiˈrø partiˈrã partiˈre partiˈrã
partiˈjø partiˈja: partiˈjø partiˈjã partiˈja: partiˈjã
Table 17.5. vendre [ˈvãdrə] ‘sell’, Gartempe (Quint 1996: 123)
1 2 3 1 2 3
.
.
.
.
ˈvãdə vãˈde vã vãˈdã vãˈde vãˈdã
ˈvãdə vãˈda: ˈvãdə vãˈdã vãˈde vãˈdã
ˈvãdjə vãˈdja: ˈvãdjə vãˈdjã vãˈdja: vãˈdjã
vãˈdi vãdəˈte vãˈde vãdəˈtã vãdəˈte vãdəˈtã
vãˈdœsə vãdəˈsa: vãˈdœsə vãdəˈsã vãdəˈsa: vãdəˈsã
vãˈdraj vãˈdra: vãˈdrø vãˈdrã vãˈdre vãˈdrã
vãˈdjø vãˈdja: vãˈdjø vãˈdjã vãˈdja: vãˈdjã
While it is conventional for Occitan grammars to distinguish three conjugations,⁶ a practice followed by this study and its source materials, the inflexional desinences of the second and third classes are largely identical to each other in the Lemosin varieties under discussion. The second and third classes are instead differentiated by thematic elements: the second conjugation (e.g. partir ‘leave’, Table 17.4) is characterized by the theme vowel /i/ and the presence of a thematic element /is/ (Maiden 2003; Esher 2016),⁷ while the third conjugation (e.g. vendre ‘sell’, Table 17.5) lacks these formatives. Although certain of the desinences and inflexional formatives in the secondand third-conjugation paradigms differ from those in the first conjugation, the patterns of person syncretism observed are almost identical. In all conjugations, the present subjunctive, imperfect indicative, imperfect subjunctive and conditional present the patterns 1=3, 2=2 and 1=3, while the future presents a single pattern, 1=3. The only differences concern the present indicative and preterite, which both present two patterns in non-first-conjugation verbs, 1=3 and 2=2. ⁶ These are respectively the continuants of Latin conjugations I (thematic vowel ), IV (thematic vowel ), and III (short thematic vowel); modern Occitan varieties do not continue Latin conjugation II, members of which were assimilated to III. Note that several authors, including Ronjat (1937) number the continuants of IV ‘third conjugation’ and the continuants of III ‘second conjugation’, as is conventional for Catalan; this study and its source material number the continuants of IV ‘second conjugation’, as is conventional for French. See also Maiden (2018a: 38f.). ⁷ In the variety of Gartempe, original [is] in the imperfect indicative of second-conjugation verbs has developed to [iʃ] as a result of assimilation: *partisjam > [partiʃã].
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
368
TAM syncretism between the present indicative and present subjunctive in the second and third conjugations (as well as for the many irregular verbs which do not present a distinctive stem in the present subjunctive) applies only in the first person singular and in the plural. In regular third-conjugation verbs, such as vendre ‘sell’ (Table 17.5), an additional pattern of TAM syncretism, between the imperfect indicative and the conditional, occurs in the second person singular and all plural forms.
17.3 The origin of the syncretism patterns 17.3.1 Introduction Almost all cases of syncretism in the regular verbs shown above are the expected result of regular sound changes, in particular the loss of final consonants, and the attraction of stress to the resulting long and nasal vowels. Thus, in diachrony, these syncretisms result more frequently from cases of phonological merger than from analogical change involving replacement of one morphological form by another. Mediaeval Occitan forms in this section are taken from Skårup (1997) unless otherwise indicated.
17.3.2 Person syncretism: 1=3 1=3 is the syncretism most widely found in Lemosin varieties. It is found systematically across all TAM categories and conjugational classes in the variety of Sanilhac, excepting the future (Marshall 1984: 48); across all TAM categories and conjugational classes in the variety of Nontron (Reydy 2008: 100–9; with the possible exception of the imperfect subjunctive, for which no forms are given); and among the many variant forms given by Lavalade (1987) for the Lemosin dialect area in general. In the variety of Gartempe, 1=3 is a case of ‘complete syncretism’ (Baerman et al. 2005: 59), since it applies to all paradigms, occurring in all synthetic paradigm categories (present indicative, present subjunctive, imperfect indicative, imperfect subjunctive, conditional, preterite, future) for all conjugational classes.⁸ Furthermore, all first person plural and third person plural forms in this variety share the desinence [ã].
⁸ The only exception given in Quint’s grammar concerns the present indicative of plaire ‘please’: two stem variants are available to the first person plural cell, only one of which is found for the third person plural cell.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
369
For first person plural forms, the desinence [ã] < - is etymological in the first-conjugation present indicative, in the second- and third-conjugation present subjunctive, and in the imperfect indicative and conditional of all conjugations: e.g. > [tsãˈtã] ‘sing...1’, > vendam > [vãˈdã] ‘sell. ..1’, > [tsãtãˈvã] ‘sing...1’, > vendiam > [vãˈdjã] ‘sell...1’, > cantariam > [tsãtəˈjã] ‘sing..1’. In all these forms, the final unstressed syllable undergoes deletion (- > -am), and the vowel of the new final syllable is nasalized by regressive assimilation to the following nasal consonant. As in French (Ohala 1989), the sequence of nasalized vowel and nasal consonant ultimately develops to a nasal vowel, -am [am, an] > [ã]; indeed, the same process can be seen to have applied to the root of chantar (- > [tsãt]). The root of vendre ‘sell’ demonstrates that the tautosyllabic sequence of front mid vowel + nasal consonant also develops to [ã]: vendiam > [vãˈdjã] (Quint 1996: 13). In early Occitan the contrast between mid-high and mid-low front vowels was neutralized under nasality (Sampson 1999: 141), with both [εn] and [en] yielding [ẽn]. Subsequently, in the variety of Gartempe, [ẽ] has lowered to [ã]; this development resembles that found in the history of French, where [ẽ] lowers and merges with [æ̃] < [ã] (Sampson 1999: 68–70). The regular development of [en] to [ã] explains the presence of the desinence [ã] in the first-conjugation present subjunctive, e.g. > cantem > [tsãˈtã] ‘sing...1’ and in the future of all lexemes, e.g. > [tsãtəˈrã] ‘sing..1’. For the imperfect subjunctive, while the development > cantessem > [tsãtəsã] would be expected, textual evidence shows that the etymological -e- of the desinences was typically replaced in mediaeval Lemosin by -a-; thus [tsãtəˈsã] is more likely to continue analogical chantessam than etymological chantessem (Chabaneau 1876: 283; Ronjat 1937: 196). The preterite has undergone significant analogical remodelling, in which the third person singular form is taken as a stem for the second person singular form and the plural forms (Ronjat 1937: 193; Bybee and Brewer 1980).⁹ The final vowel [ã] may continue etymological -am and -em in the first and third conjugations respectively, but in the second conjugation must be due to analogical extension of either -am or -em (compare modern [partisətã] ‘leave..1’ with its mediaeval equivalent partim).
⁹ According to Quint (1996: 105), the element [ət] is the regular reflex of Latin -- in the second person singular and second person plural perfect, e.g. > cantetas ‘sing..2’. Although this reconstruction is plausible from a phonological point of view, it is not supported by the textual evidence: Chabaneau (1876: 278), in a survey of fourteenth- to sixteenth-century texts from the Limousin area, finds only what he terms ‘classical’ preterite forms (i.e. without -et- or its more common equivalent -er-), while the earliest attestations which he notes for preterites in -et- are from seventeenth-century carols in the Auvergne region.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
370
For third person plural forms, final [t] is lost (Ronjat 1932: 266) and the resulting final sequence V+[n] develops to a nasal vowel, e.g. > cantan > [tsãˈtã] ‘sing...3’. Given comparative and diachronic evidence for Occitan, stress in third person plural forms of the present indicative and present subjunctive would be expected to fall on the penult, and it continues to do so in the variety of Sanilhac (Marshall 1984: 48), as well as in some of the data cited by Javanaud (1981: 68). A significant feature of Lemosin varieties is the presence of distinctive vowel quantity, to which stress assignment is sensitive (Javanaud 1981; Lavalade 1986). As nasal vowels count as long (or half-long, Javanaud 1981: 68), they can attract stress:¹⁰ in the varieties of Nontron (Reydy 2008: 117, 119) and Montembœuf (Dourdet 2015: 256), as in that of Gartempe, stress systematically falls on the final (always nasal) vowel of third person plural forms. Stressed [ã] in Gartempe is thus the expected reflex of stressed and unstressed [an] and [en], e.g. > [tsãˈtã] ‘sing...3’, > [vãˈdã] ‘sell...3’, > [tsãtãˈvã] ‘sing...3’, > [vãˈdjã] ‘sell.. .3’, > [tsãtəˈrã] ‘sing..3’, > [tsãtəˈjã] ‘sing..3’.¹¹ Final stress in third person plural forms also has consequences for stem distribution, which, in the present indicative and present subjunctive, is correlated with stress placement (see Tables 17.2 and 17.3): the alternants [aw], [aj] occur only in stressed roots. In most Occitan varieties, a stressed root, and thus a distinctive alternant in relevant lexemes, would be expected in the third person plural forms of the present indicative and present subjunctive forms. By contrast, in the variety of Gartempe,¹² primary stress in these forms has shifted to the desinence due to the long, nasalized vowel in the final syllable; since a diphthong would not be licit in an unstressed syllable (Quint 1996: 30, 119), the shift of stress is accompanied by a change in stem vowel, replacing the alternants [aw], [aj] with their unstressed counterparts [o], [e] respectively. In summary, the systematic syncretism of desinences, stress pattern and (where relevant) stem alternants between first person plural and third person plural forms in regular verbs, across all tenses and all conjugations, is almost entirely attributable to regular sound change, the only exception being the preterite, where some analogical extension of desinences is found.
¹⁰ Note also that where the penult and final syllable are of equal length (as here), stress falls on the final syllable (Javanaud 1981: 53). ¹¹ The desinences [an]/[en], etymological in these examples, were further extended by morphological analogy: the reflex of etymological - in non-first-conjugation third person plural present indicative forms was replaced in early Romance by the reflex of - (Maiden 2009a: 48), yielding [en] > [ã]; in the preterite, [ã] results from analogical extension of -an and/or -en (Ronjat 1937:179); and in the imperfect subjunctive, [ã] continues analogical -an, as described above for the first person plural form. ¹² And in some other Lemosin varieties, though not all: see, e.g. Javanaud (1981: 68), Marshall (1984: 48).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
371
17.3.3 Person syncretism: 2=2 In the variety of Gartempe, the syncretism pattern 2=2 occurs systematically in the present indicative, present subjunctive, imperfect indicative, imperfect subjunctive, and conditional of first-conjugation verbs; and in the present indicative, imperfect indicative, preterite, imperfect subjunctive, and conditional of nonfirst-conjugation verbs. Similar distributions are found in other Lemosin varieties (e.g. Marshall 1984: 48; Reydy 2008: 100–9). The syncretism of second person singular and second person plural forms in the present indicative of the first conjugation results from regular sound change. In the second person plural, the final unstressed vowel falls (e.g. > cantatz ‘sing...2’), the resulting sequence [ts] reduces to [s], and the final [s] falls with compensatory lengthening (Ronjat 1932: 275, 283f.; compare the similar development in French, Pope 1934: 206f.; De Chene and Anderson 1979: 520f.; Kavitskaya 2017), giving chantâ with a long,¹³ stressed final vowel (Chabaneau 1876: 233). In the second person singular, final [s] also falls with compensatory lengthening (Ronjat 1932: 275), e.g. > cantas > chantā with a long, originally unstressed final vowel (Chabaneau 1876: 233). As long vowels attract stress in Lemosin varieties (Chabaneau 1876: 9; Javanaud 1981: 53), primary stress in the second person singular form shifts from the root to the final syllable, resulting in syncretism with the second person plural form. The issue of stem vowel quality in the present indicative and present subjunctive, discussed above for 1=3, is equally relevant for the syncretism of second person forms. Most varieties of Occitan present a contrast between root-stressed second person singular forms and non-root-stressed second person plural forms, correlated with stem alternation. In Lemosin varieties such as that of Gartempe, the shift of stress in the second person singular forms is accompanied by the replacement of the original rhizotonic stem alternants (here [aw], [aj]) with their arrhizotonic counterparts (here [o], [e] respectively). In the imperfect indicative, conditional, and imperfect subjunctive, the syncretism 2=2 can be attributed to the same processes of deletion and compensatory lengthening, albeit applying to non-etymological desinences 2 -ssas, 2 -ssatz in the case of the imperfect subjunctive (Chabaneau 1876: 283; Ronjat 1937: 196). In the preterite, all second person singular and second person plural forms present the element -et- extended by analogy from the third person singular form. Non-first-conjugation second person singular and second person plural forms, and first-conjugation second person plural forms, all share the desinence [e].
¹³ This form, like others from the same work, is given according to Chabaneau’s original (somewhat idiosyncratic) transcription. In Chabaneau’s system (1876: 6), a circumflex is used to indicate a long stressed vowel, a macron to indicate a long unstressed vowel, and an acute accent to indicate a short stressed vowel.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
372
This desinence continues mediaeval Occitan third-conjugation preterite desinences with theme vowel -e- (Ronjat 1937: 177), which have been generalized across conjugations in most varieties of Occitan. Through the loss of final [t] and [s], without compensatory lengthening, 2 -es and 2 -etz fall together as [e], as happens in non-first-conjugation present indicative forms. In first conjugation verbs, second person singular preterite forms present the desinence [a:], which is most plausibly due to analogy from other TAM categories, as the desinence [a:] is shared by all second person singular forms in the first conjugation. In the present subjunctive, regular sound change would ordinarily produce syncretism between second person singular and second person plural forms in all conjugations: one would expect, e.g. , > ˈkãntes, kãnˈtes > chantei (Chabaneau 1876: 274; Ronjat 1932: 275) in the first conjugation, and forms in -ā or -â (as for the first-conjugation present indicative) in the other conjugations (Chabaneau 1876: 235). However, the observed forms frequently present inflexional desinences originally characteristic either of a different conjugation or of the present indicative. For example, the variants attested by Chabaneau for the present subjunctive forms of second- and third-conjugation verbs include 2 -ei, and 2 -ê or -ei (1876: 238–9, 244, 249)—the desinences expected for the firstconjugation present subjunctive—while Ronjat (1937: 165) finds analogical second person singular and second person plural first-conjugation present subjunctive forms in -ā in the area of Périgueux. In the modern variety of Gartempe, there is syncretism between present indicative and present subjunctive forms in all six person/number combinations for first-conjugation verbs (Table 17.1), and in four person/number combinations for non-first-conjugation verbs (Tables 17.4 and 17.5). Thus, while the expected outcome, 2=2 syncretism, is consistently found, the forms instantiating this relationship are commonly subject to analogical redistribution, which may involve TAM syncretism (§17.3.5) as well as person syncretism.
17.3.4 Person syncretism: 1=3 In the variety of Gartempe, the syncretism pattern 1=3 is found, for all verbs, in the present subjunctive, imperfect indicative, imperfect subjunctive, and conditional, and, for first-conjugation verbs, in the present indicative as well. Outside the present indicative, all such syncretism results from regular sound change in early Romance. There is no difference in stress placement or number of syllables between Latin first person singular and third person singular forms in the relevant TAM categories, nor is there a difference in vowel quality in the desinences of these forms. The only distinction is between the final consonants, which have been lost by the mediaeval period: , > venda ‘sell..’; , > cantava ‘sing..’; , > vendia ‘sell..’; , > cantesse ‘sing..’;
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
373
, > cantaria ‘sing.’ in mediaeval Occitan (Skårup 1997). According to Lavalade (1987), the majority of Lemosin varieties maintain etymological 1=3 syncretism in the imperfect indicative, conditional, present subjunctive, and imperfect subjunctive; in the varieties of Sanilhac (Marshall 1984: 48) and Nontron (Reydy 2008: 100–9), such syncretism is confined to the present subjunctive, imperfect indicative, and conditional. Presence of the syncretism pattern 1=3 in the present indicative is unusual among Occitan varieties, but can be traced to a much more widespread development. It is common in Occitan varieties for the first person singular form to be differentiated from the third person singular form in one or more TAM categories, by analogical extension of a desinence -i or -e (Ronjat 1937: 170, 172, 196; Esher 2017b). Thus, in the first person singular present indicative form, three variants occur in mediaeval Occitan (no desinence, -e, and -i), all three of which are attested for the variety of Nontron in the nineteenth century: chant, chante, chanti ‘sing...1’ (Chabaneau 1876: 232). In the variety of Gartempe, first person singular forms such as [tsãtə] are the regular continuants of forms in -e (e.g. chante), as unstressed [e] develops to schwa (compare present subjunctive forms). As final unstressed [a] also develops to schwa, third person singular present indicative forms in the first conjugation (e.g. chanta > [tsãtə]) become identical to the corresponding first person singular forms.
17.3.5 TAM syncretism: imperfect indicative and conditional forms Identity between the desinences of imperfect indicative and conditional forms of second- and third-conjugation verbs is common in Occitan for etymological reasons:¹⁴ the imperfect indicative forms of second- and third-conjugation verbs continue Latin imperfect indicative forms in -(), etc., while the synthetic conditional originates in a periphrasis collocating the infinitive of a lexical verb with the imperfect indicative of the auxiliary ‘have’, i.e. , etc. (Ronjat 1937: 171; Esher 2018). In general, conditional forms may nevertheless be distinguished from imperfect indicative forms by a unique stem or the presence of thematic elements (theme vowel and / or formative [r]). In regular third-conjugation verbs, which present neither a unique stem nor a theme vowel in the conditional, the contrast crucially depends on the formative [r]: compare vendiá ‘sell...3’ vs vendriá ‘sell..3’ in modern standard varieties of the Languedoc and Provence (Alibèrt 1976; Martin and Moulin 2007). But the sequence [rj], historically found throughout conditional forms, is unstable and vulnerable to change. In several varieties of the Languedoc, ¹⁴ With the notable exception of varieties spoken in Gascony.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
374
Auvergne, and Limousin areas, [rj] is commonly reduced to [j] in intervocalic contexts (cantariá [kantarjɔ] > [kantajɔ] ‘sing..3’; dormiriá [durmirjɔ] > [durmijɔ] ‘sleep..3’, Esher 2015a; see also Lanly 1971); while regular thirdconjugation verbs, which have consonant-final roots, show regional variation as to whether [r] or [j] is deleted from the original cluster [Crj]: [bẽnˈdra] ‘sell..3’ contrasts with [bẽnˈdrɔ] ‘sell..3’ in the variety of Molleville (Aude, ALLOc survey point 11.01), but with [bẽnˈdjo] ‘sell..3’ in the variety of Loubens (Ariège, ALLOc survey point 09.02). In the variety of Gartempe, [rj] reduces to [j] in all conjugations (Tables 17.1, 17.4 and 17.5), giving rise in the third conjugation to conditional forms syncretic with the corresponding person / number forms of the imperfect indicative (e.g. [vãˈdjã] ‘sell../.1/3’). There is, however, no evidence in the source material for productivity of this syncretism, which is not replicated by analogical change. On the contrary, the redistribution of stress disrupts the etymological identity between the desinences of the conditional and of nonfirst-conjugation imperfect indicative forms. For first person singular and third person singular forms of the imperfect indicative, stress shifts from the final syllable to the penult: thus, for example, [parˈtiʃə] replaces expected [partiˈʃø] ‘leave...1/3’.¹⁵ In some verbs, such as vendre ‘sell’ (Table 17.5), such a change may be phonologically motivated due to the long vowel in the penult, but in many others, including partir ‘leave’ (Table 17.4), it can only be analogical, extending the majority pattern of stress assignment in which first person singular and third person singular forms receive stress on the penult, while all other forms bear stress on the final syllable. By contrast, the first person singular and third person singular forms of the conditional retain their final stress (as does the future) rather than undergoing stress retraction (as do the corresponding imperfect indicative forms); as a result, syncretism between conditional and imperfect indicative forms is confined to the plural and the second person singular.
17.3.6 TAM syncretism: present indicative and present subjunctive forms Diverse patterns of syncretism between present indicative and present subjunctive forms occur, according to variety and conjugational class (see also Chabaneau ¹⁵ The stress retraction in the . is also associated with stem alternation. In some cases, e.g. [vˈœnjə] ‘come...1/3’ vs [vəˈnja:] ‘come...2’ (Quint 1996: 139) the alternation can be straightforwardly attributed to phonological restrictions: schwa occurs only in unstressed syllables (Quint 1996: 7f.) and thus a realization *[ˈvənjə] ‘come...1/3’ would not be licit. In other cases, e.g. [kuˈnajʃə] ‘know...1/3’ vs [kuneˈʃa:] ‘know...2’ (Quint 1996: 127), it appears that the etymologically regular, unstressed alternant (in this case [e]: Quint 1996: 6) can occur in a stressed syllable: thus the introduction of [aj] in the stressed syllable, replicating the alternant pair tonic [aj] and unstressed [e] (Table 17.3), is plausibly analogical.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
375
1867: 244). Tables 17.6 and 17.7 illustrate the extent of TAM syncretism between present indicative and present subjunctive forms in the Lemosin varieties of Gartempe (Creuse), Nontron (Dordogne), and Sanilhac (Dordogne), for firstconjugation and non-first-conjugation verbs. In all three varieties, third person singular forms of the present indicative and present subjunctive remain distinct in non-first-conjugation verbs, reflecting the regular deletion of final unstressed vowels other than [a] (e.g. > ven ‘sell...3’ vs > venda ‘sell...3’, Anglade 1921: 294); and syncretism is found between first person singular forms of the present indicative and present subjunctive forms across the regular conjugations, due to the analogical generalization of -e across first person singular forms (Ronjat 1937: 170, 172, 196; Esher 2017b). Syncretism between present indicative and present subjunctive forms is noticeably most extensive in the variety of Gartempe, chiefly due to this variety having undergone two sound changes which did not occur in more southerly varieties such as those of Nontron and Sanilhac: the reduction of unstressed final vowels to [ə], causing syncretism between third person singular forms of the present indicative and present subjunctive in the first conjugation; and the merger of [ẽ] with [ã], causing syncretism between the present indicative Table 17.6. . and . forms of first-conjugation verbs in the varieties of Gartempe (Quint 1996: 115f.), Nontron (Reydy 2008: 100), and Sanilhac (Marshall 1984: 48)
1 2 3 1 2 3
Gartempe (chantar ‘sing’)
Nontron (parlar ‘speak’)
Sanilhac (chabar ‘finish’)
. tsãtə tsãta: tsãtə tsãtã tsãta: tsãtã
. parlε parla: parlɔ parlẽm parla: parlẽn(m)
. sabε sɔba, sɔbei sabɔ sabε̃n sɔba sabε̃n
. tsãtə tsãta: tsãtə tsãtã tsãta: tsãtã
. parlε parlei parlε parl(j)ãm parle: parlãn(m)
. sabε sɔba sabε sɔbɔ̃n sɔba sɔbɔ̃n
Table 17.7. . and . forms of third-conjugation verbs in the varieties of Gartempe (Quint 1996: 123), Nontron (Reydy 2008: 105), and Sanilhac (Marshall 1984: 48)
1 2 3 1 2 3
Gartempe (vendre ‘sell’)
Nontron (metre ‘put’)
Sanilhac (vendre ‘sell’)
. vãdə vãde vã vãdã vãde vãdã
. me:tε mεtei mei mεtẽm mεte: mεtẽn(m)
. vε̃ndε vε̃ndei vε̃n vε̃ndε̃n vε̃nde vε̃ndε̃n
. vãdə vãda: vãdə vãdã vãde vãdã
. me:tε mεtei me:tε mεt(j)ãm mεte: mεtãm
. vε̃ndε vε̃nda vε̃ndε vε̃ndɔ̃n vε̃nda vε̃ndɔ̃n
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
376
and present subjunctive for first person plural and third person plural forms across conjugations. The major locus of variation is second-person forms. Syncretism between second person singular forms of the present indicative and present subjunctive is found in the first conjugation in Gartempe and Sanilhac, but in the third conjugation in Nontron; while syncretism between second person plural forms of the present indicative and present subjunctive occurs in all conjugations in Gartempe, in the first conjugation in Sanilhac and in the third conjugation in Nontron. In the varieties of Nontron and Sanilhac, the present subjunctive desinences are constant across conjugations, although observation of regular sound changes predicts that they should remain distinct (as reflexes of Latin and do not merge in these varieties). The modern distribution of desinences is due to analogical levelling across conjugations, in opposite directions: in Nontron, first-conjugation present subjunctive desinences with theme vowel -e- have been generalized across conjugations, causing incidental syncretism with non-first-conjugation present indicative desinences; while in Sanilhac, non-first-conjugation present subjunctive desinences with theme vowel -a- have been generalized across conjugations, causing incidental syncretism with first-conjugation present indicative desinences. In Gartempe, by contrast, present subjunctive desinences maintain limited contrast between conjugations, and syncretism of present indicative and present subjunctive forms has two distinct causes. In the first and third persons, syncretism of present indicative and present subjunctive forms is due to regular sound change ([am, an, em, en] > [ã]; §17.3.1). In the second person, however, such syncretism results from analogical remodelling of present subjunctive forms on the basis of present indicative forms. This process is most clearly visible in the second person plural present subjunctive form, which receives the desinence -[e] in regular non-first-conjugation verbs such as partir ‘leave’ and vendre ‘sell’ but retains its historically expected form in -[a:] in a number of irregular verbs. Quint’s (1996) grammar includes eight verb lexemes with a second person plural present subjunctive form in [a:]; second person plural forms of the present indicative and present subjunctive for these lexemes are given in Table 17.8.¹⁶ The variety of Gartempe has few lexemes for which Quint’s grammar attests a difference of stem between present indicative and present subjunctive forms,¹⁷ and it is notable that almost all such lexemes figure in Table 17.8, with the only exceptions being the impersonal verbs fâlér ‘be necessary’ and plòure ‘rain’, which by definition do not have a second person plural form, together with savêr ‘know’.¹⁸ ¹⁶ Also by implication válér / vòlêr ‘be worth’ for which a full paradigm is not given since all forms except the infinitive are syncretic with those of vòlêr (Quint 1996: 138). ¹⁷ Comparison with the dialectal variants listed by Lavalade (1987) shows that this is a common situation in Lemosin varieties, contrasting with southern Occitan varieties in which non-first-conjugation present subjunctive forms tend to present a stem (often shared with the preterite and imperfect subjunctive) distinct from that found in the present indicative (Wheeler 2011; Esher 2016). ¹⁸ Present subjunctive forms for anar/nar ‘go’ were not available (Quint 1996: 125).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
377
Table 17.8. ..2 and ..2 forms of irregular verbs retaining -[a:] for ..2 in the variety of Gartempe (Quint 1996: 108f., 127, 133f., 141) lexeme
..2
..2
ètre ‘be’ avêr ‘have’ corrêr ‘run’ plâire ‘please’ pòdêr ‘be able’ quêure ‘cook’ vòlêr ‘want’
se a: kuˈre pleˈze, pləˈze poˈde kəˈze voˈle
ʃa: aˈja: kuˈre, kuˈra: pleˈza: pyˈtse, pyˈtsa: kəˈza: vyˈtsa:
This distribution suggests that the spread of -[e] to forms originally presenting -[a:] has been favoured by pre-existing identity of stem between second person plural forms of the present indicative and present subjunctive. Such a development would be consistent with analogical changes observed elsewhere in Romance, where thematic and desinential material is redistributed according to established patterns of stem distribution (Maiden 2009b; O’Neill 2014). Stem identity acts as one factor among others in the spread of -[e], for which it is neither necessary nor sufficient. The second person plural present subjunctive forms of corrêr ‘run’, plâire ‘please’, and quêure ‘cook’ all maintain -[a:] despite presenting no difference of stem, while the second person plural present subjunctive form of pòdêr ‘be able’ shows extension of -[e] as a variant alongside -[a:] despite a difference of stem, and in the case of savêr ‘know’, the second person plural present subjunctive form [saje] has -[e] although the corresponding second person plural present indicative form [sa:] has -[a:]. The spread of -[e] in the cases of pòdêr ‘be able’ and savêr ‘know’ may be attributable to the high lexical type frequency attained by -[e] in the second person plural present subjunctive, favouring further generalization of -[e]; the retention of -[a:] in corrêr ‘run’, plâire ‘please’ and quêure ‘cook’ is more difficult to motivate. Non-first-conjugation second person singular present subjunctive forms, meanwhile, overwhelmingly retain their historically expected form in -[a:], despite pressure from the corresponding second person singular present indicative forms and second person plural present subjunctive forms, both in -[e].
17.3.7 Summary The single most frequent source of syncretism is regular sound change, which causes previously distinct forms to fall together. The most systematic syncretisms concern person-number combinations: 2=2 (the natural class of secondperson forms), 1=3 and 1=3 (neither of which is a natural class, though
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
378
each pair of forms shares a number value). TAM syncretism is also attested in the Lemosin data examined here (though less prominently), supporting the generalization made by Baerman et al. (2005: 120, 124) that the presence of TAM syncretism entails that of person syncretism. Among person syncretisms, the syncretisms 2=2 and 1=3 are instances of what Baerman et al. (2005: 59) term ‘partial syncretism’, as they occur in most, but not all, TAM categories and conjugational classes, whereas the syncretism 1=3 is a case of ‘complete syncretism’, as it applies to all paradigms.¹⁹ The Lemosin syncretisms uphold the cross-linguistic generalization that complete syncretism occurs either solely in the non-singular (as here) or in both the singular and the non-singular; complete syncretism between first and third person is cross-linguistically rare, but not without precedent (Baerman et al. 2005: 59, 62). The sources identified for the Lemosin syncretism patterns are also consistent with those proposed by Baerman et al. (2005: 71f.): the majority are due to sound changes which cause two or more forms to fall together. The Lemosin data diverge somewhat from the generalizations in Baerman et al. (2005) in the relative prominence and resilience of the syncretism patterns. Baerman et al. suggest a number of explanations for the statistical prominence of patterns arising from feature structure in their sample: such patterns ‘are available to all languages’, ‘can arise spontaneously’ and ‘are self-regenerating in case of disruptions’ (2005: 170), whereas patterns arising from sound change (which often involve functionally unnatural groupings of cells) are ‘languagespecific, and always in competition with morphological patterns based on feature structure’ (2005: 170). In the variety of Gartempe, 2=2 reflects feature structure and is manifestly available, but is relatively unusual among Occitan varieties (see also Hinzelin 2012), and is not reasserted when compromised by the analogical generalization of -[e] in non-first-conjugation second person plural present subjunctive forms;²⁰ in general, as 1=3, 1=3, inherited patterns of stem distribution and most cases of TAM syncretism in Gartempe are also due to regular sound change, morphological patterns based on feature structure are of low lexical and paradigmatic type frequency, and thus unlikely to prevail in competition. These remarks do not, of course, invalidate the proposals of Baerman et al., which remain as observed statistical tendencies.
¹⁹ The only exception given in Quint’s grammar concerns the present indicative of plâire ‘please’: two stem variants are available to the second person singular, second person plural, and first person plural cells, while only one of these stem variants is found for the third person plural cell. ²⁰ One possible interpretation is that although corresponding to feature structure, this particular syncretism is motivated by sound change and thus displays the characteristics of patterns arising from sound change; but such an analysis is unsatisfactory, as it would require native speaker grammars to contain knowledge of the origin of the various patterns, knowledge to which speakers do not have access in acquisition.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
379
17.4 The relationship between patterns of syncretism and metamorphomes 17.4.1 Introduction Both syncretism (of the systematic type discussed here) and metamorphomes (recurrent patterns of paradigmatic distribution) are phenomena in which a given set of distinct paradigm cells consistently share inflexional exponents: in the case of metamorphomes, any type of inflexional exponent may be shared (including but not limited to roots, thematic elements, inflexional desinences, entire wordforms), while, in the case of syncretism, entire inflected wordforms are shared. In the history of Romance languages, there is substantial evidence for the psychological reality of metamorphomes as groupings of mutually predictive cells, since speakers repeatedly exploit established metamorphomes as productive templates for the distribution of novel alternation patterns (Maiden 2018a). For example, the ‘N-pattern’ (Maiden 2009a, 2018a),²¹ which comprises the cells {..1, ..2, ..3, ..3, ..1, ..2, ..3, ..3, .2}, originates via the shift from phonologically predictable stress to lexically specified stress, coupled with segmental allomorphy arising from the differential development of stressed and unstressed vowels in early Romance; on the basis of the observed patterns of alternation, speakers infer a morphological generalization about the distribution of inflected forms. The reality and abstract nature of this generalization are demonstrated by the range of morphological analogies which crucially depend on it: speakers do not merely extend the existing vowel alternation patterns to additional lexemes, but also assign an N-pattern distribution to suppletive roots and thematic elements unrelated to the original phonological alternation (Maiden 2018a: 175–209). In southern varieties of Occitan, such as the variety of Graulhet exemplified in Table 17.9, the N-pattern retains its traditional Romance shape, while in French it has been reduced to {..1, ..2, ..3, .2}. The change of shape undergone by the N-pattern in French is, like its original emergence in Romance, due to the morphologization of alternations resulting from regular sound change; moreover, the sound changes involved in the French development cause systematic syncretism between the remaining N-pattern cells (Esher 2017a). In the Lemosin varieties discussed here, the N-pattern is likewise compromised by the results of regular sound change, since the stress pattern and stem allomorphy of the present indicative second person singular and present indicative third person plural cells systematically differentiate these cells from the remaining N-pattern cells, aligning them instead with cells outside the N-pattern. ²¹ It is desirable to give morphomes abstract labels, in order to refer to them independently of any phonological or functional content which may be associated with them.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
380
Table 17.9. morir [muˈri] ‘die’, Graulhet (Lieutard 2004: 230), showing etymological N-pattern alternation
1 2 3 1 2 3
.
.
.
.
ˈmɔri ˈmɔres ˈmɔr muˈrɛn muˈrɛs ˈmɔru
ˈmɔre ˈmɔres ˈmɔre muˈren muˈres ˈmɔren
muriˈsjɔ muriˈsjɔs muriˈsjɔ muriˈsjan muriˈsjas muriˈsjɔw
muriˈɛri muriˈɛres muriˈɛt muriˈɛren muriˈɛres muriˈɛru
muriˈɣɛsi muriˈɣɛsɔs muriˈɣɛsɔ muriˈɣɛsen muriˈɣɛses muriˈɣɛsu
muriˈrɛj muriˈras muriˈra muriˈren muriˈres muriˈrɔw
muriˈrjɔ muriˈrjɔs muriˈrjɔ muriˈrjan muriˈrjas muriˈrjɔw
For northern Gallo-Romance (including Lemosin) varieties, Hinzelin (2011a,b) considers syncretism ‘a major paradigm-structuring principle’ (2011b: 297) on a par with, or even capable of overriding, metamorphomic patterns such as the N-pattern. Hinzelin makes a number of theoretical conjectures about the origin and status of the syncretism patterns, three of which are examined below as relevant to their relationship with metamorphomic phenomena: namely, that patterns of syncretism can act as templates for the distribution of allomorphy; that the patterns may be due to sound change; and that metamorphomes which align with TAM category distinctions are more resilient in cases of person syncretism than metamorphomes which do not.
17.4.2 Syncretism patterns as templates A robust source of evidence for the productivity of local syncretism patterns is offered by the redistribution of existing suppletive roots (reflexes of , and ) in the verb ‘go’, a process which cannot be attributed to phonological motivations. In Occitan varieties, the v-stem continuing typically has an N-pattern distribution, occurring in the singular and third person plural forms of the present indicative and in the second person singular imperative, while other present indicative and imperative forms continue .²² However, in the variety of Gartempe, the v-stem has spread to vam [vã] ‘go...1’, vatz [va:] ‘go...2’ (Quint 1996: 125; Hinzelin 2011a: 725), replicating the firstconjugation syncretism patterns 2=2 and 1=3. The same development is attested in the Lemosin data given by Ruben (1866), and is analysed by Hinzelin (2011b: 298) as ‘take-over’ (in the sense of Carstairs[-McCarthy] 1987) of the second person singular and third person plural forms by the second person plural ²² For absence of the v-stem from the present subjunctive, see Maiden (2018a: 199f.).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
381
and first person plural respectively; the direction can be identified since the spread of suppletive forms extends reflexes of rather than reflexes of . The direction of change in these cases of suppletion is the exact opposite of that observed for the origin of the syncretism patterns. The patterns arise due to second person singular and third person plural forms adopting stem alternants and stress patterns previously characteristic of first person plural and second person plural forms; whereas, in the cited cases of suppletion, the first person plural and second person plural forms are remodelled on the second person singular and third person plural forms. This contrast indicates that, once the syncretism is established, its original directionality becomes opaque to speakers (as the forms are now, by definition, identical, neither is synchronically identifiable as the source of the other): in the Lemosin case, the data are compatible with analysis as a nondirectional syncretism, or with reversal of the historical directionality. In the varieties discussed by Hinzelin, the analogical extension of the v-stem through the present indicative compromises the pre-existing syncretism patterns ..1=.1 and ..2=.2 found elsewhere in the verb system: for example, in the data provided by Ruben (1866), nan ‘go..1’, nâ ‘go..2’ retain reflexes of (Hinzelin 2011a: 725), a distribution also attested by Chabaneau (1876: 236) and Reydy (2008). In the variety of Gartempe, this clash of innovative and inherited patterns is definitively resolved in favour of the novel pattern: a further suppletive form marchetz ‘go..2’ is co-opted from marchar ‘walk’ (Quint 1996: 125). By contrast, in other varieties, the original syncretism patterns ..1=.1 and ..2=.2 survive or are reasserted: for the first person plural imperative and second person plural imperative, Lavalade (1987: 53) includes vam, vatz among regional variants, and Benoît (1932: 96) for the Périgord region gives only vam, vàs. Hinzelin’s interpretation of the suppletive patterns is that syncretism constrains suppletion (2011b: 305) and that ‘the new syncretic stem distribution is again morphomic in nature and thus a symptom of autonomous morphology’ (2011b: 310).²³ These conclusions are entirely consistent with the findings of the present study.
17.4.3 The source of syncretism patterns Not only do the syncretism patterns serve as templates for the (re)distribution of alternation patterns, just as metamorphomes are documented to do, but the ²³ The claim that syncretism is ‘a symptom of an autonomous morphological component’ is restated by Hinzelin (2012: 70), who also describes syncretism patterns as ‘an expression of an autonomous morphological component in the mental grammar’ (2012: 77). In the absence of an explicit statement on the conceptual relation between syncretism patterns and the metamorphomic patterns identified by Maiden, the impression given is that Hinzelin views syncretism and metamorphomes as two distinct phenomena.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
382
syncretism patterns arise in the same way as metamorphomes elsewhere in Romance. Hinzelin suggests that ‘sound change may contribute to or perhaps even trigger the creation of syncretism patterns’ (2011b: 309), a conjecture supported by this study’s demonstration that the majority of syncretism patterns present in Lemosin varieties result from sound changes causing extensive homonymy between inflexional forms. As in the case of the patterns described by Maiden (e.g. 2009a, 2016a, 2018a) and Esher (2015a, 2017a), sound change produces a novel distribution which is consistent and recurrent across the lexicon, and forms the basis for a novel generalization about the paradigmatic distribution of morphological formatives. The interest of the Lemosin data is that they illustrate both split and merger in distributions of cells: within the domain of the N-pattern, the second person singular and third person plural cells are differentiated from the others, just as the N-pattern cells were differentiated from the other infectum cells; but these same cells merge with another existing distribution outside the N-pattern. As in modern French, the change promotes identity between entire wordforms rather than solely inflexional formatives, but the fundamental mechanism is the same: individual paradigm cells are reassigned from one existing metamorphomic pattern to another, resulting in a change of shape for each existing pattern (a process termed ‘transfiguration’ by Esher 2017a).
17.4.4 Interaction between syncretism patterns and metamorphomes On the basis of the Gallo-Romance data in his study, Hinzelin (2011b) makes a more general theoretical claim about the resilience of metamorphomes in diachrony: There is another tendency to dispense with suppletion inside partial paradigms but to maintain different stems across them, e.g. for , , , . [ . . . ] The evidence suggests that patterns following category lines like conditional and future, imperfect and (plural) imperative are more likely to survive abundant syncretism than more idiosyncratic distributions. (2011b: 310)
This claim is surprising in the wider Romance context, since metamorphomic patterns which follow TAM category lines, labelled ‘TAM morphomes’ by Smith (2013), are generally no more resilient than metamorphomic patterns which do not, labelled ‘person-number morphomes’ by Smith (2013). In Italo-Romance, for example, the etymological perfectum stem, originally present throughout the preterite and imperfect subjunctive, is retained only in the first person singular, third person singular and third person plural preterite forms (Maiden 2000,
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
383
2018b; also first person plural preterite in some varieties), while in some Occitan varieties, the formal identity between future and conditional forms, which do share some semantic values, breaks down (Esher 2012, 2015a): in all these varieties, the N-pattern remains systematically and robustly intact. On closer examination, the disparity between Hinzelin’s findings in relation to syncretism and the general behaviour of Romance metamorphomes turns out to be an artefact of Hinzelin’s data set, in which all cases of syncretism considered involve person syncretism within a given TAM category. By definition, person syncretism within a ‘TAM morphome’ cannot compromise the integrity of that metamorphome, since the distributional pattern is defined independently of person features: the syncretisms .1=.3, .2=.2 and .1=.3 have no bearing on stem distribution within the set of future and conditional forms overall. The case of the N-pattern is particularly instructive. As a metamorphome crucially defined with reference to person features, the N-pattern is potentially compromised by some person syncretisms—but not all. In the present indicative and present subjunctive, the syncretism 1=3 has precisely no effect on the coherence of the N-pattern: as Tables 17.2 and 17.3 show, the distinctive alternant historically characteristic of the N-pattern is retained in first person singular and third person singular forms. It is the syncretisms 2=2 and 1=3 which compromise the N-pattern: in both these cases, the domain of the syncretism overlaps with the domain of the N-pattern, and the two templates directly conflict. These data indicate a different conclusion: whether an inherited metamorphome is at risk from novel syncretism depends not on the type of metamorphome per se, but on whether or not the potential domains of the existing and novel metamorphomic patterns clash.²⁴ This conclusion, which increases the generality and accuracy of predictions about the susceptibility of metamorphomes to change, is entirely consistent with developments observed elsewhere in Romance, such as the case of Italo-Romance preterite forms, where the novel metamorphomic template corresponds to the intersection of two existing paradigmatic distributions (Maiden 2000, 2018b; Esher 2015b).
17.5 Conclusions This study provides an overview of the patterns of syncretism encountered in Lemosin varieties of Occitan, with a focus on the variety of Gartempe described by Quint (1996), which exhibits a particularly high incidence of syncretism. Person syncretism is shown to be a systematic feature of Lemosin conjugation, and TAM ²⁴ The term ‘clash’ referring to partial overlap of morphomic patterns is introduced by Maiden (2009a: 64); see also Maiden (2018a: 288f.) for discussion.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
384
syncretism is also found, with the patterns attested supporting the cross-linguistic generalizations of Baerman et al. (2005). The study demonstrates that almost all cases of syncretism in the variety of Gartempe can be traced to regular sound change: a number of such changes in this variety create homophony between inflexional forms, from which speakers can deduce a morphological generalization that the forms realizing a given pair of cells are identical. This mechanism is almost identical to that by which several prominent metamorphomic patterns have arisen in Romance, and it is proposed here that the behaviour of the syncretism patterns is best captured by considering them to be metamorphomic. Further evidence supporting this view is offered by the fact that, like established Romance metamorphomes (for which see Maiden 2018a), the Lemosin patterns can be exploited as templates for analogical redistribution of forms, as highlighted by Hinzelin (2011a,b, 2012). As the rise of the syncretism patterns observed in Lemosin affects stem alternation patterns and stress alternation patterns as well as desinences, it can involve change to the classic Romance metamorphomes identified by Maiden (2009a, 2016a, 2018a). Such change is a further example of ordinary change to metamorphomes, in which paradigm cells are reassigned from one metamorphomic template to another; it does not represent a conflict between qualitatively different phenomena. Change to metamorphomic templates is predicted to occur where there is overlap or clash between templates (e.g. a given cell patterns with one metamorphome in some lexemes or with respect to a given exponent, but with another metamorphome in other lexemes or with respect to a different exponent). The distinction between metamorphomes defined solely in terms of TAM categories, and metamorphomes the definition of which additionally requires reference to person and number features, is not found to be a predictor of diachronic resilience or of susceptibility to change.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
18 The verbs ‘rain’ and ‘snow’ in GalloRomance, and other morphological mismatches in diachrony Martin Maiden
18.1 The morphology of ‘rain’ and ‘snow’ in Romance This study will demonstrate, in effect, that the French word for ‘(to) rain’ is pleuvoir. Now, this less-than-earth-shattering revelation—perfectly well known not only to every native speaker of French but to any schoolchild beginning French as a foreign language—may scarcely seem to merit a place in a volume of studies in Gallo-Romance linguistics dedicated to a distinguished expert in the field. Nonetheless we shall see not only that it is, at least to a morphologist, not half as trivial as it sounds, but that it may be only one-eighth of the truth. It should be said immediately that most of the analysis will focus on Francoprovençal rather than French, but the implications hold no less for the latter. We need to begin, however, by looking at the diachronic morphology not only of ‘rain’, but also of ‘snow’, among the Romance languages. The meaning ‘to rain’ is expressed across the Romance languages by continuants of the Latin third conjugation verb (3.. ): e.g. Pt. chover (chove), Sp. llover (llueve), Cat., Oc. ploure (plou), Fr. pleuvoir ((il) pleut), Srs. plóver, Log. plóere, It. piòvere (piove), Vgl. pluf (3..), Ro. ploua (plouă). With the exceptions of French, Romanian,¹ and Istro-Romanian, this verb generally retains third conjugation morphology (chiefly distinguished by the presence of an unstressed, rather than a stressed, infinitive ending), or by the local continuants of third conjugation morphology (in Ibero-Romance the third conjugation systematically merges into the second). It is not clear why the French verb for ‘rain’ (and its cognate in some other Gallo-Romance varieties, as we shall see)
¹ For the particular combination of phonological and morphological circumstances which determined exceptional reassignment of this verb to the first conjugation in Romanian, see, e.g. Maiden et al. (forthcoming, §6.2). In Istro-Romanian it is assigned to the default fourth conjugation. Variation and Change in Gallo-Romance Grammar. First edition. Sam Wolfe and Martin Maiden (eds). This chapter © Martin Maiden 2020. First published in 2020 by Oxford University Press.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
386
has passed to the second conjugation,² characterized by infinitives in -oir. A number of French verbs in -oir happen to have root-final -v-, such as avoir ‘have’, pouvoir ‘be able’, mouvoir ‘move’, devoir ‘must’, savoir ‘know’, OFr. estovoir ‘be necessary’—and perhaps this phonological similarity exerted some influence on pleuvoir (see further §18.4). Note also verbs in -cevoir (e.g. recevoir ‘receive’), whose counterparts in Latin and most Romance languages have third conjugation morphology (e.g. > Cat. rebre, It. ricèvere). The salient point for the ensuing discussion, however, will be that the verb ‘rain’ is restricted, according to dialect, to one of two inflexion classes, the third or the second; these classes are famously ‘closed’ and ‘unproductive’ across the Romance languages, and comprise inherited, usually quite basic, verbs virtually all of them characterized by notable morphological irregularity. As for the verb ‘snow’, there is far less historical continuity than for the verb ‘rain’. Romanian ninge and the central Italo-Romance nengue do continue Latin third conjugation ‘snow’, but in other Romance languages this meaning is expressed by neologistic verbs historically derived mainly from the reflexes of the Latin noun (accusative of ) ‘snow’. The Romance reflex of is generally *ˈneve and occasionally, *ˈnɛve (see, e.g. Maiden 2017: 20), and the relation between this noun and the verb derived from it remains fairly transparent, as may be seen from Table 18.1. In many Italo-Romance dialects (of northwestern, and also of parts of central and southern, Italy; see AIS map 377), the verb ‘snow’ is derived instead from the noun meaning ‘snowflake’ (cf. It. fiocco < ): e.g. Milanese fjoˈka ‘to snow’. Table 18.1. Verb ‘snow’ as derived from noun ‘snow’ in Romance
Portuguese Spanish Catalan Occitan Venetian (AIS point 376) Friulian (Udine AIS point 339) Emilian (Tizzano AIS point 433) Italian Sicilian (Baucina AIS point 824) Sardinian (Villacidro AIS point 973)
‘snow’ (noun)
‘snow’ (infinitive)
neve nieve neu nèu ˈneve nɛf ˈneːva neve ˈnivi ni
nevar nevar nevar nevar neveˈga neveˈa nˈvaːr nevicare niviˈkari niˈai
² I depart in this study from the designations of conjugation classes usual in the French grammatical tradition (where, for example, verbs in -ir are described as belonging to the ‘second’ conjugation). To ensure comparability across Romance varieties and with Latin, I number Romance conjugation classes according to the system traditional for their Latin antecedents. Thus, since -oir is the continuant of Latin second conjugation -̅, I number such verbs as second conjugation in French as well.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
‘’ ‘’ -
387
In some of the verbs given in Table 18.1, a derivational affix *-ik- is also involved, yielding originally *nevi’kare,³ whence the Italo-Romance forms neveˈga, nevicare, niviˈkari, and also, ultimately, French neiger.⁴ Crucially, what all these neologisms have in common, whatever their structure and origin, is that the verb derivationally created from the noun ‘snow’ (or ‘snowflake’) is assigned to the first conjugation, that characterized historically by the thematic vowel [a] following the lexical root. Now, that verbs created from nouns should be assigned to the first conjugation is wholly unsurprising: across the Romance languages, with the exception of Romanian,⁵ the first conjugation is the default inflexional class in the verb to which neologisms of whatever kind (see, e.g. Nyrop 1960: 46–9 for French), including loanwords, are assigned.⁶ While there are occasional cases of assignment of new verbs to the fourth conjugation (cf., e.g. Nyrop 1960: 49–51 for French), what is virtually unknown in Romance languages is the introduction of a borrowed form, or a product of novel word-formation, into the continuants of the Latin second or third conjugations; these latter inflexion classes typically comprise words of high frequency and basic meaning (including, of course, ‘rain’), and are virtually all inherited from Latin. While there is some shifting between second and third conjugation,⁷ and while original fourth conjugation verbs (because of extensive morphological similarities with the third conjugation) are sometimes reassigned to the third conjugation, notably in Italo-Romance,⁸ the second and third conjugations remain, overwhelmingly, hermetically closed to new members. It is against this background that the behaviour of some eastern Gallo-Romance varieties in respect of the verb ‘snow’ stands out as remarkable. As in so many other Romance languages, the verb is derived from the reflex of the Latin noun . Exceptionally, however, the result is not assigned in these varieties to the first conjugation (or even to the weakly productive fourth), but to the second or to the third. Whether the verb ‘snow’ shows second or third conjugation morphology is in turn strictly dependent on whether, in the relevant dialects, the verb ‘rain’ has second
³ Friulian neveˈa involves a different derivational affix, historically *-edj-. ⁴ The old French noun is noif ‘snow’, from ; the modern noun neige is a back-formation from the verb neiger. ⁵ This is also true of modern Romanian, although historically, and apparently to this day in IstroRomanian, the fourth conjugation, with thematic [i], is the default. See Iordan (1935: 50–64); Pușcariu (1926). ⁶ For some possible reasons, see, e.g. Maiden (2018: 277–83). See also Schorta (1938: 132) and Maiden (2018: 278) for a systematic transfer of first conjugation infinitives to the third conjugation in the Romansh of Val Müstair. ⁷ See, e.g. Lausberg (1966: 258–61); Maiden (2011a: 204f.). ⁸ This comes about because some fourth conjugation verbs show a high degree of inflexional identity with the third conjugation: see, for example, Maiden (2003: 17–19).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
388
or third conjugation morphology. In fact, the desinential morphology of the verb ‘snow’ in these cases seems to mimic that of the semantically similar verb ‘rain’.⁹
18.2 How and why ‘snow’ follows ‘rain’ in Gallo-Romance Table 18.2 reproduces data from AIS maps 366 ‘piovere’ and ‘piove?’ (‘to rain’, ‘is it raining?’), 367 ‘piovuto’ (‘rained’ past participle), 377 ‘nevicare’, ‘nevica’, and ‘nevicato’ (‘to snow’, ‘it’s snowing’, ‘snowed’ (past participle)), 1390 ‘cavare’ (‘to dig’), 1682 ‘lava’ (‘s/he washes’), and 397 ‘tonato’ (‘thundered’, past participle) for the Francoprovençal dialects of Aosta (points 121 Rhêmes-Saint-Georges, 122 Saint Marcel, 123 Brusson, 131 Noasca, 132 Ronco Canavese, 143 Ala di Stura). The forms for ‘cavare’, ‘lava’, and ‘tonato’, show the typical morphology of a first conjugation verb: it is clear that, in contrast, the verb ‘snow’ in these examples, just like ‘rain’, generally shows distinctively non-first conjugation morphology: The distinctive characteristics of non-first (and in these examples, specifically third) conjugation morphology as illustrated in Table 18.2 are: Table 18.2. Non-first conjugation morphology of the Francoprovençal verb ‘snow’ (AIS) ‘rain’ and ‘snow’
121 122 123 131 132 143
‘rain’ ˈpluwrə ˈpluwwɛ ˈpjovre ˈpjovri ˈpjeːvre ˈpjɔwri
‘snow’ ˈnɛjvrə ˈnɛjvre ˈnɛjvre ˈnɛvri ˈnɛvre
first conjugation verbs 3. . ‘rain’ ‘snow’ ˈplukti nɛ ˈplutwej nɛj ˈpjoti nɛj ˈpjoːti nɛːt ˈpiːat nɛjt pjɔwt nɛt
. ‘rain’ ‘snow’ pluˈyj nɛˈyj pluˈvɛt nɛˈvy pjoˈy nɛˈy pjuˈvoj nuˈoj pjeˈvy neˈy pjuˈvy nyˈvy
‘dig’ ʦe: ˈe
3. . ‘wash’ ˈlaːa ˈlaːve ʧaˈvi: ˈllaːve gaˈtaːr ˈlaːvat gaˈvaːr ˈlaːvat gaˈvaːlas ˈlaːvat
. ‘thunder’ tũˈũ trũˈũ trɔˈna: truˈna: troˈna truˈna:
⁹ It is sometimes argued that the appearance of the verb ‘snow’ in the third conjugation is not a Romance development, as I have just suggested, but an inheritance from Latin. Wartburg (FEW, s.v. *nĭvĭcare) derives it from an alleged Latin third conjugation etymon *nivere (as does Meyer-Lübke 1895: 171); the Oxford Latin Dictionary, s.v. nīuit (to be corrected to niuit, since the line which this example commences is an iambic senarius) also postulates *niuere. The sole basis for this postulation, however, is one occurrence of a third person singular present indicative form niuit in a historical play by Pacuvius (who lived in the late third to early second century): Niuit sagittis, plumbo et saxis grandinat ‘It is snowing with arrows and hailing with lead and rocks’. This third conjugation present indicative form niuit is entirely isolated in the historical record of Latin, and there appear to be no other contemporary or subsequent attestations of this verb. It may also be relevant that Pacuvius apparently had a penchant for inventing neologisms (see, e.g. Boyle 2006: 92). In short, there is no reason to believe that a putative Latin verb niuere is the ancestor of the Romance forms at issue in this study (see also Gardette 1983: 549 n.2), and we may safely assume that the phenomena we consider here really are Romance innovations. I am very grateful to Wolfgang De Melo for his advice on the status of Pacuvius’ niuit, and it is to him that I owe most of the foregoing observations.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
‘’ ‘’ -
389
(i) infinitives with the unstressed ending -re (-rə, etc.), as opposed to the stressed infinitive desinence of the infinitives of the first, second, and fourth conjugations; (ii) third person singular present indicative forms without inflexional vocalic ending, resulting in monosyllabic forms (as opposed to typically disyllabic first conjugation forrms with the ending -e);¹⁰ (iii) past participles historically derivable from forms ending in *-ˈutu (which usually yields -ˈy) (rather than first conjugation endings derived from *-ˈatu) From the Atlas linguistique de la France, it is possible to see similar patterns in the wider Francoprovençal domain, although the range of morphologically relevant forms available in that atlas is smaller than that offered by the AIS, being limited to the third person singular present indicative. These patterns can be seen (Table 18.3) from ALF maps 904 ‘il neige’, 1035 ‘il pleut’, as contrasted with examples of first conjugation morphology maps 632 ‘il gèle’ and 1340 ‘tu trouves’.¹¹ Some examples are given in Table 18.3:¹² Table 18.3. Non-first conjugation morphology of the verb ‘snow’ (and ‘rain’) in Francoprovençal first conjugation 61 Estavayer (Fribourg) 70 Gruyeres (Fribourg) 935 Surjoux (Ain) 947 Bons-en-Chablais (H.-Savoie) 957 Le Biot (H.-Savoie) 964 Bozel (Savoie) 967 Chamonix-Mont-Blanc (H.-Savoie) 969 L’Etivaz (Vaud) 973 Lanslebourg (Savoie) 976 Bourg-Saint-Pierre (Valais) 982 Maïsette / Faët (Pignerol) 985 Champorcher (Aosta) 987 Ayas (Aosta)
‘it rains’ pja: pʎa:o pʎu plu plu: ple plu psao plot plø plou pjøt pjo
‘it snows’ na: ne: ne ne ne: ne ne na:e net ne: neu nejt nej
‘it freezes’ ʣaːle ʣaːle ðeːle ʣaːle ðaːle ʣeːle ðaːle ʣaːle vaːle ʣaːle ʤjalo ʣeːle ʤjala
‘you find’ trovo trove truve truve truve truve truve trove trove trøve trube truve truve
¹⁰ The final vocalic element found in some of the forms given here for ‘rain’, such as ˈpjoti, is a clitic pronoun. The preservation of historically underlying final -t in some of the ‘rain’ examples is also connected with the presence of the following clitic, in the interrogative forms: cf. Fr. il pleut (il plø) but interrogative pleut-il? (pløtil). ¹¹ The ALF does not provide a suitable example of a third person singular first conjugation verb in root-final -v, so I give here a second person singular form, historically in *-as. ¹² See also ALF points 40, 62, 924, 926, 927, 933, 936, 937, 938, 939, 946, 954, 956, 958, 959, 965, 966, 968, 977, 986.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
390
Further examples of distinctively non-first conjugation morphology in the Francoprovençal verb ‘snow’ may be found in Illiez (Fankhauser 1911: 160), Genevan (Keller 1928: 156), Sottens (Jaquenod 1931: 83), Vaux (Duraffour 1932: 77), Ruffieu (Ahlborn 1946: 45). Most strikingly, the infinitive of this verb in these dialects ends in -a or -ai (e.g. Vaux nevai, or Illiez nevaː, Ruffieu nevai),¹³ the characteristic local ending of the second conjugation, which is the ending also shown by the verb ‘rain’ (e.g. Vaux pjovai, Illiez pθøvaː; Ruffieu pʎovai; cf. Fr. pleuvoir). That is to say that the verb ‘snow’ follows the verb ‘rain’ not only in the general respect of being a non-first conjugation verb, but in the specific respect of showing second conjugation morphology in the infinitive wherever the verb ‘rain’ also does so. For further confirmation of this observation see also Jaberg (1906: 68 n.7). While I focus here on the manifestation of the phenomenon of third (or second) conjugation membership of the verb ‘snow’ in Gallo-Romance, it must be observed that the same phenomenon is also observable in Ladin (see ALDI maps 530 and 607, and AIS maps 366, 367, 377 for points 314 San Vigilio, 314 Colfosco in Badia, 315 Arabba, 312 Selva in Gardena, and 313 Penia), while in most Romansh varieties (see, e.g. AIS map 377) the infinitive of ‘snow’ is rhizotonic, like that of ‘rain’. ‘Snow’ also follows the morphology of ‘rain’, rather than that of the first conjugation, in a remote and isolated outcrop in the Val d’Antrona (see AIS map 377, for point 115 Antronapiana; also Nicolet 1929: 159, 166). Now these areas lie on the periphery of, and are separated from each other by, a large area of northern Italy (Lombardy, Piedmont) in which ‘snow’ is expressed by what is presumably a recent first conjugation neologism derived metonymically from the word for ‘snowflake’ (see §18.1 above). This geographical distribution invites two different kinds of interpretation: one is that our phenomenon is polygenetic (i.e. it has occurred several times independently, in different Romance languages), and the other is that what we observe today in Francoprovençal, Antrona, Romansh, and Ladin are actually the peripheral remnants of a now largely submerged area, comprising many of the northern Italian varieties where, today, the type derived from the word for ‘snowflake’ prevails. I incline to the latter view, but only more detailed examination of the Ladin and Italo-Romance data could establish this. Why does ‘snow’ morphologically follow ‘rain’? In some respects, the answer seems obvious. This is surely a case of what Paul ([1890] 1970: 161) called ‘contamination’, such that the exponent of one lexical meaning undergoes partial
¹³ Keller (1928: 156) gives the infinitive of ‘rain’ in the dialect of Hermance as third conjugation, but that of ‘snow’ as (apparently) second, while for the dialect of Certoux the infinitive of ‘rain’ is given with both third and second conjugation endings. For further illustrations of the correlation of conjugation-class between ‘rain’ and ‘snow’ in Francoprovençal see Keller (1928: 21).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
‘’ ‘’ -
391
formal modification under the influence of the exponent of some other, closely associated, meaning. A typical example is the development of Latin ‘fat’ under the influence of the semantically similar *’grossu ‘large’, yielding, e.g. Fr. gras ‘fat’. Similarly, Fr. framboise ‘raspberry’ reflects the influence on original bramboise (< old Frankish *brambasi) of the reflex of (the source of Fr. fraise) ‘strawberry’. In the verb, the dialectal cognates of Fr. pouvoir ‘be able’, are contaminated by cognates of Fr. vouloir ‘want’, in such a way that a [l] appears as the root-final consonant, producing the type pouloir. ALF maps 1444 / 1081 (‘vouloir et pouvoir sont deux choses’), 1083 (‘je ne pouvais ni avancer ni reculer’), and 1086 (‘il a pu’), reveal this type to be widespread in Wallon dialects: for example, Dolhain (point 193) volœːr e polœːr, 1.. polœːv, . polu. Further examples may be seen in Ahlborn (1946:45) for Ruffieu-enValromey, or Niederländer (1900: 252) for Namur. This same development is also quite widely attested in northern Italo-Romance (see, for example, Maiden 2018:209 and references therein; also ALF points 991, 899, 897 in the Alpes Maritimes).¹⁴ In our case, the close semantic connexion between ‘raining’ (the commonest form of precipitation) and ‘snowing’ (perhaps the second commonest form) seems to have favoured ‘contamination’ of the lexeme meaning ‘to snow’ by that meaning ‘to rain’. An initial problem with this line of explanation, however, is that it fails to explain why, of all ‘precipitation’ verbs, only ‘snow’ is so affected: examination of ALF map 667 (‘il grêle’) or AIS map 372 (‘grandinare’) for forms of the verb ‘hail’ resolutely show only the expected first conjugation inflexional morphology. One possibility which I will consider again below is that an accidental partial phonological resemblance between these two verbs (they both have roots in final [v]) may favour and facilitate contaminatory effects where there is also some semantic resemblance. What is most striking about case of ‘rain’ and ‘snow’ however, is that contamination is manifest not in the lexical root of these verbs, ostensibly and distinctively the bearer of the relevant lexical meanings, but in their desinential morphology which, one would assume, is independent of lexical meaning. Before addressing the theoretical significance of this development, we need to consider some other such cases in Gallo-Romance, and the light they throw on the role played by phonological similarity.
¹⁴ Gilliéron (1919: 88) proposes that the emergence of the type pouloir is somehow motivated by avoidance of homonymic clash between the second syllable of pouvoir ‘to be able’ and the verb voir ‘see’. But this idea is obviously falsified by the occurrence of the same phenomenon in Italo-Romance varieties, where no such risk of homophony exists (cf. It. potere vs vedere), and indeed by many GalloRomance varieties where the verb ‘be able’ contains no [v]. See also Gauchat (1930: 135). The nature of the association between these two verbs is, admittedly, hard to define but, as Gauchat (1930: 136) notes, there is a pragmatic link (cf. Fr. ‘vouloir c’est pouvoir’ = ‘where there’s a will there’s a way’).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
392
18.3 Other Gallo-Romance cases of lexical contamination manifested in desinential morphology 18.3.1 Osoir, falloir, and estovoir The verb ‘to dare’, apparently from proto-Romance *auˈsare ‘dare’, appears as a first conjugation verb virtually everywhere in Gallo-Romance and beyond (e.g. Fr. oser, It. osare). However, in some northeastern Gallo-Romance varieties (e.g. Wallon, Pas-de-Calais, Somme) we find that this verb has an infinitive with a second conjugation ending: e.g. in the dialect of Mons, osoir, ozwar (cf. FEW s.v. ausare); distinctively second / third conjugation past participles are also attested for this verb (e.g. Neufchâtel oˈzy). Wartburg comments (FEW s.v. ausare, p1044f.) that the meaning of this verb is sometimes ‘weakened’ to become equivalent to that of pouvoir ‘be able, may’, a sense which is particularly well represented in Swiss Romandy, Belgium, and Alsace. This may provide the semantic link between the two verbs which facilitated the morphological attraction of one on the other. The complex etymological connexion between Eng. dare and German dürfen ‘be able, allowed’ (see OED s.v. dare v.¹) is also suggestive in this respect. Wartburg (ibid. 1045 n.11) invokes the more general influence of ‘modal’ verbs in -oir (or of their local equivalents: cf. Fr. devoir, vouloir, pouvoir) to explain the aberrant conjugation assignment. Whether this conjugation shift is explicable in the general respect that oser has become a modal verb (see Meyer-Lübke 1895: 166), or in the more specific respect that it has become a near-synonym of pouvoir, it is clear that the change is attributable to the lexical meaning of the verb, and the apparent locus of that meaning lies in the easily segmentable lexical root os-. A possible case more familiar to historians of French is falloir ‘be necessary’. Its old French antecedent was fourth conjugation faillir ‘be lacking’ whence, in certain contexts, the sense ‘be necessary’ (see FEW s.v. fallĕre₂).¹⁵ Gauchat (1930: 134) says that this shift to second conjugation reflects the verb’s semantic resemblance to devoir ‘must’. FEW suggests that the new infinitive falloir, which emerges alongside original faillir, is modelled on the phonological resemblance between faut and the third person singular vaut (‘it is worth’); to this we may add other resemblances, such as subjunctives faille and vaille, imperfect indicatives fallait, valait. Just as vaut, etc., has infinitive valoir, so faut etc. acquires a new infinitive falloir (and a new past participle fallu = valu).¹⁶ This happens only when the sense is ‘be necessary’, faillir surviving in other senses, and the implication is
¹⁵ How Latin third conjugation emerged as a fourth conjugation -ir verb in French is a separate problem which lies outside the scope of this study. ¹⁶ The difference between l and ll in modern French is in this case purely orthographic and probably reflects etymologizing spelling (cf. Lat. and ).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
‘’ ‘’ -
393
thus that the emergence of falloir is also motivated by the semantic similarity with valoir (in the broadly deontic sense of that which is ‘valid’, ‘appropriate’, ‘the right thing’—FEW also cites occasional cases of use of faut for vaut, for example il faut mieux for il vaut mieux ‘it is better, more correct, more fitting’). In the case of falloir, it is hard to tell whether extensive prior morphological resemblance to valoir is the main determinant, or whether ‘modal’ meaning in a more general way was sufficient to determine the inflexional changes, as Gauchat believed. But in either case is seems that semantic resemblance with other verbs has played some role in bringing about second conjugation falloir. Similar considerations hold for old French estovoir ‘must, be necessary’. In origin (like *neˈvare ‘snow’) it is a Romance neologism, derived from the Latin phrase ‘it is necessary’: for the history of this verb in French and in other Romance varieties, and its complex morphological and semantic interaction with other verbs meaning ‘must’, see e.g. Jud (1946–9); Maiden (2011a,b). It may have been assigned to the usually closed second conjugation purely because it is almost synonymous with reflexes of second conjugation ‘owe, must’.¹⁷ But this development may also have been favoured by the fact that and come to resemble each other phonologically (cf. §18.4), the root-final intervocalic labial in both regularly becoming [v] in the relevant Romance dialects (devoir, estovoir).
18.3.2 The type il trouverra A curious example of contamination is presented by the old French future (and conditional) forms of trouver ‘to find’ (< *troˈpare).¹⁸ The modern future of this verb, e.g. 3.. trouvera [tʁuvәʁa] or [tʁuvʁa], is the historically expected form, and has been the form predominant throughout the history of French and the Oïl dialects (see also ALF map 1341). However, in old French we also find (see, e.g. Nyrop 1960: 158) sporadic examples of the type trouverra, still in use in the seventeenth century (it is criticized, for example, by Vaugelas): e.g. Rabelais ‘Là trouverrez tesmoings vieulx de renom’ ‘There you will find old and renowned witnesses’, Le Miroir d’astrologie naturelle (1711) ‘Il ne sera pas beaucoup porté pour le mariage, mais ses parens lui porteront & lui trouverront un honnête Parti.’ ‘He will not be much inclined to marriage, but his parents will bring to him and ¹⁷ There is a deeper issue here which, while extremely relevant to the theoretical concerns of this chapter, would take us outside the Gallo-Romance domain. In Romance generally, the continuants of the morphologically idiosyncratic Latin modal verbs ‘want’ and ‘be able’ are assigned to the second conjugation (Fr. vouloir, pouvoir; It. volére, potére; Ro. vrea, putea). Third conjugation ‘know’, itself often used modally in the sense ‘know how to, be able’, is also reassigned to the second conjugation (Fr. savoir; It. sapére). Given that modal already belonged to this class (and that second conjugation ‘have’ was also associated with modal uses), was second conjugation membership somehow reanalysed as a general characteristic of modal verbs, attracting other modals? ¹⁸ The etymology of this verb is problematic. See, e.g. Grzega (2003).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
394
will find him an honest party’. FEW s.v. *tropare also cites modern dialect examples from Vendôme and Blois of the type je trouverrai (see also Flutre 1955: 68). Nyrop ascribes these forms of trouver to the influence of the future (and conditional) of the verb voir ‘to see’, e.g. 3.. verra. There is an associative link between ‘finding’ something and ‘seeing’ it, in that one typically ‘finds’ something when one manages to ‘see’ it, but this is not enough to motivate the observed restriction of the apparent influence of voir to the future and conditional tenses. After all, the semantic associative link is equally present throughout the paradigms of both trouver and voir, so why would contamination only operate in the future and conditional? Why, given nous voyons ‘we see’, do we not also find, say, **nous trouvoyons ‘we find’?¹⁹ I suggest that such restriction is additionally motivated by the fact that in the future and conditional trouver presents a portion (namely [vәʁ]) phonologically resembling the future of voir (in [veʁ]) as well as presenting a semantic association with voir. The semantic association has to be invoked, because if the phenomenon were motivated purely by phonological resemblance we would expect such replacement also to operate in verbs such as arriver ‘arrive’ or laver ‘wash’, which (to the best of my knowledge) is never the case.²⁰ It looks as if, in the type trouverra, a partial phonological resemblance to the future form of the verb ‘see’, together with a semantic associative link between them, has determined the emergence of what looks like a novel inflexional ending -err-. Nyrop (1960: 158) also cites the remarkable example of second person plural trouviendrez, where the future of the verb ‘come’ (e.g. 2 viendrez) seems to have intruded into the future form of the verb ‘find’. The semantic connexion between ‘finding’ and ‘coming’ is obvious, and is lexically explicit in some languages, as in Lat. ‘come’, ‘find’.
18.3.3 The type amerrir, alunir I have mentioned that the French fourth conjugation in -ir retained a limited degree of productivity in old French. Smith (2013a: 249) observes that:
¹⁹ Actually, FEW cites a middle French infinitive with a second conjugation ending trouvoir; FEW also cites a dialectal past participle trouvu, invoking the analogical influence of voir (. vu). Note also ‘comme il n’avait pas trouvue Mandane a Babilone quand il la prit’, ‘since he had not found Mandane a in Babylon when he took it’ in a mid-sixteenth century text by Madeleine de Scudéry. In my view, and as is discussed below, this is more likely to reflect the influence of pouvoir ‘be able’ (. pu). ²⁰ The first conjugation verb envoyer ‘to send’ also has a future / conditional in enverr-, rather than expected **envoier-, already attested in Chrétien de Troyes. Nyrop (1960: 157) attributes this development to the influence of the future of voir. If in this case there is no compelling argument for a semantic association with voir, there is a strong phonological association with that verb, in that in old and middle French futures of the type envoira were common, apparently containing the infinitive form of voir.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
‘’ ‘’ -
395
It is technically possible for noun or verb classes to have some semantic or other extramorphological homogeneity. [ . . . ] More often, correlations exist merely as informal and partial generalizations. For instance, in French, intransitive or transitive ‘change-of-state’ verbs derived from an adjective will tend to belong to the regular -ir class [ . . . ]—thus intransitive grandir ‘become big, grow’, transitive agrandir ‘enlarge’ (compare grand ‘big’); intransitive faiblir ‘weaken’, transitive affaiblir (compare faible ‘weak’) [ . . . ], etc. But this conjugation class is no longer productive (the last new forms are, famously, alunir ‘land on the moon’, first attested in 1921, and amerrir ‘land on the sea’, first attested in 1928, both by analogy with existing atterrir ‘land’; the last ‘change-of-state’ verb derived from an adjective appears to have been rosir ‘grow pink’ (compare rose ‘pink’), first attested in 1823 [ . . . ]; even before it ceased to be productive, there were ‘change-of-state’ verbs derived from adjectives which belonged to other conjugations [ . . . ]; and many verbs in the regular -ir class are not ‘changeof-state’ verbs or derived from an adjective [ . . . ].
The fact that a verb derived from the noun terre ‘land, earth’ (< Lat. ()), atterrir, is assigned to the fourth conjugation rather than to the first, is itself surprising (cf. Italian first conjugation atterrare), and there is a thirteenth-century attestation of a first conjugation aterer, atterrir being first attested in the early seventeenth century; indeed, atterrer survived into modern French as a rare synonym of atterrir. However we explain the emergence of atterrir itself, amerrir and alunir emerge centuries later, and it seems uncontroversial that they are modelled on atterrir due to their lexical semantic resemblance with the latter (as also reflected in the perfectly natural, yet in fact paradoxical, English gloss ‘land on the sea’).²¹ In the case of amerrir, the lexical influence of atterrir is also evident in a type of ‘contamination’ on the orthographic rather than the phonological level. Amerrir is based on the noun mer ‘sea’ (< Lat. ), which rhymes in modern French with terre ‘land’ (compare [la mɛʁ] ‘the sea’ and [la tɛʁ] ‘the land’), but there is no phonological or etymological justification for spelling amerrir with rr (indeed, the spelling amérir is sometimes recommended), and orthographic ‘contamination’ from atterrir is obvious. Conceivably, the sometimes proposed alternative spelling allunir could reflect a kind of orthographical contamination from atterrir, involving the presence of an orthographical double letter at the beginning of the lexical root. On the morphological level, however, the influence of atterrir [atɛʁiʁ] on amerrir [amɛʁiʁ] is manifest not in the lexical root, which appears to be the sole exponent of lexical meaning, but in the inflexional ending -ir, contrary to the expected, ‘default’, assignment of neologisms to the first conjugation; this is obviously also the case with alunir [alyniʁ] (based on lune
²¹ For the facts discussed in this paragraph, see TFLi s.v. , , and .
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
396
‘moon’). Yet the ending -ir (and of course the entire array of associated inflexional forms, such as . atterri, 1.. (nous) atterrissons) is the common property of hundreds of French verbs and certainly has no general connexion with the meaning of ‘landing’ on anything.
18.3.4 First conjugation reflexes of third conjugation and third conjugation reflexes of first conjugation *boˈtare The Romance languages widely retain reflexes of Lat. ‘send’, as a third conjugation verb meaning ‘put’: e.g. It. méttere, Fr. mettre. However, some Francoprovençal varieties display first conjugation morphology in this verb (cf. ALF maps 847, 1627; Keller 1928: 20, 26): thus, Certoux mɛtɔu ‘put’—cf. first conjugation ɔmɔu ‘love’ < and third conjugation peːrdr ‘lose’ < . This shift of conjugation class is generally attributed (e.g. Keller 1928: 20) to the influence of a nearly synonymous reflex of a Romance first conjugation verb, of Frankish origin, *boˈtare (cf. It. buttare ‘throw’), also extensively attested in the relevant regions (cf. Jaberg 1906: 88). Elsewhere in GalloRomance (see Meyer-Lübke 1895: 167) the conjugational attraction between these two verbs has operated in the opposite direction, the reflex of *boˈtare moving into the third conjugation under the influence of the reflex of , to yield the type bütre in Lorrain, Picard, Norman, Poitevin, and elsewhere.
18.4 The role of phonological resemblance There is indeed some evidence from Gallo-Romance that rhyme between the lexical roots of different Romance verb lexemes may be sufficient to determine shifts of inflexion class. If the reflexes of Latin second conjugation ‘hold’ come to join the inflexion class of the reflexes of Latin ‘come’ in GalloRomance (cf. Fr. tenir and venir) this is presumably not because of any semantic resemblance, but solely because the root-forms of the inflexional paradigms of one verb are already identical to the corresponding word-forms of the other verb, save for the difference in their initial consonant (e.g. modern French 3.. tient vient; 3.. tienne vienne; 3.. tenait venait; 3. tiendra viendra; 3. tint vint; tenant venant).²² Similarly, the fact that French verbs in -cevoir (e.g. recevoir ‘receive’, percevoir ‘perceive’, décevoir ‘deceive’) have second
²² The historically underlying situation is well represented by Italian, where all word-forms of these two verbs rhyme except for those showing the distinctive thematic vowel: e.g. 3.. tiene viene; 3.. tenga venga; 3. terrà verrà; 3. tenne venne; tenendo venendo, but still tenere venire; 3.. teneva, veniva.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
‘’ ‘’ -
397
conjugation infinitives despite their Latin third conjugation origins (, , ; cf. It. ricévere, etc., Ro. percepe, etc.) reflects the fact that all of these verbs’ word-forms rhyme with those of devoir ‘must’ (e.g. Fr. 3.. doit reçoit; 3.. doive reçoive; 3.. devait recevait; 3. devra recevra; 3. dut reçut; devant recevant), and the second conjugation membership is probably not attributable to semantic association.²³ It is clear that simply sharing a root-final consonant with devoir is not sufficient to determine a shift of conjugation class: there seem to be no instances of Fr. vivre ‘to live’ or suivre ‘to follow’ becoming **vivoir or **suivoir, for example. In Pas-de-Calais and Somme, the equivalent of Fr. trouver ‘find’ (< *troˈpare) displays an infinitive truvwar or truvwer with a distinctively second conjugation ending corresponding to that of the local equivalent of pouvoir (compare ALF maps 1339 and 1081).²⁴ The most likely explanation is that the close phonological resemblance between the roots of these two verbs has determined such reanalysis. Nyrop (1960:59) mentions, alongside trouvoir ‘find’, also esprouvoir ‘experience’ and resprouvoir ‘reprove’.²⁵ Indeed, pouvoir has exercised a more complex morphological effect on verbs with which it shares a rhyming root. The inherited first person singular present indicative, and present subjunctive, root of pouvoir is [pɥi(s)] (e.g. 1.. puis, 1.. puisse, etc.). This is a type of root allomorphy historically unique to this verb, yet the rhyme of the root allomorph pouv- (or pov-) with that of old French first conjugation trouver ‘find’, prouver ‘prove’, rover ‘ask’, and also second conjugation estovoir ‘be necessary’ (whose etymology I discussed earlier) yielded old French analogical present subjunctive and first person singular forms such as 1.. truis, pruis, ruis 1/3.. truisse, pruisse, ruisse, estuisse (see, e.g. Pope 1934: 35; Jud 1946–9: 32, 35f.; Nyrop 1960: 91, 111; Fouché 1967: 93f., 113; Andrieux and Baumgartner 1983: 101).²⁶ The most interesting detail, from our present point of view, is that a phonological resemblance in respect of the lexical root not only favours the creation of novel allomorphy in the root of some first conjugation verbs, but also leads to a shift in inflexion class in the relevant forms, for subjunctive forms such
²³ There may be an association between devoir in the sense ‘owe’ and recevoir ‘receive’, in that the normal resolution of a debt is receipt of what is due. That this association is relevant to the observed morphological change could only be plausibly claimed, however, if it could be shown that the novel second conjugation forms emerged first in the verb ‘receive’. This remains to be investigated. ²⁴ But not to that of voir ‘see’ (cf ALF map 1408): this detail is significant, because FEW attributes some old French attestations of trouvoir to the influence of voir. ²⁵ However, the form sauvoir also cited by Nyrop is not an infinitive of the verb sauver ‘save’, but a noun meaning ‘reservoir for fish’; cf. FEW s.v., salvare. ²⁶ For example: ‘Par ce le sai, par ce le pruis / Que an vos dras et es suens truis / Le sanc qui cheï de ses plaies’, lit. ‘by this I know it, by this I prove it, that in your sheets and in his I find the blood which fell from his wounds’, and ‘[ . . . ] Que nule espie ne vos truisse. / -Dame, fet il, la ou je puisse / Ne me verra ja nule espie’, lit. ‘[ . . . ] that no spy should find you. “Lady”, says he, “wherever I can, no spy will ever see me.” ’ (Chrétien de Troyes, Le chevalier de la charrette (Lancelot)). These examples occur, it is true, in rhyme, but they are by no means limited to that context.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
398
as truisse bear the final -e distinctive in old French of non-first conjugation verbs (the corresponding forms of the first conjugation originally lacked a vocalic inflexion: see Pope 1952: 342–4), including pouvoir. It is clear that lexical root and inflexional desinence cannot easily be separated or ‘segmented’ in diachronic change. In a general sense, it seems that ‘resemblance’ in respect of that portion of a word-form identifiable as its lexical ‘root’, can determine changes in lexical class. The resemblance may be just semantic or both semantic and phonological, and the changes in lexical class can be manifested in inflexional endings which otherwise have no inherent connexion with the lexical root. Phonological resemblance between lexical roots certainly plays a role, but where the phonological resemblance is slight or even non-existent (as, for example, in osoir), semantic resemblance—whose sole apparent exponent is the lexical root—can also produce a morphological shift, manifest not in the root but in inflexion class as reflected in the desinences. Thus, for example, proto-Romance *ˈplɔvere would have been readily analysable as comprising a lexical root *plɔv meaning ‘rain’ followed by a conjugation-class marker -e- and an infinitive desinence -re, endings shared with dozens of other verbs (*ˈvend-ere ‘sell’, *ˈmett-ere ‘put’, and so forth). If the verb ‘rain’ has influenced the inflexional behaviour of a novel verb meaning ‘snow’ and having the root *nev-, the locus of the influence surely ought to be that root, not the ending? Might we not expect to see, for example, the emergence of a form such as **noˈvare ‘snow’, where the vowel of the root has been contaminated by the root vowel of the verb ‘rain’? Actually, in the Gallo-Romance domain I can see no case in which the phonological content of the root of the verb ‘rain’ has modified the phonological identity of the lexical root of the verb ‘snow’, and the effect of the former on the latter is always to be found in the ending.
18.5 The Gallo-Romance verbs ‘snow’ and ‘rain’ and their incompatibility with ‘constructivist’ models of morphology Gallo-Romance varieties (and especially their verbs) are of a largely ‘fusional’ morphological type. A characteristic of this type is the ‘ineffability’ of lexemes (cf. Bonami et al. 2018: vi): it is simply not possible in French, for example, to utter or write in isolation a form which bears purely the lexical meaning of some verb in abstraction from any of the grammatical meanings which the dozens of wordforms of the associated inflexional paradigms also bear. In the lexicographical tradition of Romance languages which have infinitives, it is the infinitive that is specified as reference-form for the relevant lexeme, and educated speakers of those languages will generally offer the infinitive as the citation-form of the verb: ‘the French for “find” is trouver [trʁuve]’, never ‘the French for “find” is trouv- [trʁuv]’. In reality, this convention of selecting infinitives as citation forms may owe more
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
‘’ ‘’ -
399
to the usefulness of infinitives as predictors of other inflected forms in the paradigm (their ‘low conditional entropy’—cf. Ackerman and Malouf 2013), and to their relatively simple morphosyntactic descriptions (they encode no information about person, number, mood, or tense), than to any inherent status on their part as fundamental or default exponents of the lexical meaning of the verb. In Romance varieties that have lost the morphological infinitive from the inflexional paradigm, such as Aromanian,²⁷ the lexicographical reference form is conventionally the first person singular present indicative (as it also was for Latin, despite the existence of infinitives in that language). There is even evidence that formally imperative word-forms may sometimes serve as minimal exponents of lexical meaning in abstraction from their grammatical meaning: see, e.g. Maiden (2008: 380–2), where entities that are precisely identifiable as morphological imperatives are shown to be deployed as the bearers of lexical meaning in compound-formation and elsewhere. The point is that whatever part of the paradigm one chooses as a basic or citation form, the expression of lexical meaning cannot be extricated from concomitant specification of morphosyntactic information, and the minimal utterance of that meaning will always be a complete word-form of the inflexional paradigm of the relevant lexeme. An important implication of this observation is that there is, for example, in modern spoken (and in this case also written) French no one word for ‘rain’, but actually at least eight:²⁸ pleuvoir, pleuvant, . plu, 3.. pleut, 3. pleuvra, 3. pleuvrait, 3.. pleuvait, 3.. pleuve. To come back to our Francoprovençal examples, if the verb meaning ‘rain’ contaminates the verb meaning ‘snowing’ by virtue of their semantic similarity, but does so in respect of what is identifiable as an inflexional ending, not a lexical root, it must be because the entire word-forms, say ˈpjovre, . pjoˈy, mean ‘rain’, thereby yielding ˈnɛjvre, or . nɛˈy as expressions of ‘snowing’. The type (il) trouverra ‘(he) will find’, instead of (il) trouvera, discussed in §18.3.2, seems to reflect contamination from the verb-form (il) verra ‘(he) will see’, motivated by a semantic connexion and a particular phonological resemblance between these two verbs in their future (and conditional) tense-forms, but only if verra and trouvera are viewed, respectively, as holistic exponents of the meanings ‘find’ and ‘see’, thereby making it possible for the contamination to be manifest in a portion of the latter analysable as its inflexional desinence. It is not my intention here to rehearse the by now well-known differences between so-called ‘morpheme-based’ and ‘word-and-paradigm’ approaches in ²⁷ See, for example, Papahagi (1974). ²⁸ A mere eight, because meteorological verbs obviously lack first and second person, and plural, forms. For some French verbs the number of word-forms expressing representing the lexeme may run into dozens. For modern French vouloir ‘to want’ I count, for example, eighteen distinct word-forms in the spoken language, and twenty-seven in the written language (leaving aside the forms of the now obsolescent preterite and imperfect subjunctive).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
400
morphological analysis (see, for example, Blevins 2006, 2016). Put very simply, the former is an atomistic framework in which word-structure is viewed as exhaustively compositional, being constructed from ‘morphemes’, understood as minimal signs pairing form with meaning. In this, ‘constructivist’, perspective, ‘words’ (not to mention inflexional paradigms) become no more than epiphenomena of the concatenation of morphemes. ‘Word-and-paradigm’ approaches,²⁹ in contrast, are essentially ‘abstractivist’, agnostic as to the nature of the basic unit of linguistic analysis, but generally foregrounding full word-forms, and treating sub-word elements as possible abstractions from word-forms, which have no fundamental status as unique units of storage. The significance of all the Gallo-Romance data I have reviewed in this study is that they seem inherently incompatible with a ‘constructivist’, morpheme-based, mode of analysis, while they appear strongly consistent with an ‘abstractive’ perspective. The recurrent locus of inflexion-class (‘conjugation’) marking in the relevant languages is the inflexional desinence. Even closed classes with few tokens, such as the third and especially the second conjugations in Romance, are the common property of dozens of lexemes, and thereby in principle independent of the lexical meanings of the set of lexemes with which they are associated. On the whole, the word-forms of the verbs with which I have been concerned here happen to lend themselves extremely easily to the ‘abstraction’, or ‘segmentation’, of a clearly profiled lexical root ‘morpheme’, which might be identified as the unique exponent of the verb’s lexical meaning.30 Yet each of the entire word-forms of the paradigm of the lexeme ‘rain’ must serve for speakers as an exponent of that meaning, not just the lexical root which may be identifiable within that word-form; if this were not so, then we would have no way of explaining how ‘rain’ could have contaminated ‘snow’, in some Gallo-Romance varieties, in a way that affects material outside the lexical root. At the same time, there is precise morphosyntactic mapping of the inflexional morphology of ‘rain’ on to the corresponding cells of the paradigm of ‘snow’: the infinitive ending of ‘rain’ turns up in the infinitive of ‘snow’ and nowhere else, the past participle ending of ‘rain’ in the past participle of ‘snow’, and so forth. The fine details of the historical mechanism of contamination of ‘snow’ by ‘rain’ in Gallo-Romance do not seem directly observable: it is difficult to discern whether the verb ‘rain’ acted on an existing (first conjugation) verb ‘snow’ of the Romance type *neˈvare, so that, so to speak, the original first conjugation inflexional desinences were each ‘ousted’ by the corresponding second / third conjugation endings, or whether the
²⁹ See Blevins, Ackerman, and Malouf (2019: 277) for reasons why these may be better called ‘itemand-pattern’ approaches. ³⁰ A problem in this respect, but one which I will not pursue here, is presented by the role of stress, which is, historically, what differentiates second from third conjugation verbs in the infinitive. Stress, after all, must be defined over an entire word-form.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
‘’ ‘’ -
401
verb ‘rain’ served as the model for novel derivational creation of a verb ‘snow’, speakers assigning the appropriate inflexional endings to the lexical root found in the noun *ˈneve.³¹ Either scenario naturally presupposes that speakers can also clearly identify and isolate, in each of the word-forms of the verb ‘rain’, the exponents of the relevant morphosyntactic features, which they can then map on to the forms of the verb ‘snow’—but what we observe cannot be satisfactorily explained in a model of morphology where word-forms implicitly have no psychological reality whatever. Finally, in the case of future-tense forms of trouver ‘find’, such as il trouverra, vous trouviendrez, discussed in §18.3.2, we seem to have cases where the verb ‘find’ is contaminated by extension of elements which comprise the lexical root of ‘see’ and ‘come’, yet there is an awkward mismatch of segmentation, in that forms meaning ‘see’ and ‘come’ intrude not into the lexical root of ‘find’, but into the desinential material to its right—again suggesting that the entire word-form is an exponent of the lexical meaning ‘find’, regardless of the potential segmentability of that word-form into root and desinence.³² These Gallo-Romance facts sit very comfortably, then, with a perspective on morphological structure which puts word-forms at the centre of analysis and does not assign lexical or grammatical meaning exclusively and a priori to component ‘morphemes’ regarded as fundamental ‘atoms’ of morphological structure—even where such ‘morphemes’ may also be readily abstracted from word structure and indeed possibly exploited for the creation of neologisms. ‘Word-and-paradigm’ approaches to morphology are well known to have millennia of history behind them, at least in the schoolroom. The history of morphological theory over the past few decades has shown that such approaches have theoretical validity as well. My analysis of ‘snow’, ‘rain’, and other Gallo-Romance verbs has shown that the traditional didactic approach to the representation of lexemes, namely, that they are cited as full word-forms, also reflects something theoretically significant. The word for ‘(to) rain’ in French really is pleuvoir . . . just as we all learned at school.
³¹ Evidence in support of the former hypothesis might be the existence of ‘heteroclite’ verbs meaning ‘snow’, such that some parts of their paradigm retained older first conjugation forms, while others showed newer second or third conjugation morphology. I am unaware of any such cases, however. The examples of estovoir in §18.3.1 and amerrir and alunir in §18.3.3 clearly show that lexically motivated contamination can influence conjugation-class assignment in the derivation of new forms. ³² For some other examples of the fluid and secondary nature of word-structure segmentation in diachrony, cf. also Maiden (2016c).
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
References and bibliographical abbreviations Abadal i de Vinyals, R. d’ (1952). Catalunya carolingia. II. Els diplomes carolingis a Catalunya. Barcelona: Institució Patxot. Abeillé, A. and Godard, D. (2000). ‘Varieties of in the Romance languages’. In Flickinger, D. and Kathol, A. (eds), Berkeley Formal Grammar Conference. Stanford, CA: CSLI, 2–22. Académie Française (2019). ‘La féminisation des noms de métiers et de fonctions’. http:// www.academiefrancaise.fr/sites/academiefrancaise.fr/files/rapport_feminisation_noms_ de_metier_et_de_fonction.pdf Ackema, P. and Sorace, A. (2017). ‘Auxiliary selection’. In Everaert, M. and van Riemsdijk, H. (eds), The Wiley Blackwell Companion to Syntax. Oxford: Wiley, 424–55. Ackerman, F. and Malouf, R. (2013). ‘Morphological organization: the low conditional entropy conjecture’. Language 89: 429–64. Adams, J. (2013). Social Variation and the Latin Language. Cambridge: Cambridge University Press. Adams, M. (1987a). ‘Old French, null subjects, and verb second phenomena’. UCLA: thesis. Adams, M. (1987b). ‘Embedded pro’. In Blevins, J. and Carter, J. (eds), Proceedings of NELS 18, I, 1–22. Adams, M. (1987c). ‘From Old French to the theory of Pro-Drop’. Natural Language and Linguistic Theory 5: 132. Adams, M. (1988). ‘Les effets V2 en ancien et en moyen français’. Revue québécoise de linguistique théorique et appliqué 7: 13–39. Adler, S. (2008). ‘French compound prepositions, prepositional locutions and prepositional phrases in the scope of the absolute use’. In Kurzon, D. and Adler, S. (eds) Adpositions: pragmatic, semantic and syntactic perspectives. Amsterdam / Philadelphia, PA: Benjamins, 17–35. Ahlborn, G. (1946). Le Patois de Ruffieu-en-Valromey (Ain). Gothenburg: Wettergren and Kerbers. Aikhenvald, A. (2006). ‘Grammars in contact: a cross-linguistic perspective’. In Aikhenvald, A. and Dixon, R. (eds), Grammars in Contact: A Cross-Linguistic Typology. Oxford: Oxford University Press, 1–66. Aikhenvald, S. (2008). ‘Imperative: the grammar of a clause type’. Presentation at conference Omotic Utterance type, Mood and Attitude Markers and Linguistic Typology. Leiden University, 23–25 October 2008. AIS = Jaberg, K. and Jud, J. (1928–40). Sprach- und Sachatlas Italiens und der Südschweiz. Zofingen: Ringier. Alcover, A. (1903). Questions de llengua y literatura catalana. Palma de Mallorca: Estampa de can Amengual y Muntaner. ALDI = Goebl, H., Böhmer, H., Gislimberti, S., Kattenbusch, D., Perini, E., Szekely, T., Dautermann, I., Heissmann, S., Hofmann, U., Kozak, A., Pamminger, H, Rössler, H.-M., Bauer, R., and Haimerl, E. (1998). Atlant linguistich dl ladin dolomitich y di dialec vejins/
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
404
Atlante linguistico del ladino dolomitico e dei dialetti limitrofi/Sprachatlas des Dolomitenladinischen und angrenzender Dialekte, Part 1. Wiesbaden: Reichert. http://ald1.sbg.ac.at ALEN = Brasseur 1980, 1984, 1997, 2010. Alex, P. (1965). Le Patois de Naisey, Canton de Roulans, arrondissement de Besançon: phonétique, morphologie, syntaxe, léxique, suivis d’une étude des lieux-dits de la commune. Paris: Voisin. ALF = Gilliéron, J. and Edmont, E. (1902–10). Atlas linguistique de la France. Paris: Imprimerie Nationale. Alibert, L. (1965). Dictionnaire occitan-français d’après les parlers languedociens. Toulouse: Institut d’Etudes Occitanes. Alibèrt, L. (1976). Gramatica occitana segon los parlars lengadocians, 2nd edn. Montpelhièr: CEO. ALLOc = Atlas Linguistique et Ethnographique du Languedoc Occidental (unpublished fieldwork descriptions). Alsina, A. (2016). ‘Catalan’. In Ledgeway and Maiden, 363–81. Alturo Monné, N. (1995). ‘La variació d’haver auxiliar en el català nord-occidental’. In Turell Julià, M. T. (ed.), La Sociolingüística de la variació. Barcelona: Promociones y Publicaciones Universitarias, 221–55. Ambar, M. (1992). Para uma sintaxe da inversão sujeito-verbo em português. Lisbon: Colibri. Anagnostopoulou, E. (1997). ‘Clitic left-dislocation and contrastive clitic left-dislocation’. In Anagnostopoulou, E., Riemsdijk, H. van, and Zwarts, F. (eds), Materials on Leftdislocation. Amsterdam: Benjamins, 151–92. Andersen, H. (1990). ‘The structure of drift’. In Andersen, H. and Koerner, K. (eds), Historical Linguistics 1987: Papers from the 8th International Conference on Historical Linguistics. Amsterdam / Philadelphia, PA: Benjamins, 1–20. Anderson, S. (2016). ‘Romansh (Rumantsch)’. In Ledgeway and Maiden (eds), 169–84. Andrieux-Rieux, N. and Baumgartner, E. (1983). Manuel du français du Moyen Age. III. Systèmes morphologiques de l’ancien français. A. Le verbe. Bordeaux: Sobodi. Anglade, J. (1921). Grammaire de l’ancien provençal ou ancienne langue d’oc: phonétique et morphologie. Paris: Klincksieck. Anglade, J. (1930). Grammaire élémentaire de l’ancien français. Paris: Armand Colin. Antonini, A. (1753). Principes de la grammaire françoise, pratique et raisonnée. Paris: Duschesne. Aranovich, R. (2003). ‘The semantics of auxiliary selection in Old Spanish’. Studies in Language 27: 1–37. Aranovich, R. (2007). ‘Split auxiliary selection from a cross-linguistic perspective’. In Aranovich, R. (ed.), Split Auxiliary Systems: A Cross-Linguistic Perspective. Amsterdam: Benjamins. Ashdowne, R. and Smith, J. C. (2007). ‘Some semantic and pragmatic aspects of case-loss in Old French’. In Salmons, J. and Dubenion-Smith, S. (eds), Historical Linguistics 2005. Amsterdam: John Benjamins, 191–205. Aronoff, M. (1994). Morphology by Itself. Cambridge, MA: MIT Press. Ashby, W. (1981). ‘The loss of the negative particle ne in French: a syntactic change in progress’. Language 57: 674–87. Asher, N. (2013). ‘Temporal modification’. In Jaszczolt and de Saussure, 17–36. Askedal, J. O. (1995). ‘Geographical and typological description of verbal constructions in the modern Germanic languages’. In Askedal, J. O. and Bjorvand, H. (eds), Drei Studien zum germanischen in alter und neuer Zeit. Odense, Denmark: Odense University Press, 95–146.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
405
Aub-Büscher, G. (1962). Le Parler rural de Ranrupt (Bas Rhin): essai de dialectologie vosgienne. Paris: Klincksieck. Audi, R. (ed.). (1999). The Cambridge Dictionary of Philosophy. Cambridge: Cambridge University Press. Auger, J. (2003). ‘The development of a literary standard: the case of Picard in VimeuPonthieu, France’. In Joseph, B., DeStefano, J., Jacobs, N., and Lehiste, I. (eds), When Languages Collide. Columbus: Ohio State University Press, 141–64. Avram, L. and Hill, V. (2007). ‘An irrealis BE auxiliary in Romanian’. In Aranovich, R. (ed.) Split Auxiliary Systems: A Crosslinguistic Perspective. Amsterdam: Benjamins, 47–64. Axel, K. (2007). Studies on Old High German Syntax: Left Sentence Periphery, Verb Placement and Verb-Second. Amsterdam / Philadelphia, PA: Benjamins. Ayres-Bennett, W. (2004). Sociolinguistic Variation in Seventeenth-Century France: Methodology and Case Studies. Cambridge: Cambridge University Press. Ayres-Bennett, W. (1996). A History of the French Language Through Texts. London / New York: Routledge. Ayres-Bennett, W. and Caron, P. (2016). ‘Periodization, translation, prescription and the emergence of classical French’. TPS 114: 339–90. Ayres-Bennett, W., Carlier, A., Glikman, J., Rainsford, T., Siouffi, Skupien, G., and Dekens, C. (eds) (2018). Nouvelles voies d’accès au changement linguistique. Paris: Classiques Garnier. Backus, A. (1996). Two in One. Bilingual Speech of Turkish Immigrants in the Netherlands. Tilburg: Tilburg University Press. Baerman, M. (2007). ‘Syncretism’. Language and Linguistics Compass 1: 539–51. Baerman, M., Brown, D., and Corbett, G. (2005). The Syntax-Morphology Interface. Cambridge: Cambridge University Press. Baker, M. (2003). Lexical Categories: Verbs, Nouns, and Adjectives. Cambridge: Cambridge University Press. Bakker, C. de. (1997). Germanic and Romance Inversion in French. The Hague: Holland Academic Graphics. Balcom, P. (2008). ‘On the learning of auxiliary use in the referential variety by speakers of New Brunswick Acadian French’. Canadian Journal of Linguistics 53: 7–34. Baldinger, K. (1958). ‘La position du gascon entre la Galloromania et l’ibéroromania’. Revue de linguistique romane 22: 241–92. Balon, L. and Larrivée, P. (2016). ‘L’ancien français n’est déjà plus une langue à sujet nul— nouveau témoignage des textes légaux’. JFLS 26: 221–37. Barbe, L. (1907). Dictionnaire du patois normand en usage à Louvier et dans les environs. Louviers: Izambert. Bardovi-Harlig, K. (2012). ‘Second language acquisition’. In Binnick, R. (ed.) The Oxford Handbook of Tense and Aspect. Oxford / New York: Oxford University Press, 481–503. Barthélemy, P. (1839). L’Omnibus du langage ou le Régulateur. Dijon: V. Lagier. Battye, A. and Roberts, I. (1994) (eds), Clause Structure and Language Change. Oxford: Oxford University Press. Bauche, H. (1946). Le Langage populaire. Grammaire, syntaxe et dictionnaire du français tel qu’on le parle dans le peuple avec tous les termes d’argot usuel. Paris: Payot. Bauer, B. (1995). The Emergence and Development of SVO Patterning in Latin and French. Diachronic and Psycholinguistic Perspectives. Oxford: Oxford University Press. Bauer, B. (2011). ‘Word formation in Romance’. In Maiden et al., 532–63. Bauer, B. (2017). Nominal Apposition in Indo-European. Its Forms and Functions, and its Evolution in Latin-Romance. Berlin: Mouton de Gruyter. Bauer, R., and Haimerl, E. (1998). Atlant linguistich dl ladin dolomitich y di dialec vejins / Atlante linguistico del ladino dolomitico e dei dialetti limitrofi / Sprachatlas des Dolomitenladinischen und angrenzender Dialekte, I. Wiesbaden: Reichert. http://ald1.sbg.ac.at
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
406
Bauman, J. and Torres Cacoullos, R. (2016). ‘The generalization of preposition para via fusion and ensuing loss of compositionality’. In Tortora, C., den Dikken, M., Montoya, I., and O’Neill, T. (eds), Romance Linguistics 2013. Selected Papers from the 43rd Linguistic Symposium on Romance Languages (LSRL), New York, 17–19 April 2013. Amsterdam / Philadelphia, PA: Benjamins, 39–58. Baunaz, L. and Eric, L. (2018). ‘Nanosyntax: the basics’. In Baunaz et al., 3–56. Baunaz, L., De Clercq, K., Haegeman, L., and Lander, E. (eds) (2018). Exploring Nanosyntax. Oxford / New York: Oxford University Press. Beaulieu, L. (1995). ‘The social function of linguistic variation: a sociolinguistic study in a fishing community of the north-eastern coast of New Brunswick’. University of South Carolina: thesis. Bec, P. (1963 / 1967). La Langue occitane. Paris: Presses Universitaires de France. Bec, P. (1970). Manuel pratique de philologie romane, I. Italien, espagnol, portugais, occitan, catalan, gascon. Paris: A. & J. Picard. Bec, P. (1973). Manuel pratique d’occitan moderne. Paris: Picard. Beigbeder, F. (2012). ‘Quand les écrivains font leur cinéma’. Le Nouvel observateur 1–12–2012. Belletti, A. (1990). Generalised Verb Movement. Turin: Rosenberg and Sellier. Belletti, A. (1994). ‘Verb positions: evidence from Italian’. In Lightfoot, D. and Hornstein, N. (eds), Verb Movement. Cambridge: Cambridge University Press, 19–40. Belletti, A. (2005). ‘Extended doubling and the VP periphery’. Probus 17: 1–35. Beniak, E. and Mougeon, R. (1989). ‘Recherches sociolinguistiques sur la variabilité en français ontarien’. In Beniak, E. and Mougeon, R. (eds), Le Français canadien parlé hors Québec: aperçu sociolinguistique. Sainte-Foy, Quebec: Les Presses de l’Université Laval, 69–104. Benincà, P. (1983a). ‘Osservazioni sulla sintassi dei testi di Lio Mazor’. Langue, dialecte, littérature. Etudes romanes à la mémoire de Hugo Plomteux. Louvain: Leuven University Press, 187–97. Benincà, P. (1983b). ‘Un’ipotesi sulla sintassi delle lingue romanze medievali’. Quaderni patavini di linguistica 4: 3–19. Benincà, P. (1995). ‘Complement clitics in medieval Romance: the Tobler-Mussafia law’. In Battye and Roberts, 325–44. Benincà, P. (2001). ‘On the position of topic and focus in the left periphery’. In Cinque, G. and Salvi, G. (eds), Current Studies in Italian Syntax: Essays Offered to Lorenzo Renzi. London / New York: Elsevier, 39–64. Benincà, P. (2004). ‘The left periphery of medieval Romance’. Studi linguistici e filologici online 2: 243–97. Benincà, P. (2006). ‘A detailed map of the Left Periphery of Medieval Romance’. In Zanuttini, R., Campos, H., Herburger, E., and Portner, P. (eds), Crosslinguistic Research in Syntax and Semantics: Negation, Tense, and Clausal Architecture. Washington, DC: Georgetown University Press, 53–86. Benincà, P. (2013). ‘Caratteristiche del V2 romanzo. Lingue romanze antiche, ladino dolomitico e portoghese’. In Cognola, F. and Bidese, E. (eds), Introduzione alla linguistica del mòcheno. Turin: Rosenberg and Sellier, 65–84. Benincà, P. (2014). ‘Subject clitics and particles in Provençal’. Probus 26: 183–215. Benincà, P. (2017). ‘A comparison of northern Romance and Occitan “subject” clitic systems. The null subject parameter and the residues of V2 as V-to-C’. Annali di Ca’ Foscari. Serie occidentale 51: 151–93. Benincà, P., Ledgeway, A., and Vincent, N. (2014). Diachrony and Dialects: Grammatical Change in the Dialects of Italy. Oxford: Oxford University Press.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
407
Benincà, P. and Poletto, C. (2004). ‘Topic, focus, and V2’. In Rizzi, L. (ed.), The Structure of CP and IP: The Cartography of Syntactic Structures. Oxford: Oxford University Press, 52–75. Benincà, P. and Vanelli, L. (2016). ‘Friulian’. In Ledgeway and Maiden, 139–53. Benincà, P., Parry, M., and Pescarini, D. (2016). ‘The dialects of northern Italy’. In Ledgeway and Maiden, 185–205. Benoît, R. (1932). Abrégé de grammaire périgourdine. Paris: Droz. Bentahila, A. and Davies, E. (1995). ‘Patterns of code-switching and patterns of language contact’. Lingua 96: 75–93. Bentley, D. (2004). ‘Il partitivo INDE nel siciliano e nel sardo delle origini’. In Dardano, M. and Frenguelli, G. (eds), SintAnt. La sintassi dell’italiano antico. Atti del Convegno Internazionale di Studi. Università Roma Tre, Roma, 18–21 settembre 2002. Rome: Aracne, 529–51. Bentley, D. (2006). Split Intransitivity in Italian. Berlin: Mouton de Gruyter. Bentley, D. (2018a). ‘Monotonicity in word formation: the case of Italo-Romance result state adjectives’. TPS 116: 285–319. Bentley, D. (2018b). ‘Grammaticalization of subject agreement on evidence from ItaloRomance’. Linguistics 56: 1246–301. Bentley, D. and Eyþórsson, Þ. (2003). ‘Auxiliary selection and the semantics of unaccusativity’. Lingua 114: 447–71. Bentley, D. and Ledgeway, A. (2014) ‘Manciati siti? Les constructions moyennes avec les participes résultatifs statifs dans l’italien et les variétés italo-romanes méridionales’. Langages 194: 63–80. Benveniste, E. (1962). ‘Hittite et Indo-Européen’. Paris: Maisonneuve. Benveniste, E. (1965). ‘Structure des relations d’auxiliarité’, Acta Linguistica Hungariae 9: 1–15. Benveniste, E. (1966–74). Problèmes de linguistique Générale, I-I., Paris: Gallimard. Benveniste, E. (1974). ‘Fondements syntaxiques de la composition nominale’. Problèmes de linguistique générale II. Paris: Gallimard, 45–62. Benzing, J. (1931). ‘Zur Geschichte von ser als Hilfszeitwort bei den intransitiven Verben im Spanischen’. ZRP 51: 385–460. Berk-Seligson, S. (1986). ‘Linguistic constraints on intra-sentential code-switching: a study of Spanish / Hebrew bilingualism’. Language in Society 15: 313–48. Bernini, G. (2012). ‘Tra struttura dell’informazione e finitezza: gli enunciativi’. In Orletti, F. (ed.), Grammatica e pragmatica. Atti del XXXIV Convegno Annuale della Società Italiana di Glottologia, Roma, 22–24 ottobre 2009. Rome: Il Calamo, 119–46. Berschin, H. and Berschin, W. (2011). ‘Augsburger Passionslied: die romanistische und mittellateinische Diskussion’. Zeitschrift für romanische Philologie 127: 209–19. Berschin, H., Berschin, W., and Schmidt, R. (1981). ‘Augsburger Passionslied: ein neuer romanischer Text des X. Jahrhunderts’. In Berschin, W. and Düchting, R. (eds), Lateinische Dichtungen des X. und XI. Jahrhunderts: Festgabe für Wather Bulst zum 80. Geburtstag. Heidelberg: Schneider, 251–79 and plates VI–VIII. Bertinetto, P. M. (2003). Tempi verbali e narrativa italiana dell’Otto / Novecento: quattro esercizi di stilistica della lingua. Alessandria, Italy: dell’Orso. Bertinetto, P. M. and Squartini, M. (2016). ‘Tense and aspect’. In Ledgeway and Maiden, 939–53. BFM = Base de Français Médiéval (2019). Lyon: ENS de Lyon, Laboratoire IHRIM. http://txm.bfm-corpus.org Bhat, G., Choudhury, M., and Bali, K. (2016). ‘Grammatical constraints on intrasentential codeswitching: from theories to working models’. https://arxiv.org/pdf/1612.04538v1.pdf Biberauer, T. (2003). ‘Verb second (V2) in Afrikaans: a minimalist investigation of word order variation’. University of Cambridge: doctoral thesis.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
408
Biberauer, T. (2004). ‘Reconsidering the EPP and Spec-TP in Germanic’. Cambridge Occasional Papers in Linguistics 1: 15–40. Biberauer, T. and Roberts, I. (2005). ‘Changing EPP-parameters in the history of English: accounting for variation and change’. English Language and Linguistics 9: 5–46. Biberauer, T. and Roberts, I. (2012). ‘Towards a parameter hierarchy for auxiliaries: diachronic considerations’. In Chancharu, J., Hu, X., and Mitrović, M. (eds), Cambridge Occasional Papers in Linguistics 6: 209–36. Biberauer, T. and Roberts, I. (2017). ‘Conditional inversion and types of parametric change’. In Los, B. and de Haan, P. (eds), Verb-Second Languages: Essays in Honour of Ans van Kemenade. Amsterdam: Benjamins, 55–77. Bigot, D. (2011). ‘De la norme grammaticale du français parlé au Québec’. Arborescences 1: Identités linguistiques, langues identitaires: à la croisée du prescriptivisme et du patriotisme. 0–0. https://doi.org/10.7202/1001939ar Birner, B. (2006). ‘Semantic and pragmatic contributions to information status’. In Mosegaard Hansen, M.-B. and Turner, A. (eds), Explorations in the Semantic/ Pragmatics Interface. Acta Linguistica Hafniensa 38: 14–32. Birt, P. (1985). Lé Jèrriais pour tous. A Complete Course on the Jersey Language. Jersey: Don Balleine. Bittner, M. and Hale, K. (1996). ‘The structural determination of case and agreement’. Linguistic Inquiry 27: 1–68. Bjerrome, G. (1957). Le Patois de Bagnes (Valais). Stockholm: Almqvist & Wiksell. Blasco Ferrer, E. (1984). Grammatica storica del catalano e dei suoi dialetti con speciale riguardo all’algherese. Tübingen: Narr. Blevins, J. (2016). Word and Paradigm Morphology. Oxford: Oxford University Press. Blevins, J. Ackerman, F., and Malouf, R. (2019). ‘Word and paradigm morphology’. In Audring, J. and Masini, F. (eds), The Oxford Handbook of Morphological Theory. Oxford: Oxford University Press, 264–84. Bloch, O. and von Wartburg, W. (1996 [1932]). Dictionnaire étymologique de la langue française. Paris: Presses Universitaires de France. Blondeau, H. (2006). ‘La trajectoire de l’emploi du futur chez une cohorte de Montréalais francophones entre 1971 et 1995’. Canadian Journal of Applied Linguistics 9: 73–98. Blondeau, H. and Labeau, E. (2016). ‘La référence temporelle au futur dans les bulletins météo en France et au Québec: regard variationniste sur l’oral préparé’. Revue canadienne de linguistique 61: 240–58. Blondeau, H., Frenette, Y., Martineau, F., and Tremblay, M. (2012). Sous-corpus variationniste Hochelaga-Maisonneuve et Saint-Michel, FRAN Corpus (as part of the project Le français à la mesure d’un continent: un patrimoine en partage, directed by F. Martineau). http://continent.uottawa.ca/fr/corpus-et-ressources-electroniques/corpus Blondin, J. (1823). Manuel de la pureté de la langue. Paris: [self-published]. Bonami, O. and Godard, D. (2008). ‘On the syntax of quotation in French’. In Müller, S. (ed.), 15th International Connference on HPSG, 2008. Stanford, CA: CSLI Publications, 358–77. Bonami, O., Boyé, G., Dal, G., Giraudo, H., and Namer, F. (2018). ‘Introduction’. In Bonami, O., Boyé, G., Dal, G., Giraudo, H., and Namer, F. (eds) The Lexeme in Descriptive and Theoretical Morphology. Berlin: Language Science Press, v–xiv. Bonaparte, L.-L. (1878). ‘Sur le caractère pronominal du monosyllabe béarnais «que»’. [no publication details available] Bonaparte, L.-L. (1879). ‘Troisième note sur le “que” pronominal propre au béarnais, et sur la probabilité de son origine basque’. [no publication details available] Bonfante, G. (1968). ‘Quando si è cominciato a parlare itliano?’ In Baldinger, K. (ed.). Festschrift Walther von Wartburg zum 80. Geburtstag. I. Tübingen: Niemeyer, 21–46.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
409
Bongelli, R., Riccioni, I. Vincze, L., and Zuczkowski, A. (2018). ‘Questions and epistemic stance: some examples from Italian conversations’. Ampersand 5: 29–44. Bonnard, H. and Régnier, C. (2008). Petite grammaire de l’ancien français. Paris: Magnard. Börjars, K. and Vincent, N. (2011). ‘Grammaticalization and directionality’. In Narrog, H. and Heine, B. (eds) (2011). The Oxford Handbook of Grammaticalization. Oxford: Oxford University Press, 163–76. Bosque, I. (2017). ‘Spanish exclamatives in perspective: a survey of properties, classes, and current theoretical issues.’ In Bosque, I. (ed.), Advances in the Analysis of Spanish Exclamatives. Columbus: Ohio State University Press, 1–52. Bossong, G. (1998). ‘Typologie der romanischen Sprachen. La typologie des langues romanes’. In Holtus, G., Metzeltin, M., and Schmitt, C. (eds), Lexikon der romanistischen Linguistik. VII: Kontakt, Migration und Kunstsprachen. Kontrastivität, Klassifikation und Typologie. Tübingen: Niemeyer, 1003–19. Bossong, G. (2016). ‘Classifications’. In Ledgeway and Maiden, 63–72. Bossong, Georg (1998). ‘Le marquage différentiel de l’objet dans le langues d’Europe’. In Feuillet, J. (ed.), Actance et valence dans les langues de l’Europe. Berlin: Mouton de Gruyter, 193–258. Bourciez, E. (1892). La Langue gasconne à Bordeaux. Bordeaux: Gounouilhou. Bourciez, E. (1896). ‘La conjugaison dans la Gavache du sud’. Revue des Universités du Midi 2: 142–82. Bourciez, E. (1946). Eléments de linguistique romane. Paris: Klincksiek. Bourdon, J.-P., Cournée, A., and Charpentier, Y. (1993). Dictionnaire normand-français. Paris: Presses Universitaires de France. Boutière, J. (1964). Biographies des troubadours : textes provençaux des XIIIe et XIVe siècles. Paris: A.-G. Nizet. Boutruche, R. (1935). ‘Les courants de peuplement dans l’Entre-Deux-Mers: étude sur le brassage de la population rurale: Première Partie: Du XIe au XVe Siècle. Deuxième Partie: Du XVe au XXe Siècle’. Annales d’histoire économique et sociale 7(31–32): 13–37, 124–54. Bouzet, J. (1932). ‘Les particules énonciatives en béarnais’. Oc 4: 40–53. Bouzet, J. (1933). ‘Les particules énonciatives en béarnais’, Oc 10–11: 21–38. Bouzet, J. (1951). ‘Les particules énonciatives en béarnais’. In Mélanges de linguistique offerts à Albert Dauzat par ses élèves et ses amis. Paris: Éditions d’Artrey, 47–54. Bouzet, J. (1963). Syntaxe béarnaise et gasconne. Pau, France: Marrimpouey. Bouzet, J. (1975). Manuel de grammaire béarnaise. Billère : Edicioûs de l’Escole Gastoû Febus. Bowern, C. (2010). ‘Fieldwork in contact situations’. In Hickey, 340–58. Boyle, A. (2006). An Introduction to Roman Tragedy. Oxford / New York: Taylor & Francis. Brandi, L. and Cordin, P. (1981). ‘Dialetti e italiano: un confronto sul parametro del soggetto nullo.’ Rivista di grammatica generativa 6: 36–87. Brasseur, P. (1977). ‘Le français dans les îles anglo-normandes’. Travaux de linguistique et de littérature 16: 97–104. Brasseur, P. (1980, 1984, 1997, 2010). Atlas linguistique et ethnographique normand. Paris: CNRS / (2010: Caen, France: Presses Universitaires de Caen). Brasseur, P. (1995). ‘Les pronoms personnels dans l’Atlas linguistique normand’. In Simoni, M.-R. (ed.), Dialectologie et littérature du domaine d’oïl occidental. Fontaine-lès-Dijon: A.B.D.O., 323–51. Brasseur, P. (2000). ‘La mise en œuvre du corpus dans le traitement des données lexicographiques: l’exemple du franco-terre-neuvien’. In Bavoux, C., Dupuis, R., and Kasbarian, J.-M. (eds), Images de la variation, perception et modélisation de la variation du français, en hommage à Daniel Baggioni. Paris : L’Harmattan, 193–216.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
410
Brasseur, P. and Chauveau, J.-P. (1990). Dictionnaire des régionalismes de Saint-Pierre-etMiquelon. Tübingen: Niemeyer. Bronckart, J.-P. and Sinclair, H. (1973). ‘Time, tense and aspect’. Cognition 2: 107–30. Brugmann, K. (1906). Grundriss der vergleichenden Grammatik der indogermanischen Sprachen. 2.1. Lehre von den Wortformen und ihrem Gebrauch. Allgemeines— Zusammensetzung (Komposita)—Nominalstämme. Strasbourg: Trübner. Brun-Trigaud, G. (1990). ‘Le Croissant: le concept et le mot’. Université Lyon III: thesis. Brun-Trigaud, G., le Berre, Y., and Le Dû, J. (2005). Lectures de l’Atlas linguistique de la France de Gilliéron et Edmont. Du temps dans l’espace. Editions du CTHS: Paris. Brunel, C. (1935). Bibliographie des manuscrits littéraires en ancien provençal. Paris: Droz. Brunot, F (1967). Histoire de la langue française des origines à nos jours. Paris: Colin. Brusewitz, V. (1905). ‘Étude historique sur la syntaxe des pronoms personnels dans la langue des félibres’. University of Uppsala: thesis. Buchard, A. and Carlier, A. (2008). ‘La forme verbale « être + participe passé » en tant que marqueur d’aspect et de la structure argumentale : une typologie graduée’. In Durand, J., Habert, B., and Laks, B. (eds), Congrès Mondial de Linguistique Française—CMLF 08. Paris: Institut de Linguistique Française, 2421–37. Buchi, E. (2009). ‘La dérivation en */de-/ et en */dis-/ en protoroman. Contribution à la morphologie constructionnelle de l’ancêtre commun des langues romanes’. Recherches linguistiques de Vincennes 38: 139–59. Buridant, C. (2000). Grammaire nouvelle de l’ancien français. Paris: Sedes. Buridant, C. (2010). Grammaire nouvelle de l’ancien français. Paris: Sedes. Burzio, L. (1986). Italian Syntax: a Government-Binding approach. Dordrecht: Foris. Bybee, J. (1985). Morphology: A Study of the Relation between Meaning and Form. Amsterdam: Benjamins. Bybee, J. (2010). ‘Markedness: iconicity, economy and frequency’. In Jung Song, J. (ed.), The Oxford Handbook of Linguistic Typology. Oxford: Oxford University Press, 131–47. Bybee, J. and Brewer, M. (1980). ‘Explanation in morphophonemics: changes in Provençal and Spanish preterite forms’. Lingua 52: 201–42. Camproux, C. (1958). Etude syntaxique des parlers gévaudanais. Paris: Presses Universitaires de France. Canale, M., Mougeon, R., and Bélanger, M. (1977a). Aspects of the Use of the Auxiliary AVOIR in Ontarian French. Toronto: Centre de recherches en éducation franco-ontarienne, Institut d’études pédagogiques de l’Ontario / The Ontario Institute for Studies in Education. Canale, M., Mougeon, R., and Bélanger, M. (1977b). ‘Aspects de l’usage de la préposition POUR en français ontarien: interférence et/ou surgénéralisation?’ Working Papers on Bilingualism 12: 61–78. Canale, M., Mougeon, R., and Bélanger, M. (1978). ‘Analogical leveling of the auxiliary être in Ontarian French’, In Suñer, M. (ed.), Contemporary Studies in Romance Linguistics. Washington, DC: Georgetown University Press, 41–61. Cappellaro, C., Maiden, M., and Smith, J. C. (2012). ‘The masculine wives of southern Italy’. Presentation at meeting of Societas Linguistica Europaea, Stockholm. Caradec, F. and Pouy, J.-B. (2009). Dictionnaire du français argotique et populaire. Paris: Larousse. Cardinaletti, A. (1997). ‘Subjects and clause structure’. In Haegeman, L. (ed.), The New Comparative Syntax. London: Longman, 33–63. Cardinaletti, A. (2004). ‘Toward a cartography of subject positions’. In Rizzi, L. (ed.). The Structure of CP and IP. The Cartography of Syntactic Structures. II. Oxford: Oxford University Press, 115–65.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
411
Cardinaletti, A. (2010). ‘On a (wh-)moved topic in Italian compared to Germanic’. In Alexiadou, A., Hankmaer, J., McFadden, T., Nuger, J., and Schaeffer, F. (eds), Advances in Comparative Germanic Syntax. Amsterdam: Benjamins, 3–40. Cardinaletti, A. and Repetti, L. (2010). ‘Proclitic vs enclitic pronouns in northern Italian varieties and the null-subject parameter’. In D’Alessandro, R. Ledgeway, A., and Roberts, I. (eds), Syntactic Variation: The Dialects of Italy. Cambridge: Cambridge University Press, 119–34. Cardinaletti, A. and Roberts, I. (2002). ‘Clause structure and X-second’. In Cinque, G. (ed.), Functional Structure in DP and IP. Oxford: Oxford University Press, 123–66. Cardinaletti, A. and Starke, M. (1999). ‘The typology of structural deficiency: a case study in the three classes of pronouns’. In van Riemsdijk, H. (ed.), Clitics in the Languages of Europe. Berlin: Mouton de Gruyter, 145–233. Carlier, A., Danckaert, L., Guillot-Barbance, C., Selig, M., and Vangaever, J. (forthcoming). Bridging the Gap between Late Latin and Early Old French, Empirical Investigations of (Morpho)syntactic Changes. Berlin / Boston, MA: De Gruyter. Carlin, E. (2006). ‘Feeling the need. The borrowing of Cariban functional categories into Mawayana (Arawak)’. In Aikenvald, A. and Dixon, R. (eds), Grammars in Contact: A Cross-Linguistic Typology. Oxford: Oxford University Press, 313–32. Carr, C. (1939). Nominal Compounding in Germanic. London: Oxford University Press. Carstairs[-McCarthy], A. (1987). Allomorphy in Inflexion. London: Croom Helm. Casalicchio, J. and Cognola, F. (2015). ‘On the left periphery of relaxed V2 languages: a comparison between Rhaetoromance and Mòcheno’. University of Trento: thesis. Cénac-Moncaut, J. (1863). Dictionnaire gascon-français dialecte du Départment du Gers. Suivi d’un abrégé de grammaire. Paris: Librairie archéologique de Didron. Cennamo, M. (2001a). ‘L’inaccusatività in alcune varietà campane: teorie e dati a confronto’. In Albano Leoni, F., Sornicola, R., Krosbakken, E., and Stromboli, C. (eds), Dati empirici e teorie linguistiche: atti del XIII congresso internazionale di studi della società di linguistica italiana. Rome / Cambridge: Bulzoni / Cambridge University Press, 427–53. Cennamo, M. (2001b). ‘L’extended accusative e le nozioni di voce e relazione grammaticale nel latino tardo e medievale’. In Viparelli, V. (ed.), Ricerche linguistiche tra antico e moderno. Naples: Liguori, 3–27. Cennamo, M. (2009). ‘Argument structure and alignment variations and changes in Late Latin’, In Barðdal, J. and Chelliah, S. (eds) The Role of Semantic, Pragmatic, and Discourse Factors in the Development of Case. Amsterdam / Philadelphia, PA: Benjamins, 307–46. Centineo, G. (1986). ‘A lexical theory of auxiliary selection in Italian’. Davies Working Papers in Linguistics 1: 1–35. Chabaneau, C. (1876). Grammaire limousine. Paris: Maisonneuve. Chambon, J.-P. and Olivier, P. (2000). ‘L’histoire linguistique de l’Auvergne et du Velay: notes pour une synthese provisoire’, Travaux de linguistique et de philologie 38: 83–153. Chaudenson, R., Mougeon, R., and Beniak, E. (1993). Vers une approche panlectale de la variation du français. Institut d’Études Créoles et Francophones, Didier Érudition, Coll. «Langues et développement». Chevalier, G. (1996). ‘L’emploi des formes du futur dans le parler acadien du sud-est du Nouveau-Brunswick’. In Dubois, L. and Boudreau, A. (eds), Les Acadiens et leur(s) langue(s: quand le français est minoritaire. Moncton, NB: Éditions d’Acadie, 75–89. Chiasson-Léger, M. (2014). ‘Future temporal reference in New Brunswick Acadian French’. Presentation at NWAV 43, Chicago, IL. 23–26 October 2014.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
412
Chomsky, N. (1981). Lectures on Government and Binding. Dordrecht: Foris. Chomsky, N. (1982). Some Concepts and Consequences of the Theory of Government and Binding. Cambridge, MA: MIT Press. Chomsky, N. (1986). Barriers. Cambridge, MA: MIT Press. Chomsky, N. (1995). The Minimalist Program. Cambridge, MA: MIT Press. Chomsky, N. (2007). ‘Approaching UG from below’. In Sauerland, U. and Gärtner, H.-M. (eds), Interfaces + Recursion = Language? New York: De Gruyter, 1–29. Chomsky, N. (2008). ‘On phases’. In Freidin, R. Otero, and Zubizarreta, M. L. (eds), Foundational Issues in Linguistic Theory. Essays in Honor of Jean-Roger Vergnaud. Cambridge, MA: MIT Press, 133–66. Cinque, G. (1999). Adverbs and Functional Heads: A Cross-Linguistic Perspective. New York: Oxford University Press. Cinque, G. (2004). ‘ “Restructuring” and functional structure’. In Belletti, A. (ed.), Structures and Beyond: The Cartography of Syntactic Structures, III. Oxford: Oxford University Press, 132–91. Cinque, G. (2010). ‘Introduction’. In Cinque, G. and Rizzi, L. (eds), Mapping Spatial PPs. Oxford: Oxford University Press, 3–25. Clackson, J. and Horrocks, G. (2007). The Blackwell History of the Latin language. Malden, MA / Oxford: Wiley-Blackwell. Clark, E. (1987). ‘The principle of contrast: a constraint on language acquisition’. In MacWhinney, B. (ed.), Mechanisms of Language Aquisition. London: Erlbaum. Clark, E. (1993). The Lexicon in Acquisition. Cambridge: Cambridge University Press. Cognola, F. and Schifano, N. (2018). ‘On the distribution and interpretation of the discourse particle ben across some regional varieties of Italian’. In Pană Dindelegan, G., Dragomirescu, A., Nicula, I., and Nicolae, A. (eds), Comparative and Diachronic Perspectives on Romance Syntax. Newcastle upon Tyne: Cambridge Scholars Publishing, 435–55. Cole, M. (2018). ‘A native origin for present-day English they, their, them’. Diachronica 35: 165–209. Collas, J. (1931). ‘A critical examination of the “Atlas Linguistique de la France” in so far as it concerns the Island of Guernsey’. University of Oxford: dissertation. Collins, R. (1990). ‘Charles the Bald and Wilfred the Hairy’. In Gibson, M. and Nelson, J. (eds), Charles the Bald: Court and Kingdom. Aldershot, Hampshire: Ashgate, 169–88. COM2 = Ricketts (2005). Comeau, P. (2015). ‘Vestiges from the grammaticalization path: the expression of future temporal reference in Acadian French’. JFLS 25: 330–65. Comeau, P. and Villeneuve, A.-J. (2016). ‘Future temporal reference in French: an introduction’. Canadian Journal of Linguistics 61: 231–39. Comeau, P., King, R., and Butler, G. (2012). ‘New insights on an old rivalry. The passé simple and the passé composé in spoken Acadian French’, JFLS 22: 315–43. Comrie, B. (1976). Aspect. Cambridge: Cambridge University Press. Comrie, B. (1982). ‘Remarks on clitic climbing in Brazilian Portuguese’. Lingua 58: 243–65. Corr, A. (2017). ‘Ibero-Romance and the syntax of the utterance’. University of Cambridge: thesis. Coseriu, E. (1988). ‘Der romanische Sprachtypus: Versuch einer neuen Typologisierung der romanischen Sprachen’. In Albrecht, J. (ed.), Energeia und Ergon: Sprachliche Variation, Sprachgeschichte, Sprachtypologie. Tübingen: Narr, 207–24. Coveney, A. (2010). ‘Vouvoiement and tutoiement: sociolinguistic reflections’. Journal of French Language Studies 20: 127–50.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
413
Croft, W. (2003). Typology and universals. Cambridge: Cambridge University Press. Cruschina, S. (2011). ‘Focalization and word order in old Italo-Romance’. Catalan Journal of Linguistics 10: 95–132. Cruschina, S. (2012). Discourse-related Features and Functional Projections. Oxford / New York: Oxford University Press. Cruschina, S. and Ledgeway, A. (2016). ‘Structure of the clause’. In Ledgeway and Maiden, 556–74. Cueillens, F. (1689). La Vie de Fr. Matthieu Viste. Toulouse: Louis Bosc. Cukor-Avila, P. and Bailey, G. (2013). ‘Real time and apparent time’. In Chambers, J. and Schilling, N. (eds), The Handbook of Language Variation and Change. Malden, MA / Oxford: Wiley, 239–62. Culbertson, A. (2010). ‘Convergent evidence for categorical change in French: from subject clitic to agreement marker’. Language 86: 85–132. Dahl, Ö. and Velupillai, V. 2013. ‘Perfects’. In Dryer and Haspelmath. Dahmen, W. (1995). ‘Français parlé québecois – français parlé de France: Konvergenz und Divergenz’. In Dahmen, W., Holtus, G., Kramer, J., Metzeltin, M., Schweickard, W., and Winkelmann, O. (eds), Konvergenz und Divergenz in den romanischen Sprachen. Tübingen: Narr, 223–37. Dalgish, G. (1979). ‘Morphophonemics of the OluTsootso dialect of OluLuyia: issues and implications’. University of Illinois, Urbana: thesis. Damourette, J. and Pichon, E. (1911–1936). Des mots à la pensée. Essai de grammaire de la langue française. Paris: D’Artrey. Danckaert, L. and Haegeman, L. (2012). ‘Conditional clauses, main clause phenomena and the syntax of polarity emphasis’. In Ackema, P., Alcorn, R., Heycock, C., Jaspers, D., van Craenenbroeck, J., and Vanden Wyngaerd, G. (eds), Comparative Germanic Syntax: the state of the art. Amsterdam: Benjamin, 133–67. Dardel, R. de. (1996). A La Recherche du protoroman. Tübingen: Niemeyer. Darrigrand, R. (1974). Initiation au gascon. Tours: Per Noste. Darrigrand, R. and Grosclaude, M. (1971). Abrégé de gramaire béarnaise et gasconne. Tours: Per Noste. Dauby, J. (1979). Le Livre du “Rouchi”. Parler picard de Valenciennes. Amiens: Musée de Picardie. Dauzat, A. (1906). Essai de méthodologie linguistique dans le domaine des langues et des patois romans. Paris: Champions. Davidson, J. (forthcoming). ‘The influence of Catalan laterals in Barcelonan Spanish: majority-language sociophonetic variation from minority-language contact’. Davies, M. (1996). ‘A corpus-based approach to diachronic clitic climbing in Portuguese’. Hispanic Journal 17: 93–111. Davies, M. (1997). ‘The evolution of Spanish clitic climbing: a corpus-based approach’. Studia Neophilologica 69: 251–63. Dawson, A., Auran, C., Bouzon, C., Delrue, L., Loock, R., O’Connor, K., and Patin, C. (2016). ‘French in Nord (Nord-Pas-de-Calais): a speaker from La Madeleine’. In Detey, S., Durand, J., Laks, B., and Lyche, C. (eds) (2016). Varieties of Spoken French. Oxford: Oxford University Press, 148–58. De Caillabère, T. (1965). Yocs de sourelh e d’oumpres. Pau, France: Marrimpouey. de Camelat, M. (1916). Lou darrè calhabari. Debertissance ne ue hèyte. Pau, France: Marrimpouey. de Camelat, M. (1933). Petit libiè d’ensegnance poupulàire. II. Garbe de proses 1775–1930. Pau, France: Marrimpouey.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
414
de Camelat, M. (no date). A l’aygue douce nou-b hidét! Arrepourè em dues hèytes. Pau, France: Marrimpouey. De Cat, C. (2009). French Dislocation: interpretation, syntax, acquisition. New York: Oxford University Press. De Chene, B. and Anderson, S. (1979). ‘Compensatory lengthening’. Language 55: 505–35. De Cia, S. (2019). The Syntactic-Pragmatic Interface in North-Eastern Italian Dialects: consequences for the geometry of the left periphery. University of Manchester: MS. De Garis, M. (1982). Dictiounnaire Angllais – Guernesiais. Chichester: Phillimore. De Garis, M. (1983). ‘Guernesiais: a grammatical survey’, Report and Transactions of La Société Guernesiaise 21: 319–53. De Grateloup, ([1734] 1887). Grammaire gasconne et Françoise. [Dax] Paris: Maisonneuve et Charles Leclerc. De Lacoux, H. (1687). La Vie et les actions de la venerable Mere Marie Alvequin de Jesus, Religeuse de Monmartre, Superieure et Reformatrice des Dames Augustines de S. Magloire dites Penitentes. Paris: Jean-François Dubois. De Poerck, G. (1963). ‘Les plus anciens textes de la langue française comme témoins de l’époque’. In Straka, G. (ed.), Les Anciens Textes romans non littéraires: leur apport à la connaissance de la langue au moyen âge. Actes et Colloques, 1. Paris: Klincksieck, 129–62. De Poerck, G. (1969). ‘Le Ms. Paris, B. N., lat. 1139: étude codicologique d’un recueil factice de pièces paraliturgiques (XIe–XIIIe siècle)’. Scriptorium 23: 298–312. Déchaine, R.-M. (2005). ‘Grammar at the borderline. A case study of P as a lexical category’. In Alderete, J., Han, C., and Kochetov, A. (eds) Proceedings of the 24th West Coast Conference on Formal Linguistics. Somerville, MA: Cascadilla Proceedings Project, 1–18. Dees, A. (1980). Atlas des formes et des constructions des chartes françaises du 13e siècle. Tübingen: Niemeyer. Dees, A. (1987). Atlas des formes linguistiques des textes littéraires de l’ancien français. Tübingen: Niemeyer. Den Besten, H. (1983). ‘On the interaction of root transformations and lexical deletive rules’. In Abraham, W. (ed.), On the Formal Syntax of the Westgermania. Amsterdam: Benjamins, 47–61. Descusses, M. (1986). Le Patois ardennais de Gespunsart. Paris: Société d’études linguistiques et anthropologiques de France. Deshaies, D. and Laforge, E. (1981). ‘Le futur simple et le futur proche dans le français parlé dans la ville du Québec’. Langues et linguistique 7: 21–37. Devine, A. and Stephens, L. (2006). Latin Word Order: structured meaning and information. Oxford / New York: Oxford University Press. Dickson, P. (2007). War Slang. American Fighting Words and Phrases from the Civil War to the War in Iraq. New York: Bristol Park Books. Diez, F. (1882). Grammatik der romanischen Sprachen. Bonn: Weber. Dixon, R. (1997). The Rise and Fall of Languages. Cambridge: Cambridge University Press. Donaldson, B. (2012). ‘Initial subordinate claues in Old French: syntactic variation and the clausal left periphery’. Lingua 122: 1021–46. Donaldson, B. (2015). ‘Discourse functions of subject left dislocation in old Occitan’. Journal of Historical Pragmatics 16: 159–86. Donaldson, B. (2016). ‘Preverbal subjects, information structure, and object clitic position in old Occitan’. Journal of Linguistics 52: 37–69. Donaldson, B. (2019). ‘The variable position of initial subordinate clauses in Old French: arguments against a semantic account’. In Arteaga, D. (ed.), Contributions of Romance Languages to Current Linguistic Theory. Berlin: Springer, 177–203.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
415
Donohue, M. and Smith, J. C. (1998). ‘What’s happened to us? Some developments in the Malay pronoun system’. Oceanic Linguistics 37: 65–84. Dorian, N. (1994). ‘Varieties of variation in a very small place: social homogeneity, prestige norms and linguistic variation’. Language 70: 631–96. Dorian, N. (2010). Investigating Variation. Oxford: Oxford University Press. Dourdet, J.-C. (2015). ‘Variation accentuelle dans l’espace dialectal nord-occitan limousin: éléments de problématique’. In Pfeffer, W. and Thomas, J. (eds), Nouvelles recherches en domaine occitan. Turnhout: Brepols, 237–58. Dowty, D. (1991). ‘Thematic proto-roles and argument selection’, Language 67: 547–619. Drinka, B. (2015). ‘Sources of auxiliation in the perfects of Europe’. In De Smet, H., Ghesquière, L. and Van de Velde, F. (eds), ‘On multiple source constructions in language change’. Amsterdam / Philadelphia, PA: Benjamins, 129–74. Drinka, B. (2017). Language Contact in Europe. The Periphrastic Perfect through History. Cambridge: Cambridge University Press. Drinka, B. (2019). ‘Perfecting the notion of Sprachbund: perfects and resultatives in the ‘Stratified Convergence Zones’ of Europe’. In Cennamo, M. and and Fabrizio, C. (eds), Historical linguistics 2015: Selected papers from the 22nd International Conference on Historical Linguistics. Amsterdam: Benjamins, 319–41. Dryer, M. and Haspelmath, M (eds) (2013). The World Atlas of Language Structures Online. Leipzig: Max Planck Institute for Evolutionary Anthropology. Dupuis, F. (1988). ‘Pro-drop dans les subordonnées en ancien français’. Revue québécoise de linguistique théorique et appliqué 7: 41–62. Dupuis, F. (1989). ‘L’Expression du sujet dans les subordonnées en ancien français’. University of Montreal: thesis. Duraffour, A. (1932). Description morphologique avec notes syntaxiques du parler francoprovençal de Vaux (Ain) en 1919–1931. Grenoble: Institut phonétique de Grenoble. Eberenz, R. (1977). ‘Zur Entwicklung der Opposition perfecto simple vs. perfecto compuesto im Spanischen und Katalanischen der Renaissance’. ZRP 93: 518–27. Eckardt, R. (2006). Meaning Change in Grammaticalization: an enquiry into semantic reanalysis. Oxford: Oxford University Press. Egerland, V. and Falk, C. (2010). ‘Si och så. Mellan narrativitet och grammatik’. In Byrman, G., Gustafsson, A., and Rahm, H. (eds), Svensson och svenskan. Med sinnen känsliga för språk. Festskrift till Jan Svensson den 24 januari 2010. Lund: Lunds universitets publikationer. Eide, K. (2011). ‘Norwegian (non-V2) declaratives, resumptive elements, and the Wackernagel position’. Nordic Journal of Linguistics 34: 179–213. Eliordeta, A. (2001). Verb Movement and Constituent Permutation in Basque. Leiden: LOT. Emirkanian, L. and Sankoff, D. (1986). ‘Le futur simple et le futur périphrastique’. In Lemieux, M. and Cedergren, H. (eds), Les Tendances dynamiques du français parlé à Montréal. Montréal: Office de la langue française, 189–204. Emonds, J. (1985) A Unified Theory of Syntactic Categories. Dordrecht: Foris. Esher, L. (2012). ‘The morphological relation of future, conditional and infinitive in Occitan’. In Kemenade, A. van and Haas, N. de (eds), Historical Linguistics 2009. Amsterdam: Benjamins, 315–32. Esher, L. (2015a). ‘Formal asymmetries between the Romance synthetic future and conditional in the Occitan varieties of the Western Languedoc’. TPS 113: 249–70. Esher, L. (2015b). ‘Morphomes and predictability in the history of Romance perfects’. Diachronica 32: 494–529.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
416
Esher, L. (2016). ‘Morphomic distribution of augments in varieties of Occitan’. Revue Romane 51: 271–306. Esher, L. (2017a). ‘Morphome death and transfiguration in the history of French’. Journal of Linguistics 53: 51–84. Esher, L. (2017b). ‘L’aventura de la desinéncia 1 -i dins les parlars occitans de Lengadòc’. Tenso 32: 19–40. Esher, L. (2018). ‘Implicational relationships between desinences in Occitan imperfect and conditional forms’. Lexique 53: 9–32. Etiemble, R. (1965). ‘La langue de la publicité’, Les Cahiers de la publicité 15: 105–12. Wagner, S. and Sankoff, G. (2011). ‘Age grading in the Montréal inflected future’. Language Variation and Change 23: 275–313. Fagard, B. and De Mulder, W. (2007). ‘La Formation des prépositions complexes: grammaticalisation ou lexicalisation?’. Langue Française 156: 9–29. Fagard, B. and Mardale, A. (2012). ‘The pace of grammaticalization and the evolution of prepositional systems: data from Romance’. Folia Linguistica 46: 303–40. Fagard, B., Pinto de Lima, J., Stosic, D., and Smironova, E. (eds) (forthcoming). Complex Adpositions in European Languages. Berlin: De Gruyter. Fankhauser, F. (1911). Das Patois von Val d’Illiez (Unterwallis). Halle, Germany: Karras. Fauconnier, G. (1975). ‘Pragmatic Scales and Logical Structure’. Linguistic Inquiry 6: 353–75. Faure, R. and Oliviéri, M. (2013). ‘Stratégies de topicalisation en occitan’. Corpus 12: 231–70. Ferraresi, G. and Goldbach, M. (2002). ‘V2 Syntax and topicalization in old French’. Linguistiche Berichte 189: 2–25. Ferraresi, G. and Goldbach, M. (2003). ‘Particles and sentence structure: a historical perspective’. In Junghanns, U. and Szucsich, L. (eds), Syntactic Structures and Morphological Information. Berlin: Mouton de Gruyter, 101–27. Feyel, G. (1999). La Presse en France des origines à 1944: histoire politique et matérielle. Paris: Ellipses. Field, T. (1985). ‘Speech act markers in modern Gascon’. In King, L. and Maley, C. (eds), Selected Papers from the IIIrd Linguistic Symposium on Romance Languages: Chapel Hill N.C. 24–26 March 1983. Amsterdam: Benjamins, 77–97. Fischer, S. (2002). The Catalan Clitic System: a diachronic perspective on its syntax and phonology. Berlin / New York: Mouton de Gruyter. Flege, J. (1988). ‘The production and perception of speech sounds in a foreign language’. In Winitz, H. (ed.), Human Communication and Its Disorders: A Review 1988. Norwood, NJ: Ablex, 224–401. Flege, J. (1990). ‘The intelligibility of English vowels spoken by British and Dutch talkers’. In Kent, R. (ed.), Intelligibility in Speech Disorders: Theory, Measurement, and Management. Amsterdam: Benjamins, 157–232. Flege, J. (1995). ‘Second-language speech learning: theory, findings, and problems’. In Strange, W. (ed.), Speech Perception and Linguistic Experience: Issues in CrossLanguage Research. Timonium, MD: York Press, 229–73. Flege, J. (2007). ‘Language contact in bilingualism: phonetic system interactions’. In Cole, J., and Hualde, J. (eds), Laboratory Phonology 9. Berlin: Mouton de Gruyter, 353–80. Fleischer, F. (1913). Studien zur Sprachgeographie des Gascogne. Halle, Germany: Niemeyer. Fleischman, S. (1982). The Future in Thought and Language: Diachronic Evidence from Romance. Cambridge: Cambridge University Press. Fleischman, S. (1985). ‘Discourse functions of tense-aspect oppositions in narratives’. Linguistics 23: 851–82.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
417
Fleischman, S. (1991). ‘Discourse pragmatics and the grammar of old French: a functional reinterpretation of si and the personal pronouns’. Romance Philology 44: 251–83. Fleischman, S. (1995). ‘The non-lyric texts’. In Akehurst, F. and Davis, J. (eds), A Handbook of the Troubadours. Berkeley / London: University of California Press, 176–84. Fleury, J. (1883). ‘No normand et on français’. Romania 12: 342–45. Fleury, J. (1886). Essai sur le patois normand de la Hague. Paris: Maisonneuve et Leclerc. Flikeid, K. and Péronnet, L. (1989). ‘N’est-ce pas vrai qu’il faut dire “J’avons été”?: divergences régionales en acadien’. Français Moderne 57: 219–28. Flobert, P. (1975). Les Verbes déponents latins des origins à Charlemagne. Paris: Les Belles Lettres. Fløgstad, G. (2016). Preterite Expansion and Perfect Demise in Porteño Spanish and Beyond. Leiden: Brill. Floricic, F. (2008). ‘The Italian verb-noun anthroponymic compounds at the syntax / morphology interface’. Morphology 18: 167–93. Floricic, F. (2009). ‘Negation and focus clash in Sardinian’. In Mereu, L. (ed.), Information Structure and Its Interfaces. Berlin: Mouton de Gruyter, 129–52. Floricic, F. (2012). ‘Dialectological evidence for a predicate focus analysis of Gascon que’. Paris 3—Sorbonne Nouvelle: MS. Flutre, L.-F. (1955). Le Parler picard de Mesnil-Martinsart (Somme). Geneva: Droz. Flydal, L. (1943). Aller et venir de suivis de l’infinitif comme expressions de rapports temporels. Oslo: Jacob Dybwad. Fossat, J.-L. (2006). ‘Les “petits mots” énonciatifs gascons: le cas de bè énonciatif ’. Cahiers de grammaire 30: 159–74. Fouché, P. (1967). Morphologie historique du français: le verbe. Paris: Klincksieck. Foulet, L. (1919 / 1928 / 1930). Petite syntaxe de l’ancien français. Paris: Champion. Foulet, L. (1967). Petite syntaxe de l’ancien français. Paris: Champion. Fournier, N. (1998 / 2002). La Grammaire du français classique. Paris: Belin. Fox, V. (2003). Dictionnaire gascon-français. Suivi de son lexique français-gascon et d’éléments d’un Thésaurus gascon. Bordeaux: Presses Universitaires de Bordeaux. Franco, L. (2013). ‘A typological rarum in Sogdian: overt complementizers in indicative root clauses’. Lingua Posnaniensis 55: 55–67. François, D. (1974). Français parlé: analyse des unités phoniques et significatives d’un corpus recueilli dans la région parisienne. Paris: Société d’études linguistiques et anthropologiques de France. Frappier, J. (ed.) (1954). Le Mort le Roi Artu. Geneva: Droz. Frascarelli, M. and Hinterhölzl, R. (2007). ‘Types of topics in German and Italian’. In Schwabe, K. and Winkler, S. (eds), On Information Structure, Meaning and Form. Amsterdam: Benjamins, 87–116. Frei, H. (1929). La Grammaire des fautes. Geneva: Slatkine. Frei, H. (1971 [1929]). La Grammaire des fautes. Geneva: Slatkine. Frey, W. (2004). ‘Notes on the syntax and the pragmatics of German left dislocation’. In Lohnstein, H. and Trissler, S. (eds), The Syntax and Semantics of the Left Sentence Periphery. Berlin: De Gruyter, 203–33. Friedman, V. and Joseph, B. (forthcoming). The Balkan Languages. Cambridge: Cambridge University Press. Frota, S. and Prieto, P. (2015). Intonation in Romance. Oxford: Oxford University Press. Gadet, F. and Hambye, P. (2018). ‘The metropolization of French worldwide’. In AyresBennett, W. and Carruthers, J. (eds), Manual of Romance Sociolinguistics. Berlin: Mouton, 724–44.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
418
Gal, S. (1989). ‘Language and political economy’. Annual Review of Anthropology 18: 345–67. Galet, Y. (1971). L’Evolution de l’ordre des mots dans la phrase française de 1600 à 1700. La place du pronom personnel complément d’un infinitif régime. Paris: Presses Universitaires de France. Gallego, Á. (2016). ‘A phase-theoretic approach to cliticization in Romance’. Studies in Hispanic and Lusophone linguistics 9: 67–94. Galves, C., Ribeiro, I., and Torres Moraes, M. A. (2005). ‘Syntax and morphology in the placement of clitics in European and Brazilian Portuguese’. Journal of Portuguese Linguistics 4: 143–77. Gardette, P. (1983). ‘Aspects géographiques du lexique francoprovençal (à propos de la publication du premier volume de l’Atlas linguistique du Jura et des Alpes du Nord)’. In Horiot, B, Simoni, M.-R., and Straka, G. (eds), Pierre Gardette. Etudes de géographie linguistique. Paris: Klincksieck, 547–67. Gardner-Chloros, P. (2009). Code-Switching. Cambridge: Cambridge University Press. Garzonio, J. and Poletto, C. (2008). ‘Minimizers and quantifier: a window on the development of negative markers’, Studies in Linguistics, CISCL Working Papers 2: 59–80. Garzonio, J. and Poletto, C. (2010). ‘Quantifiers as negative markers in Italian dialects’. In Van Croenenbroeck, J. (ed.), Linguistics Variation Yearbook 2009. Amsterdam: Benjamins, 127–52. Gauchat, L. (1930). ‘Vom morphologischen Denken’. In Maync, H. (ed.). Festgabe für Samuel Singer. Tübingen: Mohr, 126–43. Gavel, H. (1920). Eléments de phonétique basque. Paris: Champion. Gelderen, E. van (2004). Grammaticalization as Economy. Amsterdam: Benjamins. Gesner, E. (1978). ‘L’emploi des auxiliaires avoir et être dans le parler acadien de la Baie Sainte-Marie, Nouvelle-Écosse’, In Patterson, G. (ed.), Papers from the Second Annual Meeting of the Atlantic Provinces Linguistics Association. Halifax, NS: Mount Saint Vincent University, 16–22. Giancarli, P.-D. (2011). Les Auxiliaires être et avoir: étude comparée corse, français, acadien et anglais. Rennes: Presses Université de Rennes. Giancarli, P.-D. (2015). ‘Auxiliary selection with intransitive and reflexive verbs: the limits of gradience and scalarity, followed by a proposal’. In Kailuweit, R. and Rosemeyer, M. (eds), Auxiliary Selection Revisited: Gradience and Gradualness. Berlin / Munich / Boston, MA: De Gruyter, 79–120. Gilliéron, J. (1919). Etude sur la défectivité des verbes: la faillite de l’étymologie phonétique. Neuveville, Switzerland: Macon, Protat Frères. Gilliéron, J. and Edmond E. (1902–10). Atlas linguistique de la France. Paris: Honoré Champion. Giorgi, A. (2010). About the Speaker. Towards a Syntax of Indexicality. Oxford: Oxford University Press. Giorgi, A. and Pianesi, F. (1996). Tense and Aspect: From Semantics to Morphosyntax. Oxford: Oxford University Press. Giurgea, I. and Remberger, E. M. (2016). ‘Illocutionary force’. In Ledgeway and Maiden, 863–78. Gondon, G. (1662). L’Imitateur de Jesus Christ, ou la Vie du Venerable Pere Antoine Yvan, Prestre, Instituteur de l’Ordre des Religieuses de Nostre-Dame de Misericorde. Et l’histoire de la fondation du même ordre. Paris: Jean Bouillard. González Campos, G. and Lehmann, C. (forthcoming). The Cabecar Language. González i Planas, F. (2009). ‘Cartografia de les posicions de subjecte en gascó’. Linguistica Occitana 7: 83–98.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
419
Gosselin, L. (1996). Sémantique de la temporalité en français. Un modèle calculatoire et cognitif du temps et de l’aspect. Louvain-la-Neuve, Belgium: Duculot. Gosselin, L. (2005). Temporalité et modalité. Brussels: De Boeck. Gosselin, L. (2013). ‘The interaction of time and modality in French’. In Jaszczolt and de Saussure, 98–127. Gossen, C. (1967). Französische Skriptastudien: Untersuchungen zu den nordfranzösischen Urkundensprachen des Mittelalters. Graz, Austria: Böhlaus Nachfolger. Graffi, G. (1996). ‘Alcune riflessioni sugli imperativi italiani’. In Benincà, P., Cinque, G., De Mauro, T., and Vincent, N. (eds), Italiano e dialetto nel tempo. Saggi di grammatica per Giulio C. Lepschy. Rome: Bulzoni, 143–48. Greenberg, J. (1963). ‘Some universals of grammar with particular reference to the order of meaningful elements’. Universals of language. Cambridge, MA: MIT Press, 56–90. Grevisse, M. (2011). Le Bon usage: grammaire française refondue par André Goosse. Gembloux, Belgium: Duculot. Grevisse, M. and Goosse, A. (2008). Le Bon usage. Grammaire française avec des remarques sur la langue française d’aujourd’hui. Gembloux, Belgium: Duculot. Grewendorf, G. (2002). ‘Left dislocation as movement’. Georgetown University Working Papers in Theoretical Linguistics 2: 31–81. Grieves-Smith, A. (2009). The Spread of Change in French Negation. University of New Mexico. https://www.academia.edu/203559/The_Spread_of_Change_in_French_Negation Grimm, R. (2010). ‘A real-time study of future temporal reference in spoken Ontario French’. University of Pennsylvania Working Papers in Linguistics 16: 83–92. Grimm, R. and Nadasdi, T. (2011). ‘The future of Ontario French’. JFLS 21: 173–89. Grohmann, K. (1998). ‘Syntactic inquiries into discourse restrictions on multiple interrogatives’. Groninger Arbeiten zur germanischen Linguistik 42: 1–60. Grohmann, K. (2000). ‘Copy left dislocation’. In Billerey, R. and Lillehaugen, D. (eds), Proceedings of the Nineteenth West Coast Conference on Formal Linguistics. Somerville, MA: Cascadilla Press, 139–52. Grohmann, K. (2003). ‘German is a multiple wh-fronting language!’ In Boeckx, C. and Grohmann, K. (eds), Multiple Wh-fronting. Amsterdam: Benjamins, 99–130. Grosclaude, M. (1979). Le Béarn: témoignages sur mille ans d’histoire. Pau, France: Per Noste. Grosclaude, M. (1986). La Langue béarnaise et son histoire: études sur l’évolution de l’occitan du Béarn depuis leIIIe siècle, suivi de quatre analyses de textes desVIIe etVIIIe. Orthez, France: Per Noste. Grzega, J. (2003). ‘Zur Geschichte von fr. trouver und seinen Verwandten: Anmerkungen zu einem alten Problem’. ZRP 119: 222–31. Guardiano, C. and Longobardi, G. (2005). ‘Parametric comparison and language taxonomy’. In Batllori, M., Hernanz, M.-Ll., Picallo, C., and Roca, F. (eds), Grammaticalization and Parametric Variation. Oxford: Oxford University Press, 149–74. Gudmestad A. and Edmonds, A. (2016). ‘Variable future-time reference in French: a comparison of learners in a study-abroad and foreign-language context’. Canadian Journal of Linguistics 61: 259–85. Guerlin De Guer, C. (1901). Le Parler populaire dans la commune de Thaon (Calvados). Paris: Bouillon. Guilhemjoan, P. (2006), Nouvelle grammaire abrégée du gascon. Orthez, France: Per Noste. Guiraud, P. (1969). Le Français populaire. Paris: Presses Universitaires de France. Gvozdanović, J. (1995). ‘Western South Slavic tenses in a typological perspective’. In Thieroff, R. (ed.), Tense Systems in European Languages II. Tübingen: Niemeyer, 181–94.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
420
Haase, M. (1994). ‘L’énonciatif gascon et le substrat basque’. In Cierbide, R. (ed.), Actes du IV Congrès International de l’Association Internationale d’Études Occitanes, VitoriaGasteiz, 22–28 août 1993. Vitoria-Gasteiz: Universidad del País Vasco / Euskal Herriko Unibertsitatea, 801–19. Haase, M. (1994). ‘L’énonciatif gascon et le substrat basque’. In Cierbide, R. (ed.), Actes du IV congrès international de l’AIEO: Vitoria-Gasteiz, 22–28 août 1993. Vitoria-Gasteiz: UPV/EHU, II, 801–19. Haase, M. (1997). ‘Gascon et basque: bilinguisme et substrat’. Sprachtypologie und Universalienforschung 50: 189–228. Haddican, B. (2004). ‘Sentence polarity and word order in Basque’. The Linguistic Review 21: 87–124. Haddican, B. and Eldorieta, A. (2013). ‘V2 in Basque’. University of Pennsylvania Working Papers in Linguistics 19: 59–68. Haegeman, L. (1996). ‘Verb second, the split CP, and null subjects in early Dutch finite clauses’. Geneva Generative Papers 4: 135–75. Haegeman, L. (2003). ‘Conditional clauses: external and internal syntax’. Mind and Language 18: 317–39. Haegeman, L. (2007). ‘Operator movement and topicalisation in adverbial clauses (clitic left dislocation)’. Folia Linguistica 41: 279–325. Haegeman, L. (2009). ‘The movement analysis of temporal adverbial clauses’. English Language and Linguistics 13: 385–408. Haegeman, L. (2010a). ‘The internal syntax of adverbial clauses’. Lingua 120: 628–48. Haegeman, L. (2010b). ‘The movement derivation of conditional clauses’. Linguistic Inquiry 41: 595–621. Haegeman, L. (2012a). ‘The syntax of main clause phenomena: deriving truncation’. In Aelbrecht, L, Haegeman, L., and Nye, R. (eds), Main Clause Phenomena: New Horizons. Amsterdam: Benjamins, 113–34. Haegeman, L. (2012b). Adverbial Clauses, Main Clause Phenomena and the Composition of the Left Periphery: The Cartography of Syntactic Structures. Oxford: Oxford University Press. Haegeman, L. (2013). ‘The syntax of adverbial clauses’. In Anderson, S., Moeschler, J., and Reboul, F. (eds), The Language-Cognition Interface. Paris / Geneva: Droz, 135–65. Haegeman, L. and Ürögdi, B. (2010). ‘Referential CPs and DPs: an operator movement account’. Theoretical Linguistics 36: 111–52. Haegeman, L. (2014). ‘Locality and the distribution of main clause phenomena’. In Aboh, E., Guasti, M. T., and Roberts, I. (eds), Locality. New York: Oxford Univeristy Press, 186–222. Hagège, C. (2010). Adpositions. Oxford: Oxford University Press. Haider, H. (1993). Deutsche Syntax Generativ. Tübingen: Narr. Haiman, J. and Benincà, P. (1992). The Rhaeto-Romance Languages. London: Routledge. Hallion Bres, S. (2006). ‘Similarités morphosyntaxiques des parlers français de l’ouest canadien’. Revue canadienne de linguistique appliquée-Revue de l’Université de Moncton 37: 111–31. Hallion, S. (2000). ‘Étude du français parlé au Manitoba’. Université de Provence: thesis. Hamlin, F. R., Ricketts, P., and Hathaway, J. (1967). Introduction à l’étude de l’ancien provençal. Geneva: Droz. Han, C. (2002). ‘Interpreting interrogatives as rhetorical questions’. Lingua 112: 201–29. Hansch, A. (2014). ‘Germanic properties in the left periphery of Old French: V-to-Cmovement’. University of Ottowa: thesis.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
421
Hansen, M.-B. Mosegaard (2009). ‘The grammaticalization of negative reinforcers in Old and Middle French: a discourse-functional approach’. In Hansen, M.-B. M. and Visconti, J. (ed.), Current Trends in Diachronic Semantics and Pragmatics. Leiden: Brill, 227–51. Hansen, M.-B. Mosegaard (2013). ‘Negation in the history of French’. In Willis, D., Lucas, C., and Breitbarth, A. (eds), The History of Negation in the Languages of Europe and the Mediterranean. Oxford: Oxford University Press, 51–76. Hansen, M.-B. Mosegaard (2018). ‘The expression of clause negation: from Latin to early French’. In Carlier, A. and Guillot-Barbance, C. (eds), Latin tardif, français ancien: continuités et ruptures. Berlin / Boston, MA: De Gruyter, 269–98. Hansen, M.-B. Mosegaard and Visconti, J. (2009). ‘On the diachrony of “reinforced” negation in French and Italian’. In Rossari, C., Ricci, C., and Spiridon, A. (eds), Grammaticalization and Pragmatics. Facts, Approaches, Theoretical Issues. Leiden: Brill, 137–71. Harling-Kranck, G. (1998). Från Pyttis til Nedervetil. Tjugonio dialektprov från Nyland. Åboland: Åland och Österbotten. Helsinki: Svenska litteratursällskapet i Finland. Harris, A. and Campbell, L. (1994). Historical syntax in cross-linguistic perspective. Cambridge: Cambridge University Press. Harris, M. (1978). The Evolution of French Syntax: A Comparative Approach. London: Longman. Harris, M. (1984). ‘On the causes of word order change’. Lingua 63: 175–204. Hasselrot, B. (1972). Etude sur la vitalité de la formation diminutive française au XXe siècle. Uppsala, Sweden: Almqvist and Wiksell. Haugen, E. (1950). ‘The analysis of linguistic borrowing’. Language 26: 210–32. Hébert, J. (1984). Présentation du dialecte cauchois. Yvetot: Fédération des Foyers Ruraux de la Seine-Maritime. Heine, B. and Kuteva, T. (2005). Language Contact and Grammatical Change. Cambridge: Cambridge University Press. Heine, B. and Reh, M. (1983). ‘Diachronic observations on completive focus marking in some African languages’. Sprache und Geschichte in Afrika 5: 7–44. Heine, B. and Song, K.-A. (2010). ‘On the genesis of personal pronouns: some conceptual sources’. Language and Cognition 2: 117–48. Heinz, M., Filipponio, L., and Hinzelin, M.-O. (forthcoming). Manual of Classification and Typology of the Romance Languages. Berlin / Boston, MA: De Gruyter. Hendersson, C. (2010). El pretérito compuesto del español de Chile, Paraguay y Uruguay. Stockholm: Stockholm University. Hendschel, L. (2012). Li Croejhete walone: contribution à une grammaire de la langue wallonne. Herman, J. (1954). ‘Recherches sur l’ordre des mots dans les plus ancien textes français en prose (deuxième partie)’. Acta Linguistica Scientiarum Hungaricae 4: 351–82. Hernanz, M. Ll. (2007). ‘From polarity to modality: some (a)symmetries between bien and si in Spanish’. In Eguren, L. and Fernández-Soriano, O. (eds), Coreference, Modality, and Focus: Studies on the Syntax-Semantics Interface. Amsterdam: Benjamins, 133–69. Hernanz, M. Ll. (2010). ‘Assertive bien in Spanish and the left periphery’. In Benincà, P. and Munaro, N. (eds), Mapping the Left Periphery. Oxford: Oxford University Press, 19–62. Hetrzon, R. (1977). ‘La particule énonciative qué en gascon’. Studi italiani di linguistica teorica ed applicata 6: 161–221. Hickey, R. (2010). The Handbook of Language Contact. Oxford: Blackwell. Hill, V. (2001). ‘Verb restructuring with infinitives and subjunctives’. Linguistic Analysis 29: 304–21.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
422
Hilty, G. (1994). ‘La “Passion d’Augsbourg”, reflet d’un poème occitan du Xe siècle’. In Cerquiglini-Toulet, J. and Collet, O. (eds), Mélanges de philologie et de littérature médiévales offerts à Michel Burger. Geneva: Droz, 231–43. Hilty, G. (1995). ‘Les plus anciens monuments de la langue occitane’. In Rossi, L. (ed.), Cantarem d’aquestz trobadors: Studi occitanici in onore de Giuseppe Tavani. Alessandria, Italy: Edizioni dell’Orso, 25–45. Hinzelin, M.-O. (2004). ‘Die Stellung der Objektpronomina in frühen okzitanischen und katalanischen Texten im Vergleich’. Zeitschrift für Katalanistik 17: 111–29. Hinzelin, M.-O. (2007). Die Stellung der klitischen Objektpronomina in den romanischen Sprachen. Tübingen: Narr. Hinzelin, M.-O. (2011a). ‘La morphologie verbale du nord-occitan dans l’ensemble galloroman’. In Rieger, A. and Sumien, D. (eds), Actes du Neuvième Congrès International de l’AIEO. II. Aachen, Germany: Shaker, 719–32. Hinzelin, M.-O. (2011b). ‘Syncretism and suppletion in Gallo-Romance verb paradigms’. In Maiden et al., 287–310. Hinzelin, M.-O. (2012). ‘Verb morphology gone astray: syncretism patterns in GalloRomance’. In Gaglia, S. and Hinzelin, M.-O. (eds), Inflection and Word Formation in Romance Languages. Amsterdam / Philadelphia, PA: Benjamins, 55–81. Hinzelin, M.-O. and Kaiser, G. (2012). ‘Le paramètre du sujet nul dans les variétés dialectales de l’occitan et du francoprovençal’. In Barra-Jover, M., Brun-Trigaud, G., Dalbera, J.-P., Sauzet, P., and Scheer, T. (eds), Études de linguistique gallo-romane. Paris: Presses Universitaires de Vincennes, 247–60. Hirschbühler, P. (1990). ‘La légitimation de la construction V1 à sujet nul dans la prose et la vers en ancien français’. Revue québécoise de linguistique théorique et appliqué 19: 32–55. Hirschbühler, P. (1995). ‘Null subjects in V1 embedded clauses in Philippe de Vigneulles’ Cent Nouvelles Nouvelles’. In Battye and Roberts, 257–91. Hirschbühler, P. and Junker, J. (1988). ‘Remarques sur les sujets nuls en subordonnées en ancien et en moyen français’. Revue québécoise de linguistique théorique et appliquée 7: 63–84. Hock, H. H. and Joseph, B. (1996). Language History, Language Change and Language Relationship. An Introduction to Historical and Comparative Linguistics. Berlin: Mouton de Gruyter. Holmberg, A. (1986). ‘Word order and syntactic features in the Scandinavian languages and English’. University of Stockholm: thesis. Holmberg, A. (2002). ‘Prepositions and PPs in Zina Kotoko’. In Kappel Schmidt, B., Odden, D., and Holmberg, A. (eds), Some Aspects of the Grammar of Zina Kotoko. Munich: Lincom, 162–74. Holmberg, A. (2015). ‘Verb second’. In Kiss, T. and Alexiadou, A. (eds), Syntax—Theory and Analysis. An International Handbook of Contemporary Syntactic Research. Berlin: De Gruyter, 242–83. Holmberg, A. (forthcoming). ‘On the bottleneck hypothesis in Swedish’. In Woods, R. and Wolfe, S. (eds), Rethinking Verb Second. Oxford: Oxford University Press. Hooper, J. and Thompson, S. (1973). ‘On the applicability of root transformations’. Linguistic Inquiry 4: 465–97. Hornsby, D. (2006). Redefining Regional French. Koinéization and Dialect Levelling in Northern France. Oxford: Legenda. Hulk, A. and Pollock, J.-Y. (2001). Subject Inversion in Romance and the Theory of Universal Grammar. Oxford: Oxford: Oxford University Press.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
423
Hulk, A. and van Kemenade, A. (1995). ‘Verb-second, pro-drop, functional projections and language change’. In Battye and Roberts, 227–56. Hyman, L. (1979). ‘Phonology and noun structure’. In Hyman, L. (ed.), Aghem Grammatical Structure with Special Reference to Noun Classes, Tense-aspect and Focus Marking, 1–72. Hyman, L. (1984). ‘Auxiliary focus’. Studies in African Linguistics 15: 233–73. Iemmolo, G. (2010). ‘Topicality and differential object marking. Evidence from Romance and beyond’. Studies in Language 34: 239–72. Iglesias, O. (2015). ‘Le placement des pronoms clitiques dans les complexes verbaux en français au VIIe siècle: une approche idiolectale’. Travaux de linguistique 70: 91–120. Iordan, I. (1935). ‘Forme de conjugare mixtă în limba română’. Buletinul Institutului de Filologie Romînă Alexandru Philippide 2: 47–127. Israel, M. (1996). ‘Polarity sensitivity and lexical semantics’. Linguistics and Philosophy 19: 619–66. Jaberg, K. (1906). Ueber die assoziativen Erscheinungen in der Verbalflexion einer südostfranzösischen Dialektgruppe. Eine prinzipielle Untersuchung. Aarau, Switzerland: Sauerländer. Jackendoff, R. (1977) X’ Syntax: a study of phrase structure. Cambridge, MA: MIT Press. Jake, J. (1994). ‘Intrasentential code switching and pronouns: on the categorial status of functional elements’. Linguistics 32: 271–98. Jaquenod, F. (1931). Essai sur le verbe dans le patois de Sottens. Lausanne: Payot. Jaszczolt, K. and de Saussure, L. (eds) (2013) Time. Language, Cognition, and Reality, Oxford: Oxford University Press. Javanaud, P. (1981). ‘The vowel system of Lemosin’. University of Gothenburg: thesis. Jean-Christophe Pellets, J.-C., and René, R. (2009) Grammaire méthodique du français. Paris: Presses Universitaires de France. Jensen, F. (1986). The Syntax of Medieval Occitan. Tübingen: Niemeyer. Jensen, F. (1990). Old French and Comparative Gallo-Romance Syntax. Tübingen: Niemeyer. Jensen, J. (1994). Syntaxe de l’ancien occitan. Tübingen: Niemeyer. Jespersen, O. (1917). Negation in English and Other Languages. Copenhagen: A. F. Høst. Jespersen, O. ([1924] 1992) The Philosophy of Grammar. Reprinted with an introduction by James McCawley. Chicago: Chicago University Press. JFLS = Journal of French Language Studies. Johanisson, T. (1945). Hava och vara. Som tempusbildande hjälpverb i de nordiska språken. Lund: C. W. K. Gleerup / Leipzig: Harrassowitz. Johnson, D. (2008). ‘Rbrul’. http://www.danielezrajohnson.com/rbrul.html Johnson, D. (2009). ‘Getting off the Goldvarb standard: introducing Rbrul for mixed-effects variable rule analysis’. Language and Linguistics Compass 3: 359–83. Joly, A. (2013). ‘De la subordination en béarnais’. Studia UBB Philologia 4: 245–59. Jones, M. A. (1988). ‘Ergative and impersonal constructions in Sardinian’, paper presented at the Sixteenth Cambridge Romance Linguistics Seminar (January), Trinity Hall, Cambridge. Jones, M. A. (1993) Sardinian Syntax. London: Longman. Jones, M. C. (2018). ‘Does language loss follow a principled structural path? Evidence from Jersey Norman French’. JFLS 28: 399–429. Jones, M. C. (1998). Language Obsolescence and Revitalisation: Linguistic Change in Two Sociolinguistically Contrasting Welsh Communities. Oxford: Clarendon Press. Jones, M. C. (2001). Jersey Norman French: A Linguistic Study of an Obsolescent Dialect. Oxford: Blackwell.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
424
Jones, M. C. (2008). The Guernsey Norman French Translations of Thomas Martin: A Linguistic Study of an Unpublished Archive. Leuven: Peeters. Jones, M. C. (2011). ‘Diatopic variation and the study of regional French’. French Studies 65: 505–14. Jones, M. C. (2015). Variation and Change in Mainland and Insular Norman. A Study of Superstrate Influence. Leiden: Brill. Jones, M. C. (2017). ‘Norman forms—French norms: diaphasic variation and language contact’. In Tyne, H., Bilger, M., Cappeau, P., and Guerin, E. (eds), La Variation en question(s). Brussels: Lang, 111–34. Joseph, J. (1992). ‘The Gascon enunciative as syntactic solution’. In Laeufer, C. and Morgan, T. (eds), Theoretical Analyses in Romance Linguistics: Selected Papers from the 19th Linguistic Symposium on Romance Languages (LSRLIX), The Ohio State University, 21–29 April 1989. Amsterdam: Benjamins, 481–95. Josephson, F. (2006). ‘La langue poétique indo-européenne’. In Pinault, G.-J. and Petit, D. (eds) Actes du Colloque de travail de la Société des Études Indo-Européennes (Indogermanische Gesellschaft/Society for Indo-European Studies), Paris, 22–24 octobre 2003. Louvain, Belgium / Paris: Peeters, 141–56. Jud, K. (1946–49). ‘Altfrz. estuet; bündnerrom. stuver, stuvair’. Vox Romanica 9: 29–56. Kahn, F. (1954). Le Système des temps de l’indicatif chez un Parisien et chez une Bâloise. Geneva: Droz. Kailuweit, R. (2015). ‘BE or HAVE in contemporary Standard French—residua of semantic motivation’. In Kailuweit, R. and Rosemeyer, M. (eds), Auxiliary Selection Revisited: Gradience and Gradualness. Berlin / Munich / Boston, MA: de Gruyter, 249–76. Kaiser, G. (2002a). ‘Die Verb-Zweit-Stellung im Raetoromanischen. Ein typologischer Vergleich’. Ladinia 26–27: 313–34. Kaiser, G. (2002b). Verbstellung und Verbstellungswandel in den Romanischen Sprachen. Tübingen: Niemeyer. Karenova, K. (2006). ‘Théories linguistiques et phénomènes discursifs: interface complexe en gascon’. In Gurski, C. (ed.), Actes de l’Atelier bilingue en linguistique théorique 2005. University of Western Ontario, 1–12. Karenova, K. (2008). ‘Variation à la périphérie gauche du gascon’. Revue des étudiants en linguistique du Québec 1: 46–60. Karttunen, L. (1977). ‘The syntax and semantics of questions’. Linguistics and Philosophy 1: 3–44. Kato, M. and Martins, A. M. (2016). ‘European Portuguese and Brazilian Portuguese: an overview on word order’. In Wetzels, L., Menuzzi, S., and Costa, J. (eds), The Handbook of Portuguese Linguistics. Malden, MA: Wiley-Blackwell, 15–40. Kavitskaya, D. (2017). ‘Compensatory lengthening and structure preservation revisited yet again’. In Bowern, C., Horn, L., and Zanuttini, R. (eds), On Looking into Words (and Beyond). Berlin: Language Science Press, 41–58. Kayne, R. (1975). French Syntax: The Transformational Cycle. Cambridge, MA: MIT Press. Kayne, R. (1980). ‘Vers une solution d’un problème grammatical: *Je l’ai voulu lire, j’ai tout voulu lire’. Langue française 46: 32–40. Kayne, R. (1989a). ‘Facets of Romance past-participle agreement’. In Benincà, P. (ed.) Dialect Variation and the Theory of Grammar. Dordrecht: Foris, 85–103. Kayne, R. (1989b). ‘Null subjects and clitic climbing’. In Jaeggli, O. and Safir, K. (eds), The Null Subject Parameter. Dordrecht / Boston, MA: Kluwer, 239–61. Kayne, R. (1993). ‘Toward a modular theory of auxiliary selection’. Studia Linguistica 47: 3–31.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
425
Kayne, R. (2000). Parameters and Universals. Oxford: Oxford: Oxford University Press. Kayne, R. (2005). Movement and Silence. Oxford: Oxford: Oxford University Press. Kayne, R. and Pollock, J.-Y. (2012). ‘Toward an analysis of French hyper-complex inversion’. In Brugé, L., Cardinaletti, A., Giusti, G., Munaro, N., and Poletto, P. (eds), Functional Heads. Oxford: Oxford: Oxford University Press, 150–67. Keller, O. (1928). La Flexion du verbe dans le patois genevois. Geneva: Olschki. King, R. and Butler, G. (2005). ‘Les Franco-Terreneuviens et le franco-terreneuvien’. In Valdman, A., Auger, J., and Piston-Hatten, D. (eds), Le Français en Amérique du Nord. Sainte-Foy, Quebec: Presses de l’Université de Laval, 169–85. King, R. and Nadasdi, T. (2003). ‘Back to the future in Acadian French’. JFLS 13: 323–37. King, R. and Nadasdi, T. (2005). ‘Deux auxiliaires qui voulaient mourir en français acadien’. In Brasseur, P. and Falkert, A. (eds), Français d’Amérique: approches morphosyntaxiques. Paris: L’Harmattan, 103–11. Kiparsky, P. (1973). ‘Productivity in phonology’. In Kenstowitz, M. and Kisseberth, C. (eds), Issues in Phonological Theory. The Hague: Mouton, 169–76. Kiparsky, P. (1995). ‘Indo-European Origins of Germanic Syntax’. In Battye and Roberts, 140–69. Kiparsky, P. (1997). ‘The rise of positional licensing’. In Kemenade, A. van and Vincent, N. (eds), Parameters of Morphosyntactic Change. Cambridge: Cambridge University Press, 460–94. Kiparsky, P. (2008). ‘Universals constrain change: change results in typological generalizations’. In Good, J. (ed.), Linguistic Universals and Language Change. Oxford: Oxford University Press, 23–53. Kiparsky, P. and Condoravdi, C. (2006). ‘Tracking Jespersen’s Cycle’. In Janse, M., Joseph, B. D., and Ralli, A. (eds), Proceedings of the 2nd International Conference of Modern Greek Dialects and Linguistic Theory. Mytilene, Greece: University of Patras Press, 172–97. Klævik-Pettersen, E. (2018). ‘Inversion, V-to-C, and verb-second: an investigation into the syntax and word order of Old French and Late Latin’. University of Oslo: thesis. Klein, W. (1994). Time in Language. London: Routledge. Knaus, V. and Nadasdi, T. (2001). ‘Être ou ne pas être en immersion’. La Revue canadienne des langues vivantes 58: 287–306. Koschwitz, E. (1913). Les Plus Anciens Monuments de la langue française publiés pour les cours universitaires. Leipzig: Reisland. Kremnitz, G. (1974). Versuch zur Kodifizierung des Okzitanischen seit dem 19. Jahrhundert und ihre Annahme durch die Sprecher. Tübingen: Narr. Kristol, A. (2009). ‘Syntaxe variationnelle du clitique sujet en francoprovençal valaisan contemporain: un modèle pour la diachronie du galloroman septentrional?’. Travaux de linguistique 59: 47–76. Kristol, A. (2016). ‘Francoprovençal’. In Ledgeway and Maiden, 350–62. Kukenheim, L. (1967). Grammaire historique de la langue française: les parties du discours. Leiden: Universitaire Pers. Kunert, H. (2003). ‘L’ordine degli elementi della frase in occitano antico’. Romanische Forschungen 115: 194–209. L’Huillier, M. (1999). Advanced French Grammar. Oxford: Oxford University Press. La Fauci, N. (1988). Oggetti e soggetti nella formazione della morfosintassi romanza. Pisa: Giardini. La Fauci, N. (1989). ‘Ausiliari perfettivi e accordo del participio passato in italiano e francese’. In Foresti, F., Rizzi, E., and Benedini, P. (eds), L’italiano tra le lingue romanze. Atti del X Congresso Internazionale di Studi della Società Linguistica Italiana. Bologna 25–27 settembre 1986. Rome: Bulzoni, 213–42.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
426
La Fauci, N. (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. La Fauci, N. (2011). Relazioni e differenze: questioni di linguistica razionale. Palermo: Sellerio. Labeau, E. (2015). ‘Il était une fois le passé simple . . . ’. JFLS 25: 165–87. Labelle, M. (1992). ‘Change of state and valency’. Journal of Linguistics 28: 375–414. Labelle, M. (2007). ‘Clausal architecture in Early Old French’. Lingua 117: 289–316. Labelle, M. (2016). ‘Participle fronting and clause structure in Old and Middle French’. In Tortora, C., den Dikken, M., Montoya, I. L., and O’Neill, T. (eds), Romance Languages and Linguistic Theory. Amsterdam: Benjamins, 213–32. Labelle, M. and Hirschbühler, P. (2005). ‘Changes in clausal structure and the position of clitics in old French’. In Batllori, M., Hernanz, M. Ll., Picallo, C., and Roca, F. (eds), Grammaticalization and Parametric Variation. Oxford: Oxford University Press, 149–78. Labelle, M. and Hirschbühler, P. (2017). ‘Leftward Stylistic Displacement (LSD) in medieval French’. In Mathieu, E. and Truswell, R. (eds), From Micro-change to Macro-change. Oxford: Oxford University Press. Labelle, M. and Hirschbühler, P. (2018). ‘Topic and focus in old French V1 and V2 structures’. Canadian Journal of Linguistics/Revue canadienne de linguistique 63: 264–87. Labov, W. (1966). The Social Stratification of English in New York City. Washington, DC: Center for Applied Linguistics. Labov, W. (1972). Sociolinguistic Patterns. Philadelphia, PA: University of Pennsylvania Press. Labov, W. (1994). Principles of Linguistic Change I: Internal Factors. Oxford: Blackwell. Labov, W. (2001). Principles of Linguistic Change II: Social Factors. Oxford: Blackwell. Labov, W. (2010). Principles of Linguistic Change III: Cognitive and Cultural Factors. Oxford: Blackwell. Lafitte, J. (2005), Situation sociolinguistique et écriture du gascon d’aujourd’hui. Universite Rennes 2: doctoral dissertation. Lafitte, J. (2019). ‘Le que énonciatif gascon dans l’histoire’. Enstitut biarnes e gascoun 1–7. https://www.institut-bearnaisgascon.com/le-que-du-gascon Lafont, R. (1964). ‘Remarques sur l’emploi de e introductif du verbe principal en ancien occitan’. Revue de linguistique romane 28: 34–41. Lafont, R. (1967). La Phrase occitane: essai d’analyse systématique. Paris: Presses Universitaires de France. Lagueux, P.-A. (2005). ‘Caractéristiques morpho-syntaxiques du français québécois’. In Brasseur, P. and Falkert, A. (eds), Français d’Amérique: approches morphosyntaxiques. Paris: L’Harmattan, 57–69. Lahousse, K. (2003). ‘The distribution of postverbal nominal subjects in French’. Université de Paris 8: thesis. Laka, I. (1990). ‘On the nature of functional categories and projections’. MIT: thesis. Lalanne, T. (1951). ‘La limite nord du gascon’. Le Français moderne 19: 135–52. Lamiroy, B. and De Mulder, W. (2011). ‘Degrees of grammaticalization across languages’. In Harrog, H. and Heine, B. (eds), The Oxford Handbook of Grammaticalization. Oxford: Oxford University Press, 302–17. Langacker, R. (1987). The Foundations of Cognitive Gramm. I: Theoretical Prerequisites. Stanford, CA: Stanford University Press. Lanly, A. (1971). ‘Sur des formes occitanes de conditionnel sans -r-’. In Cluzel, I. and Pirot, F. (eds), Mélanges de philologie romane dédiés à la mémoire de Jean Boutière (1899–1967). Liège: Soledi, 795–800.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
427
Larrivée, P. (2011). ‘The role of pragmatics in grammatical change: the case of French preverbal non’. Journal of Pragmatics 43: 1987–96. Larrivée, P. (2014). ‘The syntax of pragmatics: the case of presuppositional negatives’. Syntaxe et Sémantique 15: 115–37. Lartigue, P. (1998), Les racines de la langue gasconne. Monein, France: Princi negue. Lartigue, P. (2006). ‘Le gascon “noir” des Landes de Gascogne’. In Boyrie-Fenié, A.-B. and Goulaze, H. (eds), Un pays dans sa langue: le gascon dans l’ensemble d’Oc. Sabres, France: Parc Naturel des Landes de Gascogne, 23–37. Lass, R. (2000). ‘Perodization in English and the concept “Middle” ’. In Taavitsainen, I., Pahta, P., and Rissanen, M. (eds), Placing Middle English in Context. Berlin / New York: Mouton, 7–42. Lausberg, H. (1966). Lingüística románica, II: Morfología. Madrid: Gredos. Lavalade, Y. (1986). L’Accentuation occitane. Limoges: La Clau lemosina. Lavalade, Y. (1987). La Conjugaison occitane (Limousin). Limoges: La Clau Lemosina. Law, D. (2014). Language Contact, Inherited Similarity and Social Difference: The Story of Linguistic Interaction in the Maya Lowlands. Amsterdam / Philadelphia, PA: Benjamins. Le Gallo à Fulgence Bienvenüe. http://welcome2bienvenue.free.fr/LV_LYCEE Le Gallo, la langue romane de Bretagne. http://www.cndp.fr/crdp-rennes/crdp/crdp_dos siers/dossiers/gallo/methode/12.htm Le Gallo, la langue romane de Bretagne. http://www.cndp.fr/crdp-rennes/crdp Le Maistre, F. (1966). Dictionnaire Jersiais–Français. Jersey: Don Balleine. Ledgeway, A. (2003). ‘L’estensione dell’ausiliare perfettivo avere nell’antico napoletano: intransitività scissa condizionata da fattori modali’. Archivio glottologico italiano 88: 27–71. Ledgeway, A. (2007). ‘Old Neapolitan word order: some initial observations’. In Lepschy, A. L. and Tosi, A. (eds), Histories and Dictionaries of the Languages of Italy. Ravenna: Longo, 121–49. Ledgeway, A. (2008). ‘Satisfying V2 in early Romance: merge vs move’. Journal of Linguistics 44: 437–70. Ledgeway, A. (2009). ‘Aspetti della sintassi della periferia sinistra del cosentino’. In Pescarini, D. (ed.), Studi sui dialetti della Calabria. Padua: Unipress, 3–24. Ledgeway, A. (2012). From Latin to Romance. Morphosyntactic Typology and Change. Oxford: Oxford University Press. Ledgeway, A. (2014a). ‘La morphologie flexionnelle de l’impératif des dialectes de l’Italie méridionale: la distribution de la métaphonie’. In Léonard, J. L. (ed.), Morphologie flexionnelle et dialectologie romane: typologie(s) et modélisation(s). Mémoires de la Société linguistique de Paris XII. Louvain, Belgium: Peeters, 13–33. Ledgeway, A. (2014b). ‘Romance auxiliary selection in light of Romanian evidence’. In Pană Dindelegan, G., Zafiu, R., and Dragomirescu, A. (eds), Diachronic Variation in Romanian. Newcastle upon Tyne: Cambridge Scholars Publishing, 3–35. Ledgeway, A. (2015). ‘Parallels in Romance nominal and clausal microvariation’. Revue roumaine de linguistique 60: 105–27. Ledgeway, A. (2016a). ‘Functional categories’. In Ledgeway and Maiden, 761–71. Ledgeway, A. (2016b). ‘Italian, Tuscan, Corsican’. In Ledgeway and Maiden, 206–27. Ledgeway, A. (2017a). ‘Marking presuppositional negation in the dialects of southern Italy’. In Cruschina, S., Hartmann, K., and Remberger, E.-M. (eds), Studies on Negation: Syntax, Semantics and Variation. Vienna: Vienna University Press, 105–30. Ledgeway, A. (2017b). ‘Late Latin verb second: the sentential word order of the Itinerarium Egeriae’. Catalan Journal of Linguistics 16: 163–216.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
428
Ledgeway, A. (2019). ‘Word order in the Histoire ancienne jusqu’à César’. Presentation at the Secrets of Success Workshop, Oslo, Norway. Ledgeway, A. (forthcoming). ‘The causative construction in the dialects of southern Italy and the phonology–syntax interface’. In Bellucci, G., Franco, L., and Lorusso, P. (eds), Linguistic Variation: Structure and Interpretation. Berlin: De Gruyter. Ledgeway, A. (forthcoming). ‘From Latin to Romance Syntax: the great leap’. In Crisma, P. and Longobardi, G. (eds), The Oxford Handbook of Diachronic and Historical Linguistics. Oxford: Oxford University Press. Ledgeway, A. and Lombardi, A. (2005). ‘Verb movement, adverbs and clitic positions in Romance’. Probus 17: 79–114. Ledgeway, A. and Lombardi, A. (2014). ‘The development of the southern subjunctive: morphological loss and syntactic gain’. In Benincà, P., Ledgeway, A., and Vincent, N. (eds), Diachrony and Dialects. Grammatical Change in the Dialects of Italy. Oxford: Oxford University Press, 25–47. Ledgeway, A. and Maiden, M. (eds) (2016). The Oxford Guide to the Romance Languages. Oxford: Oxford University Press. Legendre, G. (2007). ‘Optimizing auxiliary selection in Romance’. In Aranovich, R. (ed.), Split Auxiliary Systems: A Cross-Linguistic Perspective. Amsterdam: Benjamins, 145–80. Legendre, G. (2017). ‘Auxiliaries’. In Dufter, A. and Stark, E. (eds), Manual of Romance Morphosyntax and Syntax. Berlin: De Gruyter, 272–98. Legendre, G. and Sorace, A. (2003). ‘Auxiliaires et intransitivité en français et dans les langues romanes’. In Godard, D. (ed.), Les Langues romanes: problèmes de la phrase Simple. Paris: CNRS, 185–234. Lehmann, C. (1995). Thoughts on Grammaticalization. Munich: LINCOM Europa. Lehmann, C. (2002a). ‘New reflections on grammaticalization and lexicalization’. In Wischer, I. and Diewald, G. (eds), New Reflections on Grammaticalization. Amsterdam / Philadelphia, PA: Benjamins, 1–18. Lehmann, C. (2002b). Thoughts on Grammaticalization. Erfurt, Germany: Seminar für Sprachwissenschaft der Universität Erfurt. Lehmann, C. (2019). ‘Complex spatial prepositions from Latin to Castilian’. Revue romane 54: 93–125. Lemieux, M. and Dupuis, F. (1995). ‘The locus of verb movement in non-asymmetric verbsecond languages: the case of middle French’. In Roberts and Battye, 7–42. Lepelley, R. (1974). Le Parler normand du Val de Saire (Manche). Phonétique Morphologie Syntaxe Vocabulaire de la vie rurale. Caen, France: Musée de Normandie. Lepelley, R. (1999). La Normandie dialectale. Petite encyclopédie des langages et mots régionaux de la province de Normandie et des îles anglo-normandes. Condé-surNoireau, France: Presses Universitaires de Caen. Lesage, R. and Gagnon, S. (1993). ‘Futur simple et futur périphrastique dans la presse québécoise’. In Crochetière, A., Boulanger, J-C., and Ouellon, C. (eds), Les Langues menacées: actes du XVe Congrès international des linguistes. Sainte-Foy, Quebec: Les Presses de l’Université Laval, 367–70. Lespy, V. (1858). Grammaire béarnaise suivie d’un vocabulaire français-béarnais. Pau, France: Typographie et Lithographie Veronese. Lespy, V. (1876). Proverbes du pays de Béarn. Enigmes et contes populaires. Paris: Maisonneuve. Levitt, J. (1979). ‘Variations in the use of the auxiliairies avoir and être in modern French’. Geolinguistics 5: 25–32. Libert, A. (2013). Adpositions and Other Parts of Speech. Bern: Peter Lang.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
429
Liddicoat, A. (1994). A Grammar of the Norman French of the Channel Islands. The Dialects of Jersey and Sark. Berlin: Mouton de Gruyter. Liddicoat, A. (2011). A Grammar of the Norman French of the Channel Islands: The dialects of Jersey and Sark. Berlin: De Gruyter. Lieutard, H. (2004). Phonologie et morphologie du parlar occitan de Graulhet (Tarn). Montpellier, France: CEO. Lightfoot, D. (1979). Principles of Diachronic Syntax. Cambridge: Cambridge University Press. Lobo, M. and Martins, A. M. (2017). ‘Subjects’. In Dufter, A. and Stark, E. (eds), Manual of Romance Morphosyntax and Syntax. Berlin / Boston, MA: De Gruyter, 27–88. Lodge, R. A. (1993). French: From Dialect to Standard. London: Routledge. Lombardi, A. (2007). ‘Posizione dei clitici e ordine dei constituenti della lingua sarda medievale’. In Bentley, D. and Ledgeway, A. (eds), Sui dialetti italoromanzi: saggi in onore di Nigel B. Vincent. King’s Lynn, Norfolk: Biddles, 133–48. Lombardi, A. and Middleton, R. (2004). ‘Alcune osservazioni sull’ordine delle parole negli antichi volgari italiani’. In Dardano, M. and Frenguelli, G. (eds), Sint Ant: la sintassi dell’italiano antico. Atti del convegno internazionale di studi. Rome: Arcane, 553–82. Loporcaro, M. (1998). Sintassi comparata dell’accordo participiale romanzo. Turin: Rosenberg and Sellier. Loporcaro, M. (2007). ‘On triple auxiliation in Romance’, Linguistics 45: 173–222. Loporcaro, M. (2016). ‘Auxiliary selection and participial agreement.’ In Ledgeway and Maiden, 802–18. Loporcaro, M. and Putzu, I. (2013). ‘Variation in auxiliary selection, syntactic change, and the internal classification of Campidanese Sardinian’. In Paulis, G., Pinto, I., and Putzu, I. (eds), Repertorio plurilingue e variazione linguistica a Cagliari. Milan: Angeli, 200–44. Luchaire, A. (1877). De Lingua Aquitanica. Hachette: Paris. Luchaire, A. (1879). Etudes sur les idiomes pyrénéens de la région française. Maisonneuve: Paris. Lusignan, S. (2004). La Langue des rois au Moyen Âge: le français en France et en Angleterre. Paris: Presses universitaires de France. Lyons, C. (1999). Definiteness. Cambridge: Cambridge University Press. Lyons, J. (1968). Introduction to Theoretical Linguistics. Cambridge: Cambridge Univeristy Press. Mackenzie, I. (2006). Unaccusative Verbs in Romance Languages. Basingstoke: Palgrave Macmillan. MacLean, S. (2018). ‘Waltharius: treasure, revenge and kingship in the Ottonian Wild West’. In Gilbert, K. and White, S. (eds), Emotion, Violence, Vengeance and Law in the Middle Ages: Essays in Honour of William Ian Miller. Leiden: Brill, 225–51. MacSwan, J. (2005). ‘Codeswitching and generative grammar: a critique of the MLF model and some remarks on “modified minimalism” ’. Bilingualism: Language and Cognition 8: 1–22 Maiden, M. (1994). A Linguistic History of Italian. London: Longman. Maiden, M. (2000). ‘Di un cambiamento intramorfologico: origini del tipo dissi dicesti ecc., nell’italoromanzo’. Archivio glottologico italiano 85: 137–71. Maiden, M. (2003). ‘Verb augments and meaninglessness in Romance morphology’. Studi di grammatica italiana 22: 1–61. Maiden, M. (2008). ‘Effects of word-formation processes in Italian. Reflections on Maria Grossmann and Franz Rainer (eds) 2004 La formazione delle parole in italiano. Tübingen: Narr’. Rivista di linguistica 20: 375–400.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
430
Maiden, M. (2009a). ‘From pure phonology to pure morphology. The reshaping of the Romance verb’. Recherches linguistiques de Vincennes 38: 45–82. Maiden, M. (2009b). ‘Un capitolo di morfologia storica del romeno: preterito e tempi affini’. ZRP 125: 273–309. Maiden, M. (2011a). ‘Morphomes and “stress-conditioned allomorphy” in Romansh’. In Maiden et al., 36–50. Maiden, M. (2011b). ‘Instabilità e variazione dialettale nella storia di un verbo trascurato: dovere nelle varietà italoromanze e “retoromanze” ’. In Marcato, G. (ed.) Le nuove forme del dialetto. Padua, Italy: Unipress, 97–104. Maiden, M. (2016a). ‘Morphomes’. In Ledgeway and Maiden, 708–21. Maiden, M. (2016b). ‘Some lessons from history: morphomes in diachrony’. In Luís, A. and Bermúdez-Otero, R. (eds), The Morphome Debate. Oxford: Oxford University Press, 33–63. Maiden, M. (2016c). ‘Ambiguity in Romanian word-structure. The structure of plurals in . . . uri’. Revue roumaine de linguistique 61: 3–12. Maiden, M. (2017). ‘Italo-Romance metaphony and the Tuscan diphthongs’. Transactions of the Philological Society 114: 198–232. Maiden, M. (2018a). The Romance Verb. Morphomic Structure and Diachrony. Oxford: Oxford University Press. Maiden, M. (2018b). ‘New thoughts on an old puzzle. The Italian alternation type dissi, dicesti, feci, facesti, etc.’. Revue romane 52: 217–60. Maiden, M. and Robustelli, C. (2007). A Reference Grammar of Modern Italian. London: Hodder Arnold. Maiden, M., Smith, J. C., and Ledgeway, A. (eds) (2011). The Cambridge History of the Romance Languages I: Structures. Cambridge: Cambridge University Press. Maiden, M., Smith, J. C., Goldbach, M., and Hinzelin, M.-O. (eds) (2011). Morphological Autonomy. Oxford: Oxford University Press. Maldonado, C. (1999). ‘Discurso directo y discurso indirecto’. In Bosque, I. and Demonte, V. (eds), Gramática descriptiva de la lengua española. Madrid: Espasa, 3549–95. Maling, J. (1990). ‘Inversion in embedded clauses in modern icelandic’. In Maling, J. and Zaenen, A. (eds), Modern Icelandic Syntax. New York: Academic Press, 71–91. Manente, M. (2008). ‘L’Aspect, les auxiliaires “être” et “avoir” et l’hypothèse inaccusative dans une perspective comparative français/italien’. Universities of Venice and Paris VIII: thesis. Manzano, F. (2018). ‘Romance dialectology: from the nineteenth century to the era of sociolinguistics’. In Ayres-Bennett, W. and Carruthers, J. (eds), Manual of Romance Sociolinguistics. Berlin: Mouton, 134–72. Manzini, M. R. (1996). ‘Il congiuntivo’. In Benincà, P., Cinque, G., De Mauro, T., and Vincent, N. (eds), Italiano e dialetti nel tempo. Saggi di grammatica per Giulio C. Lepschy. Rome: Bulzoni, 175–89. Manzini, M. R. and Savoia, L. (2003). ‘The nature of complementizers’. Rivista di grammatica generativa 28: 87–110. Manzini, M. R. and Savoia, L. (2005). I dialetti italiani e romanci: morfosintassi generativa. Alessandria, Italy: Edizioni dell’Orso. Manzini, M. R. and Savoia, L. (2011). Grammatical Categories. Cambridge: Cambridge University Press. Marchello-Nizia, C. (1980). La Langue française aux IVe et Ve siècles. Paris: Nathan. Marchello-Nizia, C. (1985). Dire le vrai: l’adverbe ‘si’ en français médiéval. Geneva: Droz. Marchello-Nizia, C. (1995). L’Evolution du français: ordre des mots, démonstratifs, accent tonique. Paris: Armand Colin.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
431
Marchello-Nizia, C. (1999). Le Français en diachronie: douze siècles d’évolution. Paris: Ophrys. Marchese, L. (1983). ‘On assertive focus and the inherent focus nature of negatives and imperatives: evidence from Kru’. Journal of African Languages and Linguistics 5: 115–29. Marcus, N. (2010). ‘The Gascon énonciatif system: past, present, and future. a study of language contact, change, endangerment, and maintenance’. University of California, Berkeley: thesis. Marie, É. (2012). Dictionnaire normand-français. Bayeux: OREP. Marshall, M. (1984). The Dialect of Notre-Dame-de-Sanilhac. Saratoga, NY: ANMA Libri. Martin, G. and Moulin, B. (2007). Grammaire provençale et cartes linguistiques. Aix-enProvence: Comitat Sestian d’Estudis Occitans. Martineau, F. (1991). ‘Clitic climbing in infinitival constructions of middle French’. In Wanner, D. and Kibbee, D. (eds), New Analyses in Romance Linguistics: Selected Papers from the VIII Linguistic Symposium on Romance Languages, Urbana-Champaign, April 7–9, 1988. Amsterdam / Philadelphia, PA: Benjamins, 235–51. Martineau, F. (2009). ‘A distance de Paris: usages linguistiques en France et en Nouvelle France à l’époque classique’. In Aquino-Weber, D., Cotelli, S., and Kristol, A. (eds), Sociolinguistique historique du domaine gallo-roman: enjeux et méthodologies. Bern: Lang, 221–42. Martinet, A. (1969). Le Français sans fard. Paris: Presses Universitaires de France. Massourre, J.-L. (2005). Le Gascon. ‘Lengatge estranh’. Pech de Rayssac, France: Jean-Louis Massourre. Massourre, J.-L. (2012). Le Gascon, les mots et le système. Geneva: Champion / Slatkine. Mateu Fontanals, J. (2014). ‘La selecció de l’auxiliar en els verbs intransitius del català antic: alguns aspectes descriptius i explicatius’. Caplletra: Revista Internacional de Filologia 38: 211–40. Mathieu, E. (2006). ‘Stylistic fronting in Old French’. Probus 18: 219–66. Mathieu, E. (2009). ‘On the Germanic properties of old French’. In Crisma, P. and Longobardi, G. (eds), Historical Syntax and Linguistic Theory. Oxford: Oxford University Press, 344–57. Mathieu, E. (2012). ‘The left periphery in Old French’. In Arteaga, D. (ed.), Research in Old French: The State of the Art. Dordrecht: Springer, 327–50. Matos, G. (2013). ‘Quotative inversion in peninsular Portuguese and Spanish, and in English’. Catalan Journal of Linguistics 12: 111–30. Matras, Y. (2007). ‘The borrowability of grammatical categories’. In Matras, Y. and Sakel, J. (eds), Grammatical Borrowing in Cross-linguistic Perspective. Berlin: Mouton de Gruyter, 31–74. Matras, Y. (2009). Language Contact. Cambridge: Cambridge University Press. Maupas, C. (1607). Grammaire et syntaxe françoise. Paris: Adrian Bacot. McLaughlin, M. (forthcoming). La Presse française historique: Histoire d’un genre et l’histoire de la langue. Paris: Garnier. McMahon, A. (1994). Understanding Language Change. Cambridge: Cambridge University Press. McWhorter, J. (2002). ‘What happened to English?’ Diachronica 19: 217–72. Medina Granda, R. M. (1999). Polaridad negativa en occitano antiguo: elementos de comparación con otros romances medievales. Oviedo, Spain: Universidad de Oviedo. Medina Granda, R. M. (2007). ‘Occitano antiguo ge(n)s: su ausencia en ciertos contextos negativos’. In Floricic, F. (ed.), La Négation dans les langues romanes. Amsterdam: Benjamins, 1–27.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
432
Meillet, A. (1909/1982). Linguistique historique et linguistique générale. Paris: Champion. Meklenborg Salvesen, C. (2009). ‘Le CP interrogatif—une étude diachronique du français’. University of Oslo: thesis. Meklenborg Salvesen, C. (2011). ‘Stylistic fronting and remnant movement in Old French’. In Berns, J. Jacobs, H., and Scheer, T. (eds), Romance Languages and Linguistic Theory 2009. Amsterdam: Benjamins, 323–42. Meklenborg Salvesen, C. (2013). ‘Topics and the left periphery: a comparison of Old French and modern Germanic’. In Lohndal, T. (ed.), In Search of Universal Grammar: from Old Norse to Zoque Amsterdam: Benjamins, 131–72. Meklenborg Salvesen, C. (2014). ‘Le complémenteur que et la périphérie gauche. Analyse diachronique’. Syntaxe et Sémantique 15: 47–80. Meklenborg Salvesen, C. (forthcoming). ‘Resumptive particles and verb second’. In Woods, R. and Wolfe, S (eds), Rethinking Verb Second. Oxford: Oxford University Press. Ménard, P. (1988). Syntaxe de l’ancien français. Bordeaux: Bière. Ménard, P. (1994). Syntaxe de l’ancien français. Bordeaux: Broché. Mensching, G. and Remberger, E.-M. (2016). ‘Sardinian’. In Ledgeway and Maiden, 270–91. Meyer-Lübke, W. (1895). Grammaire des langues romanes. Morphologie. Paris: Welter. Meyer, P. (1880). ‘Les troisièmes personnes du pluriel en provençal’, Romania 9: 192–215. Millardet, G. (1923). Linguistique et dialectologie romanes. Problèmes et méthodes. Montpellier / Paris: Société des langues romanes / Champion. Milroy, L. and Gordon, M. (2003). Sociolinguistics: Method and Interpretation. Oxford: Blackwell. Miremont, P. (1976). La Syntaxe occitane du Périgord. Aurillac, France: Gerbert. Mithun, M. (2012). ‘Morphologies in contact: form, meaning, and use in the grammar of reference’. In Stolz, T., Vanhove, M., Otsuka, H., and Urdzu, A. (eds), Morphologies in Contact. Berlin: Akademia, 15–36. Moignet, G. (1973). Grammaire de l’ancien français: morphologie—syntaxe. Paris: Klincksieck. Moll, F. de B. (1952). Gramática histórica catalana. Madrid: Editorial Gredos. Monachesi, P. (2005). The Verbal Complex in Romance: A Case Study in Grammatical Interfaces. Oxford: Oxford University Press. Mooney, D. (2016). Southern Regional French: A Linguistic Analysis of Language and Dialect Contact. Oxford: Legenda. Mooney, D. (2018). ‘Quantitative approaches for modelling variation and change: a case study of sociophonetic data from Occitan’. In Ayres-Bennett, W. and Carruthers, J. (eds), Manual of Romance Sociolinguistics. Berlin: Mouton, 59–90. Mooney, D. (2019). ‘Phonetic transfer in language contact: evidence for equivalence classification in the mid-vowels of Occitan-French bilinguals’. Journal of the International Phonetic Association 49: 53–85. Moreux, B. (2004). ‘Béarnais and Gascon today’. International Journal of the Sociology of Language 169: 25–62. Moreux, B. and Moreux, C. (1989). ‘La transmission du béarnaise en milieu rural aujourd’hui’. In Moreux, B. (ed.), Langues en Béarn. Toulouse: Presses Universitaires du Mirail, 235–56. Morin, A. (2005). ‘Syntaxe de la particule que en gascon’. Linguistica occitana 4: 60–7. Morin, A. (2006). ‘On the syntax of clause type particles: evidence from Gascon, Innu and Quebec French’. Concordia University, Montreal: thesis. Morin, A. (2008). ‘The syntax of Gascon clause-type particles’. Linguistica Atlantica 29: 137–55.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
433
Mortensen, L. (2018). ‘Latin as vernacular: critical mass and the “librarization” of New Book Languages’. In Kössinger, N., Krotz, E., Müller, S., and Rychterová, P. (eds), Anfangsgeschichten Origin Stories: der Beginn volkssprachiger Schriftlichkeit in komparatistischer Perspektive. The Rise of Vernacular Literacy in a Comparative Perspective. Paderborn, Germany: Fink, 71–90. Muller, C. (2005). ‘Evolution de la syntaxe sujet-verbe et réanalyse’. Verbum 25: 481–93. Muller, C. (2009). ‘Major constituent order, information packaging and narrative structure in two Middle French texts’. In Dufter, A. and Daniel, J. (eds), Focus and Background in the Romance Language. Amsterdam / Philadelphia, PA: Benjamins, 239–80. Munaro, N. (2004). ‘Computational puzzles of conditional clause preposing’. In Di Sciullo, A. M. and Delmonte, R. (eds), UG and External Systems. Amsterdam: Benjamins, 73–94. Munaro, N. (2010). ‘Toward a hierarchy of clause types’. In Benincà, P. and Munaro, N. (eds), Mapping the Left Periphery. The Cartography of Syntactic Structures, V. Oxford: Oxford University Press, 125–62. Muysken, P. (1981). ‘Halfway between Quechua and Spanish: the case for relexification’. In Highfield, A. and Valdman, A. (eds), Historicity and Variation in Creole Studies. Ann Arbor, MI: Karma, 52–78. Muysken, P. (2000). Bilingual Speech. A Typology of Code-Mixing. Cambridge: Cambridge University Press. Myers-Scotton, C. (2002). Contact Linguistics. Oxford: Oxford University Press. Nelde, P. H. (1984). ‘Aspects of linguistic determination along the Germanic–Romance linguistic boundary’. Journal of Multilingual and Multicultural Development 5: 217–24. Neumann-Holzschuh, I., Brasseur, P., and Wiesmath, R. (2005). ‘Le français acadien au Canada et en Louisiane’. In Valdman, A., Auger, J., and Piston-Hatten, D. (eds), Le Français en Amérique du Nord. Sainte-Foy, Quebec: Presses de l’Université de Laval, 479–503. Nichols, J. (1986). ‘Head-marking and dependent-marking grammar’. Language 62: 56–119. Nicolet, N. (1929). Der Dialekt des Antronatales. Lautlehre, Formenlehre, Texte, Glossar. Halle, Germany: Niemeyer. Niederländer, J. (1900). ‘Die Mundart von Namur’. ZRP 24: 1–32; 251–309. Nurse, D. (2006). ‘Focus in Bantu: verbal morphology and function’. ZAS Papers in Linguistics 43: 189–207. Nyrop, K. (1930). Grammaire historique de la langue française. Copenhagen: Gyldendal. Nyrop, K. (1960). Grammaire historique de la langue française, II. Copenhagen: Gyldenhal. O’Neill, P. (2014). ‘The morphome in constructive and abstractive models of morphology’. Morphology 24: 25–70. O’Rourke, B., Pujolar, J., and Ramallo, F. (2015). ‘New speakers of minority languages: the challenging opportunity—foreword’. International Journal of the Sociology of Language 231: 1–20. OED = Oxford English Dictionary. Ohala, J. (1989). ‘Sound change is drawn from a pool of synchronic variation’. In Breivik, L. E., and Jahr, E. (eds), Language Change: Contributions to the Study of Its Causes. Berlin: Mouton de Gruyter, 173–98. Öhl, P. (2010). ‘Formal and functional constraints on constituent order and their universality’. In Breul, C. and Göbbel, E. (eds), Comparative and Contrastive Studies of Information Structure. Amsterdam: Benjamins, 231–76. OLD = Glare, P. (ed.) (1982). The Oxford Latin Dictionary. Oxford: Oxford University Press. Oliviéri, M. (2004). ‘Paramètre du sujet nul et inversion du sujet dans les dialectes italiens et occitans’. Cahiers de Grammaire 29: 105–20.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
434
Oliviéri, M. and Sauzet, P. (2016). ‘Southern Gallo-Romance (Occitan)’. In Ledgeway and Maiden, 319–49. Orr, R. (1992). ‘Slavo-Celtica‘. Canadian Slavonic Papers / Revue canadienne des slavistes 34: 245–68. Ortiz de Urbina, J. (1989). Parameters in the Grammar of Basque. Dordrecht: Foris. Otheguy, R. and Zentella, A. C. (2012). Spanish in New York: Language Contact, Dialectal Leveling, and Structural Continuity. Oxford: Oxford University Press. Paden, W. (1998). An Introduction to Old Occitan. New York: The Modern Language Association of America. Palay, S. (1927). Lou marcat de la trouje. Farcejade en u acte. Pau, France: Marrimpouey. Palay, S. (1974). Dictionnaire du béarnais et du gascon modernes. Paris: Editions de Centre national de la recherche Scientifique. Pană Dindelegan, G. (2013). ‘The subject’. In Pană Dindelegan, G. (ed.), The Grammar of Romanian. Oxford: Oxford University Press, 100–25. Paoli, S. (2006). ‘The fine structure of the Left Periphery: COMPs and subjects. Evidence from Romance’. Lingua 117: 1057–79. Paoli, S. (forthcoming). ‘A step forward in understanding pas: an investigation of the postverbal negator in Old Occitan’. Folia Linguistica Historica. Papahagi, T. (1974). Dicționarul dialectului aromân general și etimologic. Bucharest: Editura Academiei. Papen, R. (2016). ‘Les variétés de français de l’Ouest canadien: un bilan de plus de 60 ans de recherche’. Presentation at Les français d’ici, Université de Saint-Boniface (Winnipeg, Canada), 7–9 June. Papen, R. and Rottet, K. (1997). ‘A structural sketch of the Cajun French spoken in Lafourche and Terrebonne Parishes’. In Valdman, A. (ed.), French and Creole in Louisiana. New York: Plenum Press, 71–108. Paul, H. ([1890] 1970). Principles of the History of Language. College Park, MD: McGrath. Perlmutter, D. (1978). ‘Impersonal passives and the Unaccusative Hypothesis’. Proceedings of the Fourth Annual Meeting of the Berkeley Linguistics Society. Berkeley; University of California, 57–89. Perlmutter, D. (1989). ‘Multiattachment and the unaccusative hypothesis: the perfect auxiliary in Italian’. Probus 1: 63–119. Péronnet, L. (1989). Le Parler acadien du sud-est du Nouveau Brunswick: éléments grammaticaux et lexicaux. New York: Lang. Perugi, M. (2003). ‘Histoire de la réflexion sur les langues romanes: l’occitan’. In Ernst, G., Glessen, M-D., Schmitt, C., and Schweickard, W. (eds), Histoire linguistique de la Romania: manuel international d’histoire linguistique de la Romania. Berlin / Boston, MA: Mouton de Gruyter, 242–54. Pesetsky, D. (1987). ‘Wh-in-situ: movement and unselective binding’. In Reuland, E. and ter Meulen, A. (eds), The Representation of (In)definiteness. Cambridge, MA: MIT Press, 98–129. Pesetsky, D. (2000). Phrasal Movement and Its Kin. Cambridge MA: MIT Press. Phalan, B., Onial, M., Balmford, A., and Green, R. (2011). ‘Reconciling food production and biodiversity conservation: land sharing and land sparing compared’. Science 333: 1289–91. Picone, M. and Valdman, A. (2005). ‘La situation du français en Louisiane’. In Valdman, A, Auger, J., and Piston-Hatlen, D. (eds), Le Français en Amérique du Nord. Québec: Presses de l’Université Laval, 143–68. Platzack, C. (1995). ‘The loss of verb second in English and French’. In Roberts and Battye, 200–26.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
435
Plungian, V. and van der Auwera, J. (2006). ‘Towards a typology of discontinuous past marking’. Sprachtypologie und Universalienforschung 59: 317–49. Poletto, C. (1995). ‘The diachronic development of Italian subject clitics in northern eastern Italian dialects’. In Roberts and Battye, 295–324. Poletto, C. (2000). The Higher Functional Field. Oxford: Oxford University Press. Poletto, C. (2002). ‘The left periphery of a V2–Rhaetoromance dialect: a new perspective on V2 and V3’. In Barbiers, S., Cornips, L., and van der Kleij, S. (eds), Syntactic Microvariation. Proceedings of the Workshop on Syntactic Microvariation, Amsterdam, August 2000. Amsterdam: Meertens Institute. http://www.meertens.knaw.nl/me dewerkers/marc.van.oostendorp/microvariation/poletto.pdf Poletto, C. (2005). ‘ “Si” and “e” as CP expletives in Old Italian’. In Batllori, M., Hernanz, M.-Ll., Picallo, C., and Roca, F. (eds), Grammaticalization and Parametric Variation. Oxford: Oxford University Press, 206–35. Poletto, C. (2006). ‘Parallel phases: a study of the high and low periphery of Old Italian’. In Frascarelli, M. (ed.), Phases of Interpretation. Berlin: Mouton de Gruyter, 261–95. Poletto, C. (2013). ‘On V2 types’. In Luraghi, S. and Parodi, C. (eds), The Bloomsbury Companion to Syntax. London: Bloomsbury, 154–64. Poletto, C. (2014). Word Order in Old Italian. Oxford: Oxford University Press. Poletto, C. (2016). ‘Negation’. In Ledgeway and Maiden, 833–46. Poletto, C. and Pollock, J.-Y. (2004). ‘On the left periphery of some Romance whquestions’. In Rizzi, L. (ed.), The Structure of CP and IP. Oxford: Oxford: Oxford University Press, 251–96. Poletto, C. and Sanfelici, E. (2018). ‘On relative complementizers and relative pronouns’. In Garzonio, J. and Rossi, S. (eds), Variation in C: Comparative Approaches to the Complementizer Phrase. Amsterdam: Benjamins, 265–98. Poletto, C. and Tortora, C. (2016). ‘Subject clitics: syntax’. In Ledgeway and Maiden, 772–85. Pollock, J.-Y. (1989). ‘Verb movement, UG and the structure of IP’. Linguistic Inquiry 20: 365–425. Pooley, T. (1988). ‘Grammatical and phonological variation in the working-class French of Roubaix’. University of London: thesis. Pooley, T. (2004). Language, Dialect and Identity in Lille. Lewiston, NY / Queenston, ON / Lampeter, Wales: Edwin Mellen Press. Pooley, T. (2006). ‘On the geographical spread of Oïl French in France’. Journal of French Language Studies 16: 357–90. Pope, M. (1934). From Latin to Modern French with Especial Consideration of AngloNorman: Phonology and Morphology. Manchester: Manchester University Press. Poplack, S. (2001). ‘Variability, frequency and productivity in the irrealis domain of French’. In Bybee, J. and Hopper, P. (eds), Frequency and the Emergence of Linguistic Structure. Amsterdam: Benjamins, 405–28. Poplack, S. and Dion, N. (2009). ‘Prescription vs praxis: the evolution of future temporal reference in French’. Language 85: 557–87. Poplack, S. and Turpin, D. (1999). ‘Does the FUTUR have a future in (Canadian) French?’ Probus 11: 133–64. Portes, J. (1857). Fablos caousidos de Lafountaino libromen traduitos en patouès pyrénéen et enrichidos dous éléments de la grammairo d’aquéro lengo. Bagnères-de-Bigorre, France: Plassot. Posner, R. (1996). The Romance Languages. Cambridge: Cambridge University Press. Prévost, S. (2002). Postposition du sujet en français aux Ve et VIe siècles: analyse sémanticopragmatique. Paris: CNRS.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
436
Price, G. (1971). The French Language: Present and Past. London: Arnold. Price, G. (1978). ‘L’interrogation négative sans ne’. Studii și cercetări lingvistice 29: 599–606. Price, G. (1993). ‘Pas (point) without ne in interrogative clauses’, French Language Studies 3: 191–5. Price, G. (1997). ‘Negative particles in French’. In Gregory, S., Trotter, D., and Rothwell, W. (eds), De mot en mot. Aspects of Medieval Linguistics. Cardiff: University of Wales Press, 173–90. Price, G. (2003). A Comprehensive French Grammar. Oxford: Blackwell. Priestley, L. (1955). ‘Reprise constructions in French’. Archivum Linguisticum 7: 1–28. Pullum, G. and Huddleston, R. (2002). ‘Prepositions and preposition phrases’. In Pullum, G. and Huddleston, R. (eds) The Cambridge Grammar of the English Language. Cambridge: Cambridge University Press, 597–661. Puscariu, S. (1926). Studii istroromâne, II. Bucharest: Cultura naţională. Pusch, C. (1998a). ‘L’énonciatif gascon entre pragmatique et grammaire: analyse d’un corpus oral’. Presentation at XXIIe Congrès International de Linguistique et Philologie Romanes (Brussels, 23–29 July 1998). Pusch, C. (1998b). ‘Präverbale Marker im gesprochenen Gaskognisch. Zur Distribution, Genese und Pragmatik des Enunziativs’. Freiburg im Breisgau: doctoral thesis. Pusch, C. (1999). ‘La phrase principale affirmative en gascon: un cas de focalisation figée?’ In Guimier, C. (ed.), La Thématisation dans les langues: actes du colloque de Caen, octobre 1997. Bern: Lang, 107–19. Pusch, C. (2000a). ‘The attitudinal meaning of preverbal markers in Gascon: insights from the analysis of literary and spoken language data’. In Andersen, G. and Fretheim, T. (eds), Pragmatic Markers and Propositional Attitude. Amsterdam: Benjamins, 189–207. Pusch, C. (2000b). ‘L’énonciatif gascon entre pragmatique et grammaire: analyse d’un corpus oral’. In Englbert, A., Pierrand, M., Rosier, L., and van Raemdonck, D. (eds), Actes duXIIe congrès international de linguistique et de philologie romanes: Bruxelles, 23–29 juillet 1998. Tübingen: Niemeyer, 625–30. Pusch, C. (2001). ‘Prosthèse proclitique et morphogénèse: le cas de l’énonciatif e en gascon’. In Muller, C., de Carvalho, P., Labrune, L., Lambert, F., and Plogg, K. (eds), Clitiques et cliticisation: actes de colloque de Bordeaux, octobre 1998. Paris: Champion, 381–93. Pusch, C. (2002). ‘Preverbal modal particles in Gascony Occitan’. Belgian Journal of Linguistics 16: 105–18. Pusch, C. (2007). ‘Is there evidence for evidentiality in Gascony Occitan?’ Rivista di linguistica 19: 91–108. Puyau, J. (2013). Comprendre, parler, lire, écrire le béarnais. Cressé, France: Éditions des Régionalismes. Queyron, P. (1907). ‘La “Gavacherie” de Monségur (Arrondissement de La Réole)’. Revue de l’Agenais 34: 327–39, 436–46. Quint, N. (1996). Grammaire du parler occitan nord-limousin marchois de Gartempe et de Saint-Sylvain-Montaigut. Limoges: La Clau Lemosina. Raible, W. (2001). ‘Linking clauses’. In Haspelmath, M., König, E., Oesterreicher, W., and Raible, W. (eds), Language Typology and Language Universals—Sprachtypologie und sprachliche Universalien—La Typologie des langues et les universaux linguistiques. An International Handbook—Ein internationales Handbuch—Manuel international. Berlin / New York: De Gruyter, 590–617. Ravier, X. (1991). ‘Okzitanisch: Areallinguistik. Les aires linguistiques’. In Holtus, G., Metzeltin, M., and Schmitt, C. (eds), Lexikon der romanistischen Linguistik. V,2: Okzitanisch, Katalanisch. L’Occitan, Le catalan. Tübingen: Niemeyer, 80–105. Ravier, X. and Cursente, B. (2005). Le Cartulaire de Bigorre (XIe–XIIIe siècle). Paris: CTHS.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
437
Raynaud de Lage, G. (1983). Manuel pratique d’ancien français. Paris: Picard. Rea, B. (2014). ‘Aspects of pronoun and auxiliary morphology in French, with particular reference to spoken Montréal French’. University of Oxford: dissertation. Rea, B. (2017). ‘Challenges of analysing linguistic variation in a growing metropolis: a trend study of auxiliary alternation in Montréal French (1971–2016)’. Presentation at New Ways of Analyzing Variation (NWAV 46), Madison University, 2–5 November. Rea, B. (2018a). ‘Variabilité dans la sélection de l’auxiliaire: résultats d’un nouveau corpus de français montréalais (2016)’. Presentation at Les français d’ici, Concordia University, Montréal (Canada), 23–25 May. Rea, B. (2018b). ‘Le retour d’un “être” cher? Étude sociolinguistique de l’alternance des auxiliaires dans le français de Montréal (1971–2016)’. Congrès Mondial de Linguistique Française 6 (Université de Mons, Belgium). https://doi.org/10.1051/shsconf/20184602002 Rea, B. (forthcoming). ‘ “Je m’ai fait mal quand j’ai tombé”: A real- and apparent-time study of auxiliary alternation in intransitive and pronominal verbs in spoken Montréal French (1971–2016)’. University of Oxford : thesis. Reenen, P. v. and Schøsler, L. (1992). ‘Ancien et Moyen Français: SI “thématique”, analyse exhaustive d’une série de textes’. Vox Romanica 51: 101–27. Reenen, P. v. and Schøsler, L. (1993). ‘SI “thématique”, étude de SI en ancien et moyen français, discours direct’. In Hilty, G. (ed.), Actes du Xe congrès international de linguistique et philologie romanes. Zürich: Francke, 617–28. Reenen, P. v. and Schøsler, L. (2000). ‘The pragmatic functions of the Old French particles ainz, apres, donc, lors, or, pluis, and si’. In Herring, S., Reenen, P. v., and Schøsler, L. (eds), Textual Parameters in Older Languages. Amsterdam: Benjamins, 59–105. Régnier, C. (1990). Aliscans. Paris: Champion. Reichenbach, H. (1947). Elements of Symbolic Logic. New York: Macmillan. Remacle, L. (1948). Le Problème de l’ancien wallon. Liège: Faculté de Philosophie et Lettres. Remacle, L. (1952). Syntaxe du parler wallon de La Gleize. I. Noms et articles—Adjectif et pronoms. Paris: Les Belles Lettres. Remacle, L. (1956). Syntaxe du parler wallon de La Gleize. II. Verbes, adverbes, prépositions. Paris: Les Belles Lettres. Remberger, E.-M. (2010). ‘Left-peripheral interactions in Sardinian’. Lingua 120: 555–81. Renaud, C. and Villeneuve, A.-J. (2008). ‘L’alternance des auxiliaires avoir et être à Chicoutimi-Jonquière’. Presentation at Les français d’ici, Ottawa (Canada), 23 May. Renault, G. (1919). Le Poilu tel qu’il se parle. Paris: Bossard. Renzi, L. and Vanelli, L. (1983). ‘I pronomi soggetti in alcune varietà romanze’. In Benincà, P., Cortelazzo, M., Prosdocimi, A., Vanelli, L., and Zamboni, A. (eds), Studi in onore di Giovan Battista Pellegrini. Pisa: Pacini, 121–45. Reydy, J.-P. (2008). Notre occitan. Limoges: IEO Limousin. Ricketts, P. (2005). Concordance de l’Occitan médiéval. Turnhout, Belgium: Brepols. Rideout, D. (2011). ‘Auxiliary selection in 16th century French: imposing norms in the face of language change’. In Johanson, S., Hazenberg, E., and Power, S. (eds), Proceedings of the 33rd Atlantic Provinces Linguistics Association Memorial University of Newfoundland: Memorial University of Newfoundland Occasional Papers in Linguistics, vol. 2. Riegel, M., Pellat, J.-C., and Rioul, R. (1994). Grammaire méthodique du français. Paris: Presses Universitaires de France. Rigau, G. and Suïls, J. (2010). ‘Microvariation in Catalan and Occitan complementizers: the so-called expletive se’. Catalan Journal of Linguistics 9: 151–65. Rijk, R. de (1969). ‘Is Basque an SOV language?’ Fontes Linguae Vasconum 3: 319–51.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
438
Ringe, D. and Eska, J. F. (2013). Historical Linguistics: Toward a Twenty-First Century Reintegration. Cambridge: Cambridge University Press. Rinke, E. and Meisel, J. (2009). ‘Subject inversion in old French: syntax and information structure’. In Kaiser, G. A. and Remberger, E.-M. (eds), Proceedings of the Workshop on Null Subjects, Expletives and Locatives in Romance. Konstanz, Germany: Fachbereich Sprachwissenschaft, 93–130. Rivero, M.-L. (1994). ‘Negation, imperatives and Wackernagel effects’. Rivista di linguistica 6: 39–66. Rizzi, L. (1978). ‘A restructuring rule in Italian syntax’. In Keyser, S. (ed.), Recent Ttransformational Studies in European Languages. Cambridge, MA: MIT Press, 113–58. Rizzi, L. (1982). Issues in Italian Syntax. Dordrecht: Foris. Rizzi, L. (1986). ‘On the status of subject clitics in Romance’. In Jaeggli, O. and SilvaCorvalan, C. (eds), Studies in Romance Linguistics. Dordrecht: Foris, 391–419. Rizzi, L. (1986b). ‘Null objects in Italian and the theory of pro’. Linguistic Inquiry 17: 501–57. Rizzi, L. (1990a). Relativized Minimality. Cambridge MA: MIT Press. Rizzi, L. (1990b). ‘Speculations on verb-second’. In Mascaró, J. and Nespor, M. (eds), Grammar in Progress: Essays in Honour of Henk van Riemsdijk. Groningen: Foris, 375–86. Rizzi, L. (1997). ‘The fine structure of the left periphery’. In Haegeman, L. (ed.), Elements of Grammar. Dordrecht: Kluwer, 281–37. Rizzi, L. (2001a). ‘On the position “int(errogative)” in the Left Periphery of the clause’. In Cinque, G. and Salvi, G. (eds), Current Studies in Italian Syntax. Essays Offered to Lorenzo Renzi. Amsterdam: Elsevier, 287–96. Rizzi, L. (2001b). ‘Extraction from weak islands, reconstruction, and agreement’. In Chierchia, G., Guasti, M.T., and Cechetto, C. (eds), Semantic Interfaces. Reference, Anaphora and Aspect. Stanford, CA: CSLI Publications, 155–76. Rizzi, L. (2004). ‘On the cartography of syntactic structures’. In Rizzi, L. (ed.), The Structure of CP and IP. Oxford: Oxford University Press, 3–15. Rizzi, L. (2013). ‘The functional structure of the sentence, and cartography’. In Dikken, M. (ed.), The Cambridge Handbook of Generative Syntax. Cambridge: Cambridge University Press, 425–57. Rizzi, L. (2015). ‘Notes on labeling and subject positions’. In Di Domenico, E., Hamann, C. and Matteini, S. (eds), Structures, Strategies and Beyond: Studies in Honour of Adriana Belletti. Amsterdam: Benjamins, 17–46. Rizzi, L. and Roberts, I. (1989). ‘Complex inversion in French’. Probus 1: 1–30. Robert, M., Dhéralde, L., and Duclou, L. (1968–1969). Dictionnaire de la langue limousine. Diciounāri de lo lingo limousino. Limoges: Société d’ethnographie du Limousin. Roberts, I. (1993). Verbs and Diachronic Syntax: A Comparative History of English and French. Dordrecht: Kluwer. Roberts, I. (1997). ‘Restructuring, head movement, and locality’. Linguistic Inquiry 28: 423–60. Roberts, I. (2004). ‘The C-system in Brythonic Celtic languages, V2, and the EPP’. In Rizzi, L. (ed.), The Structure of CP and IP. Oxford: Oxford University Press, 297–328. Roberts, I. (2007). Diachronic Syntax. Oxford: Oxford University Press. Roberts, I. (2010). Varieties of French and the null subject parameter. In Roberts, I., Holmberg, A., Biberauer, T., and Sheehan, M., Parametric Variation : Null Subjects in Minimalist Theory. Cambridge: Cambridge University Press, 303–28.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
439
Roberts, I. (2019). Parameter Hierarchies and Universal Grammar. Oxford: Oxford University Press. Roberts, I. and Roussou, A. (2002). ‘The extended projection principle as a condition on the Tense dependency’. In Svenonius, P. (ed.), Subjects, Expletives, and the EPP. Oxford: Oxford University Press, 125–55. Roberts, I. and Roussou, A. (2003). Syntactic Change: A Minimalist Approach to Grammaticalization. Cambridge: Cambridge University Press. Roberts, N. (2012). ‘Future temporal reference in hexagonal French’. University of Pennsylvania Working Papers in Linguistics 18: 96–106. Roberts, N. (2013). ‘The influence of linguistic factors on the expression of futurity in Martinique French’. Newcastle Working Papers in Linguistics 19: 138–51. Roberts, N. (2014). ‘A sociolinguistic study of grammatical variation in Martinique French’. Newcastle University: thesis. Roberts, N. (2016). ‘The future of Martinique French: the role of random effects on the variable expression of futurity’. Canadian Journal of Linguistics 61: 286–313. Robin, P., Le Prévost, A., Passy, A.-F., and De Blosseville, E. (1879). Dictionnaire du patois normand en usage dans le département de l’Eure. Evreux, France: Hérissey. Rohde, H. (2006). ‘Rhetorical questions as redundant questions’. San Diego Linguistics Papers 2: 134–68. Rohlfs, G. (1968). Grammatica storica della lingua italiana e dei suoi dialetti. Morfologia. Turin: Einaudi. Rohlfs, G. (1969). Grammatica storica della lingua italiana e dei suoi dialetti. Sintassi e formazione delle parole. Turin: Einaudi. Rohlfs, G. (1970a). Le Gascon: étude de philologie pyrénéenne. Tübingen: Niemeyer. Rohlfs, G. (1970b). ‘Catalan, provençal, gascon et espagnol’. Estudis romànics 12: 7–17. Romdahl, A. (1881). Glossaire du patois du Val de Saire (Manche). Linköping, Sweden: Ridderstads Boktryckeri. Romieu, M. and Bianchi, A. (1999). La lenga del trobar: precis de gramatica d’occitan ancian = La langue du trobar. Pessac, France: Presses Universitaires de Bordeaux. Romieu, M. and Bianchi, A. (2006). Iniciacion a l’occitan ancian: initiation à l’ancien occitan; dètz e nòu tèxtes de l’edat mejana comentats. Pessac, France: Presses Universitaires de Bordeaux. Ronjat, J. (1913). Essai de syntaxe des parlers provençaux modernes. Macon, France: Protat Frères. Ronjat, J. (1930). Grammaire istorique des parlers provençaux modernes. I. Introduction. Première partie. Fonétique. Montpellier: Société des langues romanes. Ronjat, J. (1932). Grammaire istorique des parlers provençaux modernes. II. Montpellier: Société des Langues Romanes. Ronjat, J. (1937). Grammaire istorique des parlers provençaux modernes. III. Syntaxe. Montpellier: Société des langues modernes. Rose, F. (2015). ‘Innovative complexity in the promoninal paradigm of Mojeño: a result of contact?’ In Gardani, F., Arkadiev, P., and Amrididze, N. (eds), Borrowed Morphology. Berlin: De Gruyter, 241–67. Rosemeyer, M. (2013). ‘Tornar and volver: the interplay of frequency and semantics in compound tense auxiliary selection in Medieval and Classical Spanish’. In van Gelderen, E., Cennamo, M., and Barðdal, J. (eds), Argument Structure in Flux. The Naples–Capri Papers. Amsterdam / Philadelphia, PA: Benjamins, 435–57. Rosen, C. (1984). ‘The interface between semantic roles and initial grammatical relations’. In Perlmutter, D. and Rosen, C. (eds), Studies in Relational Grammar 2. Chicago: Chicago University Press, 38–77.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
440
Rossi-Gensane, N. (2016). ‘Syntactic variation in spoken French’. In Detey, S., Durand, J., Laks, B., and Lyche, C. (eds), Varieties of Spoken French. Oxford: Oxford University Press, 86–107. Round, E. (2015). ‘Rhizomorphomes, meromorphomes and metamorphomes’. In Corbett, G., Brown, D., and Baerman, M. (eds), Understanding and Measuring Morphological Complexity. Oxford: Oxford University Press, 29–52. Roussel, B. (2016). ‘Deux auxiliaires pour le prix d’un: l’alternance entre avoir et être en français acadien’.Presentation at Les français d’ici (LFDI 6), Université de Saint-Boniface (Winnipeg, Canada), 7–9 June. Rouveret, A. (2004). ‘Les clitiques pronominaux et la périphérie gauche en ancien français’. Bulletin de la Société de Linguistique de Paris 99: 181–237. Roux, J. (1895). Grammaire limousine. Brive, France: École limousine. Rowlett, P. (2007). The Syntax of French. Cambridge / New York: Cambridge University Press. Roy, I. and Svenonius, P. (2009). ‘Complex prepositions’. In François, J., Gilbert, E., Guimier, C. and, Krause, M. (eds) Autour de la préposition. Caen, France: Presses Universitaires de Caen, 105–16. Ruben, E. (1866). J. Foucaud. Poésies en patois limousin. Paris: Firmin Didot. Russo, M. and Roberts, J. (1999). ‘Linguistic change in endangered dialects: the case of alternation between avoir and être in Vermont French’. Language Variation and Change 11: 67–85. Ryckeboer, H. (2002). ‘Dutch / Flemish in the North of France’. In Treffers-Daller, J. and and Willemyns, R. (eds), Language Contact at the Romance–Germanic Language Border. Clevedon, Bristol: Multilingual Matters, 22–35. Martin, S. (1659). La Vie de S. Thomas de Vil-Neuve. Toulouse: Jean Boude. Sadock, J. (1971). ‘Queclaratives’. Papers from the Seventh Regional Meeting of the Chicago Linguistics Society. Chicago: Chicago Linguistic Society, 223–32. Sadock, J. (1974). Toward a Linguistic Theory of Speech Acts. New York: Academic Press. Salvesen, C. (2011). ‘Stylistic Fronting and remnant movement in Old French’. In Berns, J., Jacobs, H., and Scheer, T. (eds), Romance Languages and Linguistic Theory 2009: Selected Papers from ‘Going Romance’ Nice 2009. Amsterdam / Philadelphia, PA: Benjamins, 323–42. Salvesen, C. (2013). ‘Topics and the left periphery: a comparison of old French and modern Germanic’. In Lohndal, T. (ed.), In Search of Universal Grammar: From Old Norse to Zoque, 131–72. Amsterdam: Benjamins. Salvesen, C. (2019). ‘Resumptive particles and verb second’. In Woods, R. and Wolfe, S. (eds), Rethinking Verb Second. Oxford: Oxford University Press. Salvesen, C. and Bech, K. (2014). ‘Postverbal subjects in old English and old French’. Oslo Studies in Language 6: 201–28. Salvesen, C. and George Walkden (2017). ‘Diagnosing embedded V2 in Old French and Old English’. In Mathieu, E. and Truswell, R. (eds), From Micro-change to Macro-change. Oxford: Oxford University Press. Salvesen, C. and Walkden, G. (2017). ‘Diagnosing embedded V2 in old English and old French’. In Truswell, R. and Mathieu, E. (eds), Micro-change and Macro-change in Diachronic Syntax. Oxford: Oxford University Press, 168–81. Salvi, G. (2000a). ‘Il ladino. Schizzo linguistico’. Verbum 1: 151–69. Salvi, G. (2000b). ‘La formazione della sistema V2 delle lingue romanze antiche’. Lingua e Stile 35: 665–92. Salvi, G. (2011). ‘Morphosyntactic persistence’. In Maiden, M., Smith, J. C., and Ledgeway, A. (eds), The Cambridge History of the Romance Languages. I. Structures. Cambridge: Cambridge University Press, 318–81.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
441
Salvi, G. (2012). ‘On the nature of the V2 system of medieval romance’. In Brugè, L., Cardinaletti, A., Giusti, G., Munaro, N., and Poletto, C. (eds), Functional Heads. Oxford / New York: Oxford University Press, 103–11. Salvi, G. (2016). ‘Ladin’. In Ledgeway and Maiden, 154–68. Sammons, O., Nadasdi, T., and Mougeon, R. (2015). ‘Moving through the past: thirty years of avoir été in Ontario French’. Journal of French Language Studies 25: 397–422. Sampson, R. (1999). Nasal Vowel Evolution in Romance. Oxford: Oxford University Press. Sankoff, D. (1988). ‘Problems of representativeness’. In Ammon, U., Dittmar, N., and Mattheier, K. (eds), Sociolinguistics: An International Handbook of the Science of Language and Society. Berlin: De Gruyter, 899–903. Sankoff, D. and Laberge, S. (1978). ‘The linguistic market and the statistical explanation of variability’. In Sankoff, D. (ed.), Linguistic Variation: Models and Methods. New York: Academic Press, 239–50. Sankoff, G. and Thibault, P. (1977). ‘L’alternance entre les auxiliaires avoir et être en français parlé à Montréal’. Langue française 43: 81–108. Sankoff, G. and Thibault, P. (1980). ‘The alternation between the auxiliaries avoir and être in Montreal French’. In Sankoff, G. (ed.), The Social Life of Language. Philadelphia, PA: University of Pennsylvania Press, 311–45. Sankoff, G. and Wagner, S. (2006). ‘Age-grading in retrograde movement: the inflected future in Montréal French’. University of Pennsylvania Working Papers in Linguistics 12: 203–16. Sankoff, G., Thibault, P., and Evans Wagner, S. (2004). ‘An apparent time paradox: change in Montréal French auxiliary selection, 1971–1995’. Paper presented at New Ways of Analyzing Variation (NWAV 33), Ann Arbor, MI. Sankoff, G., Wagner, S., and Jensen, L. (2012). ‘The long tail of language change: québécois French futures in real time’. University of Pennsylvania Working Papers in Linguistics 18: 107–16. Sapir, E. ([1921] 1979). Language: An Introduction to the Study of Speech. London: Granada. Saussure, L. de (2013). ‘Perspectival interprétations of tenses’. In Jaszczolt and de Saussure, 46–69. Sauzet, P. (1989). ‘Topicalisation et prolepse en occitan’. Revue des langues romanes 93: 235–73. Sauzet, P. and Oliviéri, M. (2016). ‘Southern Gallo-Romance (Occitan)’. In Ledgeway and Maiden, 319–49. Scalise, S. and Bisetto, A. (2009). ‘The classification of compounds’. In Lieber, R. and Stekauer, P. (eds), The Oxford Handbook of Compounding. Oxford: Oxford University Press, 34–53. Schaden, G. (2012). ‘Modelling the “aoristic drift of the Present Perfect” as inflation. An essay in historical pragmatics’. International Review of Pragmatics 4: 261–92. Schifano, N. (2015). ‘Il posizionamento del verbo nei dialetti romanzi d’Italia’. The Italianist 35: 121–38. Schifano, N. (2018). Verb Movement in Romance: A Comparative Study. Oxford: Oxford University Press. Schilling, N. (2013). Sociolinguistic Fieldwork. Cambridge: Cambridge University Press. Schlegel, A. W. von (1818). Observations sur la langue et la littérature provençales. Paris: Librairie grecque-latine-allemande. Schlieben-Lange, B. (1971). Okzitanische und Katalanische Verbprobleme. Ein Beitrag zur funktionellen synchronischen Untersuchung des Verbalsystems der beiden Sprachen (Tempus und Aspekt). Tübingen: Niemeyer.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
442
Schorta, A. (1938). Lautlehre der Mundart von Müstair. Paris: Droz. Schwegler, A. (1990). Analyticity and Syntheticity: A Diachronic Perspective with Special Reference to Romance Languages. Berlin / New York: Mouton de Gruyter. Schwenter, S. (2000). ‘Viewpoints and polysemy: linking adversative and causal meanings of discourse markers’. In Couper-Kuhlen, E. and Kortmann, B. (eds), Cause, Condition, Concession, Contrast: Cognitive and Discourse Perspectives. Berlin: Mouton de Gruyter, 257–81. Schwenter, S. (2005). ‘The pragmatics of negation in Brazilian Portuguese’, Lingua 115: 1427–56. Seijido, M. (2014). ‘Les questions syntaxiques chez les remarqueurs du XVIIe siècle: quel apport?’ In Ayres-Bennett, W. and Rainsford, T. (eds), L’Histoire du français. Etat des lieux et perspectives. Paris: Garnier, 93–110. Seutin, É. (1975). Description grammaticale du parler de l’Ile-aux-Coudres. Montreal: Presses de l’Université de Montréal. Sichel-Bazin, R. (2011). ‘Prosodic markers of focus in French and Occitan statements and yes-no questions’. Presentation at Phonetics and Phonology in Iberia 2011, Universitat Rovira i Virgili, Tarragona, 21–22 June. Sicre, P. (1909). Éléments de grammaire du dialecte de Foix. Foix, France: Gadrat aîné. Simonenko, A. and Hirschbühler, P. (2012). ‘Placement de clitiques dans les propositions V1 et évolution de la structure de la proposition en ancien français’. In Dufresne, M. (ed.), Typologie, ordre des mots et groupe verbal en français médiéval. Sainte-Foy, Quebec: Les Presses de l’Université Laval, 11–53. Simonenko, A., Crabbé, B., and Prévost, S. (forthcoming). ‘Agreement syncretisation and the loss of null subjects’. Language Variation and Change. Sitaridou, I. (2005). ‘A corpus-based study of null subjects in Old French and Old Occitan’. In Putsch, C., Kabatek, J., and Raible, W. (eds), Corpora and Diachronic Linguistics. Tübingen: Narr, 359–74. Sitaridou, I. (2012). ‘A comparative study of word order in Old Romance’. In Carlier, A., Lamiroy, B., and De Mulder, W. (eds), Folia Linguistica 46: 1–51. Skårup, P. (1975). Les Premières Zones de la proposition en ancien français: essai de syntaxe de position. Copenhagen: Revue romane. Skårup, P. (1997). Morphologie élémentaire de l’ancien occitan. Copenhagen: Museum Tusculanum Press. Smith, J. C. (1989a). ‘A pragmatic view of French deixis’. York Papers in Linguistics 14: 263–78. Smith, J. C. (1989b). ‘Deixis et personne en français’. Sigma 12: 1–14. Smith, J. C. (1989c). ‘Actualization reanalyzed: evidence from the Romance compound past tenses’. In Walsh, T. (ed.). Synchronic and Diachronic Approaches to Language Variation and Change. Washington, D.C.: Georgetown University Press, 310–25. Smith, J. C. (1991a). ‘Thematicity and “object”-participle agreement in Romance’. In Wanner, D. and Kibbee, D. (eds), New Analyses in Romance Linguistics: selected papers from the Eighteenth Linguistic Symposium on Romance Languages, April 7–9, 1988. Amsterdam: Benjamins, 335–52. Smith, J. C. (1991b). ‘Problemi dell’accordo del participio passato coll’oggetto diretto nei tempi composti coniugati con avere in italiano, con speciale riferimento ai dialetti’. In Gianelli, L., Maraschio, N., Salani, T. P., and Vedovelli, M. (eds), Tra Rinascimento e strutture attuali: saggi di linguistica italiana (Atti del Primo Convegno della Società Internazionale di Linguistica e Filologia Italiana, Siena, 28–31 marzo 1989, volume 1 ). Turin: Rosenberg and Sellier, 365–71.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
443
Smith, J. C. (1992a). ‘Circumstantial complements and direct objects in the Romance languages: configuration, Case, and thematic structure’. In Roca, I. M. (ed.), Thematic Structure: its role in grammar. Berlin: Foris, 293–316. Smith, J. C. (1992b). ‘Traits, marques et sous-spécification: application à la deixis’. In Morel, M-A. and Danon-Boileau, L. (eds), La Deixis: colloque en Sorbonne, 8–9 juin 1990. Paris: Presses Universitaires de France, 257–64. Smith, J. C. (1993a). ‘The agreement of the past participle conjugated with avoir and a preceding direct object: a brief history of prescriptive attitudes’. In Sampson, R. (ed.), Authority and the French Language: papers from a conference at the University of Bristol. Münster: Nodus, 87–125. Smith, J. C. (1993b). ‘La desaparición de la concordancia entre participio de pasado y objeto directo en castellano y catalán: aspectos geográficos e históricos’. In Penny, R. (ed.), Actas del Primer Congreso Anglo-Hispano: tomo I, Lingüística. Madrid: Castalia, 275–85. Smith, J. C. (1995a).‘Agreement between past participle and direct object in Catalan: the hypothesis of Castilian influence revisited’. In Fisiak, J. (ed.), Linguistic Change under Contact Conditions. Berlin: Mouton, 271–89. Smith, J. C. (1995b). ‘Perceptual factors and the disappearance of agreement between past participle and direct object in Romance’. In Smith, J. C. and Maiden, M. (eds), Linguistic Theory and the Romance Languages. Amsterdam: John Benjamins, 161–80. Smith, J. C. (1995c). ‘L’évolution sémantique et pragmatique des adverbes déictiques ici, là et là-bas’. Langue française 107: 43–57. Smith, J. C. (1996). ‘Surfonctionnalité et hyperanalyse: l’accord du participe passé dans les langues romanes à la lumière de deux théories récentes’. Faits de langues 8: 113–20. Smith, J. C. (1997). ‘Types and tokens in language change: some evidence from Romance’, In Hickey, R. and Puppel, S. (eds), Language History and Linguistic Modelling: a festschrift for Jacek Fisiak on his 60th birthday. Berlin: Mouton, 1099–111. Smith, J. C. (1998). ‘Gothic Language’. In Mulvey-Roberts, M. (ed.), The Handbook to Gothic Literature. New York: University Press, 269f. Smith, J. C. (1999a). ‘The refunctionalization of a pronominal subsystem between Latin and Romance’. Oxford University Working Papers in Linguistics, Philology and Phonetics 4: 141–56. Smith, J. C. (1999b). ‘Markedness and morphosyntactic change revisited: the case of Romance past participle agreement’. In Embleton, S., Joseph, J., and Niederehe, H.-J. (eds), The Emergence of the Modern Language Sciences: studies in the transition from historical-comparative to structural linguistics in honour of E. F. K. Koerner - II: methodological perspectives and applications. Amsterdam / Philadelphia, PA: Benjamins, 203–16. Smith, J. C. (2000). ‘Reanalysis and extension: notes and queries’. In Kovačič, I., MilojevićSheppard, M., Orel-Kos, S., and Orešnik, J. (eds), Linguistics and Language Studies: exploring language from different perspectives. Ljubljana: Filozofska Fakulteta Univerze v Ljubljani, 45–60. Smith, J. C. (2001a). ‘Illocutionary conversion, bystander deixis, and Romance “ethic” pronouns’. Working Papers in Functional Grammar 74: 1–24. Smith, J. C. (2001b). ‘Markedness, functionality, and perseveration in the actualization of a morphosyntactic change’. In Andersen, H. (ed.), Actualization: linguistic change in progress. Amsterdam: John Benjamins, 203–23. Smith, J. C. (2002). ‘Middle French: When? What? Why?’. Language Sciences 24: 423–45. Smith, J. C. (2003). ‘Romance Linguistics: future perfect or future in the past?’. La corónica 31: 97–102.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
444
Smith, J. C. (2004). ‘The nominative-accusative opposition between Latin and GalloRomance: a study in refunctionalization’. Oxford University Working Papers in Linguistics, Philology and Phonetics 9: 117–31. Smith, J. C. (2005). ‘Some refunctionalizations of the nominative-accusative opposition between Latin and Gallo- Romance’. In Smelik, B., Hofman, R., Hamans, C., and Cram, D. (eds), A Companion in Linguistics: a festschrift for Anders Ahlqvist on the occasion of his sixtieth birthday. Nijmegen: De Keltische Draak, 269–85. Smith, J. C. (2006a). ‘French’. In Brown, K. (ed.), The Encyclopedia of Language and Linguistics Volume IV. Oxford: Elsevier, 629–32. Smith, J. C. (2006b). ‘How to do things without junk: the refunctionalization of a pronominal subsystem between Latin and Romance’. In Montreuil, J-P Y. (ed.), New Perspectives on Romance Linguistics: selected papers from the 35th Linguistic Symposium on Romance Languages (LSRL), Austin, Texas, February 2005. Volume II: Phonetics, Phonology and Dialectology. Amsterdam: John Benjamins, 2006, 183–205. Smith, J. C. (2008). ‘The refunctionalisation of first-person plural inflection in Tiwi’. In Bowern, C., Evans, B., and Miceli, L (eds), Morphology and Language History. In Honour of Harold Koch. Amsterdam: John Benjamins, 341–8. Smith, J. C. (2011a). ‘Change and continuity in form-function relationships’. In Maiden et al., 268–317. Smith, J. C. (2011b). ‘Variable analyses of verbal inflection in (mainly) Canadian French. In Maiden, M., Smith, J.C., Goldbach, M., and Hinzelin, M-O. (eds) Morphological Autonomy: perspectives from Romance inflectional morphology. Oxford: Oxford: Oxford University Press, 311–26. Smith, J. C. (2013). ‘The morphome as a gradient phenomenon: evidence from Romance’. In Cruschina, S., Maiden, M., and Smith, J. C. (eds) (2013). The Boundaries of Pure Morphology. Diachronic and Synchronic Perspectives. Oxford: Oxford University Press, 246–61. Smith, J. C. (2016). ‘French and Northern Gallo-Romance’. In Ledgeway and Maiden, 292–318. Smith, J. C. (2019). ‘Diachrony without history? Venetan and other languages from a Romance typological perspective’. Presentation at A Comparative Perspective on the Languages of the Veneto, St Catherine’s College, Oxford, 3–4 April 2019. Smith, J. C. Forthcoming. ‘Social factors in language change and variation’, in Ledgeway, A. and Maiden, M. (eds), The Cambridge Handbook of Romance Linguistics. Cambridge: Cambridge University Press. Smith, J. C. and Sneddon, C. (2002). ‘Further evidence for a “Middle French” koine: the fusion of “preposition + article” sequences’. University of Pennsylvania Working Papers in Linguistics 8: 265–80. Smith, N. and Bergin, T. (1984). An Old Provençal Primer. New York: Garland. Société Jersiaise (2008). Dictionnaithe Jèrriais-Angliais. Jersey: Don Balleine. Söhrman, I. (2013). ‘Reference, aspectuality and modality in ante-preterit (pluperfect) in Romance Languages’. In Josefson, F. and Söhrman, I. (eds), Diachronic and Typological Perspectives on Verbs. Amsterdam: Benjamins, 173–209. Sorace, A. (2000). ‘Gradience in auxiliary selection with intransitive verbs’. Language 76: 859–90. Sorace, A. (2004). ‘Gradience at the lexicon-syntax interface: evidence from auxiliary selection’. In Alexiadou, A., Anagnostopoulou, E., and Everaert, M. (eds), The Unaccusativity Puzzle. Oxford: Oxford University Press, 243–68. Speas, M. and Tenny, C. (2003). ‘Configurational properties of point of view roles’. In Di Sciullo, A. M. (ed.), Asymmetry in Grammar. Amsterdam: Benjamins, 315–44.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
445
Spence, N. (1993). A Brief History of Jèrriais. Jersey: Don Balleine. Spevak, O. (2010). Constituent Order in Classical Latin Prose. Amsterdam / Philadelphia, PA: Benjamins. Sportiche, D. (1996). ‘Clitic constructions’. In Rooryck, J. and Zaring, L. (eds), Phrase Structure and the Lexicon. Dordrecht: Kluwer, 213–76. Sprouse, R. and Vance, B. (1999). ‘An explanation for the decline of null pronouns in certain Germanic and Romance languages’’. In DeGraff, M. (ed.), Language Creation and Language Change: creolization, diachrony and development. Cambridge MA: MIT Press, 257–84. Squartini, M. and Bertinetto, P. M. (2000). ‘The simple and compound past in Romance languages’. In Dahl, Ö. (ed.), Tense and Aspect in the Languages of Europe. New York: Mouton de Gruyter, 403–39. Stasse, S. (2004). Dictionnaire populaire de wallon liégeois. Liège: Société Royale Littéraire La Wallonie. Stein, D. (1989). ‘Markedness and linguistic change’. In Tomić, O. (ed.), Markedness in Synchrony and Diachrony. Berlin: Mouton de Gruyter, 67–86. Steiner, B. D. (2014). ‘The role of information structure in the loss of verb-second in the history of French’. University of Indiana: thesis. Steinkrüger, P. (1995). ‘Grammatikalisierungen von Auxiliaren und Copulae im Katalanischen der Decadència’. Zeitschrift für Katalanistik 8: 35–62. Stelling, L. (2011). ‘The effects of grammatical proscription on morphosyntactic change: Auxiliary variation in Franco-American French’. Arborescences: revue d’études françaises 1. https://id.erudit.org/iderudit/1001942ar Štichauer, P. (2018). ‘Lexical splits within periphrasis: mixed perfective auxiliation systems in Italo-Romance’. Morphology 28: 1–23. Suchier, H. (1891). Le Français et le provençal. Paris: Bouillon. Suïls Subirà, J. and Ribes, S. (2015). ‘Modality markers in Gascon, between grammar and stylistic variation’. eHumanistica/IVITRA 8: 544–76. Sundell, L.-G. (1991). Le Temps futur en français moderne. Uppsala: Almquist and Wiksell. Suñer, M. (2000). ‘The syntax of direct quotes with special reference to Spanish and English’. Natural Language and Linguistic Theory 18: 525–78. Svenonius, P. (2004). ‘Slavic prefixes inside and outside VP’. Nordlyd 32: 205–53. Sweetenham, C. and Paterson, L. (2003). The Canso a Antioca. An Occitan Epic Chronicle of the First Crusade. Ashgate: Aidenshot. Tagliamonte, S. and Baayen, R. H. (2012). ‘Models, forests and trees of York English: was / were variation as a case study for statistical practice’. Language Variation and Change 24: 135–78. Talmy, L. (2007). ‘Lexical typologies’. In Shopen, T. (ed.) Language Typology and Syntactic Description. III. Grammatical Categories and the Lexicon. Cambridge: Cambridge University Press, 66–168. Tesnière, L. (1988). Eléments de Syntaxe Structurale. Paris: Klincksieck. Thibault, P. and Sankoff, G. (1997). ‘A sociolinguistic study ofverb morphology in Montreal French: reanalyzing the alternation between avoir and être’. Presentation at Pennsylvania Linguistics Colloquium. Thielmann, P. (1885). ‘Habere mit dem Part. Perf. Pass.’. Archiv für lateinische Lexicographie und Grammatik 2: 372–423; 509–49. Thogmartin, C. (1979). ‘Old Mines, Missouri et la survivance du français dans la haute vallée de Mississippi’. In Valdman, A. (ed.), Le Français hors de la France. Paris: Champion, 111–18. Thomas, G. (2008). ‘Exploring the parameters of a Central European Sprachbund’. Canadian Slavonic Papers / Revue canadienne des slavistes 50: 123–53.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
446
Thomas, G. (2010). ‘Serbo-Croatian as a bridge between the Balkan and the Central European Sprachbünde’. Balkanistica 23: 371–88. Thomason, S. (2001). Language Contact. Washington: Georgetown University Press. Thomason, S. and Everett, D. (2001). ‘Pronoun borrowing’. In Chang, C., Houser, M., Kim, Y., Mortensen, Park-Doob, M., and Maziar, T. (eds), Proceedings of the 27th annual meeting of the Berkeley Linguistics Society. Berkeley: Berkeley Linguistics Society, 301–15. Thomason, S. and Kaufman, T. (1988). Language Contact, Creolization, and Genetic Linguistics. Berkeley / Los Angeles / Oxford: University of California Press. Thurneysen, R. (1892). ‘Zur Stellung des Verbums im Altfranzösischen’. ZRP 16: 289–307. Timberlake, A. (1977). ‘Reanalysis and actualization in syntactic change’. In Li, C. (ed.), Mechanisms of Syntactic Change. Austin and London: University of Texas Press, 141–77. TLFi = Trésor de la langue Française informatisé. ATILF—CNRS and Université de Lorraine. http://www.atilf.fr/tlfi Tobler, A. (1875). [Review of J. Le Coultre. De l’ordre des mots dans Chrétien de Troyes]. Vermischte Beiträge zur französischen Grammatik 5: 395–414. Tobler, A. and Lommatsch, E. (1936). Altfranzösisches Wörterbuch. Berlin: Weidmann. Togeby, K. (1982). Grammaire française, II, Les Formes Personnelles du Verbe. Copenhagen: Akademisk Forlag. Tolskaya, I. (2018) ‘Nanosyntax of Russian verbal prefixes’. In Baunaz et al., 205–36. Tomlinson, H. (1981). ‘Le Guernesiais—étude grammaticale et lexicale du parler normand de l’île de Guernesey’. University of Edinburgh: thesis. Tomlinson, H. (2008). A Descriptive Grammar of Guernsey French. Guernsey: Melody Press. Tortora, C. (1997). The Syntax and Semantics of the Weak Locatives. Newark, DE: University of Delaware dissertation. Tortora, C. (2014). A Comparative Grammar of Borgomanerese. New York: Oxford University Press. TPS = Transactions of the Philological Society. Trask, L. (1996). Historical linguistics. London: Arnold. Traugott, E. (2010a). ‘(Inter)subjectivity and (inter)subjectification: a reassessment’. In Davidse, K., Vandelanotte, L., and Cuyckens, H. (eds), Subjectification, Intersubjectification and Grammaticalization. Berlin / New York: Mouton de Gruyter, 29–71. Traugott, E. (2010b). ‘Dialogic contexts as motivations for syntactic change’. In Cloutier, R., Hamilton-Brehm, A., and Kretzschmar, W. (eds), Studies in the History of the English Language V: Variation and Change in English Grammar and Lexicon: Contemporary Approaches. Berlin / New York: Mouton de Gruyter, 11–27. Traugott, E. and Dasher, R. (2002). Regularity in Semantic Change. Cambridge / New York: Cambridge University Press. Treffers-Daller, J. and Willemyns, R. (2002). ‘Aspects of the Romance-Germanic language border: an introduction’. In Treffers-Daller, J. and Willemyns, R. (eds), Language Contact at the Romance-Germanic Language Border. Clevedon: Multilingual Matters, 1–8. Tremblay, M., Dupuis, F., and Dufresne, M. (2005). ‘The reanalysis of the French prepositional system: a case of grammaticalization in competing grammars.’ In Batllori, M., Hernanz, M.-Ll., Picallo, C., and Roca, F. (eds), Grammaticalization and Parametric Change. Oxford University Press, 109–23. Tristram, A. (2014). Variation and Change in French Morphosyntax: the case of collective nouns. Abindgon: Routledge.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
447
Trudgill, P. (1986). Dialects in Contact. Oxford: Blackwell. Tuten, D., Pato, E., and Schwarzwald, O. (2016). ‘Spanish, Astur-Leonese, NavarroAragonese, Judaeo-Spanish’. In Ledgeway and Maiden, 382–410. Tuttle, E. (1986). ‘The spread of esse as universal auxiliary in central Italo-Romance’, Medioevo romanzo 11: 229–87. UPNC = Université Populaire Normande du Coutançais (1995). Essai de grammaire de la langue normande. Périers: Garlan. Utheim, U. (2013). Auxiliaires et intransitivité en français et dans les langues romanes. University of Oslo: disssertation. van der Auwera, J. (ed.). (1998). Adverbial Constructions in the Languages of Europe. Berlin: De Gruyter. van der Voort, H. (2000). ‘A grammar of Kwaza’. University of Amsterdam: thesis. Van Gelderen, E. (2009). ‘Feature economy in the linguistic cycle’. In Crisma, P. and Longobardi, G. (eds), Historical Syntax and Linguistic Theory. Oxford: Oxford University Press, 93–109. van Reenen, P. and Schøsler, L. (2000). ‘The pragmatic functions of the Old French particles AINZ, APRES, DONC, LORS, OR, PUIS, and SI’. In Herring, S., van Reenen, P., and Schøsler, L. (eds), Textual Parameters in Older Languages. Amsterdam / Philadelphia, PA: Benjamins, 59–105. Van Valin, R. (1990). ‘Semantic parameters of split intransitivity’. Language 66: 221–60. Van Valin, R. and LaPolla, R. (1997). Syntax: structure, meaning and function. Cambridge: Cambridge University Press. Vance, B. (1987). ‘The evolution of prodrop in Medieval French’. Cornell, ms. Vance, B. (1988). ‘L’évolution de pro-drop en français médiéval’. Revue québécoise de linguistique théorique et appliquée 7: 85–109. Vance, B. (1997). Syntactic Change in Medieval French: verb-second and null subjects. Dordrecht: Kluwer. Vance, B., Donaldson, B., and Steiner, B. (2009). ‘V2 loss in old French and old Occitan: the role of fronted clauses’. In Colina, S., Olarrea, A., and Carvalho, A. M. (eds), Romance linguistics 2009: selected papers from the 39th Linguistic Symposium on Romance Languages (LSRL), Tucson, Arizona, March 2009. Amsterdam / Philadelphia, PA: Benjamins, 301–20. Vance, G. (1993). ‘Verb-first declaratives introduced by et and the position of pro in old and middle French’. Lingua 89: 281–314. Vance, G. (1995). ‘On the decline of verb movement to Comp in old and middle French’. In Roberts and Battye, 173–99. Vanelli, L. (1995). ‘La deissi’. In Renzi, L., Salvi, G., and Cardinaletti, A. (eds), Grande grammatica italiana di consultazione. Bologna: Mulino, 261–468. Vanelli, L., Renzi, L., and Benincà, P. (1985). ‘Typologie des pronoms sujet dans les langues romanes’. Actes due VII Congrès International de Linguistique et Philologie Romanes. III. Amsterdam: North Holland, 163–76. Vanelli, L., Renzi, L., and Benincà, P. (1986). ‘Tipologia dei pronomi soggetto nelle lingue romanze medievali’. Quaderni patavini di linguistica 5: 49–66. Vasseur, G. (1996). Grammaire des parlers picards du Vimeu (Somme) avec considération spéciale du dialecte de Nibas. Abbeville: Paillart. Veloudis, I. (2018). ‘Exploiting polar questions for expressive purposes: “queclaratives” and “whimperatives” in Modern Greek’. Lingua 204: 117–33. Vermesse, L. (1861). Vocabulaire du patois de Lille. Lille: Behague. Verne, J. (1892). Le Château des Carpathes Paris: Hetzel.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
448
Vikner, S. (1995). Verb Movement and Expletive Subjects in the Germanic Languages. New York: Oxford University Press. Vildomec, V. (1971). Multilingualism. Leiden: A.W. Sijthoff. Villa-García, J. (2015). The Syntax of Multiple-que Sentences in Spanish. Along the Left Periphery. Amsterdam: Benjamins. Villeneuve, A. J. and Auger, J. (2013). ‘ “Chtileu qu’i m’freumereu m’bouque i n’est point coér au monne”: grammatical variation and diglossia in Picardie’. JFLS 23: 109–33. Villeneuve, A.-J. (2016). ‘Entre vernaculaire et ‘standard’: le français québécois parlé en contexte d’entrevue’. Presentation at Les français d’ici, Université de Saint-Boniface (Winnipeg, Canada), June 7–9. Villeneuve, A.-J. and Comeau, P. (2016). ‘Breaking down temporal distance in a continental French variety: future temporal reference in Vimeu’. Canadian Journal of Linguistics 61: 314–36. Villers, M.-E. de (2009). Multidictionnaire de la langue française. Montréal: Québec Amérique. Vincent, N. (1982). ‘The Development of the Auxiliaries habere and esse in Romance’. In Vincent, N. and Harris, M. (eds), Studies in the Romance Verb. London: Croom Helm, 71–96. Vincent, N. (1993). ‘Head- vs. dependent-marking: the case of the clause’. In Corbett, G., Fraser, N., and McGlashan, S. (eds), Heads in Grammatical Theory. Cambridge: Cambridge University Press, 140–63. Vincent, N. (1997). ‘The emergence of the D-system in Romance’. In Kemenade and Vincent, 149–69. Vincent, N. (1999). ‘The evolution of c-structure. Prepositions and PPs from IndoEuropean to Romance’. Linguistics 37: 1111–54. Vincent, N. (2014). ‘Compositionality and change’. In Bowern, C. and Evans, B. (eds), The Routledge Handbook of Historical Linguistics. London: Routledge, 103–23. Vincent, N. (2017). ‘Parts and particles: the story of ’. In Los, B. and de Haan, P. (eds) Word Order Change in Acquisition and Language Contact: essays in honour of Ans van Kemenade. Amsterdam: Benjamins, 292–310. Voorhoeve, C. (1994). ‘Contact-induced change in the non-Austronesian languages in the north Moluccas, Indonesia’. In Dutton, T. and Darrell, T. (eds), Language Contact and Language Change in the Austronesian World. Berlin: Mouton de Gruyter, 649–74. Wales, L. (2002). ‘The relative frequency of the synthetic and composite futures in the newspaper Ouest-France and some observations on distribution’. JFLS 12: 73–93. Walkden, G. (2017). ‘Language contact and V3 in Germanic varieties new and old’. Journal of Comparative Germanic Linguistics 20: 49–81. Wanner, D. (1987). The Development of Romance Clitic Pronouns: from Latin to Old Romance. Berlin / New York: Mouton de Gruyter. Wartburg, W. von (1950) Die Ausgliederung der romanischen Sprachräume. Bern: Francke. Wartburg, W. von (1958). Évolution et structure de la langue française. Bern: Francke. Wartburg, W. von. (1922–2003). Französisches etymologisches Wörterbuch. Basel: Zbinden. Weinreich, U. (1964). Languages in contact. London / The Hague / Paris: Mouton. Weinreich, U. (1968). Languages in Contact: findings and problems. The Hague: Mouton. Wheeler, M. (1988). ‘Occitan’. In Harris, M. and Vincent, N. (eds), The Romance Languages. Oxford: Oxford University Press, 246–78. Wheeler, M. (2011). ‘The evolution of a morphome in Catalan verb inflection’. In Maiden et al., 183–209.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
449
Willis, D. (2017). ‘Degrammaticalization’. In Ledgeway, A. and Roberts, I. (eds) The Cambridge Handbook of Historical Syntax. Cambridge: Cambridge University Press, 28–48. Willis, L. (2000). “Être ou ne plus être?” Auxiliary alternation in Ottawa-Hull French. Université d’Ottawa: dissertation. Wiltschko, M. (1997). ‘Parasitic operators in German left-dislocations’. In Anagnostopoulou, E., Riemsdijk, H. van, and Zwarts, F. (eds), Materials on Left-dislocation. Amsterdam: Benjamins, 307–39. Wiltschko, M. (2014). The Universal Structure of Categories. Towards a Formal Typology. Cambridge: Cambridge University Press. Winford, D. (2003). An Introduction to Contact Linguistics. Oxford: Blackwell. Wolfe, S. (2012). ‘Word order variation in Old Sardinian’. University of Cambridge: thesis. Wolfe, S. (2014). ‘Syntactic variation in the medieval Romance languages: a new perspective on verb-initial orders’. Presentation at Current Issues in Linguistic Variation: the 14th International Conference of the Department of Linguistics’, Bucharest. Wolfe, S. (2015a). ‘Microparametric variation in old Italo-Romance syntax: the view from old Sicilian and old Sardinian’. In Aboh, E. (ed.), Romance Languages and Linguistic Theory 2013: Selected Papers from ‘Going Romance’ Amsterdam 2013. Amsterdam: John Benjamins, 51–66. Wolfe, S. (2015b). ‘The old Sardinian Condaghes. A syntactic study’. TPS 113: 177–205. Wolfe, S. (2015c). Microvariation in medieval Romance syntax: a comparative study. University of Cambridge: thesis. Wolfe, S. (2015d). ‘Microvariation in old Italo-Romance syntax: the view from old Sardinian and old Sicilian’. Archivio Glottologico Italiano 100: 3–36. Wolfe, S. (2015e). ‘Verb-initial orders in old Romance: a comparative account’. Revue roumaine de linguistique 60: 48–172. Wolfe, S. (2015f ). ‘The old Sardinian condaghes. A syntactic study’. TPS 113: 177–205. Wolfe, S. (2015g). ‘The nature of old Spanish verb second reconsidered’. Lingua 164: 132–55. Wolfe, S. (2016a). ‘On the left periphery of V2 languages’. Rivista di Grammatica Generativa: Selected Papers from the 41st Incontro di Grammatica Generativa 38: 287–310. Wolfe, S. (2016b). ‘A comparative perspective on the evolution of Romance clausal structure’. Diachronica 33: 461–502. Wolfe, S. (2017). ‘Syntactic variation in two sister languages: a study of word order in old French and old Occitan’. In Zafiu, R., Dindelegan, G., Dragomirescu, A., and Nicolae, A. (eds), Romance Syntax. Comparative and Diachronic Perspectives. Newcastle upon Tyne: Cambridge Scholars, 53–85. Wolfe, S. (2018a). ‘Probing the syntax of a problematic particle: Old French ‘si’ revisited’. TPS 116: 332–62. Wolfe, S. (2018b). ‘Syntactic variation in two sister languages: a study of word order in old French and old Occitan’. In Pană Dindelegan, G., Dragomirescu, A., Nicula, I., and Nicolae, A. (eds), Comparative and Diachronic Perspectives on Romance Syntax. Newcastle upon Tyne: Cambridge Scholars Publishing, 53–83. Wolfe, S. (2018c). ‘Occitan, verb second, and the medieval Romance word order debate’. In Berns, J., Jacobs, H., and Nouveau, D. (eds), Romance Languages and Linguistic Theory 2015: Selected Papers from Going Romance Nijmegen 2015. Amsterdam: Benjamins, 315–36. Wolfe, S. (2019). Verb Second in Medieval Romance. Oxford: Oxford University Press. Wolfe, S. (forthcoming). ‘Rethinking medieval Romance V2’. In Woods, R. and Wolfe, S. (eds), Rethinking Verb Second. Oxford: Oxford University Press.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
450
Wolfe, S. (ms). ‘Variation and change in the medieval French subject system’. University of Oxford: thesis. Wolfe, S. and Woods, R. (forthcoming). ‘Verb second in the 21st century’. In Woods, R. and Wolfe, S. (eds) Rethinking Verb Second. Oxford: Oxford University Press. Woolhiser, C. (2005). ‘Political borders and dialect divergence / convergence’. In Auer, P., Hinskens, F., and Kerswill, P. (eds), Dialect Change: Convergence and Divergence in European Languages. Cambridge: Cambridge University Press, 236–62. Wright, R. (2013). ‘Periodization’. In Maiden, Smith, and Ledgeway, 107–24. Wüest, J. (1985). ‘Les énonciatifs gascons et la théorie de l’énonciation’. In Kristol, A. and Wüest, J. (eds), Drin de tot: travaux de sociolinguistique et de dialetcologie béarnaise. Bern: Lang, 285–307. Wurmbrand, S. (2001). Infinitives: Restructuring and Clause Structure. Berlin / New York: Mouton de Gruyter. Ylikoski, J. (2016). ‘Degrammaticalization in North Saami: development of adpositions, adverbs and a free lexical noun from inflectional and derivational suffixes’. FinnischUgrische Mitteilungen 40: 113–73. Zafiu, R. (2013a). ‘Information structure’. In Pană Dindelegan, G. (ed.) The Grammar of Romanian. Oxford: Oxford University Press, 568–74. Zafiu, R. (2013b). ‘Tense, aspect, mood’. In Pană Dindelegan, G. (ed.). The Grammar of Romanian. Oxford: Oxford University Press, 24–65. Zafiu, R. (2016). ‘The syntax of moods and tenses’. In Pană Dindelegan, G. (ed.), The Syntax of Old Romanian. Oxford: Oxford University Press, 14–52. Zamboni, A. (1998). ‘Dal latino tardo al romanzo arcaico: aspetti diacronico-tipologici della flessione nominale’. In Ramat, P. and Roma, E. (eds), Sintassi storica. Atti del XXX Congresso Internazionale della Società di Linguistica Italiana, Pavia, 26–28 settembre 1996. Rome: Bulzoni, 127–46. Zamboni, A. (2000). Alle origini dell’italiano: dinamiche e tipologie della transizione dal latino. Rome: Carocci. Zanuttini, R. (1997). Negation and Clausal Structure. Oxford: Oxford University Press. Zaring, L. (1998). ‘Object shift in Old French’. In Schwegler, A., Tranel, B., and UribeEtxebarria, M. (eds), Romance Linguistics: Theoretical Perspectives. Amsterdam: Benjamins, 319–33. Zaring, L. (2010). ‘Changing from OV to VO: more evidence from Old French’. Revista Philologica Romanica 10: 1–18. Zaring, L. (2011). ‘On the nature of OV and VO order in Old French’. Lingua 121: 1831–52. Zauner, A. (1896). ‘Die Konjugation im Béarnischen’. ZRP 20: 433–70. Zimmer, D. (1994). ‘ “Ça va tu marcher, ça marchera tu pas, je le sais pas.” Le futur simple et le futur périphrastique dans le français parlé à Montréal’. Langues et linguistique 20: 213–26. Zimmermann, M. (2014). Expletive and Referential Subject Pronouns in Medieval French. Berlin / New York: Mouton. ZRP = Zeitschrift für romanische Philologie. Zubizarreta, M. L. (1998). Prosody, Focus, and Word Order. Cambridge, MA: MIT Press. Zumthor, P. (1963). Langue et techniques poétiques à l’époque romane (XIe–XIIIe siècles). Paris: Klincksieck.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
Index Note: Figures are indicated by an italic ‘f ’, and notes are indicated by ‘n’ following the page number. ab 315 abstractivist morphology 400 Académie Française 346 Ackema, P. 215 Ackerman, F. 399 Adams, J. 169 adverb locative 47 scene-setting 19 temporal 47 adverbial clause 47, 51, 67 age 222, 225, 235f. age-grading 264 agreement 165 Ahlborn, G. 390 Aikhenvald, A. 280 aimer mieux 149 AIS 388 ALF 389, 391 alignment 181, 191, 198 active-inactive 198f., 206, 208f., 211 active-middle 193f., 198f., 206–9, 211 active-stative 161 nominative-accusative 171, 193f., 207–11 aller 149 Alsace 391 amerrir 394f. analogical change, analogy 368, 374 extension 369–71, 381 levelling 376 remodelling 369, 376 morphological 365, 370 analyticity 191, 206 Anglade, J. 45–6, 312–15 animacy 232f., 237f., 240 anteriors 242 Antonini, A. 261, 263 Antrona 390 aoristic drift 191–3, 204, 208, 212 apparent time 270 apposition 326f., 334, 342 Aquitaine 318, 322 Aranovich, R. 214 argot 344–6 argument structure 218f. Armenian 192
Aromanian 399 Aronoff, M. 364 articles 73, 81 assertion 126, 129 aterrir 394f. Atlas Linguistique de la France 282, 284, 286, 289, 295f. Atlas Linguistique et Ethnographique Normand 282, 287, 289 attention to speech 227 Aub-Büscher, G. 220 Audi, R. 247 Auger, J. 220 Augsburg Passion 315, 319, 321 autonomous morphology 364, 381 Auvergne 307, 322f., 374 auxiliary Auxiliary Selection Hierarchy 215 perfect auxiliary 55, 192–3, 199–200, 208, 210 selection 10, 171–81, 214f., 217–20, 222–4, 226–40 Avanzi, M. 214, 238 Ayres-Bennet, W. 140, 144, 146, 149 Baayen, R. H. 271 Baerman, M. 364, 368, 378, 384 Bagnes, dialect of 331f., 334, 336, 338 Bantu 110–13 Bardovi-Harlig, K. 248 Basque 84f., 98f., 113–15 Bauche, H. 219 Bauer, B. 326–8, 331, 336, 344, 346 Béarn, béarnais 259, 276 Bec, P. 10, 12 Bélanger, M. 214, 220 Belgium 391 Benincà, P. 13, 18, 22, 57, 75, 82 Bentley, D. 221 Benveniste, E. 219, 242, 327f. Berschin, H. 315, 319f. Berschin, W. 315, 319f. Bigot, D. 227 bilingualism 226, 258–60, 265, 270f., 274f., 277f., 282, 296, 307f., 322f. Birt, P. 295
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
452
Blevins, J. 400 Blondeau, H. 261, 264, 266f., 268f., 273 *boˈtare 396 booklets 309–11 books 309–11, 318, 322 borrowing 279–81, 303 lexical 284, 304 bottleneck effect 44, 65, 88–9, 97 Bouzet, J. 83–4, 94 Brasseur, P. 281f., 284, 287, 289, 294f. Bronckart, J.-P. 248 Burzio, L. 215, 218 calquing 288, 298f., 304 Canadian French 213–40 Canale, M. 214, 227–9, 231 Carstairs[-McCarthy], A. 380 cartographic enterprise 51 case agreement 327 case inflexion 191, 196, 206, 209, 211 case-system 312 Castilian 217 Catalan 10, 143, 162, 177–81, 188, 211, 214–17, 367 Algherese 179 Balearic 179 causative 140, 146 central vs peripheral clauses 92, 94–5, 98 -cevoir 396f. Chabaneau, C. 369, 372, 381 Chambon, J.-P. 318f., 322f. change in apparent time 151 in real time 151 Channel Islands 5 Charlemagne Sprachbund 170, 173–81 Chaudenson, R. 220 church 321 Cinque, G. 86, 94 Clark, E. 223 clash (of morphological patterns) 381, 383f. classification 3, 10 clause-typing 87 cleft 4, 11, 102 Clermont-Ferrand manuscript 307, 315, 323 clitic, clitics (see also pronouns) climbing 142, 147–57 placement 4, 9, 76, 146 pronouns 280, 300, 304 subject 71f., 103, 193–6, 198f., 206–8, 211 Collas, J. 282, 287, 289, 295f. collinear, collinearity 268f., 271 Comeau, P. 261, 270, 273f., 276 compensatory lengthening 371f.
C(omplementizer) (Phrase) 55, 63, 73f., 79, 82, 96, 104 complex constructions 140, 147, 151 complex inversion 57 complex predicates 143, 147 composition 327, 344 compound tenses 154, 233f., 238 compounds 326–46 additional 328f.,334f., 339, 344 appositive 5, 326–46 determinative 327, 333, 340 hybrid 328f., 333–7, 339, 341f., 344 serial 331, 344, 346 type-of 328–30, 335–9, 342, 344–6 Comrie, B. 242 concessive 74 Concordance de l’Occitan médiéval (COM2) 118 conditional 365, 367–9, 371–4 configurationality 196, 199 conjugation class 368, 378, 385–401 first conjugation 365, 367, 369, 371–3, 375, 377f. second conjugation 367, 369 third conjugation 365, 367, 368f., 372–6 conservatism 40 constructivist morphology 400 contact-related change 281 contamination 390–401 contingency 270, 275 coordination 152 copula verbs 230f. corpus study 118, 138, 140, 144–5, 147, 157 Corr, A. 77 Corsican 194, 214, 219 Coseriu, E. 38 Cotentin peninsula 280, 282, 284, 286, 288, 289, 294f. Council of Tours 312 Croft, W. 218 Croissant linguistique 365 crowdsourcing 214, 238f. Cruschina, S. 11, 75 Daco-Romance 192 Damourette, J. 255 dating of change 148–9, 156, 176 Dauby, J. 220 De Garis, M. 287, 295 De Melo, W. 388 De Poerck, G. 308, 318 declarative 79, 85–6, 89 Dees, A. 309 definite article 312–15 definiteness 73
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
deontic 247, 253, 257 dependent marking 193, 199, 206–7, 211 deponent 161, 167–8, 175–6, 191 Descusses, M. 219 devoir 149, 397 dialect contact 113–15, 156 dialect mixing 2 diatopic variation 139 Differential Object Marking 194, 212 Dion, N. 262, 267, 269 directionality 381 discourse pragmatics 15, 17f., 28, 35, 50, 52–5 Dixon, R. 280 Donaldson, B. 13, 72 Dorian, N. 290 Dowty, D. 216 drift 288 Drinka, B. 171f., 242 Duraffour, A. 390 Dutch 46, 60f., 187, 215, 218 dvandva 328f., 331 écart rhétorique 309 edge feature (see also EPP) 37, 74, 98 Edmonds, A. 263, 267–9, 272 education, level of 226, 239 egressive 245, 247, 256 embedded clause 92 Emirkanian, L. 263f. emphasis 122 enclisis 74, 76, 142 England 321, 323 English 77, 279–82, 284, 296–8, 302, 304 contact with 226 entropy, conditional 399 enunciative particle 73, 84, 87, 96, 103, 109 system 71, 77–9 envoyer 394 Epître farcie de St Etienne 307 EPP (see also edge feature) 44, 60, 64, 73f., 82f. Erkanbald 319 Escolas Calandretas 259f. Esher, L. 374f., 379, 382f. 4, 176, 211 estovoir 393, 397 être 149 Eulalia sequence 308f. event 245, 247, 252, 255 Everett, D. 279f. evidential 80, 86, 89 exclamative 74–5, 77, 80–1, 86, 89 exoglossic standard 281 expletive subject clitic 196 Eythorsson, E. 221
453
falloir 391–3 Fankhauser, F. 390 Feature Scattering Hypothesis 87 feature structure 378 Fenno-Swedish 42 Ferry, J. 259 FEW 393f. fin(iteness) (phrase) 18, 21, 23, 45, 55, 87, 93–4 First World War 327, 340, 343, 345 Flege, J. 260 Fleischman, S. 21, 34, 45, 51, 242f., 261 Flemish 338f. Fleury, J. 294 Flutre, L.-F. 219 Flydal, L. 261 flyleaves 309–11 focus contrastive 125 Focus (Phrase) 13, 17–8, 22, 25, 27, 29, 44, 57, 66, 88–90, 97, 105–7, 110–13, 124 information 90, 134 Force (Phrase) 23, 45, 55, 63, 87, 93f. frame-setter 22–3, 27, 51, 63, 66 Français de nos régions 238f. Franco-American 235 Francophonie 223 Francoprovençal 72, 143, 259, 308f., 313–18, 323, 331, 385, 388–90, 399 Frascarelli, M. 44f., 57 free relative 50 free variation 217, 221–3 Frei, H. 219, 340, 345 French 1, 3–5, 55, 71–9, 81, 83, 86, 106, 128, 138–40, 142, 144, 155–7, 181, 187, 195–203, 307–23, 326–46, 364, 367, 369, 371, 379, 382, 385–7, 392–8 Acadian 214, 219, 224, 226, 263f., 267f., 272–4 of Alberta 224 Cajun 224, 228 Canadian 262–4, 267, 269f., 276 European 219, 229, 235 Laurentian 220, 224, 240, 262f., 266f., 270, 272f., 276 of Manitoba 224, 228, 232 of Martinique 263f., 267, 269, 273f., 276 middle French 12, 38, 142 Mitchif 224 of Montréal 213–40 of New England 214, 224f. non-standard 326 North American 214, 224, 232 old French 1, 9, 14–24, 41, 43, 47–51, 72, 120, 141–2, 156, 308, 312, 314 of Ontario 213f., 220, 224, 226, 227–9, 231
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
454
French (cont.) popular 219, 230 Québécois 220, 227f., 230, 239 southern 272–4, 276 standard 213, 217, 219, 221, 229, 332, 334–7, 339, 344–6 of Vermont 224f., 228f. frequency 218, 227–9, 231 Friulian 72, 198, 208 fronting 17, 48, 51, 74 Functional Grammar 3 functional syntax 191, 198, 204 fusional morphology 398 future 5, 365, 367–9, 383 epistemic 253, 255–7 future temporal reference 258–78 Gagnon, S. 263 Galet, Y. 140, 146 Gallo 202 Gallo-Italian 192, 194, 210 Gallo-Romance 5, 9, 27, 32, 138, 142–3, 144, 155–6, 157, 194, 206, 212 northern 181–8, 191, 193f., 196, 199, 204–7 southern 176–81 Gardette, P. 388 Gartempe 365–78, 380f., 383f. Garzonio, J. 117, 136 Gascon 4, 71–3, 78–85, 92, 96–8, 100–16, 142, 209, 219, 258–78, 331–4, 338 Gauchat, L. 391 gender (of speakers) 222, 225, 235, 239 Gender Paradox 225 gender 288f., 297, 304, 330f., 334, 343, 346 generativism 1, 155 Genevan 390 German 187, 192, 215, 218, 322 Germanic 16, 64, 88, 175, 328f., 335–8, 343, 345 ges 128–36 Giancarli, P.-D. 219–21 Gilliéron, J. 391 Giorgi, A. 87 Giurgea, I. 75 Goldvarb 235 Gosselin, L. 245f., 255–7 Gossen, C. 308, 318 goutte 128 Government and Binding Theory 3, 9 gradual change 140 grammaticality judgements 234f., 239 grammaticalization 21, 23, 84, 117, 169, 261, 268, 276 Graulhet 379f. Greenberg, J. 330
Grimm, R. 262f., 266, 269f. Grzega, J. 393 Guardiano, C. 73 Gudmestad, A. 263, 267–9, 272 Guernesiais 280, 282, 295f. Guernsey 280, 282, 287, 296 Guiraud, P. 219 4, 175–6, 200–3, 211 Haegeman, L. 92 hagiographical texts 145, 147–9, 157 Hallion, S. 221, 228, 232 Hanging Topic 45, 63 head marking 191, 196, 207, 210, 212 Head Movement 144 Hendschel, L. 219 Hickey, R. 260 Hilty, G. 308, 315 Hinterhölzl, R. 44f., 57 Hinzelin, M.-O. 144, 378, 380–4 hodiernal aspect 236f., 240 Holmberg, A. 74 homonymic clash 391 hortative, exhortative 74 hypothetical 74 Ibero-Romance 77, 96, 100, 142, 210, 385 Illiez 390 illocutionary force 74, 78 imminence 267f., 271, 273f., 276 imperative 74–7, 86, 89, 103f., 380f., 399 imperfect indicative 365–9, 371–4 impersonal pronoun 286–8, 291 Indo-European 326–31, 344 ineffability of lexemes 398 infectum 382 infinitive 3, 138–42, 147, 152, 398f. inflexion 192, 194, 206–8 inflexion class (see also conjugation) 385–401 default inflexion class 387, 395 I(nflexion) P(hrase) see T(ense) P(hrase) ingressive 245, 247, 256 innovation 40 Institut d’Estudis Occitans 259 intergenerational transmission 259 International Society for Historical Linguistics 3 interposition 155 interrogative 74–7, 80f., 86, 89, 126 interspeaker variation 228 intertextual variation 15, 21f. interval of enunciation 245–8, 250f., 253, 255f. interval of reference 245–8, 256 intransitive verbs 214f.,218f., 228–30, 233f. inversion 14f., 72, 74, 77
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
Italian 1, 41, 77, 142–3, 214–19, 221, 386 Italo-Romance 3, 40, 142, 162, 382f. iterative 235f. Jaberg, K. 390, 396 Jake, J. 280, 304 Jaquenod, F. 390 Javanaud, P. 370 Jensen, F. 141, 242 Jèrriais 202f., 280, 282–4, 286, 295 Jersey 280, 282 Jespersen’s Cycle 85–6 Johnson, D. 235f., 271 Jonas sermon 312 Jones, Mari 272, 295, 304 Jones, Michael 220 Josephson, F. 243 journalistic data 145, 153 Jud, K. 393 jussive 74 Kailuweit, R. 221, 229 Kaiser, G. 74 Kayne, R. 143, 155, 215 Keller, O. 396 King, R. 214, 220, 228, 262f., 266f., 276 Kiparsky, P. 51 Knaus, V. 226, 229 Koschwitz, E. 308 Kukenheim, L. 312–15 L’Huillier, M. 243f. La Gleize 331f., 338f., 345 Labeau, E. 261, 268, 273 Laberge, S. 226 Labov, W. 225, 281 Ladin 74, 204, 208, 390 language contact 2, 84, 98–9, 144, 156, 279 language death 225 language shift 258 langue d’oc, Languedocien 220, 259 langue d’oïl 259, 284 Languedoc 373 Lass, R. 38 Late Merge Principle 99 Latin 3, 45, 73, 99, 120, 171, 191f., 194, 197–9, 315f., 318f., 327f., 373, 385–8, 393, 396f., 399 Carolingian 170, 172, 175f., 178 late Latin 170, 175f. Lavalade, Y. 368, 373 Law, D. 280, 303f. Ledgeway, A. 10, 76, 81, 125, 141, 161, 165, 171, 188–90, 214, 217–20
left dislocation 11, 13, 19, 56–7, 64 Left Periphery 4, 14–21, 44–5, 125 left-branching 330f, 338f., 343, 345 Legendre, G. 215–17, 221 Lemosin 5, 220, 331–3, 337f., 364–84 Lepelley, R. 284f. Les Esseintes 106–15 Lesage, R. 263 lexical effect 222, 227 lexically specified stress 379 Liège 186f. Lightfoot, D. 240 Limousin (see Lemosin) linguistic insecurity 240 linguistic market index 226 literary texts 156 Lodge, R. A. 322 logistic regression 235 Loire Valley 322f. Lombardi, A. 76 Longobardi, G. 73 Loporcaro, M. 215, 218, 221f. Lorrain 220 Lusignan, S. 321 Lyon 145 Mackenzie, I. 218f. MacLean, S. 319 Maiden, M. 219, 364, 382–4, 393, 399 Maling, J. 55 Malouf, R. 399 Manente, M. 220 Manzini, R. 75, 94 Marcus, N. 78 Martineau, F. 139 Matras, Y. 279, 304 matrix / embedded asymmetries 13, 24, 61–2, 109 Matrix Language Frame model 296 matrix verb 138–140, 146–7, 150–1, 154 Maupas, C. 261, 263 maximal vs minimal category 58, 60 Mayan 303 McLaughlin, M. 140, 145–6, 148, 152, 157 Medina Granda, R. M. 128 Merge 73, 76–7, 79, 81–2, 84, 98–9 metaphor 326, 329, 336–9, 342–6, 344f. metonymy 326, 329, 335, 337, 339, 342, 345 Meyer-Lübke, W. 388, 392, 396 Meyer, P. 315–17, 321 microparameter 75 mie 128 minority communities 229
455
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
456
Mitchif (see French) 395 mixed-effects modelling 235 regression 271, 274 modal verbs 149 modality 253 Mons 391 mood-based variation 221 Moreux, B. 259 morphemes 399, 401 morphological derivation 231 morphomes 382 metamorphomes 364f., 379–84 Mortensen, L. 309, 311 Mougeon, R. 214, 220, 226 move 73, 77, 79, 81, 84, 98–9 multi 57 multilingualism 5 Munaro, N. 75, 87 Myers-Scotton, C. 296
317f. null subjects 23, 33, 71–2, 98–9, 138–9, 155, 181, 195, 281, 292 number 218f., 231f. Nyrop, K. 394, 397
N-pattern 379f., 382f. Nadasdi, T. 214, 226, 229, 263, 266–70, 272f., 276 nationalism 234 Neapolitan (old) 81f. negation 84–6, 96, 103f., 117–37, 153 constituent 133 emphatic 117 postverbal 4, 117–37 negative polar question 125, 127 negative polarity item 122, 128 negative wh-question 126 New France 139 new information 52, 54f., 136 new speakers 260, 275 newspapers 145, 157 Nicolet, N. 390 Niederländer, J. 391 Nithard 312 nominal apposition 334 nominative forms 317f. non-standard varieties 218, 222f. non-veridical polarity 72, 74–5, 81–2, 97 Nontron 368, 370, 373, 375f. Norman 5, 161, 182–4, 279–304 insular 280 mainland 280, 282 Norman French 203 Normandy 280, 288 North America 139 Northern Italian dialects 13, 72, 75–6, 194, 210
Pacuvius 388 Paris 72, 145, 148–50, 156, 322 participle 162, 185f., 191–2, 198, 201–5, 209, 237f., 228, 230, 240 particle 9, 16, 20–3, 41, 78 partitive 135 pas 119–28 Pas-de-Calais 397 passé composé 4, 241–57 passé simple 268, 273 passive 167f., 191, 206 past tense 192f., 203f., 208, 210 pattern replication 297, 304 Paul, H. 390 Pellat, J.-C. 242, 255 perception verb 140, 146 perfect 4, 161, 167f., 175f., 191f., 194 perfectum 382 Périgord 381 periodization 3f., 12, 38–40 periphrasis 164, 373 periphrastic tenses 213, 218, 240 Perlmutter, D. 197, 215f., 218 person 218, 221, 231f., 269–74 Pescarini, D. 75 philology 4 phonological merger 368 phonologically predictable stress 379 Picard 182–4, 200, 219, 270, 271, 276, 331 Pichon, E. 255 Picone, M. 228 Piedmontese 218
O’Neill, P. 377 Observer’s Paradox 281 Occitan 4–5, 72, 85, 102, 138–9, 141, 144, 148, 155–7, 162, 176–7, 188, 194, 209–10, 214, 218–20, 241–5, 249–54, 307–23, 331 middle Occitan 12, 38 old Occitan 9, 13, 24–36, 41, 43, 51–5, 67–9, 117–37, 156, 176, 312, 314, 368 occupational term 331, 344, 346 Oïl dialects 214, 219 old information 62, 123, 135 Olivier, P. 318f., 322 optative 74, 76–7, 80–1, 86, 89 osoir 391, 398 Oxford Latin Dictionary 388
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
pleuvoir 385f., 399, 401 Plungian, V. 247 poilu 340–5 point of reference 245 polarity projection 85 Poletto, C. 23, 55, 73–5, 85, 117, 136 Pooley, T. 220 Poplack, S. 262f., 265, 267–70, 273, 275 Portuguese 77, 142–3, 144, 170, 192, 217 Brazilian 143 European 143, 192 pouvoir 149, 391, 393, 397f. pragmatics 3, 129 predicates 138 prepositional phrase 47 prepositions 141, 146, 315 present futurate 262 indicative 365–83 présent accompli 241–57 présent inclusif 241–57 present perfect 242 present subjunctive 373, 375–80 presupposition 86, 124, 131–2, 136 preterite 172, 178, 194, 198–200, 203–5, 365, 367f., 369–71, 382f. preverbal position objects of objects 17f., 22 of subjects 82f., 97 pro-drop (see null subjects) productivity 365, 374, 380 pronominal verbs 215, 221, 231f., 234, 238 pronouns (see also clitics, impersonal, reflexive, resumptive) borrowing 280f., 303 conjunctive 284–91 disjunctive 280, 292–4, 304 object 139–40, 143 personal 279 sharing 281, 284, 286, 288f., 291f., 296f., 304 unstressed (see also clitics) 314f. propositional complements 92 proto-Indo-European 73 Provençal 32, 220, 332, 334f., 336, 345 Provence 373 Pusch, C. 78, 80, 82, 102 quand 270 317f. quantifier 133f. quantity (of vowels) 370 Quint, N. 365, 369f., 374, 376 quotative 74–7, 80, 86, 89
457
Rabelais 393 Raeto-Romance 162, 194, 198, 204, 208–10, 214, 218 Rbrul 235, 271 Rea, B. 221 real language in use 222 realis 73 recomplementation 63, 95f. reflexive pronouns 295, 300, 302f. regional variation 10 registers 227 Reichenbach, H. 245, 247 relative clause 48 Relativized Minimality 88 Remacle, L. 184, 205, 219, 308, 332, 338f. Remberger, E.-M. 75 Renaud 225, 227, 230–2 Renault, G. 340–3 restrictive relative clauses 94 restructuring 143 resumption 41 resumptive generalized resumptive 46, 58, 66 resumptive pronoun 64 specialized resumptive 46, 60f. rhetorical question 126 rhyme 396f. Riegel, M. 242, 255 right-branching 330, 339, 345 Rioul, R. 242, 255 Rizzi, L. 13, 44, 74, 87f. Roberts, I. 21, 74, 76, 81, 143 Roberts, J. 214, 225, 228f. Roberts, N. 262–4, 267, 269f., 272f., 276 Robustelli, C. 219 Rohlfs, G. 103, 332 Romance 3, 326f. medieval 74, 81f. Romance 2, 191 Northern vs Southern Romance 4, 10, 161, 191–4, 199, 207, 209–10 Romanian 3, 77, 167, 208, 214, 385–7 Romansh 387, 390 Ronjat, J. 220, 332–5, 372 roof language 189 root allomorphy 365 roots 379 Rosen, C. 215 Round, E. 364 Roussel, B. 229–32 Roussou, A. 21 Roux, J. 220 Ruffieu 390 Russo, M. 214, 225, 228f.
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
458
S-Curve 140 så 42, 64 saliency 218f. Salvi, G. 74 Sampson, R. 369 Sanfelici, E. 73 Sanilhac 368, 370, 373, 375f. Sankoff, G. 62–5, 213, 220–38, 269f. Sapir, E. 288 Sardinian 192, 194, 210, 212, 214, 218, 220 Sark 280, 282, 289 Saussure, L. de 241, 251–53, 255, Savoia, L. 75 Schilling, N. 226 segmentation 401 semantics 3 317 Sercquiais 203, 280, 282, 286, 288, 295 seventeenth century 145–6 si 4, 13, 20–3, 26, 34–6, 42, 45–68, 81 Sicilian 32, 202, 217 Sicre, P. 220 Sinclair, H. 248 skeuomorphy 186 Smith, J. C. 1, 6, 12, 38, 71, 161, 181, 220, 223, 231, 240, 365, 382, 394f. sociodemographic changes 234 socioeconomic/socioprofessional status 225f., 233, 236–8, 240 sociolinguistics 3 sociolinguistic corpus 233–8 sociolinguistic interviews 214, 216, 234f. Söhrman, I. 245 Somme 397 Sorace, A. 215–17, 221 Sottens 390 sound change 264, 370–2, 377–82, 384 southern Italian dialects 76, 136, 192, 210 Spanish 142–3, 144, 170, 197, 209 Porteño 192 Speech Act Phrase 87 speech communities 227 Speech Learning Model 260, 277 split intransitivity 4, 164, 171, 214f., 217f., 240 St Leger 307 standardization 157 Stelling, L. 226 Stichauer, P. 218 Strasbourg 319, 322 Strasbourg Oaths 307 Stratified Convergence Zone 189 stress assignment 370, 374 strong vs weak functional head 73–5, 81–3, 98f.
structural borrowing 281 stylistic fronting 55 subject 48, 53, 191, 194, 196, 198–200, 202, 204, 209 subject position 74 subjectivity 127 subjunctive 80, 92, 94, 268, 273 Sundell, L.-G. 253 superstrate 280f. suppletive roots 379–81 SVO 12, 330 Swedish 64 Swiss Romandy 391 syncretism (see also TAM) 364–84 complete 368, 378 non-directional 381 partial 378 synonymy avoidance 223 syntactic drift 304 syntheticity 191 Tagliamonte, S. 271 ‘take-over’ 380 TAM syncretism 365, 368, 372–5, 378 tautological hybrid 336 telicity 215, 221 témoigner 149 templates 365, 379–81, 383f. temporal intervals 245 temporal variation 243 tense 218f., 221, 232, 237 T(ense) P(hrase) 193 text type 155 thematic elements 373, 379 Thibault, P. 213f., 220–35, 237f., 240 third person plural ending 315f. Thomason, S. 279f., 281 Tomlinson, H. 284, 295, 300, 302f., topic 22–3, 27, 44, 51, 55, 64, 89, 93–4, 97 topic-continuity marker 21 topic-shift 52, 54, 62 Toulouse 145, 147–51 transfiguration 382 transitivity 213, 219, 229, 231 translation 140 transmission 277 Trask, L. 288 trend study 234, 264 Tristram, A. 222 trouverra 393f., 401 Turpin, D. 262, 269f. Tuscan 192, 194, 210 Tuttle, E. 214
OUP CORRECTED PROOF – FINAL, 14/4/2020, SPi
typology 2, 38 typological distance 280f., 296, 304 typological proximity 285, 296 unaccusativity 164, 176–7, 179, 216, 218f., 230 Unaccusative Hypothesis 197 unergativity 164, 215, 216, 218f. 317f. Utheim 218, 221 V-to-C movement 14f., 73–6, 81, 98f. Valdman, A. 225, 228 van der Auwera, J. 247 variation 1, 144, 155 variationist sociolinguistics 1, 223f., 229, 235 Vasseur, G. 220 Vaugelas 393 Vaux 390 Venetan 32, 192, 210 Verb First 23–4, 26f., 34 verb paradigm 364 Verb Second 9, 13–20, 22, 27, 42–4, 60, 71–5, 81, 84, 87f., 92, 98, 181, 194 relaxed 32, 44 relics 4 strict 44 verb semantics 228f., 240 verb stem regularity 268, 272f., 275, 277 Verb Third or greater 18–20, 28, 30–2
459
verbal inflexion 212 Villeneuve 225, 261–3, 267f., 270 Vincent, N. 219, 226 voice 205 voir 394 vouloir 149, 399 vouvoiement 269 VP adverbials 34 VSO 83, 98f., 161 Wagner, S. 228 Wales 262f., 266, 268 Wallon 182–4, 200f., 205f., 210, 308, 313, 331, 334, 338 Wartburg, W. von 388, 392 Weinreich, U. 260, 279f. Western provinces (Canada) 224 wh-word 56 Willis, L. 225f., 228–32, 235, 237f. Winford, D. 303 Wolfe, S. 12, 21, 27, 44–5, 58, 71–2, 134 word order 326 word-and-paradigm 399f. Wright, R. 12 Zamboni, A. 161f., 171, 188–90, 214 Zanuttini, R. 120 Zimmer, D. 263f. Zumthor, P. 309
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
OXFORD STUDIES IN DIACHRONIC AND HISTORICAL LINGUISTICS General editors Adam Ledgeway and Ian Roberts, University of Cambridge Advisory editors Cynthia L. 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; George Walkden, University of Konstanz; David Willis, University of Cambridge 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
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
9 Discourse and Pragmatic Markers from Latin to the Romance Languages Edited by Chiara Ghezzi and Piera Molinelli 10 Vowel Length from Latin to Romance Michele Loporcaro 11 The Evolution of Functional Left Peripheries in Hungarian Syntax Edited by Katalin É. Kiss 12 Syntactic Reconstruction and Proto-Germanic George Walkden 13 The History of Low German Negation Anne Breitbarth 14 Arabic Indefinites, Interrogatives, and Negators A Linguistic History of Western Dialects David Wilmsen 15 Syntax over Time Lexical, Morphological, and Information-Structural Interactions Edited by Theresa Biberauer and George Walkden 16 Syllable and Segment in Latin Ranjan Sen 17 Participles in Rigvedic Sanskrit The Syntax and Semantics of Adjectival Verb Forms John J. Lowe 18 Verb Movement and Clause Structure in Old Romanian Virginia Hill and Gabriela Alboiu 19 The Syntax of Old Romanian Edited by Gabriela Pană Dindelegan 20 Grammaticalization and the Rise of Configurationality in Indo-Aryan Uta Reinöhl 21 The Rise and Fall of Ergativity in Aramaic Cycles of Alignment Change Eleanor Coghill
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
22 Portuguese Relative Clauses in Synchrony and Diachrony Adriana Cardoso 23 Micro-change and Macro-change in Diachronic Syntax Edited by Eric Mathieu and Robert Truswell 24 The Development of Latin Clause Structure A Study of the Extended Verb Phrase Lieven Danckaert 25 Transitive Nouns and Adjectives Evidence from Early Indo-Aryan John. J. Lowe 26 Quantitative Historical Linguistics A Corpus Framework Gard B. Jenset and Barbara McGillivray 27 Gender from Latin to Romance History, Geography, Typology Michele Loporcaro 28 Clause Structure and Word Order in the History of German Edited by Agnes Jäger, Gisella Ferraresi, and Helmut Weiß 29 Word Order Change Edited by Ana Maria Martins and Adriana Cardoso 30 Arabic Historical Dialectology Linguistic and Sociolinguistic Approaches Edited by Clive Holes 31 Grammaticalization from a Typological Perspective Edited by Heiko Narrog and Bernd Heine 32 Negation and Nonveridicality in the History of Greek Katerina Chatzopoulou 33 Indefinites between Latin and Romance Chiara Gianollo 34 Verb Second in Medieval Romance Sam Wolfe
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
35 Referential Null Subjects in Early English Kristian A. Rusten 36 Word Order and Parameter Change in Romanian A Comparative Romance Perspective Alexandru Nicolae 37 Cycles in Language Change Edited by Miriam Bouzouita, Anne Breitbarth, Lieven Danckaert, and Elisabeth Witzenhausen 38 Palatal Sound Change in the Romance Languages Diachronic and Synchronic Perspectives André Zampaulo 39 Dative External Possessors in Early English Cynthia L. Allen 40 The History of Negation in the Languages of Europe and the Mediterranean Volume II: Patterns and Processes Anne Breitbarth, Christopher Lucas, and David Willis 41 Variation and Change in Gallo-Romance Grammar Edited by Sam Wolfe and Martin Maiden Noun-Based Constructions in the History of Portuguese and Spanish Patrícia Amaral and Manuel Delicado Cantero Redevelopment of Case Systems in Indo-Aryan Miriam Butt Classical Portuguese Grammar and History Charlotte Galves, Aroldo de Andrade, Christiane Namiuti, and Maria Clara Paixão de Sousa Morphological Borrowing Francesco Gardani Nominal Expressions and Language Change From Early Latin to Modern Romance Giuliana Giusti Syntactic Features and the Limits of Syntactic Change Edited by Jóhannes Gísli Jónsson and Thórhallur Eythórsson
OUP CORRECTED PROOF – FINAL, 15/4/2020, SPi
A Study in Grammatical Change The Modern Greek Weak Subject Pronoun τος and its Implications for Language Change and Structure Brian D. Joseph Reconstructing Pre-Islamic Arabic Dialects Alexander Magidow Iranian Syntax in Classical Armenian Robin Meyer Phonetic Causes of Sound Change Palatalization and Assibilation of Stop Conconants Daniel Recasens