See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/320015029

Ya les oiseaux qui chantent. A corpus analysis of French il y a clefts

Thesis · September 2017

CITATIONS READS 5 116

1 author:

Lena Karssenberg KU Leuven

13 PUBLICATIONS 79 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Presentational clefts in French and in Italian: a corpus-based, comparative analysis View project

Preverbal indefinite subjects View project

All content following this page was uploaded by Lena Karssenberg on 20 June 2019.

The user has requested enhancement of the downloaded file.

KU Leuven Faculty of Arts Linguistics department

Ya les oiseaux qui chantent

a corpus analysis of French il y a clefts

Dissertation submitted in partial fulfillment of the requirements for the degree of Doctor in Linguistics

by

Lena Karssenberg

Public defense 23 September 2017, Leuven

Supervisor Prof. Dr. Karen Lahousse

Co-supervisors Prof. Dr. Andreas Dufter Prof. Dr. Stefania Marzo

Chairman Prof. Dr. Dirk Geeraerts

Additional member of the jury

Prof. Dr. Kristin Davidse Prof. Dr. Véronique Lagae Prof. Dr. Béatrice Lamiroy Prof. Dr. Manuel Leonetti

Acknowledgments | Dankwoord

There are many people I sincerely want to thank for their support in the past four years, but the person I’m by far most grateful to is my supervisor, Karen Lahousse. Not only for her detailed and constructive criticism on all output related to the project, but also for her emotional support during difficult periods and for encouraging me to try and stay in academia after finishing the PhD. I really felt privileged to have her as my supervisor and it is unquestionable that she had a significant effect on both the quality of the dissertation and on my own well-being. I’ve said it before, but it’s worth repeating: I would not have traded you for any other supervisor and I truly hope our paths will keep crossing!

I’m also very grateful to my two co-supervisors Stefania Marzo & Andreas Dufter. Stefania for the discussions about the Italian data and the recurrent reminder “Just because it’s under-researched doesn’t mean it’s worthy of attention, give good arguments!”. Andreas for his quick feedback on all first drafts of the dissertation and for his very warm welcome in Munich. Their constructive comments and suggestions clearly helped improve the dissertation. I’m looking forward to collaborating with both of you in the future (hopefully!).

I would also like to thank Kristin Davidse & Béatrice Lamiroy for their comments on earlier drafts of certain chapters, and Manuel Leonetti & Véronique Lagae for being on my jury. I’m honored to be judged by such experts.

Furthermore, I’m thankful to Piet Mertens, Hendrik De Smet, Lyan Verwimp and Daniela Guglielmo for the extraction of the data in Le Monde, YCCQA, CFPP and La Stampa respectively.

Doing a PhD would not have been as much fun without all my colleagues at KU Leuven: my office mates (in order of appearance: Tom, Maria Chiara & Morgane) and all the others (including, but not limited to, Elwys, Laura, Melanie, Céline, Karlien, Gaétan, Isabeau, Stefano, Benedikt, Dirk, Freek…). I very much enjoyed the lunches at Alma and the coffees afterwards. What’s more, you were all very patient with me the countless times I had to run back to my office to check whether I had really locked my door, so thanks a lot! My dissertation was financed by the Research Foundation Flanders (FWO) (research project G061113N, under the supervision of Karen, Stefania & Béatrice). Moreover, several FWO grants allowed me to go to Chicago (4 weeks), Paris (3 months) and Munich (3 weeks) and organize a 2-day workshop about non-prototypical clefts at KU Leuven. I sincerely thank the FWO for enabling me to have all those enriching experiences. I’m grateful to the people at Lattice (Paris 3, especially Sophie Prévost, Benjamin Fagard & Frédéric Landragin). The same goes for all the people in Munich, especially Andreas, Thomas, Nicholas, Johannes and Sebastian.

I’m also grateful to the anonymous reviewers whose comments and suggestions helped improve the papers I submitted and, as a consequence, some of the chapters in this dissertation.

Of course, I also want to thank my family (Tjalling & Mieke, Sanne & Sara en al hun aanhang) and my non-linguist friends (Nienke, Asela, Brechtje, Leonee, Maret, Janna, Joost, Jon, Burger, Chris, Arion…). It may not have been very clear to you what I was actually doing during the past four years, but you supported me anyway, which is much appreciated!

Last but not least, I’m proud to say that the most mind-blowing, life-changing discovery that I’ve made during my PhD is Klaus Ludwig. Like the man at the printer’s said: “Your PhD is about French and Italian, you wrote it in English, in Leuven, but you’re from the Netherlands. Let me guess: your boyfriend is German.” – “Yes, he really is!” I calculated that the average waiting time in number of days between (a) hearing that one of my papers is accepted by a journal and (b) the actual publication of that paper is longer than the time it took us to (a) meet, (b) fall madly in love and (c) get married. I’d say that’s quite an achievement. Thank you for everything, Klaus. You’re so much more important than this dissertation.

Thank you everyone! Bedankt iedereen!

Voor mijn ouders, Mieke en Tjalling

Table of contents

General introduction ...... 1

1. Why study il y a clefts? ...... 1 1.1. Using more structure when you could do with less ...... 1 1.2. A (less well-known) member of the cleft family ...... 4

2. Terminology ...... 8

3. Il y a clefts in a nutshell ...... 9 3.1. A definition of il y a clefts? ...... 9 3.2. Semantics and discourse functions ...... 10 3.3. Similar il y a sentences (not considered in this dissertation) ...... 11

4. Goals and structure of the dissertation ...... 13

Part I - Delimitation of il y a clefts ...... 17

Introduction ...... 19

Chapter 1 - Criteria for clefthood in previous studies ...... 23 1.1 The decleftability criterion ...... 23 1.2 Diagnostics: type of relative clause ...... 26 1.2.1 Restrictive relative clauses (RRCs) vs. cleft relative clauses ...... 27 1.2.2 Syntactic constituency tests: RRC or not? ...... 30 1.2.2.1 Pronominalisation ...... 31 1.2.2.2 Dislocation ...... 32 1.2.2.3 Omission of the relative clause ...... 33 1.2.2.4 Adjectival/prepositional equivalent ...... 35 1.2.2.5 Interim conclusion ...... 36 1.2.3 Appositive relative clauses ...... 37 1.3 Semantico-pragmatic diagnostics ...... 40 1.3.1 Semantic classifications of il y a... qui sentences ...... 40 1.3.2 Two properties that distinguish between stative & eventive il y a... qui sentences . 42 1.3.3 Linguistic tests ...... 44 1.3.3.1 “What’s happening?” ...... 44 1.3.3.2 The interpretation of il y a: reformulation test ...... 45 1.3.3.3 Negation of il y a ...... 47 1.3.4 Specificational il y a clefts ...... 51 1.4 Overview of the diagnostics ...... 53

Chapter 2 - Corpora and extraction ...... 55 2.1 Description of the corpora ...... 55 2.2 Il y a... qui sentences: extraction of the data ...... 56 Table of contents

2.2.1 Le Monde ...... 57 2.2.2 YCCQA ...... 59 2.2.3 CFPP ...... 60 2.2.4 Overview and comparability across the corpora ...... 61 2.2.5 Frequency of il y a… qui sentences across the corpora ...... 63 2.3 Il n’y a que ‘there is only’: extraction of the data ...... 64

Chapter 3 - Application of criteria to the corpus data ...... 67 3.1 Quantitative overview ...... 68 3.2 Il y a clefts ...... 70 3.2.1 Eventive il y a clefts ...... 70 3.2.2 Specificational il y a clefts ...... 73 3.3 Il y a cleft lookalikes ...... 76 3.3.1 Cleft lookalikes with RRCs ...... 76 3.3.2 Cleft lookalikes with appositive relative clauses ...... 79 3.4 Unclear case 1: type of relative clause underspecified ...... 81 3.4.1 Corpus data: aspectually underspecified il y a... qui sentences ...... 82 3.4.2 About the non-discrete boundaries between relative clause types ...... 87 3.4.3 Hypotheses about processing il y a… qui sentences with ambiguous relatives ...... 89 3.5 Unclear case 2: decleftable + RRC ...... 91 3.6 Possible ways of defining il y a clefts ...... 96 3.7 My definition of il y a clefts ...... 100

Conclusion ...... 103

Part II - Distribution, semantic and morphosyntactic properties ...... 105

Introduction ...... 107

Chapter 4 - Il y a clefts: two semantic types (eventive vs. specificational) ...... 109 4.1 Distribution ...... 110 4.1.1 Il y a clefts in general ...... 110 4.1.2 Eventive vs. specificational il y a clefts ...... 111 4.2 Type of clefted constituent ...... 113 4.2.1 Grammatical category & frequency of proper nouns ...... 113 4.2.2 (In)definiteness of the clefted constituent ...... 115

Chapter 5 - Il n’y a que clefts ...... 119 5.1 Il n’y a que: clefts vs. cleft lookalikes ...... 120 5.2 Distribution of il n’y a que clefts ...... 122 5.3 Properties of the constituent introduced by il n’y a que ...... 123 5.3.1 Grammatical category ...... 123 5.3.2 (In)definiteness ...... 125

Conclusion ...... 127 Table of contents

Part III - Functions of il y a clefts ...... 129

Introduction ...... 131

Chapter 6 - The function of avoiding ‘bad’ preverbal subjects ...... 135 6.1 Previous studies: motivations of il y a clefts/il y a… qui sentences ...... 136 6.2 The hypothesis evaluated in this chapter ...... 139 6.3 Some subjects are better than others ...... 140 6.3.1 Factors that influence subject acceptability ...... 140 6.3.2 Underlying reasons for (in)acceptability ...... 143 6.4 Corpus data ...... 147 6.4.1 Basic requirements for successful declefting ...... 147 6.4.2 Corpus data: degrees of decleftability ...... 149 6.4.3 Declefting is possible: eventive/specificational semantics ...... 150 6.4.4 Declefting is possible due to other properties ...... 152 6.4.5 Declefting is possible, but requires a lexical adjustment ...... 155 6.5 Confirmation of the hypothesis ...... 159 6.6 Demystifying decleftability ...... 162 6.7 Conclusion ...... 162

Chapter 7 - The information structure of il y a clefts ...... 165 7.1 Background: information structure terminology ...... 166 7.1.1 Information Structure, Common Ground and QUDs ...... 167 7.1.2 The ‘referential’ and the ‘relational’ dimension of information structure ...... 168 7.1.3 Focus & background ...... 170 7.1.4 Topic & comment ...... 171 7.1.5 Thetic/‘all-focus’/‘all-comment’ sentences ...... 175 7.2 Previous studies: the information structure of il y a clefts ...... 177 7.2.1 Information structure articulations of il y a clefts ...... 177 7.2.2 Clefts in general: markers of non-topicality and/or focality ...... 178 7.3 Corpus data: different information structure articulations ...... 181 7.3.1 Quantitative overview ...... 182 7.3.2 All-focus (with/without light relative clause) ...... 182 7.3.3 Focus + background ...... 186 7.3.4 Ambiguous focus (multiple focus?) ...... 189 7.3.5 Contrastive topic + comment ...... 191 7.3.5.1 Contrastive spatiotemporal topic + comment ...... 191 7.3.5.2 Contrastive aboutness-topic + contrastive comment ...... 193 7.4 The non-topic marking hypothesis: evaluation ...... 197 7.5 Comparison with existential il y a sentences ...... 199 7.5.1 All-focus ...... 199 7.5.2 Focus + background ...... 200 7.5.3 Topic + comment (double contrast) ...... 203 7.5.4 Summary ...... 205 7.6 Hypothesis: the link between ‘non-topics’ and ‘suboptimal subjects’ ...... 206 7.7 Psycho-linguistic evidence about related constructions ...... 210 7.8 Conclusion ...... 213 Table of contents

Chapter 8 - Information structure: c’est clefts vs. il y a clefts ...... 219 8.1 Information structure: same articulations, different distribution ...... 220 8.1.1 All-focus ...... 221 8.1.2 Focus-background ...... 221 8.1.3 Multiple focus ...... 222 8.1.4 Spatiotemporal topic + comment ...... 223 8.1.5 Aboutness-topic + comment ...... 224 8.1.6 Distribution ...... 224 8.2 Differences between il y a clefts and c’est clefts ...... 225 8.2.1 Formal difference: prepositional & adverbial clefted elements ...... 226 8.2.2 All-focus il y a/c’est clefts: text type/genre ...... 227 8.2.3 Focus-background il y a/c’est clefts: exhaustiveness ...... 229 8.2.4 Spoken vs. written French ...... 231 8.2.5 Markers of non-topicality ...... 232 8.3 Conclusion ...... 233

Chapter 9 - Other pragmatic functions of il y a clefts ...... 235 9.1 Discourse coherence ...... 237 9.2 Signaling that a singular indefinite does not refer to a single referent ...... 241 9.3 Reinforcing negation ...... 244 9.4 Pragmatic functions of the cleft relative clause ...... 248 9.4.1 Previous accounts: pragmatic functions of appositive relative clauses ...... 248 9.4.2 Functions of the relative clause in il y a clefts ...... 250 9.5 Conclusion ...... 253

Conclusion ...... 255

Part IV - Italian c’è clefts (vs. French il y a clefts) ...... 257

Chapter 10 - Italian c’è clefts (vs. French il y a clefts) ...... 259 10.1 Introduction ...... 259 10.2 Previous analyses ...... 261 10.2.1 Semantic and functional classifications of c’è sentences ...... 262 10.2.2 An instance of the cleft category? ...... 264 10.2.3 My definition of c’è clefts ...... 265 10.2.4 Information structure ...... 267 10.2.5 Frequency: French clefts vs. Italian clefts ...... 268 10.3 Corpus and extraction ...... 270 10.4 Delimitation ambiguities ...... 271 10.4.1 Ambiguous relative clauses ...... 271 10.4.2 Decleftable + RRC ...... 274 10.5 Frequency: Italian vs. French ...... 277 10.6 Information structure ...... 279 10.6.1 All-focus ...... 280 10.6.2 Focus + background ...... 282 10.6.3 Contrastive/partitive topic + comment ...... 285 Table of contents

10.6.4 Interim conclusion: c’è clefts and information structure ...... 290 10.6.5 C’è clefts and existential sentences: the same articulations ...... 291 10.7 Additional functions of c’è clefts ...... 294 10.7.1 Reinforcing negation ...... 294 10.7.2 Signaling that a singular indefinite does not refer to a single referent ...... 296 10.8 The nature of the cross-linguistic label ‘existential clefts’ ...... 300 10.8.1 ‘Comparative concepts’ vs. ‘cross-linguistic categories’ ...... 300 10.8.2 Existential clefts as a ‘comparative concept’ ...... 303 10.9 Conclusion ...... 304

Part V - Il y a clefts: compositionality and status as a construction .. 307

Chapter 11 - The compositionality of il y a clefts ...... 309 11.1 Introduction ...... 309 11.2 Compositionality: previous analyses of il y a clefts ...... 311 11.2.1 Il y a clefts are non-compositional (Lambrecht) ...... 311 11.2.2 The ‘Presentational Relative Construction’ ...... 315 11.2.3 Il y a clefts are compositional (Jullien) ...... 318 11.3 Il y a clefts: further evidence for a compositional analysis ...... 320 11.3.1 Discourse functions ...... 320 11.3.2 Expressing a single proposition/decleftability ...... 322 11.3.3 No inherent semantics ...... 323 11.4 Conclusion ...... 324

Chapter 12 - No evidence (yet) for il y a clefts as a construction ...... 325 12.1 Introduction ...... 325 12.2 What are constructions? ...... 326 12.2.1 Narrow definition: constructions as necessarily non-compositional ...... 326 12.2.2 Broad definition: constructions that allow compositionality ...... 329 12.3 Clefts as compositional constructions: previous analyses ...... 333 12.3.1 Compositional constructions: there clefts and it clefts ...... 333 12.3.2 A compositional construction: that clefts ...... 334 12.3.3 A compositional construction: il y a clefts ...... 336 12.4 Il y a clefts: a construction or not? ...... 338 12.5 Il y a clefts as an instance of a higher-level (cross-linguistic) ‘cleft construction’? .. 340 12.6 Conclusion ...... 344

General conclusion ...... 345

Appendix ...... 363 References ...... 389

List of tables

Table 1: Overview of different diagnostics for the distinction il y a cleft vs. cleft lookalike ...... 54 Table 2: Corpus data: included and excluded il y a… qui sentences ...... 61 Table 3: Il y a… qui sentences in YCCQA & CFPP: comparability with the Le Monde query ...... 62 Table 4: Il y a… qui sentences that are included in quantitative comparisons between the corpora .. 62 Table 5: Normalized frequency of il y a... qui sentences throughout the corpora ...... 63 Table 6: Il y a… qui sentences appearing within a citation or not (Le Monde) ...... 63 Table 7: Il n’y a que with or without (different types of) relative clauses ...... 65 Table 8: Overview of different diagnostics for the distinction il y a cleft vs. cleft lookalike ...... 68 Table 9: Classification il y a cleft vs. cleft lookalike according to traditional criteria ...... 70 Table 10: Normalized frequency of il y a clefts ...... 110 Table 11: Il y a clefts appearing within a citation or not (Le Monde) ...... 111 Table 12: Distribution of eventive vs. specificational il y a clefts ...... 112 Table 13: Grammatical category of the clefted constituent (il y a clefts) ...... 113 Table 14: Questions leading to specificational il y a clefts with proper nouns ...... 114 Table 15: (In)definiteness of the clefted constituent in il y a clefts ...... 115 Table 16: Questions leading to specificational il y a clefts with definite NPs ...... 116 Table 17: Quantified vs. non-quantified indefinite constituents in il y a clefts ...... 116 Table 18: Il n’y a que clefts vs. cleft lookalikes with qui/que clauses ...... 120 Table 19: Normalized frequency of il n’y a que clefts ...... 122 Table 20: Il n’y a que clefts appearing within a citation or not (Le Monde) ...... 122 Table 21: Type of constituent behind il n’y a que ...... 123 Table 22: Nominal vs. non-nominal constituents behind il n’y a que ...... 124 Table 23: (In)definiteness of nominal constituents behind il n’y a que ...... 125 Table 24: Overview of hypotheses about the motivation of il y a sentences ...... 138 Table 25: Properties that influence subject acceptability ...... 141 Table 26: Indefinite pronouns in preverbal position in Cappeau (2008:75) ...... 142 Table 27: All il y a… qui sentences in the corpus data: classes of decleftability ...... 149 Table 28: Information structure articulations of il y a clefts ...... 182 Table 29: Information structure articulations of il n'y a que clefts ...... 182 Table 30: Information structure articulations of il y a clefts & c’est clefts ...... 220 Table 31: Il y a clefts vs. c’est clefts: distribution of information structure articulations ...... 225

Table 32: Properties of presentational, locative and existential c'è sentences in previous studies ... 264 Table 33: Normalized frequency of il y a clefts (Le Monde) and c'è clefts (La Stampa) ...... 277 Table 34: Normalized frequency of il y a clefts throughout the corpora ...... 349 Table 35: Types of il y a… qui sentences that were not extracted from Le Monde ...... 365 Table 36: All il y a… qui sentences: subject vs. object relative clauses ...... 371 Table 37: All il y a… qui sentences: syntactic category of the constituent introduced by il y a .... 371 Table 38: All il y a… qui sentences: types of pronouns ...... 372 Table 39: All il y a… qui sentences: (in)definiteness of the constituent introduced by il y a ...... 372 Table 40: All il y a… qui sentences: human or non-human constituent introduced by il y a ...... 372 Table 41: All il y a… qui sentences: semantically vague constituents introduced by il y a ...... 373 Table 42: All il y a… qui sentences: classes of decleftability per corpus ...... 375 Table 43: Normalized frequency of decleftable il y a... qui sentences throughout the corpora ..... 375 Table 44: (Non-)decleftable il y a... qui sentences appearing within a citation or not (Le Monde) ..... 375 Table 45: All il n’y a que sentences: type of relativizer ...... 377 Table 46: All il n’y a que sentences: infinitival clauses ...... 378 Table 47: All il n’y a que sentences: subject vs. object relative clauses ...... 379 Table 48: All il n’y a que sentences: grammatical category of introduced constituent ...... 380 Table 49: All il n’y a que sentences: nominal vs. non-nominal constituents ...... 381 Table 50: All il n’y a que sentences: (in)definiteness of (pro)nominal constituents ...... 381 Table 51: All il n’y a que sentences: determiners of indefinite NPs ...... 382 Table 52: Eventive il y a clefts: grammatical category of clefted contituent ...... 383 Table 53: Eventive il y a clefts: (in)definiteness of introduced element ...... 383 Table 54: Specificational and eventive il y a clefts: subject vs. object relative clauses ...... 383 Table 55: Il n’y a que clefts: determiners of indefinite NPs ...... 387 Table 56: Il n’y a que clefts: subject vs. object vs. adjunct relative clauses ...... 387

List of figures

Figure 1: Application of the diagnostics to the corpus data ...... 69 Figure 2: Assumption in previous studies about the correlation between decleftability & type of relative clause in il y a clefts ...... 92 Figure 3: Possible combinations of decleftability & type of relative clause in the corpus data .... 92 Figure 4: Classification of different il y a... qui sentence types ...... 100 Figure 5: Link between decleftability and subject acceptability ...... 150 Figure 6: Link between decleftability and subject acceptability ...... 159 Figure 7: Classification of different il y a... qui sentence types ...... 348 Figure 8: Link between decleftability and subject acceptability ...... 353

General introduction

Section 1 – Why study il y a clefts?

Section 2 – Terminology

Section 3 – Il y a clefts in a nutshell

Section 4 – Goals and structure of the dissertation

General Introduction 1

1. Why study il y a clefts?

There are two main reasons why il y a clefts are a worthwhile research topic: at first sight, they seem to be a redundant linguistic structure (section 1.1), and they are one of the under-researched members of a (cross-linguistic) cleft family (section 1.2).

1.1. Using more structure when you could do with less

People like attention. When we communicate, we want the listener to pay attention to what we’re saying and we don’t want him/her to lose interest in us. Therefore, we generally produce utterances that are easy to process: the listener might give up on us if it takes too much effort to follow what we’re saying. Moreover, people are also lazy. We don’t want to work too hard and so we don’t construct very complicated sentences when we don’t have to. As a consequence of these general principles, it is unlikely that a speaker will utter (1) instead of (2), unless he/she is trying to be funny. Put simply, the sentence in (1) is more difficult to process for the interlocutor and more complex to construct for the speaker, not only because it contains more complicated words (capacity, sodium chloride), but also because it’s longer and has a syntactically more complex structure (e.g. a ditransitive verb, prepositional phrases).

(1) Does it lie within your capacity to transfer the sodium chloride to me? 1 (2) Could you pass the salt?

In some cases, however, speakers do use what seems to be ‘extra’ structure when they could have expressed the same semantic proposition by means of a simpler version. Given the general preference for ‘simple’ over ‘complex’ described above, this of course raises the question why the speaker would do so. Examples of such cases are impersonal structures (3) and dislocations (4).

1 All examples taken from the corpora I used are followed by the name of the corpus in question. Examples that are not followed by any indication of a source are invented.

2 General Introduction

(3) French impersonal structure Il arrive des gens.

EXPL arrive PART people

‘People are coming.’ (Béguelin 1998:26, my translation)

(4) French right-dislocation Ça raconte quoi, ce livre ?

it tells what the book

‘What’s that book about?’ (De Cat 2007:86)

The propositions that those sentences convey could also have been expressed without the ‘extra’ structure (i.e. the expletive il in (3) and the right-dislocated constituent ce livre ‘that book’ in (4)), as shown in (5). Speakers nonetheless use impersonal structures and dislocations for a variety of (mostly pragmatic) reasons (see e.g. Malchukov & Siewierska 2011; Achard 2015 about impersonal structures; Blasco- Dulbecco 1999; De Cat 2007; De Stefani et al. 2015 about dislocations).

(5) ‘Simple’ versions a. Des gens arrivent.

PART people arrive

‘People are coming.’

b. Ce livre raconte quoi ?

that book tells what

‘What’s that book about?’

Another sentence type that is argued to use ‘extra’ structure in order to achieve certain pragmatic effects is the so-called ‘cleft sentence’, or ‘cleft’ for short. Clefts are generally defined as sentences that have a biclausal structure (a main clause + a relative or relative-like clause) and which express a single proposition that could also be expressed by means of a canonical word order pattern (i.e. Subject Verb in French). The most well-known type of French cleft is the so-called ‘c’est cleft’ (‘it is’), illustrated

General Introduction 3

in (6).2 This cleft expresses the same semantics as the Subject Verb word order equivalent in (7).

(6) French c’est cleft C’est les chats qui aiment les pelotes de laine.

‘CATS love balls of wool.’ (Lit. ‘It’s cats who love balls of wool.’)

(Destruel 2013:97, literal translation mine)

(7) ‘Simple’ version: semantic equivalent with Subject Verb word order Les chats aiment les pelotes de laine.

‘Cats love balls of wool.’

In this dissertation, I examine a less well-known type of cleft, namely the so-called ‘il y a cleft’ (‘there is’) (8).3 This cleft also corresponds to a semantic equivalent with Subject Verb word order, as shown in (9).

(8) French il y a cleft Maman ! Il y a le chat qui déchire les rideaux !

‘Mom! The cat is tearing up the curtains!’

(Lit. ‘There’s the cat that’s tearing up the curtains!’)

(Culioli 1990:138, in Pitavy 2009:127, my translation)

2 On French c’est ‘it is’ clefts: e.g. Hupet & Tilmant 1986; Clech-Darbon et al. 1999; Katz 2000a, 2000b; Lambrecht 2001; Rialland et al. 2002; Muller 2003b; Muller 2003a; Doetjes et al. 2004; Blanche-Benveniste 2006; Hobæk Haff 2006; Scappini 2006; Rouquier 2007; Bouchard et al. 2008; De Stefani 2008; Dufter 2008; Jullien 2008; Mertens 2008; Carter-Thomas 2009; Dufter 2009b, 2009a; Destruel 2012; Mertens 2012; Destruel 2013; Lahousse & Borremans 2014; Lahousse & Lamiroy 2014, 2015; Rouquier 2015; Karssenberg & Lahousse forthcoming-a. 3 On French il y a clefts: Lambrecht 1986; Giry-Schneider 1988; Lambrecht 1988a; Léard 1992; Lambrecht 1994; Ashby 1995; Blanche-Benveniste 1997; Lagae & Rouget 1998; Ashby 1999; Katz 2000a; Lambrecht 2000a, 2000b, 2001, 2002; Choi-Jonin & Lagae 2005; Jullien 2007; Avanzi 2008; Jullien 2008; Pitavy 2009; Secova 2010; Willems & Meulleman 2010; Lamiroy & De Mulder 2011:section 4; Sabio & Benzitoun 2013; Verwimp 2013; Hamann & Tuller 2015; Pekarek Doehler 2015; Karssenberg 2016a, 2016b; Verwimp & Lahousse 2016; Karssenberg & Lahousse 2017; Karssenberg accepted; Karssenberg & Lahousse forthcoming-a. Note that not all of these authors consider il y a… qui sentences like (8) as cleft structures. For instance, Lagae & Rouget (1998) and Choi-Jonin & Lagae (2005) use the term ‘presentational il y a sentence’ and only analyze c’est clefts such as (6) and il n’y a que… qui sentences such as (i) as clefts. (i) Il n’y a que Paul qui puisse le faire. ‘There’s only Paul who can do it.’

4 General Introduction

(9) ‘Simple’ version: semantic equivalent with Subject Verb word order Maman ! Le chat déchire les rideaux !

‘Mom! The cat is tearing up the curtains!’

So why would the speaker use extra structure (i.e. (8)) instead of just uttering the simple version that expresses the same semantic content (i.e. (9))? One predominant answer to this question in the literature on clefts in general is that clefts are used to indicate that the utterance has certain ‘information structure’ properties that are normally not associated with canonical word order sentences, or which are signaled less clearly if only prosody is used. For instance, Lambrecht (1994, 2001) argues that il y a clefts are generally used to introduce a new referent into the discourse about which the speaker says something by means of a relative clause. According to Lambrecht, this is easier to process than a canonical sentence with SV word order, because there is a cognitive principle that states: “Do not introduce a referent and talk about it in the same clause” (Lambrecht 1994:185, Principle of the Separation of reference and Role). However, most of the studies that are specifically dedicated to il y a clefts are based on a limited set of (invented) examples (e.g. Lambrecht 1986, but see Lagae & Rouget 1998; Choi-Jonin & Lagae 2005 and Verwimp & Lahousse 2016 for corpus analyses). In other words, previous hypotheses about the functional motivations of il y a clefts are in need of empirical verification. This will be one of the main goals of this dissertation.

1.2. A (less well-known) member of the cleft family

There is also a second reason why il y a clefts make an interesting research topic. As stated in the previous section, c’est clefts have received much more attention than il y a clefts. In this sense, il y a clefts can be seen as a relatively under-researched cleft type, alongside clefts introduced by voici/voilà ‘see here/there’ (10a) and j’ai ‘I have’ (11a). One important thing all of these cleft types have in common is that they can be

General Introduction 5

‘declefted’, i.e. they can be transformed into an equivalent with Subject Verb (SV) word order.4

(10) a. Voici/voilà cleft (‘see here/see there’)5 Voilà l’horloge qui sonne un… deux… trois…

see.there the.clock that strikes one two three

‘There’s the clock that strikes one… two… three’

(Dassi 2003:517, my translation)

b. Declefted equivalent

L’horloge sonne un… deux… trois.

‘The clock strikes one… two… three.’

(11) a. Possessive cleft6 J’ai ma copine qui habite à Paris.

I.have my friend who lives in Paris

‘My friend lives in Paris’ (Conti 2010:64, my translation)

b. Declefted equivalent

Ma copine habite à Paris.

‘My friend lives in Paris.’

Moreover, clefts are a cross-linguistic phenomenon, occurring not only in French, but also in Italian ((12) & (13)), English ((14)- (15)) and many other languages (see e.g. Smits 1989a; Gundel 2006; Huber 2006; Roggia 2008; Dufter 2009b; De Cesare et al. 2016; Garassino 2016 and Wehr 2016 for comparative studies).

4 On a cross-linguistic level, not all declefted sentences have SV word order. For instance, Italian c’è clefts often correspond to VS rather than SV word order (see Chapter 10). Therefore, I will use the term ‘decleftability test/criterion’ (also used by e.g. Delin & Oberlander 1995, 2005) rather than ‘SV test/criterion’ (see e.g. Verwimp & Lahousse 2016). Note, however, that I do not take ‘declefting’ to refer to any kind of transformational process (as in generative linguistics), but as a mere diagnostic for clefthood. In other words, I see the link between clefts and their canonical equivalents as a semantic correspondence, but I do not believe clefts should be derived from their SV equivalents. 5 See also Karssenberg et al. (under review) for a corpus-based analysis of voici/voilà clefts. 6 On possessive clefts in French: e.g. Rothenberg 1979; Blanche-Benveniste 1983; Furukawa 1996a; Lagae & Rouget 1998; Lambrecht 2002; Conti 2010.

6 General Introduction

(12) a. Italian è cleft (‘[it] is’)7 È la nebbia che mi fa paura.

is the mist that me does fear

‘It’s the mist that scares me.’ (Marzo & Crocco 2015:31, my translation)

b. Declefted equivalent

La nebbia mi fa paura.

‘The mist scares me.’

(13) a. Italian c’è cleft (‘there is’)8 C’è il gatto che gioca nel giardino.

there.is the cat that plays in.the garden

‘The cat is playing in the garden.’ (De Cesare 2007:136, my translation)

b. Declefted equivalent

Il gatto gioca nel giardino.

‘The cat is playing in the garden.’

(14) a. English it cleft9 – A: Is your knee hurting?

– B: No, it’s my FOOT that hurts. (Lambrecht 2001:486)

b. Declefted equivalent

No, my FOOT hurts.

7 On Italian è ‘it is’ clefts: e.g. D'Achille et al. 2005; Belletti 2008; Roggia 2008; Dufter 2009b; De Cesare 2012; De Cesare et al. 2014; Garassino 2014b, 2014a; De Cesare & Garassino 2015; Garassino 2016. 8 On Italian c’è clefts ‘there is’: Berruto 1986; Berretta 1995; De Cesare 2007; Sornicola 2010; Marzo & Crocco 2015; Karssenberg et al. forthcoming. 9 On English it clefts: e.g. Prince 1978; Atlas & Levinson 1981; Declerck 1983, 1984, 1988; Sornicola 1988; Delin 1989; Smits 1989a; Hedberg 1990; Collins 1991; Delin 1992; Delin & Oberlander 1995; Meinunger 1997, 1998; Hedberg 1999, 2000; Lambrecht 2001; Delin & Oberlander 2005; Gundel 2006; Hedberg & Fadden 2007; Bicler & Davidse 2008; Han & Hedberg 2008; Dufter 2009b; Piotrowski 2009; Patten 2010; Reeve 2010; Drenhaus et al. 2011; Belletti 2012; Reeve 2012; Den Dikken 2013; Hedberg 2013; Destruel & Velleman 2014; Haegeman et al. 2014; De Cesare & Garassino 2015; Garassino 2016; Wehr 2016.

General Introduction 7

(15) a. English there cleft10 – A: I’ve really just got to fill them in on lexicographers’ needs just because we’ve been doing a lot of it but there’s other people that you think are doing kind of creative corpus lexicography.

– B: Well, there’s [McCarthy] [who’s just building a new one]

(Davidse & Kimps 2016:123)

b. Declefted equivalent

McCarthy is just building a new one.

It clefts and their cross-linguistic counterparts are often regarded as the most ‘prototypical’ clefts and have received most attention in the linguistic literature. In contrast, the other cleft types (there is/il y a etc.) are less well studied and their analysis is largely influenced by the literature on clefts introduced by it/c’est. Studying the lesser-known clefts introduced by expressions other than it and its cross-linguistic equivalents may shed new light on the phenomenon of clefts in general. In other words, the analysis of il y a clefts is also relevant for research on (i) other French cleft types (c’est, voici/voilà…) and on (ii) clefts in other languages.

10 On English there clefts: Huddleston 1971:325; Lambrecht 1988b; Davidse 1999, 2000; Lambrecht 2001; Piotrowski 2009; Davidse 2014; Davidse & Kimps 2016; Njende 2016.

8 General Introduction

2. Terminology

Throughout this dissertation, I will use the following terminology to refer to the different parts of cleft sentences, visualized in (16). The two clauses that clefts consist of are called the ‘main clause’ (il y a le chat ‘there’s the cat’) and the ‘cleft relative clause’ (qui déchire les rideaux ‘that’s tearing up the curtains’). The main clause can itself be divided into the ‘cleft introducer’ (il y a, or in other cases c’est ‘it is’, j’ai ‘I have’ etc.) and the ‘clefted element’ (le chat ‘the cat’).

(16) ‘Main clause’ ‘Cleft relative clause’

Il y a le chat qui déchire les rideaux

There’s the cat that is tearing up the curtains

‘clefted element’

‘cleft introducer’

It should be noted that there is no consensus about the status of the clause introduced by qui ‘that’ as a relative clause. In the linguistic literature on it/c’est clefts, it has been observed that the relative clause in clefts displays peculiarities that may question its status as a ‘real’ relative clause, which is why several authors refer to it as a ‘pseudo- relative’ or ‘relative-like structure’.11 I will follow Lambrecht and others who analyze il y a clefts (e.g. Smits 1989a; Léard 1992; Furukawa 1996b; Lambrecht 2001) by referring to the clauses introduced by qui/que/qu’ in clefts as relative clauses, aware of the fact that there is still a debate about the use of this term in the analysis of clefts more generally.

11 See e.g. Cappeau & Deulofeu 2001; Dufter 2006; Koopman & Sportiche 2007; Frascarelli & Ramaglia 2009; Casalicchio 2013c, 2013a; Cinque 2013.

General Introduction 9

3. Il y a clefts in a nutshell

In what follows, I briefly comment on the definition of il y a clefts (section 3.1). I then present the main semantic and discourse-functional properties of il y a clefts (section 3.2) and sentence types that are similar to il y a clefts in form (and function), but which will not be considered in this dissertation (section 3.3).

3.1. A definition of il y a clefts?

In his analysis of c’est clefts, Dufter (2006) states the following: “The literature on c’est clefts in contemporary French is manifold (…). Therefore, it is all the more surprising that [these] sentences are often only illustrated by means of examples and characterized vaguely, whereas an explicit definition including necessary and sufficient or at least prototypical form and function properties remains absent” (Dufter 2006:33, my translation and emphasis12). Unfortunately, the same is true of clefts introduced by il y a: there is no general consensus about which diagnostics are sufficient and/or necessary for the definition of il y a clefts. Most analyses do present a number of properties that are claimed to be ‘defining’ or ‘prototypical’ of il y a clefts, such as decleftability and eventive (as opposed to stative) semantics. However, the status of those properties as sufficient and/or necessary criteria for clefthood is often left implicit. I will illustrate and comment on the different diagnostics in detail in Chapter 1, but in the next section I already briefly present some of the semantic and discourse-functional properties typically associated with il y a clefts.

12 “Die Literatur zu c’est-Spaltsätzen im Französischen der Gegenwart ist vielfältig (…). Umso mehr erstaunt, dass [diese] Sätzen oft nur durch Beispiele illustriert und vage charakterisiert wird, eine explizite Definition durch Angabe notwendiger und hinreichender oder zumindest prototypischer Form- und Funktionseigenschaften jedoch zumeist unterbleibt.” (Dufter 2006:33)

10 General Introduction

3.2. Semantics and discourse functions13

Clefts introduced by il y a are said to be typical of spoken language (e.g. Lambrecht 1986; Blanche-Benveniste 1997:93; Katz 2000a; Choi-Jonin & Lagae 2005; Dufter 2008; Willems & Meulleman 2010), although to the best of my knowledge, this claim has not been corroborated by corpus research comparing spoken and written French.14 Il y a clefts are often called ‘presentational’, because many authors argue that their main function is to introduce a new referent or event into the discourse (Lambrecht 1986; Giry-Schneider 1988; Lambrecht 1988a, 1994; Ashby 1995, 1999; Lambrecht 2000a, 2000b, 2001, 2002; Jullien 2007; Pitavy 2009; Secova 2010). As for the semantics of il y a clefts, most authors only present examples of il y a clefts expressing an event, such as (17).

(17) Eventive il y a cleft Il y a mes parents qui arrivent demain.

‘My parents are coming tomorrow.’

(Lit. ‘There’s my parents who are coming tomorrow.’)

(Secova 2010:190, my translation)

However, Lambrecht (1988a); Léard (1992); Lambrecht (2001) and Verwimp & Lahousse (2016)15 also acknowledge the existence of a different type of il y a cleft, which has specificational semantics. This means that the sentence expresses a value for a certain variable.16 For instance, Lambrecht (1988a) states that the cleft in (18) was uttered in a discourse context in which friends are discussing whether or not to go to the theatre. Thus, the relative clause expresses a variable (people who want to go), for which the clefted element provides a value (Beth).

13 As for the prosody of il y a clefts, Jullien (2014) observes that they have no uniform prosodic profile (Jullien 2014:316; 336-337; 483). This is in line with Avanzi’s (2008) conclusions about the prosody of different types of il y a sentences, including il y a clefts. Prosody will not be considered in this dissertation. 14 Note, however, that Cappeau & Deulofeu (2001) found that the il y en a construction is much more frequent in their informal than in their formal political corpus (126 vs. 5 tokens respectively). 15 See also Pierrard (1985) and Furukawa (1996b), who analyze specificational il n’y a que ‘there is only’ clefts, without using this terminology. 16 See e.g. Lahousse (2007, 2009); Heycock (2012); Den Dikken (2013) about specificational sentences.

General Introduction 11

(18) Specificational il y a cleft ‘Y’a Beth qui veut y aller, euh, y’a y’a Jean-Marc, y’a moi, bon.

‘There’s Beth who wants to go, uh, there’s there’s Jean-Marc, there’s me, ok.’

(Lambrecht 1988a:154)

Specificational il y a clefts have received very little attention in comparison to eventive il y a clefts (but see Verwimp & Lahousse 2016 for a corpus analysis).17

In both eventive and specificational instances of il y a clefts, the cleft introducer il y a is argued to be semantically ‘light’ and to serve a pragmatic rather than a semantic function (e.g. Lambrecht 1986; Lagae & Rouget 1998:315; Lambrecht 2000b:654; Choi-Jonin & Lagae 2005:54). However, il y a is nonetheless not entirely devoid of meaning (Lambrecht 2001:468).18

3.3. Similar il y a sentences (not considered in this dissertation)

There are a number of structures that closely resemble il y a clefts in form (and function), some of which are illustrated in (19).19 Although a comparison with il y a clefts would definitely be interesting, the related structures in (19) - (21) fall out of the scope of this dissertation.

(19) a. Il y en a qui l’utilisent c’est vrai

EXPL there of.it has that it.use it’s true

‘There are some who use it, it’s true’

(Blanche-Benveniste 1990:65, my translation)

17 In contrast, specificational clefts introduced by it and its cross-linguistic equivalents are the most well studied subtype of clefts (see references in footnote 9). See also Collins (1992); Davidse (1999, 2000); Piotrowski (2009) and Davidse (2014) about specificational there clefts in English. 18 See also Davidse (2000:1120) about the semantics of the cleft introducers there is and it is. 19 See Jeanjean (1979); Deulofeu (1989); Blanche-Benveniste (1990, 1997); Cappeau & Deulofeu (2001); Berrendonner (2003); Choi-Jonin & Lagae (2005); Avanzi (2008); Willems & Meulleman (2010); Jullien (2014); Pekarek Doehler (2015) about the sentence types in (19). See Blanche- Benveniste 1997; Cappeau & Deulofeu 2001; Choi-Jonin & Lagae 2005; Avanzi 2008 and Willems & Meulleman (2010) for comparisons between these sentence types and il y a clefts.

12 General Introduction

b. Par contre il il y en a d’autres qui ont

however EXPL EXPL there of.it has PART.others who have

récupéré euh des fauteuils des meubles

recovered uh PART chairs PART furniture

‘However, there are others who recovered uh chairs furniture’

(Cappeau & Deulofeu 2001:23, my translation)

In the two examples in (19), the referent is encoded by the pronoun en ‘of.it’, which can optionally be further specified by an expression such as d’autres ‘others’, as in (19b). In both examples, en is not used anaphorically, i.e. to refer to a previously mentioned NP, but rather designates a group of people (see Choi-Jonin & Lagae 2005:48). In other instances, en does function anaphorically. For instance, in (20), en refers to des habitants ‘inhabitants’.20

(20) Il y avait des habitants, il y en avait eu

EXPL there had PART inhabitants EXPL there of.it had been

‘There were inhabitants, there had been some.’

(Blanche-Benveniste 1997:92, my translation)

A second type of il y a sentence that is formally and functionally similar to il y a clefts is illustrated in (21). This example differs with respect to standard il y a clefts in that there is no relativizer (qui/que/qu’ ‘that’), but rather a personal pronoun that is coreferential with the referent introduced by il y a, in this case des gensi ilsi ‘peoplei theyi’ (see e.g. Choi-Jonin & Lagae 2005; Verwimp 2013).

(21) Il y a des gens ils ont mauvais caractère ‘There are people they have a bad character.’

(Choi-Jonin & Lagae 2005:39, my translation)

20 See for instance Lamiroy (1991); Lagae (1997) and Lagae (2001) about en ‘of.it’.

General Introduction 13

4. Goals and structure of the dissertation

The dissertation is subdivided into five major parts:

Part I: Delimitation of il y a clefts

In the first part of the dissertation, I summarize previous diagnostics for identifying il y a clefts (Chapter 1). I then present the corpora that I used for the analysis: spoken French (interviews; the CFPP corpus), informal written French (from an online discussion forum, YCCQA) and formal written French (from the national newspaper Le Monde), and I comment on the extraction procedure (Chapter 2). In Chapter 3, I explain how I applied the diagnostics from previous studies to the corpus data.

I will approach the corpus data in a bottom-up, inclusive way: I do not intend to use the corpora as a database for clear examples that resemble utterances presented in previous studies. Instead, I will look at the data in order to see whether the existent description of il y a clefts accurately fits the way in which speakers and writers of French use il y a clefts. In this sense, the present dissertation has the methodological goal of testing to what extent the current definition of il y a clefts can be operationalized for corpus research. This will turn out to be a difficult endeavour in certain cases: previous analyses of il y a clefts do not clearly state which diagnostics should be seen as sufficient and/or necessary, which leads to confusing results in a non-negligible number of corpus tokens.

Part II: Distribution, semantic and morphosyntactic properties

Given that there is no thorough empirical description of il y a clefts, we don’t have a precise understanding yet of the exact semantic, morphosyntactic and discursive properties of il y a clefts as they naturally occur in spoken and written French. One major goal of this dissertation is therefore to fill this descriptive gap in the linguistic literature by means of an in-depth corpus analysis of il y a clefts.

After having clearly established which tokens I consider as instances of il y a clefts in Part I, I will analyze those clefts in terms of (i) their distribution throughout different registers of French; (ii) their morphosyntactic and (iii) semantic properties (Chapter 4). A separate analysis of the same properties is presented of il y a clefts containing the ne… que ‘only’ particle (i.e. il n’y a que clefts) (Chapter 5).

14 General Introduction

Part III: Functions of il y a clefts

The third main goal of the dissertation is of an explanatory nature: to unravel the reasons why speakers and writers use il y a clefts rather than their syntactically less complex canonical equivalents. In order to do this, I will test two previous (and related) hypotheses about the functional motivations behind il y a clefts against the corpus data, i.e. (i) the idea that il y a clefts introduce constituents that would be ‘bad subjects’ if they appeared in preverbal subject position in a canonical sentence with SV word order (Chapter 6), and (ii) the hypothesis, mentioned above, that il y a clefts are used to overtly mark certain information structure properties (Chapter 7). I will also analyze differences in information structure between il y a clefts and c’est clefts, showing that they can have the same information structure interpretations, but that these interpretations are distributed very differently in corpus data of the two cleft types (Chapter 8). I present several factors that (partially) explain those differences. Moreover, I will show that il y a clefts can have several functions that have not yet been acknowledged in previous studies (Chapter 9). In this sense, my analysis is in line with a growing body of research showing that linguistic structures that used to be seen as motivated by information structure alone are actually in many cases multifunctional (e.g. Delin & Oberlander 1995; Matić & Wedgwood 2012).

Part IV: Italian c’è clefts (vs. French il y a clefts)

Another major issue that is addressed is the way il y a clefts relate to their Italian counterpart, the c’è cleft (Chapter 10). On the basis of a corpus analysis of c’è clefts, I will argue that c’è clefts and il y a clefts are strikingly similar in several respects: (i) delimitation problems, (ii) information structure potential and (iii) additional functions that help explain why speakers use a c’è cleft instead of its canonical counterpart. I will discuss whether this should or should not be taken as evidence in favor of a cross- linguistic category of ‘existential clefts’ (i.e. clefts introduced by c’è, il y a, there is…) that has the same representation in speakers’ mental grammars. It should be noted that I analyzed the French and the Italian data in a bi-directional way: insights about one of the languages led me to check whether the other language displayed the same property or not. In other words, I did not impose the analysis of the French data as a model on the Italian data in a top-down way.

General Introduction 15

Part V: Compositionality and status as a construction

In the last, more theory-oriented Part of the dissertation, I investigate the way il y a clefts may be represented in speakers’ mental grammars by analyzing their compositional or non-compositional nature (Chapter 11). I will argue against Lambrecht’s (1988a) non-compositional analysis and present arguments in favor of a compositional analysis (pace Jullien 2014). In other words, the meaning and functions of il y a clefts can be derived from their sub-components. Moreover, I evaluate whether il y a clefts should be seen as a ‘construction’ in the sense of Construction Grammar (Chapter 12) and argue that there is no conclusive evidence in this respect. I also comment on il y a clefts as instantiations of (i) a cross-linguistic construction that would include Italian c’è clefts and (ii) an overarching French cleft construction that would include clefts introduced by c’est, voici/voilà etc.

Part I

Delimitation of il y a clefts

Ya qqch qui cloche…

‘There’s something wrong…’

------

Part I: Introduction

The main goal of this part of the thesis is to delimit the phenomenon under investigation: il y a clefts (22).

(22) Il y a une voiture qui s’est arrêtée derrière moi ‘A car stopped behind me.’ (Lit. ‘There’s a car that stopped behind me.’)

(Secova 2010:37, my translation)

The first step in any corpus analysis after extraction of the data is to distinguish between the instances that will be analyzed in depth (i.e. which fall within the definition of the phenomenon) and those that will not be considered. In the case of il y a clefts, this task is quite straightforward in some cases, but more difficult in a considerable number of other cases. For instance, it has been observed that the same sequence of words may be a cleft in one context but not in another context. This is shown by the opposition between (23) and (24).21 In (23), the il y a… qui sentence expresses the existence of a certain object and the relative clause restricts the reference of the antecedent (i.e. it is a restrictive relative clause). In (24), on the other hand, the sentence is a cleft, because the whole sentence is interpreted as expressing an event, and the relative clause does not have the semantic function of restriction: it is a different type of relative clause. As I will explain shortly, it is generally assumed that the sentences in (23) and (24) differ not only in terms of their semantics, but also in the underlying syntactic structure.

21 See also Clech-Darbon et al. (1999:84); Scappini (2006:53) and Katz (2000a:260) for the same distinction concerning c’est clefts; De Cesare et al. (2016:155-157) about è clefts; Reeve (2012:1) for English it clefts and Davidse (2000:1103) for English there clefts. 20 Part I | Introduction

(23) Cleft lookalike (Question implicite: Sur la table, il y a quoi ?)

‘(Implicit question: What is on the table?)’

Sur la table, il y a le vase qui est cassé.

‘On the table, there’s the vase that’s broken.’

(Pierrard 1985:48, my translation and boldface)

(24) Il y a cleft (Question implicite: Qu’est-ce qui se passe ?)

‘(Implicit question: What’s happening?)’

Il y a le vase qui est cassé.

‘There’s the vase that’s broken.’

(Pierrard 1985:48, my translation and boldface)

For the sake of brevity, I will refer to il y a sentences that contain a relative clause but which are not traditionally considered to be clefts, such as (23), as ‘cleft lookalikes’. The term ‘il y a… qui sentence’22 includes both il y a clefts and il y a cleft lookalikes.

Examples such as (23) and (24) show that it is crucial to take into account the discourse context when analyzing clefts: without a context, it is often not possible to state whether a given sequence of words is a cleft or not. This is all the more important since several authors working on il y a clefts acknowledge the existence of ambiguous cases, i.e. il y a… qui sentences that cannot unambiguously be classified as either clefts or cleft lookalikes (Giry-Schneider 1988:90; Lagae & Rouget 1998:315- 316; Choi-Jonin & Lagae 2005:51; Jullien 2014:34&71). For instance, according to Choi- Jonin & Lagae (2005:51), there are ambiguous cases with a double interpretation, and only a sufficiently large discourse context allows to decide whether the sentence is a cleft or not.23 The authors illustrate this by means of the example in (25), which can

22 The term ‘il y a… qui sentence’ will be used to refer to il y a sentences followed by relative clauses introduced by all three relativizers qui, que and qu’. 23 “(…) il reste des cas ambigus, susceptibles d’une double interprétation. (…) Seul un contexte suffisamment vaste permettrait de trancher dans les cas difficiles” (Choi-Jonin & Lagae 2005:51).

Part I | Introduction 21

either be interpreted as a cleft lookalike with an existential interpretation or as a cleft that could answer the question “What’s happening?”.

(25) Il y a beaucoup de clients qui choisissent les spécialités régionales. ‘There are many clients who choose the regional specialties.’

(Choi-Jonin & Lagae 2005:51, my translation)

However, despite the fact that the existence of ambiguous cases has been acknowledged, the exact reasons for this ambiguity and the frequency of these unclear il y a… qui sentences are not analyzed in detail in previous works on il y a clefts.

Another potential source of confusion is the fact that previous analyses of il y a clefts describe properties that set them apart from cleft lookalikes, but without clearly stating which of these properties are sufficient and/or necessary criteria for the definition of il y a clefts. Consequently, in some cases, property A may indicate that the sentence is a cleft, while property B indicates that it is not, and it is not clear which of those properties is ‘more important’. In this dissertation, I take as an empirical point of departure all il y a… qui sentences that qualify as clefts according to at least one criterion for clefthood presented in previous analyses. By analyzing those instances systematically, I will be able to differentiate between both central and more peripheral instances of il y a clefts. In this way, I hope to come to a better understanding of il y a clefts and of il y a… qui sentences more generally.

The structure of Part I is as follows: in Chapter 1, I present and illustrate the main diagnostics that distinguish between il y a clefts and il y a cleft lookalikes, as they are reported in previous studies, and I identify the strengths and weaknesses of these diagnostics. Chapter 2 gives an overview of the corpora I used and of the extraction of the data. In Chapter 3, I test the diagnostics from previous studies against corpus data, which reveals that the standard definition of il y a clefts is in fact not clear-cut. Furthermore, my findings will support the claim that the boundaries between different types of relative clauses in general are often difficult to determine (e.g. Gapany 2004).

------

Chapter 1

Criteria for clefthood in previous studies

In this chapter, I present the diagnostics that have been proposed to distinguish between il y a clefts and their lookalikes. In section 1.1, I discuss the decleftability diagnostic. I then illustrate and comment on other main diagnostics related to the type of relative clause (section 1.2) and to the semantic properties that distinguish between il y a clefts and their lookalikes (section 1.3). An overview of the different criteria from previous works is presented in section 1.4.

1.1 The decleftability criterion

Recall from the General Introduction that all clefts, including il y a clefts, can be transformed into sentences with canonical word order (in the case of French: Subject Verb) expressing the same proposition, as shown in (26).

(26) a. Il y a cleft Y a le téléphone qui sonne.

there has the phone that rings

‘The phone is ringing.’

b. Declefted equivalent (SV word order)

Le téléphone sonne.

‘The phone is ringing.’

(Lambrecht 1988a:136) 24 Chapter 1 | Criteria for clefthood in previous studies

The decleftability criterion24 and the corresponding observation that clefts express a single proposition are often considered to be core properties of clefts. For instance, Lambrecht (2001) calls the expression of a single proposition via biclausal syntax “the defining property of clefts” (Lambrecht 2001:463); Katz (2000a) argues that the correspondence with a canonical counterpart is an “integral part of the c’est cleft construction” (Katz 2000a:254) and Dufter (2006) states that clefts “systematically correspond to a monoclausal variant with the same propositional and illocutionary content” (Dufter 2006:32, see also Dufter 2008).25 Unfortunately, however, it often remains implicit that declefting should not be possible for cleft lookalikes (e.g. Lambrecht 1986). Furthermore, there are several important aspects of the decleftability criterion one should keep in mind when applying this test to a given il y a… qui sentence. Firstly, the decleftability test is used in order to show that the sentence expresses one semantic proposition in a given context. However, this does not mean that the cleft version and the declefted version should be exactly equivalent on a pragmatic level. In fact, the declefted sentence may sound odd or even be unacceptable for pragmatic reasons. Thus, according to Lambrecht (1988a), the SV equivalents of il y a clefts are always “pragmatically unacceptable in most discourse contexts” (Lambrecht 1988a:137). The reason why this is so is that in spoken French, there is a ban on discourse-new referents in preverbal subject position. This is illustrated by the two sentences in (27). The sentence in (27a) is a cleft that can be declefted: (27b) expresses the same proposition as the cleft version. However, even though the declefted sentence in (27b) is grammatically and semantically well formed, Lambrecht argues that this sentence is unlikely to be uttered in a conversation if ‘Jean’ has not been mentioned before, because this is pragmatically unacceptable in spoken French due to a constraint on discourse-new NPs in preverbal subject position.

(27) a. Y’a Jean qui s’est cassé la jambe. ‘Jean broke his leg.’

24 The term ‘declefting’ is used by Delin & Oberlander (1995, 2005); Dufter (2009b) and De Cesare et al. (2014), among others. 25 “Ein Spaltsatz ist eine biklausale Struktur mit Kopulaprädikat im Matrixsatz und subordiniertem Satz, welche systematisch mit einer propositional und illokutiv äquivalenten monoklausalen Variante korrespondiert.” (Dufter 2006:32)

Chapter 1 | Criteria for clefthood in previous studies 25

b. Jean s’est cassé la jambe.

‘Jean broke his leg.’ (Lambrecht 1986:115)

In other words, it is important to realize that the decleftability criterion should be used to evaluate whether a given French cleft is semantically equivalent to a sentence with SV word order. However, it does not matter for the decleftability test if this SV equivalent is awkward on a pragmatic level (as in (27b)).

In this respect, it should be noted that, in languages other than French, declefting does not always lead to a pragmatically awkward result. For instance, many English clefts introduced by it or there can be declefted without leading to pragmatic infelicity, as in (28a) (= cleft) versus (28b) (= declefted equivalent). The reason for this is that in English, intonation patterns can achieve effects that are similar to the use of the cleft format. Hence, the utterance in (28b), with prosodic stress on the preverbal subject, can be seen as a pragmatically felicitous alternative to the cleft version in (28a). In contrast, such prosodic prominence is only marginally possible in French (29).

(28) a. It cleft – A: Is your knee hurting?

– B: No, it’s my FOOT that hurts. (Lambrecht 2001:486)

b. Declefted equivalent with prosodic stress

No, my FOOT hurts.

(29) Declefted sentence with prosodic stress in French (based on 28b) # Non, mon PIED me fait mal.

It has been argued that (especially spoken) French tends to mark discourse functions by means of syntactic constructions rather than by prosody (Klein 2012). Therefore, the types of constituents that can appear in preverbal subject position are much more restricted by pragmatics than in English. Cross-linguistic differences such as these are important for the analysis of clefts, because they influence the pragmatic acceptability of the declefted equivalents.

A second factor that is important to keep in mind when applying the decleftability criterion is the fact that the discourse context may determine/influence

26 Chapter 1 | Criteria for clefthood in previous studies

the semantic proposition expressed by the sentence. This was illustrated in the Introduction by means of the examples in (30) and (31): the same sequence of words may be interpreted as eventive (30) in one context, but as stative in another (31).

(30) Il y a cleft (Question implicite: Qu’est-ce qui se passe ?)

‘(Implicit question: What’s happening?)’

Il y a le vase qui est cassé.

‘There’s the vase that’s broken.’

(Pierrard 1985:48, my translation and boldface)

(31) Cleft lookalike (Question implicite: Sur la table, il y a quoi ?)

‘(Implicit question: What is on the table?)’

Sur la table, il y a le vase qui est cassé.

‘On the table, there’s the vase that’s broken.’

(Pierrard 1985:48, my translation and boldface)

This has consequences for the distinction between clefts and cleft lookalikes. Therefore, I will always apply the decleftability criterion while taking into account the original discourse context that surrounds the utterance.

1.2 Diagnostics: type of relative clause

Apart from the decleftability criterion, many diagnostics proposed in the literature to distinguish between clefts and cleft lookalikes are based on the premise that cleft lookalikes usually contain so-called restrictive relative clauses, whereas il y a clefts contain a different type of relative clause, henceforth referred to as ‘cleft relative clauses’.26 In section 1.2.1, I explain in more detail in which ways cleft relative clauses supposedly differ from restrictive relative clauses. A variety of syntactic diagnostics,

26 The relative clause in il y a clefts is referred to as ‘predicative relative clause’ by Lagae & Rouget (1998) and Choi-Jonin & Lagae (2005), among others.

Chapter 1 | Criteria for clefthood in previous studies 27

mostly constituency tests, are given in the literature to distinguish between restrictive relative clauses and cleft relative clauses. 27 I illustrate and comment on these tests in section 1.2.2. Moreover, cleft relative clauses may also resemble ‘appositive relative clauses’, as I explain in section 1.2.3.

1.2.1 Restrictive relative clauses (RRCs) vs. cleft relative clauses

There is no general consensus about the exact syntactic nature of cleft sentences, which has received a large amount of attention across languages and across frameworks.28 However, one point that most scholars agree on is the idea that the relative clause in clefts is not a restrictive relative clause (henceforth ‘RRC’). This is assumed to hold for all cleft types, including c’est clefts and il y a clefts.29 RRCs are a semantic subtype of relative clauses in general, whose main function is to restrict the reference of their antecedent.30 For instance, the relative clause qui ont la peau mate ‘who have a pale skin’ in (32) is an instance of a RRC: it limits the reference of all possible men to only those men who have a pale skin.

(32) Restrictive relative clause (RRC) Les hommes qui ont la peau mate ne craignent pas le soleil.

‘People who have a pale skin don’t fear the sun.’

(Fuchs 1987a:3, my translation and underlining)

27 In generative approaches to it clefts, syntactic diagnostics are predominant (e.g. Belletti 2008, 2012, 2013). However, almost all analyses of il y a clefts are not written within a generative framework and focus on semantic, rather than on syntactic criteria. Authors like Lambrecht (1986) often combine syntactic and semantic criteria. 28 See for instance Reeve (2010, 2011, 2012) (and references therein) for a presentation of two competing syntactico-semantic analyses of clefts (mostly discussed in generative linguistics), which he refers to as the ‘specificational analysis’ (Akmajian 1970; Percus 1997) and the ‘expletive analysis’ (Chomsky 1977; Kiss 1998). See also Davidse (1999, 2000); Lambrecht (2001); Davidse (2014); Hilpert (2014: section 5.2) and Davidse & Kimps (2016) for Construction Grammar approaches to clefts. I will present several Construction Grammar analyses of clefts in Part V. 29 In what follows, I will sometimes illustrate a certain relevant aspect of the restrictive/cleft relative clause distinction by means of examples of other clefts types (e.g. c’est clefts or there clefts). This is because certain aspects of relative clauses are not explained in detail in the literature on il y a clefts. 30 See also Fuchs (1987b); Godard (1988); Kayne (1994); Alexiadou et al. (2000); Gapany (2004); Muller (2006); Grevisse & Goosse (2008:1429-1441); Riegel et al. (2011: chapter 16).

28 Chapter 1 | Criteria for clefthood in previous studies

Several properties and tests distinguish RRCs from cleft relative clauses. Firstly, cleft relative clauses can have a proper noun as antecedent (33), contrary to RRCs (34) (e.g. Lambrecht 1986:118; 1988a:141; Anscombre 1996; Choi-Jonin & Lagae 2005). This is because the main function of RRCs is to restrict the reference of their antecedent. This restrictive function is incompatible with an antecedent whose reference is already fully determined, as is the case with a proper noun or a definite pronoun like him (see e.g. Lambrecht 1986; Langacker 1991:59; Davidse 2000:1103; Tellier 2003).

(33) Proper noun + cleft relative clause C’est bien Vladimir V. qui a mis un frein aux réformes économiques.

‘It is Vladimir V. who has stopped economic reforms.’

(Dufter 2009b:89)

(34) * Proper noun + RRC A: Who’s that talking to the police? – B: It’s *[John/him who was sacked].

(Davidse & Kimps 2016:120)

Secondly, only the relative pronouns qui ‘that’ (for subject relative clauses), que ‘that’ (for object relative clauses) and their contracted form qu’ ‘that’ are reported to appear in il y a clefts, whereas there is a wider variety of relativizers that are able to introduce RRCs (e.g. dont ‘of whom/which’, auquel ‘to/at which’ etc.).31 Furthermore, the use of object relative clauses in il y a clefts is limited to certain contexts (see section 1.3.2).

Thirdly, according to several authors, cleft relative clauses and RRCs are not in the same structural position with respect to their antecedent, as indicated by a variety of constituency tests (see next section): while RRCs modify their antecedent, with which they form a single constituent (i.e. they are ‘NP-internal’), cleft relative clauses do not form a single constituent with their antecedent, i.e. they are ‘NP- external’ (see for instance Lambrecht 1986, 2001; Huddleston & Pullum 2002; Choi- Jonin & Lagae 2005:47, inter alia). 32 This difference is illustrated by the brackets in

31 But see Dufter (2006:40) for a discussion of relativizers that occur in c’est clefts in his corpora, including où ‘where’ and dont ‘of which’. 32 Relative clauses in other types of contexts, such as the ones in (i), may also be analyzed as NP- external (see e.g. Ruwet 1978; Van Der Auwera 1985; Furukawa 1996a; Lambrecht 2000a; Furukawa

Chapter 1 | Criteria for clefthood in previous studies 29

(35)-(36): in (35), the relative clause is embedded inside the complex NP the director who was sacked, whereas in (36), the cleft relative clause does not form a constituent with its antecedent the director.33

(35) NP-internal relative clause ≠ cleft A: Who’s that talking to the police? — B: It’s [the [director who was sacked]].

(36) NP-external relative clause = cleft A: I hear they sacked the secretary. - B: No, it was [the director] [who was sacked].

(Huddleston & Pullum 2002:1416)

Lambrecht (1986) presents a similar opposition for French possessive clefts (i.e. clefts introduced by j’ai ‘I have’, il a ‘he has’ etc., illustrated in (37)), which he analyzes as a subtype of clefts with similar discourse functions as il y a clefts.34 Thus, although the sentences in (37) both contain a main clause and a relative clause, the relative clauses are in different positions: in (37a), it is NP-external, whereas in (37b), it is NP-internal, i.e. the RRC forms a single constituent with its antecedent maison ‘house’. Hence, (37a) is a possessive cleft, whereas (37b) is a cleft lookalike. The relative clauses in these examples also differ semantically: the cleft relative clause in (37a) is eventive,

2005; Casalicchio 2013c). However, as pointed out by Choi-Jonin & Lagae (2005:47), these relative clauses do not all have exactly the same properties. (i) a. Claude est là qui pleure. ‘Claude is there who cries.’ b. J’ai vu Pierre qui partait. ‘I saw Pierre who left.’ c. Avec mon chat qui est malade, je ne peux pas partir. ‘With my cat that is ill, I can’t leave.’ 33 Moreover, according to Davidse (2000:1108-1113) (following Langacker 1991), RRCs take only the nominal head (without its determiners) as antecedent. Thus, the RRC which they had tested in (i) is vehicle, not every vehicle. (i) Every vehicle which they had tested had some defect. (Davidse 2000:1110) In contrast, cleft relative clauses take the whole NP as their antecedent, not just the nominal head. For instance, the antecedent of the cleft relative clause who was sacked in (36) is the director, not just director. This distinction has not been presented for il y a clefts. 34 See e.g. Rothenberg (1979); Blanche-Benveniste (1983); Furukawa (1996a); Lambrecht (2002) and Conti (2010) about this type of cleft.

30 Chapter 1 | Criteria for clefthood in previous studies

whereas the RRC in (37b) restricts the reference of ‘house’ to the type of all contextually relevant houses that are painted red.35

(37) a. Il a [sa maison] [qui s’écroule !] ‘His house is collapsing!’ (Lit. ‘He has his house which is collapsing.’)

b. Il a [la [maison qui est peinte en rouge]]

‘He owns the house that’s painted red.’

(Based on Lambrecht 1986:123)

Given that cleft relative clauses do not form a single constituent with their antecedent, according to Lambrecht and others, one can use constituency tests to determine the type of relative clause in a given sentence (RRC/cleft relative clause), as will be illustrated in the next section (1.2.2).36

1.2.2 Syntactic constituency tests: RRC or not?

A number of constituency tests that are used in syntactic analyses in general are also applied to il y a… qui sentences in order to determine whether a given il y a… qui sentence is a cleft or contains a RRC: pronominalisation (section 1.2.2.1); dislocation (section 1.2.2.2); omission of the relative clause (section 1.2.2.3) and substitution of the relative clause by an adjectival or prepositional equivalent (section 1.2.2.4). An interim conclusion is presented in section 1.2.2.5. Note that the tests presented below only distinguish between RRCs and cleft relative clauses. I will comment on appositive relative clauses in section 1.2.3.

35 The sentence in (37b) is ambiguous, because the VP est peinte ‘is painted’ can also be interpreted as a passive form, in which case the sentence means ‘He owns the house that is being painted red’. 36 It is important to realize that not all researchers agree on the idea described above according to which there is a one-to-one pairing between the semantic function of the relative clause (restricting reference of the antecedent) and its syntactic structure (NP-internal). For instance, Gapany (2004:63) states that the function of a relative clause and its syntactic structure are two separate levels of analysis, and that a difference in semantic function does not necessarily correspond to a difference in syntax. Of course, it may still be the case that a certain form tends to appear in combination with a certain function. But seeing a form-function combination as a tendency is very different from seeing it as an implicational relationship in which one necessarily combines with the other and vice versa.

Chapter 1 | Criteria for clefthood in previous studies 31

1.2.2.1 Pronominalisation

One of the ways to determine whether a relative clause forms a single constituent with its antecedent (RRC) or not (cleft relative clause) is to pronominalize the sequence of the antecedent and the relative clause (Lambrecht 1986; Blanche- Benveniste 1990:65; Lagae & Rouget 1998:314). If pronominalisation is possible, the relative clause is restrictive, as in (38). If pronominalisation is not possible, the sentence is a cleft, as illustrated by the possessive cleft in (39):

(38) RRC + pronominalisation a. Il a [la maison [qui s’trouve à côté]]

He has [the house [that’s around the corner]]

b. Il a [celle-là].

He has [that one]. (Adapted from Lambrecht 1986:119)

(39) Possessive cleft + pronominalisation a. Il a [sa maison] [qui s’écroule].

‘His house is collapsing.’ (Lit. ‘He has his house that’s collapsing’)

b. * Il a [celle-là].

* ‘He has [that one].’ (Adapted from Lambrecht 1986:120)

Lambrecht explains the difference between (38) and (39) as follows:

“[t]he fact that pronominal representation of the sequence head noun plus qui- clause is possible in a [RRC] construction shows (i) that this sequence must be a syntactic unit, since it can be replaced by another unit which itself is a single constituent (the demonstrative celle-là), and (ii) it shows that this sequence must be a conceptual unit, since it can be represented by a pronominal morpheme which expresses a referential unit.” (Lambrecht 1986:119).

Lambrecht doesn’t apply this diagnostic to il y a clefts in particular, but does state that it applies to all cleft types.

However, a critical note is in order here: according to Tellier (2003:44), if a word sequence can be replaced or moved, this indicates that it is a constituent.

32 Chapter 1 | Criteria for clefthood in previous studies

Nonetheless, not all constituents can be replaced or moved. Consequently, if the test does not work (as in (39)) this does not necessarily mean the sequence is not a constituent: there may be other factors that block pronominalisation (or some other syntactic operation).37 For instance, the fact that the relative clause and its antecedent in (39a) cannot be pronominalized may be due to the pronoun’s inability to express an event, and, hence, pronominalisation is blocked by a semantic factor. However, the relative clause and its antecedent may still be one constituent at a syntactic level. Therefore, when a constituency test does not apply, we should not draw conclusions too quickly.

1.2.2.2 Dislocation

Choi-Jonin & Lagae (2005:51) perform the pronominalisation test on the example in (40) (the relative clause and its antecedent are replaced by en ‘of.it’), and simultaneously apply the dislocation test: des clients qui choisissent les spécialités régionales ‘clients who choose the regional specialties’ is right-dislocated in (40b).

(40) RRC + pronominalisation & dislocation a. Il y a beaucoup de clients qui choisissent les spécialités régionales.

‘There are many clients who choose the regional specialties.’

b. Il y en a beaucoup, des clients qui choisissent les spécialités régionales.

‘There are many of them, clients who choose the regional specialties.’

(Choi-Jonin & Lagae 2005:51, my translation and boldface)

The fact that the relative clause and its antecedent can be pronominalized as well as dislocated shows that they form a single constituent. However, the authors state that in a different discourse context, the sentence in (40a) could also be analyzed as a cleft,

37 “(…) si une suite de mots est déplacée ou remplacée par un pronom, nous pouvons conclure qu’elle forme un constituant. Mais l’impossibilité de déplacer ou de remplacer une suite de mots ne prouve pas que cette suite ne forme pas un constituant. Il pourrait en effet y avoir d’autres facteurs qui empêchent une opération syntaxique de s’appliquer. ” (Tellier 2003:44)

Chapter 1 | Criteria for clefthood in previous studies 33

for instance as an answer to the question ‘What’s happening?’ (41).38 Hence, Choi- Jonin & Lagae (2005) rightfully emphasize that utterances can only be classified as clefts or as cleft lookalikes when considered in their original context.

(41) – Qu’est-ce qui se passe ? ‘What’s happening?’

– Il y a beaucoup de clients qui choisissent les spécialités régionales.

‘There are many clients who choose the regional specialties.’

(Choi-Jonin & Lagae 2005:51, my translation)

In other words, the syntactic diagnostics should not be applied blindly, without considering the context that a given sentence appears in. If the test works outside of context, as in (40), indicating that the utterance is not a cleft, the sentence may still be a cleft in a different context (41). This observation will be crucial in accounting for ambiguous cases (see chapter 3.4).

1.2.2.3 Omission of the relative clause

Another test used to distinguish between clefts and cleft lookalikes is the omission of the relative clause. Lagae & Rouget (1998) and Choi-Jonin & Lagae (2005) argue that the relative clause in il y a clefts such as (42a) cannot be omitted (see (42b)), because it expresses “the assertion proper in the sentence” (Lambrecht 1986:117). RRCs such as qui sont gros ‘who are fat’ in (43a), on the other hand, can be omitted (see (43b)), because they express presupposed information, according to the authors.

(42) Cleft + omission of the relative clause a. Il y a le téléphone qui sonne.

‘The phone is ringing.’

b. # Il y a le téléphone.

‘There’s the phone.’

38 The authors do not present the original discourse context that this sentence appeared in.

34 Chapter 1 | Criteria for clefthood in previous studies

(43) RRC + omission of the relative clause a. Sur l’autre trottoir, il y a quelques hommes qui sont gros.

‘On the other sidewalk, there are some men who are fat.’

b. Sur l’autre trottoir, il y a quelques hommes.

‘On the other sidewalk, there are some men.’

(Choi-Jonin & Lagae 2005:45, my translation)

However, this test cannot easily be applied to all sentences containing RRCs. For instance, omission of the RRC in the example presented by Choi-Jonin & Lagae (2005) in (44) does not lead to an acceptable result, because it deletes crucial information expressed by the sentence. The resulting sentence in (44b) may be appropriate in certain discourse contexts (to say that people are present in a given location), but does not convey the same meaning as (44a).

(44) Omission of a RRC a. mais il y a des gens qui sont allergiques aux piqûres d’abeilles.

‘But there are people who are allergic to bee stings.’

(Choi-Jonin & Lagae 2005:43, my translation)

b. # mais il y a des gens.

# ‘but there are people.’

Indeed, it is hard to analyze the relative clause in (44) as expressing presupposed information.39 Moreover, the truth conditions change: (44a) entails that people exist, but (44b) does not entail that there exist people who are allergic to bee stings.40

39 The idea that RRCs express presupposed information is probably due to the fact that most examples used in the literature to illustrate this type of relative clause have a definite antecedent, whose existence is presupposed because of the determiner (see Gazdar 1979; Van Der Sandt 1988; Dobrovie-Sorin & Beyssade 2004:30; Delin & Oberlander 2005 about definite determiners as presupposition triggers). 40 The lack of a discourse context in (44a) also complicates the application of the tests.

Chapter 1 | Criteria for clefthood in previous studies 35

1.2.2.4 Adjectival/prepositional equivalent

Several authors argue that RRCs can be replaced by an adjectival or prepositional equivalent, contrary to cleft relative clauses (e.g. Léard 1992:31; Choi-Jonin & Lagae 2005). The example in (45) shows that the RRC qui traitent de l’uranium ‘that treat uranium’ can be replaced by de traitement d’uranium ‘of uranium treatment’ or dans lesquelles on traite l’uranium ‘in which uranium is treated’.

(45) RRC + adjectival/prepositional equivalent Il y a désormais des usines qui traitent efficacement l’uranium/de traitement d’uranium/dans lesquelles on traite l’uranium.

‘There are now factories that efficiently treat uranium/of uranium treatment/in which they treat uranium.’

(Léard 1992:31, my translation and underlining)

This test shows that RRCs are optional modifiers, just like adjectives. However, not all RRCs correspond to a lexically equivalent prepositional or adjectival phrase. For instance, according to Choi-Jonin & Lagae (2005), the relative clause in (46a) can be replaced by âgé ‘old’ or en noir ‘in black’ (46b), indicating that it’s a RRC. In contrast, the relative clause in (47a) cannot be replaced by âgé ‘old’ or de mon village ‘from my village’ (47b), indicating that this sentence is a cleft, according to the authors. However, the authors do not seem to take into account the fact that ‘old’ and ‘in black’ do not convey the same meaning as the relative clause ‘who walks by over there’.

(46) RRC + adjectival/prepositional equivalent a. Je connais le monsieur qui passe par là.

‘I know the man who walks by there.’

b. Je connais le monsieur âgé/en noir.

‘I know the old man in black.’

(Choi-Jonin & Lagae 2005:46-47, my translation and boldface)

36 Chapter 1 | Criteria for clefthood in previous studies

(47) Cleft sentence + adjectival/prepositional equivalent a. et une fois il y avait un monsieur qui passait par là.

‘And one time there was a man who walked by there.’

b. ?? et une fois il y avait un monsieur âgé/de mon village.

‘And one day there was an old man of my village.’

(Choi-Jonin & Lagae 2005:46-47, my translation and boldface)

Of course, I agree that there is a difference between RRCs, which can in some cases be replaced by a lexical equivalent, and cleft relative clauses, which seemingly cannot be replaced as easily.41 I only wish to add to this valid observation that we must be cautious when we apply the prepositional/adjectival equivalent test. Thus, as is the case with the pronominalisation and dislocation tests, if the adjectival/prepositional test works (as in (45)), this may be taken as an indication that the relative clause is restrictive, but if the test does not work, this does not entail that the relative clause is not restrictive. Hence, if a relative clause cannot be replaced by an adjectival or prepositional equivalent, this should not be taken as irrefutable evidence that the sentence is a cleft.

1.2.2.5 Interim conclusion

To conclude, constituency tests are useful tools to determine the type of relative clause in many cases, but they do not apply to all sentences without difficulty. In this respect, it is important to keep in mind that for some authors, the semantic distinction between states and events is an important additional criterion when determining whether or not a sentence is a cleft (e.g. Choi-Jonin & Lagae 2005). On the other hand, on most generative accounts, only the syntactic structure of the relative clause (NP- internal or NP-external), and not the semantics, is a valid criterion for clefthood (e.g. Belletti 2013). However, we saw that the application of syntactic constituency tests

41 Note that one could replace qui passait ‘who walked by’ in (47) by de passage ‘of passage’ (il y avait un monsieur de passage ‘there was a man who walked by’). The possibility to replace a relative clause seems to depend heavily on the type of verb. For instance, it is hard to conceive of a replacement for qui chantait ‘who was singing in (i). (i) Il y avait une femme qui chantait. ‘There was a woman who was singing.’

Chapter 1 | Criteria for clefthood in previous studies 37

may be influenced by the semantics of the sentence. Hence, semantics must be taken into account when one evaluates whether a given il y a... qui sentence is a cleft or not. I will return to the influence of semantics on the classification of il y a... qui sentences as clefts or cleft lookalikes in chapter 3.4.

1.2.3 Appositive relative clauses

Apart from RRCs, il y a cleft lookalikes may also contain appositive relative clauses, which are regularly referred to as ‘non-restrictive relative clauses’, because they do not have the restrictive function (see e.g. Gapany 2004; Cinque 2008; Grevisse & Goosse 2008; Cinque & Benincà 2010; Riegel et al. 2011:804).

An example of an appositive relative clause is qui sont mortels ‘who are mortal’ in (48), which does not restrict the reference of les hommes ‘people’: this sentence indicates that all people are mortal.

(48) Appositive (i.e. non-restrictive) relative clause Les hommes, qui sont mortels, vivent dans l’oubli de cette condition.

‘People, who are mortal, live without comprehending that condition.’

(Fuchs 1987a:3, my translation and underlining)

Appositive relative clauses share several properties with cleft relative clauses: they do not form a constituent with their antecedent and they too can refer to a proper noun, because they don’t have the restrictive function (e.g. Smits 1989a; Riegel et al. 2011:805; Loock 2013). However, it is generally argued that appositive relative clauses differ from other types of relative clauses (including RRCs and cleft relative clauses) by the fact that they are preceded by a comma (in writing) or a short pause (in speech) (see Smits 1989b:42; Riegel et al. 2011:805-806). Apart from the (optional) comma or pause, appositive relative clauses also differ from other types of relative clauses in that they present information that is interpreted as ‘extra’ or as an ‘afterthought’ that the interlocutor didn’t enquire about. In more specific terms, it has been argued that appositive relative clauses express propositions that are not “part of the at-issue

38 Chapter 1 | Criteria for clefthood in previous studies

content” (AnderBois et al. 2013:2, see also Potts 2005). 42 This means that the information is not the ‘main point’ of the utterance and does not clearly relate back to a Question Under Discussion (QUD) in prior context.43 For instance, there is a clear difference in interpretation between the RRC in (49) on the one hand, and the same sentence with an added comma in (50): the information in the appositive relative clause in (50) does not help identify the referent of the director, but rather expresses a new piece of information about the referent.

(49) RRC A: Who’s that talking to the police? — B: It’s the director who was sacked.

(Huddleston & Pullum 2002: 1416, my underlining)

(50) Appositive relative clause with comma A: Who’s that talking to the police? — B: It’s the director, who was sacked.

However, not all appositive relative clauses are preceded by a comma. Importantly, Lagae & Rouget (1998) and Choi-Jonin & Lagae (2005) show that il y a... qui sentences may also contain appositive relative clauses without a comma, as in (51). This relative clause is not restrictive (it does not restrict the reference of its antecedent), and is not a cleft relative clause, because it doesn’t express an event or a discourse-given variable (as in specificational instances of il y a clefts, see section 1.3.4). Rather, the relative clause in (51) provides additional information about Daniel Agostini.44

42 According to AnderBois et al. (2010), at-issue content and non-at-issue content differ in the way they enter the Common Ground: at-issue meaning is a proposal to update the Common Ground, whereas not-at-issue content (such as appositive relative clauses) are “updates which are imposed on the common ground and not up for negotiation” (AnderBois et al. 2010:329). This idea can also be related to the notion of ‘challengeability’ (Givón 1982), which refers to the fact that some information is open to challenge by the interlocutor while other information is not. 43 See for instance Beaver & Clark (2008) and Clifton & Frazier (2012) about the interaction between discourse coherence and QUD, which the latter characterize as follows: “The notion of QUD comes from an analysis that views discourse structure as a series of questions, implicit or explicit, which if accepted by the interlocutor are ideally followed by comments or replies that partially or completely answer them” (Clifton & Frazier 2012:352). I will return to QUDs and related notions in chapter 7.1, (‘the information structure of il y a clefts’). 44 According to Lagae & Rouget (1998), appositive relative clauses differ from predicative relative clauses (i.e. the relative clauses in il y a clefts) in that the relativizer qui can be replaced by lequel: (i) Il y avait Daniel Agostini, lequel est un DEUG d’Arles.

Chapter 1 | Criteria for clefthood in previous studies 39

(51) L1: qui il y avait à part Gilbert ? ‘Who was there except for Gilbert?’

L2: il y avait Daniel Agostini qui est un DEUG d’Arles

‘Daniel Agostini was there, who is a DEUG45 graduate from Arles.’

(Choi-Jonin & Lagae 2005:49, my translation)

In his analysis of relative clause types, Gapany (2004) even states that he does not take into account the presence or absence of a comma, because commas are a mere orthographic convention imposed by grammarians, which is not always respected (Gapany 2004:14&68). This is confirmed by examples such as (52) and (53), in which a relative clause follows a proper noun (Bou-Bou and Pin-Pon respectively), indicating that the relative clause is not restrictive. However, these sentences also do not have a cleft structure, since the proper nouns are introduced by the preposition avec ‘with’ and the main clause je dis à ‘I say to’. In other words, they are appositive relative clauses that are not preceded by a comma.

(52) Bref, Pin-Pon et Mickey reviennent, avec Bou-Bou qu’ils ont ramassé en route et qui me dit à peine bonsoir. ‘In short, Pin-Pon and Mickey come back, with Bou-Bou whom they picked up on the way and who barely says good evening to me.’

(L’été meurtrier, Sébastien Japrisot 2012 [1977], p. 118, my translation)

(53) À la fin, je dis à Pin-Pon qui respire fort, debout avec son casque à la main : « Si au moins tu me laisses parler. » ‘In the end, I say to Pin-Pon who is breathing loudly, standing upright with his hat in his hand: “If you would at least let me speak.”

(L’été meurtrier, Sébastien Japrisot 2012 [1977], p. 121, my translation)

‘Daniel Agostini was there, who is a DEUG graduate from Arles.’ (Lagae & Rouget 1998:315, my translation) See also Davidse (2000:section 2.2) and Davidse & Kimps (2016:section 3.2) about appositive relative clauses vs. cleft relative clauses in English it/there clefts. 45 ‘DEUG’ refers to a two-year university degree.

40 Chapter 1 | Criteria for clefthood in previous studies

In a similar vein, Gapany (2004) states that the presence of a pause preceding the relative clause in spoken French is not a decisive criterion for appositive relative clauses either. This is because there may be various reasons for which the speaker pauses, and hence pauses may also precede RRCs, as shown by Gapany & Zay (1995). This is confirmed by Jullien’s (2014) prosodic analysis of authentic il y a clefts: he shows that the constituent introduced by the il y a cleft may be followed by a pause, possibly in order to make it more salient (Jullien 2014:274).46

1.3 Semantico-pragmatic diagnostics

Apart from the different types of relative clauses (restrictive, appositive and cleft relative clauses) presented in the previous section, several authors argue that il y a clefts and their lookalikes also differ in terms of their semantics: typical il y a clefts express events, whereas il y a cleft lookalikes express states (e.g. Lagae & Rouget 1998; Choi-Jonin & Lagae 2005). Given this distinction, a number of diagnostics have been proposed. In what follows, I first present the predominant semantic classification of il y a... qui sentences and the distinction between states and events (section 1.3.1). Section 1.3.2 discusses discriminatory properties for il y a clefts and their lookalikes related to the state/event distinction, and section 1.3.3 is devoted to linguistic manipulations of the utterance, whose results indicate whether the sentence is a cleft (eventive interpretation) or not (stative interpretation). Section 1.3.4 is devoted to a less well-known subtype of il y a clefts with specificational rather than eventive semantics.

1.3.1 Semantic classifications of il y a... qui sentences

In most analyses of il y a... qui sentences, a three-way classification is presented that differentiates between ‘locative’ (54), ‘existential’ (55) and ‘presentational’ (or ‘eventive’) (56) uses of il y a... qui sentences (Giry-Schneider 1988; Léard 1992:31-32;

46 Interestingly, in languages such as German, all relative clauses are preceded by a comma. The presence or absence of a comma therefore does not allow to distinguish appositive relative clauses from restrictive or cleft relative clauses in this language (e.g. Loock 2013; Schubö et al. 2015).

Chapter 1 | Criteria for clefthood in previous studies 41

Choi-Jonin & Lagae 2005:43; Willems & Meulleman 2010; Verwimp & Lahousse 2016).47 Only ‘presentational’ il y a... qui sentences such as (56) are generally analyzed as clefts.

(54) Locative il y a... qui sentence Il y a ici (devant moi) les enfants qui m’ont bousculé.

‘The children who bumped into me are here (in front of me).’

(Léard 1992:31, my translation)

(55) Existential il y a... qui sentence Il y a désormais des usines qui traitent efficacement l’uranium.

‘There are now factories that efficiently treat uranium.’

(Léard 1992:31, my translation)

(56) Presentational il y a... qui sentence (= cleft) Il y a le chien qui court après un lapin.

‘The dog is chasing a rabbit.’ (Léard 1992:31, my translation)

The three categories above are usually only characterized informally (‘expression of location’, ‘expression of existence’, ‘expression of an event’), rather than defined by means of sufficient and/or necessary properties. Furthermore, there is no consensus about the question whether locative sentences and existential sentences are discrete sentence types, or whether one is a subtype of the other. Moreover, ‘presenting’ a new referent or event is also closely linked to the expression of the existence of this referent (or occurrence of an event) at a certain location (see e.g. discussion in Lyons 1967; Clark 1978; Koch 2012; Meulleman 2012b; Creissels 2014).48 Consequently, the boundaries between the different categories may rightfully be expected to become blurry when applied to authentic language data. This is also argued for by Jullien (2014), who states that the boundaries between the different (locative/presentational/

47 Similar classifications are also proposed for sentences introduced by existential expressions in other languages, such as Italian (Berruto 1986; Berretta 1995; De Cesare 2007; Sornicola 2010; Cruschina 2012; Marzo & Crocco 2015; Casalicchio 2016; Cruschina 2016), see Chapter 10. 48 I will return to existential sentences in chapter 7.5.

42 Chapter 1 | Criteria for clefthood in previous studies

existential) uses of il y a sentences in his corpus data are not as distinct as the literature seems to suggest (Jullien 2014:319).

In semantic analyses of the state/event distinction in general, it is often claimed that events are dynamic, whereas states are non-dynamic (see Husband 2010 for an overview). However, there is no general consensus about how to determine dynamicity or the state/event distinction in authentic language data. Most discussions of the state/event distinction use invented examples to illustrate the notions ‘state’, ‘event’ and ‘dynamic’, as in (57), rather than defining them in such a way that they can be operationalized for corpus data.

(57) a. He is silly (stative situation) b. He is being silly (“is… ing” constructs a dynamic situation)

(Desclés & Guentchéva 1995, text between brackets is the authors’)

In chapter 3.4, it will become clear that it is not always a straightforward task to determine whether an il y a... qui sentence is stative or eventive, which has consequences for the classification of such sentences as clefts or cleft lookalikes.

1.3.2 Two properties that distinguish between stative & eventive il y a... qui sentences

The first property that is claimed to distinguish stative il y a... qui sentences (i.e. cleft lookalikes) from eventive il y a clefts is that contrary to il y a cleft lookalikes, eventive il y a clefts do not contain a stative VP (Lagae & Rouget 1998; Choi-Jonin & Lagae 2005:44-45). According to the authors who mention this difference, the ban on lexically stative predicates in il y a clefts explains the awkwardness of the examples in (58). As the authors argue, these two examples are pragmatically odd,49 because they do not express an event and are therefore unable to answer the question ‘What’s happening?’.50

49 The asterisk and the double question marks signal pragmatic infelicity in (58a) and (58b). 50 Moreover, the sentence in (58a) is difficult to interpret as a ‘normal’ il y a cleft lookalike, because it is hard to imagine a context in which this sentence is informative.

Chapter 1 | Criteria for clefthood in previous studies 43

(58) a. *Il y a le téléphone qui est rouge. ‘There’s the phone that’s red.’

(Lagae & Rouget 1998:322, my translation)

b. Qu’est-ce qui se passe ? -?? Il y a des conducteurs de taxi qui sont gros.

‘What’s happening?’ -??’There are taxi drivers that are fat.’

(Choi-Jonin & Lagae 2005:45, my translation)

However, Léard (1992) shows that a lexically stative VP such as being blond in (59) can be interpreted as expressing an event (dying your hair blond) (see also Giry-Schneider 1988). Despite the lexically stative predicate, this sentence expresses an event and can therefore be seen as an il y a cleft.

(59) Il y a Pierre qui est blond. = ‘Un événement nouveau : Pierre se teint en blond’

‘There’s Pierre who’s blond.’

= ‘A new event: Pierre dies his hair blond.’

(Léard 1992:53, my translation)

Hence, in my view, it is not the lexical nature of the VP that is relevant but rather the opposition between ‘individual-level’ and ‘stage-level’ predicates, as also mentioned by Dobrovie-Sorin & Beyssade (2004:81) (see for instance Diesing 1992; Jäger 1996, 2001 about the individual/stage-level distinction). According to this distinction, introduced by Carlson (1977), individual-level predicates denote permanent properties (to be tall, to love pasta), whereas stage-level predicates denote temporary properties (to be coming in five minutes, to be available, examples from Leonetti 1998:142). Importantly, predicates that are lexically individual-level can be interpreted as stage- level predicates, depending on the discourse context. This is what happens in (59): ‘being blond’ is interpreted as ‘dying one’s hair blond’.51

51 See also Diesing (1992:20); Leonetti & Escandell Vidal (2002) and Husband (2010) about individual-level to stage-level coercion.

44 Chapter 1 | Criteria for clefthood in previous studies

The second property that distinguishes eventive il y a clefts from their stative lookalikes is that eventive il y a clefts cannot contain object relative clauses according to several authors (Lambrecht 1986:121; Giry-Schneider 1988:88-89; Léard 1992:36; Lagae & Rouget 1998:315; Choi-Jonin & Lagae 2005:47), as shown in (60).

(60) ?? Y’a Jean qu’une voiture a renversé ! ‘A car ran over Jean!’ (Lit. ‘There’s Jean that a car ran over!’)

(Lambrecht 1986:122, literal translation mine)

However, neither Lambrecht nor other authors give a detailed explanation of why object relative clauses are awkward in eventive il y a clefts.

1.3.3 Linguistic tests

So far, we have seen several properties that distinguish between eventive and stative il y a… qui sentences. In this section, I comment on the main linguistic tests that are applied to determine whether the il y a... qui sentence in question is eventive (i.e. a cleft) or stative (a cleft lookalike): (i) the ‘what’s happening?’ test (section 1.3.3.1), (ii) reformulation of il y a by il existe ‘there exists’ or on trouve ‘one finds’ (section 1.3.3.2), and (iii) negation of il y a (section 1.3.3.3). An overview of all relative clause-based and semantico-pragmatic diagnostics will be given in section 1.4.

1.3.3.1 “What’s happening?”

According to several authors, il y a clefts express an event, and should therefore be able to answer the question “What’s happening?” or “What happened?” (61) (e.g. Choi-Jonin & Lagae 2005:51; Pitavy 2009).

(61) – Qu’est-ce qui se passe ? ‘What’s happening?’

– Il y a des conducteurs de taxi qui font grève.

‘There are taxi drivers who are striking.’

(Anscombre 1996, in Verwimp & Lahousse 2016:6, my translation)

Chapter 1 | Criteria for clefthood in previous studies 45

However, Giry-Schneider (1988:94-95) states that any sentence introduced by il y a can answer the question ‘What’s happening?’ and be interpreted as an event, given a proper contextualisation (see discussion in section 1.3.2 about individual-level predicates receiving a stage-level reading). She illustrates this with the il y a... qui sentence in (62), which can be interpreted as expressing the existence of a certain type of people, but it can also answer the question “What’s up?”.52

(62) – Qu’est-ce qu’il y a ? ‘What’s up?’/‘What’s wrong?’

– Il y a ici des gens qui grognent tout le temps.

‘There are people here who grumble all the time.’

(Giry-Schneider 1988:94, my translation)

As a consequence, the value of the test in (61) is limited: one will always have to check whether the sentence under investigation, when considered in its original context, has an eventive interpretation.

1.3.3.2 The interpretation of il y a: reformulation test

Another test proposed by several authors is to replace il y a by an expression denoting existence or location such as il existe ‘there exists’ and on trouve ‘one finds’ (Giry- Schneider 1988; Anscombre 1996:95-97; Lagae & Rouget 1998:315; Martin 2002:80; Choi-Jonin & Lagae 2005:43). This replacement is possible in il y a cleft lookalikes, because in such sentences, il y a has a locative or existential interpretation, as in (63). If, on the other hand, such a replacement is not possible, the sentence is supposedly an il y a cleft (64). This is because the expression il y a in clefts is relatively delexicalized and no longer expresses the existence or location of a referent, contrary to il y a in cleft lookalikes.

52 Note, however, that the question Qu’est-ce qu’il y a? ‘What’s up?’ may be allowed in more discourse contexts than the question Qu’est-ce qui se passe? ‘What’s happening?’, which enquires more explicitly about an event.

46 Chapter 1 | Criteria for clefthood in previous studies

(63) Reformulation possible à il y a cleft lookalike a. Mais il y a des gens qui sont allergiques aux piqûres d’abeilles.

‘But there are people who are allergic to bee stings.’

b. Il existe/on trouve des gens qui sont allergiques.

‘There exist/one finds people who are allergic.’

(Choi-Jonin & Lagae 2005:43, my translation and boldface)

(64) No reformulation à il y a cleft – Qu’est-ce qu’il y a?

‘What’s wrong?’

– Il y a le lavabo qui déborde.

The sink is overflowing (lit. ‘There’s the sink that’s overflowing.’)

– * Il existe le lavabo qui déborde.

* ‘There exists the sink that’s overflowing.’

(Giry-Schneider 1988:88-89, my translation and boldface)

Note that this reformulation test, just like the constituency tests discussed in section 1.2.2, must be considered in the original discourse context of the il y a... qui sentence (Choi-Jonin & Lagae 2005).53

53 Moreover, according to Creissels (2014:7), the expression il y a is less semantically heavy than the expression ‘il existe’ and can therefore be used in a wider variety of contexts. Therefore, it may not always be possible to replace il y a by a more semantically specific expression of existence or location. As Creissels (2014:7) puts it, “It cannot be denied that there is a relationship between the meaning expressed by English exist or French exister and that expressed by existential predication [such as there is and il y a sentences], but the notion underlying the use of existential predication is quite obviously much wider than that underlying the use of exist. It is very easy to find existential clauses that cannot be paraphrased by clauses headed by the verb exist, whereas uses of exist that cannot be straightforwardly paraphrased by means of existential predication are relatively marginal.” (Creissels 2014:7) Hence, if it is problematic to replace il y a by il existe in a given sentence, this does not unequivocally indicate that this sentence is an il y a cleft.

Chapter 1 | Criteria for clefthood in previous studies 47

1.3.3.3 Negation of il y a

According to several authors, negating il y a is not possible in cleft instances (65) (Lambrecht 1986:118; Léard 1992:37; Lagae & Rouget 1998:322; Choi-Jonin & Lagae 2005:48).

(65) * Y’a pas le téléphone qui sonne. ‘The phone doesn’t ring/isn’t ringing.’ (Lambrecht 1986:118)

However, these authors do not present examples of il y a cleft lookalikes that do contain a negation, and hence we do not know if negation is problematic in all sentences introduced by il y a, or only in eventive il y a... qui sentences.54 Furthermore, Lagae & Rouget (1998) present the made-up il y a cleft in (66) to show that the apparent ungrammaticality of negation in il y a clefts may not be an absolute rule (see also Secova 2010:248 for criticism of the negation test).

(66) Et puis au moins, à la surface, il y a pas les feuilles qui flottent. ‘And then at least at the surface, the leaves are not floating.’

(Lit. ‘There are not the leaves that float.’)

(Lagae & Rouget 1998:323, my translation)

Their intuition is confirmed by an online search yielding several counter-examples to the ban against negation in il y a clefts.55 For instance, several properties indicate that the sentence in (67a) can be considered a cleft: it corresponds to an SV equivalent (67b) and il y a cannot be replaced by il existe ‘there exists’ (67c) or on trouve ‘one finds’. The sentence expresses an event (‘my mother yells at me’), which is negated.

54 Choi-Jonin & Lagae (2005) and Léard (1992) contrast negated il y a clefts with the acceptable sentences in (i) – (ii). However, given that the sentences in (i) and (ii) are not introduced by il y a, they do not form a minimal pair with infelicitous negated il y a clefts such as (65). (i) Je ne connais pas le monsieur qui passe par là. ‘I don’t know the man who passes by over there.’ (Choi-Jonin & Lagae 2005:48, my translation) (ii) Ce n’est pas le chien qui arrive. ‘It isn’t the dog that’s coming.’ (Léard 1992:37, my translation) 55 For more examples of eventive il y a clefts containing a negation, see Appendix, Section 3.

48 Chapter 1 | Criteria for clefthood in previous studies

(67) a. quand on surf sur le net le temps passe vite donc on passe plus de 4 h sans s’eb redre compte surtout quand y’a pas ma mere qui me cris dessus :-C et qui m dit que j’ai trop tardé lol LOL

‘When you surf online, time goes by quickly, so you spend more than 4 hours without realizing it, especially when there isn’t my mother who yells at me :-C and who tells me it’s taking too long lol LOL’ (www)

b. … surtout quand ma mère ne me crie pas dessus

‘… especially when my mother doesn’t yell at me.’

c. # … quand il n’existe pas ma mère qui me crie dessus

# ‘… when there doesn’t exist my mother who yells at me.’

Similarly, the examples in (68) and (69) express the negated events ‘my domain doesn’t appear’ and ‘my computer and the TV don’t distract me’, and, hence, they are negated il y a clefts with eventive semantics.

(68) Bonjour,déjà merci de ta réponse Christophe Effectivement en allant sur mon billing account sur Hostgator (dans manage domains) il n'y a pas mon domaine qui apparaît, pourtant avec l'abonnement que j'ai pris, un domaine était compris, il faut l'enregistrer en plus?

‘Hello, first of all, thanks for your answer, Christophe

Indeed when I go to my billing account on Hostgator (in ‘manage domains’) my domain doesn’t appear (Lit. ‘There isn’t my domain that appears’), however with the subscription that I bought, a domain is included, do you have to register it too?’ (www)

~ Mon domaine n’apparaît pas.

‘My domain doesn’t appear.’

Chapter 1 | Criteria for clefthood in previous studies 49

(69) Et moi, pourquoi je viens à la médiathèque? – « ma maman, elle me lit plein d’histoires »

– « c’est plus facile de travailler ici, il n y a pas mon ordi, la télé...qui me tente »

– And me, why do I come to the library?

– “My mother reads plenty of stories to me.”

– “It’s easier to work here, my computer, the TV… don’t tempt me

(Lit. ‘There aren’t my computer, the TV… that tempt me’) (www)

~ mon ordi, la télé ne me tentent pas.

‘My computer, the TV… don’t tempt me.’

It thus seems that negation is not completely ruled out from eventive il y a clefts. In certain cases (e.g. (66) - (69)), it is pragmatically acceptable to negate an event, while in other cases (e.g. (65): ‘there isn’t the phone that’s ringing’), negation is not acceptable. Insights in the conditions that license the negation of events are presented in Givón (2001: section 8.2.2). According to Givón (2001), the occurrence of events is cognitively salient, whereas we perceive states as the ‘norm’. Therefore, telling someone about an event that took place is informative, but telling him or her that an event did not take place is uninformative, because the absence of an event is the norm. This explains the unacceptability of (70b), as opposed to (70a): “if an event did not occur at all, why should one bother to talk about a specific individual who ‘participated’ in that non-event? (…) The norm of one’s everyday routine is not ‘all people visit my office at all times’, but rather ‘most people don’t ever visit my office at all’” (Givón 2001:373).

(70) a. A man came into my office yesterday and said… b. *A man didn’t come into my office yesterday and said…

(Givón 2001:373)56

However, negating an event becomes possible if the (implied) norm is that the event does take place, for instance in the context of a constant flow of visitors at a doctor’s office in (70) (Nahajec 2012:66-67). In this context, the absence of the event becomes

56 Givón uses the asterisk to indicate pragmatic awkwardness.

50 Chapter 1 | Criteria for clefthood in previous studies

informative and the sentence in (70b) becomes acceptable, according to Nahajec (2012). In other words, a negated event requires a context in which the norm is that the event does take place. This is exactly what is evoked in the examples in (66) - (69). For instance, the sentence in (69) necessarily implies that normally, the speaker is distracted by his/her computer and TV, but at the library, this event exceptionally does not take place. Put slightly differently, the presence of a spatiotemporal domain (when I’m at the library) indicates that the negated event does take place in another spatiotemporal domain (when I’m at home), and hence facilitates processing the sentence. Indeed, without the spatiotemporal expression or any kind of context that sets the event of ‘being distracted by my computer’ as the norm and ‘not being distracted’ as the exception, the sentence seems quite odd, as shown in (71).

(71) [No prior context] ? Il n’y a pas mon ordi, la télé qui me tentent.

? ‘There aren’t my computer, the TV that tempt me.’

It therefore comes as no surprise that all four examples in (66) - (69) explicitly contain a spatiotemporal setting (at the surface in (66), the adverbial clause introduced by quand in (67), ‘when going to a certain website’ in (68), and ‘when I’m at the library’ in (69)): the presence of a spatiotemporal domain facilitates processing a negated event. All four il y a clefts express the fact that a certain event that normally does take place, does not take place in a given spatiotemporal setting.

Chapter 1 | Criteria for clefthood in previous studies 51

1.3.4 Specificational il y a clefts

The majority of analyses of il y a clefts only consider semantically eventive il y a clefts. However, as explained in the General Introduction, there is also a subtype of il y a cleft that does not express an event, but rather has a specificational interpretation (Léard 1992; Lambrecht 2001; Verwimp & Lahousse 2016). Although not all diagnostics for identifying il y a clefts presented in previous studies apply to specificational il y a clefts, I will follow Léard, Lambrecht and Verwimp & Lahousse in considering them as clefts.

Specificational sentences present a certain value for a given variable. For instance, in (72), the relative clause expresses the variable ‘x wants to go to the theater’, and the clefted element expresses the value (x = Beth).

(72) [Context: Discussion about who wants to go to the theater] Y’a Beth qui veut y aller, euh, y’a y’a Jean-Marc, y’a moi, bon.

‘There’s Beth who wants to go, uh, there’s there’s Jean-Marc, there’s me, ok.’

(Lambrecht 1988a:154, my underlining)

All the semantic criteria for il y a clefts evaluate whether or not a given il y a... qui sentence is eventive or not, and hence, those criteria cannot easily be applied to specificational il y a clefts, since those clefts are not eventive. For instance, Verwimp & Lahousse (2016) show that specificational il y a clefts (73) cannot always answer the question ‘What’s happening?’ (74).

(73) – B: mm + donc + et c’est limité par quoi? par les grands axes routiers qui l’encerclent? euh ++

‘mm so and what delimits it? The highways that encircle it? Uh’

– A: euh ben y a oui enfin y a (mm) à côté là y a l’autoroute qu’est pas (mm) effectivement qu’est pas loin (mm) qu’est pas loin d’ici + donc euh oui +

‘uh well there’s yes well there’s (mm) next to it there there’s the highway that isn’t (mm) indeed that isn’t far way (mm) that isn’t far from here + so uh yeah.’

(Verwimp & Lahousse 2016:19, my translation)

52 Chapter 1 | Criteria for clefthood in previous studies

(74) – Q: Qu’est-ce qui se passe? ‘What’s happening?’

– #A: y a l’autoroute qu’est pas (mm) effectivement qu’est pas loin (mm) qu’est pas loin d’ici

‘There’s the highway that isn’t (mm) indeed that isn’t far way (mm) that isn’t far from here’

(Verwimp & Lahousse 2016:20, my translation)

Verwimp & Lahousse (2016) also argue that specificational clefts cannot always be declefted. However, the authors do not apply the decleftability criterion in the same way as I do, and as was intended by Lambrecht (see section 1.1). According to Verwimp & Lahousse (2016), the decleftability test doesn’t work in (75), because “the SV(O) paraphrase does not have the same discourse interpretation as the construction with il y a” (Verwimp & Lahousse 2016:13).

(75) Specificational il y a cleft + decleftability test ?# Euh oui oui y a celui-là celui d’la Roquette est pas loin, l’Père-Lachaise est un bel espace vert

?# ‘Uh yes yes there’s the one the one at la Roquette isn’t far away, the Père Lachaise is a nice green space.’ (Verwimp & Lahousse 2016:13, my translation)

However, in Lambrecht’s view, the goal of the decleftability test is not to check whether the declefted equivalent has the same discourse-pragmatic interpretation, but whether it denotes the same proposition on a semantic level. Indeed, Lambrecht (1988a) states that the SV equivalent of il y a clefts will almost always be pragmatically inacceptable in spoken French precisely because clefts are used for pragmatic functions that their SV equivalent cannot perform (see section 1.1, I will analyze these pragmatic motivations in detail in Part III). In this view, and contrary to the claim made by Verwimp & Lahousse (2016), the test in (75) does not fail: the declefted sentence denotes the same proposition, namely that the small Roquette green spot is not far away. In other words, specificational clefts do pass the decleftability test, but do not satisfy all criteria that identify eventive semantics (e.g. the ‘What’s happening test’). However, the fact that semantic criteria focusing on the state/event distinction are

Chapter 1 | Criteria for clefthood in previous studies 53

irrelevant for specificational clefts does not mean that these criteria are necessarily invalid for identifying il y a clefts of the eventive type. In chapter 3.2.2, I further analyze the diagnostic problems that may arise in specificational il y a clefts.

1.4 Overview of the diagnostics

The diagnostics presented in the previous sections that serve to distinguish between clefts and cleft lookalikes differ on two levels. Firstly, a distinction can be made between ‘positive’ and ‘negative’ criteria: positive criteria identify cleft sentences, while negative criteria identify cleft lookalikes (and, hence, indirectly indicate that the sentence is not a cleft). For instance, the presence of a proper noun shows that the relative clause cannot be a RRC, and therefore indicates that the sentence may be a cleft (i.e. a positive criterion). All criteria that identify RRCs, on the other hand, are negative: they rule out a cleft interpretation of a given sentence.

Secondly, a methodological distinction can be made between properties that are visible to the naked eye on the one hand and linguistic manipulations (or ‘tests’) on the other. The proper noun criterion is an example of a property, whereas the decleftability test and all syntactic constituency tests are linguistic manipulations that alter the original sentence.57

Table 1 presents an overview of the different diagnostics presented in this chapter, indicating whether they are positive (indicating clefthood) or negative (indicating a cleft lookalike, e.g. the presence of a RRC), and a property or a manipulation of the original.

57 Of course, the goal of the manipulations is to reveal a property of the sentence, and in this sense they do not differ fundamentally from the proper name criterion. However, manipulations may sometimes be more problematic to apply, because one must judge to what extent the manipulated outcome still corresponds to the original sentence it is based on.

54 Chapter 1 | Criteria for clefthood in previous studies

Test or Section property

Positive diagnostics

Declefting test 1.1

Presence of proper noun property 1.2.1

‘What's happening?’ test 1.3.3

Negative diagnostics

Pronominalisation of relative clause + its antecedent test 1.2.2.1

Dislocation of relative clause + its antecedent test 1.2.2.2

Omission of relative clause test 1.2.2.3

Adjectival/prep. equivalent of relative clause test 1.2.2.4

Individual-level predicate property 1.3.2

Object relative clause property 1.3.2

Reformulation of il y a test 1.3.3.2

Negation (il n'y a pas) test/property 1.3.3.3

Table 1: Overview of different diagnostics for the distinction il y a cleft vs. cleft lookalike

In sum, the different diagnostics are often illustrated by convincing examples. However, it has also become clear that these diagnostics cannot always be applied straightforwardly to all il y a... qui sentences. This will be further confirmed in Chapter 3, where I show that some il y a… qui sentences can easily be classified as clefts or as cleft lookalikes on the basis of these tests, but others remain problematic, even when considered in their original discourse context.58 But first I will present the corpora and the extraction of the data (Chapter 2).

58 See chapter 3.1 for more details about the way in which I applied the diagnostics to classify the corpus data.

------

Chapter 2

Corpora and extraction

In the previous chapter, I presented the most important properties that characterize il y a clefts and their lookalikes according to previous studies. In the next chapter, I will check to what extent these properties hold for authentic language data. But I will first describe the corpora that were used (section 2.1) and the way in which the data were extracted (section 2.2). Section 2.3 presents the separate query that was used to extract il y a... qui sentences with the restrictive particle ne… que ‘only’ (i.e. il n’y a que ‘there is only’). After having made clear in this chapter what the dataset is, I will apply the different diagnostics to distinguish between clefts and cleft lookalikes in the next chapter (Chapter 3).

2.1 Description of the corpora

The Le Monde corpus consists of all newspaper articles published in 1998 in Le Monde and contains approximately 25.7 million lemmatized words. The corpus was tagged and lemmatised with Cordial software (see Verlinde & Selva 2001a, 2001b for more details). This corpus is characterized by more formal language use than the other two corpora that were used. However, I do not wish to imply that Le Monde is representative of one uniform discourse register of ‘formal French’: it is important to acknowledge that linguistic tendencies or properties holding for the Le Monde corpus may not hold for other subgenres of ‘formal’ French such as political speeches. In this respect, I agree with Deulofeu (2000:273) when he emphasizes the importance of recognizing linguistic tendencies related to certain traditions, such as the journalistic one. 56 Chapter 2 | Corpora and extraction

YCCQA (Yahoo-based Contrastive Corpus of Questions and Answers) is a corpus of informal written French, English, Spanish and German, based on the Internet forum Yahoo Questions and Answers (2006-2009). It was compiled and is managed by Hendrik De Smet (KU Leuven). On the online forum, people ask and answer questions about a wide range of subjects, such as cars, politics, love and restaurants. One person will ask a question, but several people may answer it, sometimes leading to long discussions (especially when the topic is religion or politics). The French part comprises approximately 6.1 million words. The language used on the forum is mostly very informal. I did not correct the many spelling mistakes in the examples taken from this corpus. The website that this corpus is based on is still online and can be consulted at https://fr.answers.yahoo.com.

The CFPP corpus (‘Corpus de Français Parlé Parisien des années 2000’, also referred to as ‘Discours sur la Ville’) consists of (spoken) interviews with inhabitants of several Paris districts about their neighbourhood (changes, shops, atmosphere, etc.), and is available online in various formats, including the original audio files (for more details, see Branca-Rosoff et al. 2012). For the extraction, the Transcriber format was used. The entire corpus contains approximately 550.000 words, corresponding to 38.9 recorded hours.

2.2 Il y a... qui sentences: extraction of the data59

Several different queries were used in order to extract all il y a... qui sentences followed by a relative clause introduced by the relativizers qui/que/qu’, regardless of whether these sentences are clefts or cleft lookalikes (the cleft/lookalike distinction was coded for manually after the initial extraction, see next chapter). In what follows, I explain how the data were extracted from each corpus.

59 I am very grateful to several colleagues who extracted data for me from the corpora: Piet Mertens (KU Leuven) extracted all tokens from Le Monde, Hendrik De Smet (KU Leuven) extracted the il y a… qui dataset from YCCQA, and Lyan Verwimp (KU Leuven) extracted the data from the CFPP corpus (see Verwimp 2013; Verwimp & Lahousse 2016). I did all manual (re)coding after the initial extractions myself, and any errors in the coding process or in the interpretation of the data remain my own.

Chapter 2 | Corpora and extraction 57

2.2.1 Le Monde

The Le Monde query extracted all instances of il y a (in different tenses), followed by at least one and a maximum of four words, followed by the relative pronouns qui, que or qu’. This resulted in 693 hits, which were all manually analyzed and coded. A group of 271 tokens was identified as noise and hence excluded from the rest of the analysis. In these sentences, there is no relative clause that takes the element introduced by il y a as antecedent. In some cases of noise, the que clause is part of a comparative construction such as autres … que ‘other… than’ in (76). In other cases, the antecedent of the relative clause precedes il y a, as in (77), where il y a is used temporally and is syntactically independent of the relative clause that follows it.

(76) Excluded: comparative expression Mais il y a d'autres valeurs en cause que les droits de l'homme (…).

‘But there are other values at stake than human rights.’ (Le Monde)

(77) Excluded: antecedent of relative is in a different clause (…) le Camerounais Pierre Njanka, inconnu il y a quelques semaines et qui marque un but décisif ; (…).

‘the Camaroonian Pierre Njanka, unknown a few weeks ago and who scores an important goal.’ (Le Monde)

Furthermore, qui sometimes functions as an interrogative pronoun (see (78)) and que sometimes introduces a complement clause, such as toutes les chances que ‘every chance that’ in (79). I also excluded il y a sentences with the atonic pronoun ce followed by qui/que (80) and instances in which il y a… que is used temporally (81).60

(78) Excluded: qui ‘who’ is an interrogative pronoun Car s’il y a un problème, qui sera en première ligne ?

‘Because if there’s a problem, who will be on the front line?’ (Le Monde)

60 Such sentences are not analyzed as clefts in the relevant French literature. See, however, Karssenberg et al. (under review) for arguments why temporal uses of voici/voilà… que ‘see here/there… that’ sentences can be regarded as clefts. See also De Cesare et al. (2016:165) and Valentini (2016) about temporal è clefts (‘it is’) in Italian.

58 Chapter 2 | Corpora and extraction

(79) Excluded: complement clause (…) il y avait toutes les chances que nous parvenions à un accord (…).

‘there was every chance that we would come to a deal (…).’ (Le Monde)

(80) Excluded: ce que ‘that which’ A l'origine de cette stratégie radicale, il y a ce que Peter Balasz (…) appelle les trois “M” : la monnaie, les marchés et le management.

‘At the origin of this radical strategy, there is that which Peter Balasz calls the three “M’s”: money, markets and management.’ (Le Monde)

(81) Excluded: temporal use of il y a… que On a mis en doute ma compétence, alors qu'il y a 30 ans que je travaille.

‘They question my competence, even though I’ve been working for 30 years.’ (Lit. ‘… even though there’s 30 years that I work’) (Le Monde)

Lastly, I will not take into consideration instances with relativizers others than qui, que and qu’ or sentences in which these relativizers are preceded by a preposition. The Le Monde query yielded three instances containing à qui ‘to whom’ (82) and pour qui ‘for whom’ (83).

(82) Excluded: relativizer à qui ‘to whom’ La nature comporte une inégalité majeure : il y a les pays à qui elle fournit l'eau et il y a les pays à qui elle n'en offre pas ou très peu.

‘Nature presents a major inequality: there are countries she provides water to and there are countries that she offers none or very little.’ (Le Monde)

(83) Excluded: relativizers pour qui ‘for whom’ (…) il y a encore des gens pour qui trousser une jeune stagiaire à la Maison Blanche [est] un détournement de pouvoir de la part du président.

‘There are still people for whom chasing a young intern at the White House [is] an abuse of power by the President.’ (Le Monde)

After manually eliminating the sentences of the types illustrated above, 422 tokens were retained, analyzed and coded manually.

Chapter 2 | Corpora and extraction 59

2.2.2 YCCQA

The query used for YCCQA is deliberately more inclusive than the Le Monde query, 61 most notably because it does not limit the number of words between il y a and the relativizer: it retrieved all instances of y a (in different tenses), followed by at least one word, followed by the relative pronoun qui, que or qu’ before the end of the same writer’s post. Furthermore, the expletive il was not included in the query, because it is often dropped in informal registers of French (such as online forums).62 The query resulted in 6455 hits, which were all manually analyzed and coded. By using a more inclusive query than the one used in Le Monde, I was able to identify a number of interesting il y a... qui sentences that were not extracted from Le Monde. However, the query also inevitably yielded considerably more noise tokens (5032 in total), which were all manually excluded from the rest of the analysis. In most of these cases, the relative pronoun is not syntactically linked to the il y a... qui sentence, as in (84).

(84) “Tant qu’il y a de la Vie, il y a de l'espoir !" Courage! Il ne vous reste qu’à changer le Monde ! ‘“As long as there is life, there is hope!” Hang in there! You only have to change the world!’ (YCCQA)

The reason why punctuation was not taken into account in the query in order to automatically exclude sentences such as (84) is that punctuation is known to be unreliable in informal written registers.

Other types of noise in the YCCQA data include the same types that were found in the Le Monde corpus (see previous section), as well as sentences in which ya is part of a word (e.g. alléluya ‘hallelujah’ in (85)) and a few sentences in which il y en a ‘there are people’ (lit. ‘there are of it’) is misspelled, as in (86).

(85) Savez-vous qu'Obama a copié notre bonne Ségolène ? Alléluya !? ‘Did you know that Obama copied our good Ségolène? Hallelujah!?’ (YCCQA)

61 See section 2.2.4 about comparability across the different corpora and the influence of the different queries on the quantitative comparison of the data. 62 See e.g. Lambrecht (1988a:137); Lamiroy & De Mulder (2011:311); Pekarek Doehler (2015:177) about the elision of the expletive il.

60 Chapter 2 | Corpora and extraction

(86) Excluded: il y en a Tôt ou tard y'a n'a qui bosse pendent que d'autre dorment.

‘Early or late, there are people who work while others are sleeping.’

(Normative spelling: il y en a qui bossent pendant que d’autres dorment)

(YCCQA)

Furthermore, a range of different relativizers were found in the data (pour ‘for’, à qui ‘to whom’, où ‘where, dont ‘of which’, avec qui ‘with whom’, sur lequel ‘on which’, par qui ‘by whom’ and pour qui ‘for whom’).63 These relativizers are relatively infrequent (22 in total) with respect to qui (1300 tokens) and que (123 tokens).

Lastly, the query gave rise to a number of il y a... qui sentences containing the exclusive expression ne… que ‘only’, which were separated from the il y a… qui data. A separate targeted query extracted il n’y a que sentences (see section 2.3).

All in all, 1423 il y a… qui sentences from YCCQA were retained and subsequently manually analyzed and coded in more detail.

2.2.3 CFPP

The query that was used for this corpus is y a + at least one word + qui/que/qu’. This resulted in 802 tokens. The number of retained sentences in the CFPP corpus is 666.

Apart from the types of excluded sentences already discussed in the two previous sections, the CFPP corpus contains sentences in which the speaker doesn’t finish his/her relative clause, such as (87). These four instances were also excluded from the analysis.

(87) Excluded: unfinished relative clause oh: un un libraire mais enfin c'est pas il y a les kiosques qui sont euh

‘Oh a a librarian but well it’s not there’s the kiosks that are uh’ (CFPP)

63 Examples of all these different relativizers can be found in the Appendix (Section 1).

Chapter 2 | Corpora and extraction 61

2.2.4 Overview and comparability across the corpora

Table 2 presents an overview of the number of included and excluded il y a… qui sentences in the three corpora, as explained in the previous sections. All the ‘included’ sentences were manually analyzed for several parameters.

Le Monde YCCQA CFPP Total

n % n % n % n % Included il y a… 422 61% 1423 22% 666 83% 2511 32% qui sentences Excluded il y a… 271 39% 5032 78% 136 17% 5439 68% qui sentences Total 693 100% 6455 100% 802 100% 7950 100% Table 2: Corpus data: included and excluded il y a… qui sentences

It should be noted, however, that the queries used for YCCQA and CFPP are more permissive than the one used for Le Monde. Therefore, the numbers in Table 2 also include il y a… qui sentences that were not extracted from Le Monde. For instance, the extracted YCCQA and CFPP data include il y a... qui sentences containing a negation, such as (88), but these were not extracted from Le Monde. Detailed information about which types of sentences are yielded by the YCCQA and CFPP queries but not by the Le Monde query can be found in the Appendix (Section 2).

(88) Element between il and y: ne Il n’y a aucun aliment par lui-même qui fait maigrir.

EXPL neg.there has no food by itself that makes lose.weight

‘There is no food product that makes you lose weight by itself.’ (YCCQA)

In order to guarantee comparability, I identified all types of il y a… qui sentences in YCCQA and CFPP that were not extracted from the Le Monde corpus, such as (88). Those sentences will not be considered when the corpora are compared with each other on a quantitative level. However, I did analyze and code those sentences and will consider some of them separately in other sections when relevant. For example, I analyze il y a... qui sentences containing a negation in chapter 9.3. Table 3 presents an

62 Chapter 2 | Corpora and extraction

overview of the number of il y a… qui sentences that are either comparable or incomparable to the Le Monde data. What this table shows is that 77% of the extracted il y a… qui sentences from YCCQA would also have been extracted from Le Monde, and the percentage of comparable il y a… qui sentences in CFPP is 82%. In other words, using the Le Monde query, I would not have extracted 23% and 18% of the data from YCCQA and CFPP respectively. Therefore, those sentences will not be retained in quantitative comparisons.

YCCQA CFPP Total

n % n % n % Comparable il y a… qui 1093 77% 544 82% 1637 78% sentences Incomparable il y a… qui 330 23% 122 18% 452 22% sentences Total 1423 100% 666 100% 2089 100% Table 3: Il y a… qui sentences in YCCQA & CFPP: comparability with the Le Monde query

The total number of il y a… qui sentences that are included in quantitative comparisons of the three corpora are given in Table 4.

Le Monde YCCQA CFPP Total

Total number of comparable il y a… 422 1093 544 2059 qui sentences Table 4: Il y a… qui sentences that are included in quantitative comparisons between the corpora

Chapter 2 | Corpora and extraction 63

2.2.5 Frequency of il y a… qui sentences across the corpora

Table 5 presents the normalized frequency of the il y a… qui sentences that were retained for in-depth analysis per 500.000 words per corpus. The numbers for YCCQA and CFPP only include those il y a… qui sentences that would also have been extracted by means of the Le Monde query, i.e. they do not include instances such as (88), which includes a negation (see previous section). Recall that these sentences include both clefts and cleft lookalikes.

Le Monde YCCQA CFPP Total

Corpus size in words 25700000 6100000 550.000 32.350.000

Total number of il y a… qui 422 1093 544 2059 sentences Number of included il y a... qui 8 90 495 32 sentences per 500.000 words Table 5: Normalized frequency of il y a... qui sentences throughout the corpora

Table 5 shows that there is an increasing line going from formal written (8 per 500.000 words) to informal written (90) to spoken French (495). It has been claimed that il y a clefts are a spoken phenomenon rather than a written one (Lambrecht 1986; Blanche- Benveniste 1997:93; Katz 2000a; Choi-Jonin & Lagae 2005; Dufter 2008; Willems & Meulleman 2010). Table 5 confirms that this claim holds for il y a… qui sentences in general. This is further confirmed by the fact that the majority of all il y a… qui sentences in Le Monde appear within a citation (62%), as shown in Table 6. I will return to this in chapter 4.1.1.

Le Monde

n % Occurs in citation 260 62% Does not occur in citation 162 38% Total 422 100%

Table 6: Il y a… qui sentences appearing within a citation or not (Le Monde)

64 Chapter 2 | Corpora and extraction

Note that several semantic and morphosyntactic properties of all il y a… qui sentences (clefts and cleft lookalikes) extracted from the corpora can be found in the Appendix (Section 4). These include aspects such as the distribution of subject/object relative clauses, the syntactic category (NP, pronoun, proper noun) of the constituent introduced by il y a and the (in)definiteness and semantic class of this constituent.

2.3 Il n’y a que ‘there is only’: extraction of the data

Apart from the main queries for the extraction of the il y a… qui sentences, a second query was used in order to extract the structure il n’y a que … qui/que/qu’ in YCCQA and Le Monde. One of the factors that make il n’y a que clefts interesting to consider is the fact that they can introduce non-(pro)nominal constituents, such as prepositional phrases (89), whereas this is not possible in il y a clefts without the ne… que ‘only’ particle (90). 64

(89) Il n’y a qu’à moi que Jean parle. ‘Jean only talks to me.’ (Lit. ‘There’s only to me that Jean talks’)

(Pierrard 1985:46, my translation)

(90) * Il y a en bateau que Jean voyage. ‘There’s only by boat that Jean is traveling.’ (Pierrard 1985:46, my translation)

All instances of ‘(il n’)y a que’ with all different tenses of the verb avoir were extracted from YCCQA and Le Monde and then manually coded. The query did not specify any context after que, which means that it yielded instances with and without relative clauses (see (91) - (92)).

(91) Il n’y a que – with relative clause introduced by qui y'a que toi qui comprend ou c'est moi qui pige rien à rien ?

‘Are you the only one who understands or is it me who doesn’t understand a thing?’ (Lit. ‘Is there only you who understands…’) (YCCQA)

64 English there clefts may be less restricted in this sense, as shown by the acceptable example in (i): (i) There’s on the table that you may have left it. (Davidse 2000:1106)

Chapter 2 | Corpora and extraction 65

(92) Il n’y a que – without relative clause Jusqu'à présent, il n'y eut que quelques échos assourdis.

‘Until now, there have only been a few muted echoes.’ (Le Monde)

Furthermore, if a relative clause was present, it could be introduced by qui, que, qu’ or any other expression, such as où ‘where’ (93), and the data also contain sentences with infinitival clauses such as pour savoir… ‘to know…’ in (94). More examples of extracted il n’y a que sentences with infinitival clauses and relativizers other than qui/que/qu’ can be found in the Appendix (Section 6). In the rest of the dissertation, I will limit myself to analyzing the cases with relative clauses introduced by qui/que/qu’.

(93) Il n’y a que – with clause introduced by où ‘where’ – A: Peut-on installer la climatisation dans une 206 qui ne l'a pas ?

‘Is it possible to install air conditioning in a 206 that doesn’t have it?’

– B: Il n’y a que certaine voiture japonaise où c'est possible

‘That’s only possible in certain Japanese cars.’

(Lit. ‘There is only in certain Japanese cars where it’s possible.’) (YCCQA)

(94) Pour ‘for’ + infinitive Il y a que toi pour savoir si tu doit l'acheter ou pas.

‘Only you can know if you should buy it or not.’

(Lit. ‘There is only you for know.INF if you should buy it or not’) (YCCQA)

The distribution of the different classes is presented in Table 7:

Le Monde YCCQA Total n % n % n % No relative clause 175 74% 168 44% 343 55% Relative clause (qui/que) 41 17% 179 47% 220 35% Infinitival clause 19 8% 30 8% 49 8% Other relativizer (où…) 2 1% 6 2% 8 1% Total 237 100% 383 100% 620 100% Table 7: Il n’y a que with or without (different types of) relative clauses

------

Chapter 3

Application of criteria to the corpus data

In Chapter 1, I described the diagnostics that have been proposed in previous studies to distinguish between il y a clefts and their lookalikes. In this chapter,65 I apply these diagnostics to all il y a… qui sentences extracted from the three corpora. The intention is to account for the data in a systematic way, i.e. not only select the clearest cases of il y a clefts and il y a cleft lookalikes and disregard the more difficult ones.66 In doing so, I aim to achieve two goals: (i) firstly, to determine whether the diagnostics can make a clear distinction between il y a clefts and il y a cleft lookalikes in all cases, or whether problems arise, as is often the case when diagnostics that are mainly based on relatively few, prototypical examples are confronted with authentic language data. If such problems indeed arise (and they do, as will be shown below), the question is what causes them. (ii) Secondly, once the il y a clefts are singled out from their lookalikes, I will be able to analyze these cleft instances in more detail in the rest of this dissertation.

The structure of this chapter is as follows. In section 3.1, I present a quantitative overview of the different classes (clear clefts, clear cleft lookalikes, unclear cases). I then analyze several examples in detail in order to show that the diagnostics from previous works effectively identify clear instances of il y a clefts (section 3.2) and of il y a cleft lookalikes (section 3.3). After presenting the clear cases, I show that the diagnostics also lead to unclear results. The first class of unclear cases

65 Part of this chapter will be published in Karssenberg (accepted). 66 See Marzo et al. (2012) about the difference between “corpus-illustrated analyses, [which use] hand-picked corpus examples to illustrate a given point, and corpus-based analyses, [which are] characterized by a systematic analysis of corpus instances and empirical verification of theoretically grounded hypotheses” (Marzo et al. 2012:2). 68 Chapter 3 | Application of criteria to the corpus data

contains instances that are semantically ambiguous between a cleft reading and a cleft lookalike reading (“unclear case 1”, section 3.4). In the second class of unclear cases, the diagnostics yield contradictory results (“unclear case 2”, section 3.5). In section 3.6, I propose that there are three ways of defining il y a clefts. In section 3.7, I explicitly state the way in which I define il y a clefts in this dissertation.

3.1 Quantitative overview

For the sake of ease, I repeat the overview of the different diagnostics presented in Chapter 1 below (Table 8).

Test or Section property Positive diagnostics

Declefting test 1.1 Presence of proper noun property 1.2.1 ‘What's happening?’ test 1.3.3 Negative diagnostics Pronominalisation of relative clause + its antecedent test 1.2.2.1 Dislocation of relative clause + its antecedent test 1.2.2.2 Omission of relative clause test 1.2.2.3 Adjectival/prep. equivalent of relative clause test 1.2.2.4 Individual-level predicate property 1.3.2 Object relative clause property 1.3.2 Reformulation of il y a test 1.3.3.2 Negation (il n'y a pas) test/property 1.3.3.3 Table 8: Overview of different diagnostics for the distinction il y a cleft vs. cleft lookalike

The way in which I applied these diagnostics is the following. Concerning the type of relative clause, I analyzed the relative clause as restrictive if at least one test that identifies RRCs gives a positive result. The different properties (presence of a proper noun, individual-level predicate, object relative clause) were used as indications, but

Chapter 3 | Application of criteria to the corpus data 69

not as definitive proof. Negation was not taken into account for the reasons explained in section 1.3.3.3.

As for the distinction between il y a clefts and il y a cleft lookalikes, for the time being, I remain neutral about which criteria are sufficient and/or necessary, in line with previous analyses. As I will show in sections 3.4 and 3.5, not specifying the status of the diagnostics as sufficient or necessary leads to classificatory problems. I will present my own view on the status of the different diagnostics in section 3.6. I used the diagnostics in the following way to distinguish between clefts and cleft lookalikes. If at least one positive diagnostic (e.g. decleftability) does apply, and none of the negative diagnostics apply (i.e. there is no proof that the relative clause is restrictive), I classified the instance as a cleft. If at least one of the negative diagnostics applies (indicating that the relative clause is a RRC), and none of the positive diagnostics apply (e.g. declefting is impossible), I classified the instance as a cleft lookalike. If, finally, both positive and negative diagnostics apply simultaneously, or if it is not possible to apply the diagnostics that identify RRCs with certainty, I classified the sentence as ‘unclear’. This is illustrated in Figure 1:

All il y a… qui sentences

At least one positive diagnostic None of the positive diagnostics for clefthood applies (e.g. decleftability) for clefthood apply

At least one diagnostic indicates No diagnostics indicate that the sentence that the sentence is is not a cleft not a cleft

Unclear case “Il y a cleft” “Il y a cleft lookalike”

Figure 1: Application of the diagnostics to the corpus data

70 Chapter 3 | Application of criteria to the corpus data

Applying the criteria in this way led to the results presented in Table 9.67 As can be seen, two thirds of all il y a… qui sentences can be classified as cases of clefts (28%) or cleft lookalikes (32%). However, in 40% of all cases, it is not clear how the il y a… qui sentence should be classified according to the criteria from previous analyses. The two main sources of this lack of clarity are discussed in sections 3.4 and 3.5.

Le Monde YCCQA CFPP Total n % n % n % n % Il y a clefts68 (section 3.2) 81 19% 285 26% 210 39% 576 28% Il y a cleft lookalikes (section 3.3) 190 45% 308 28% 152 28% 650 32% Unclear cases (sections 3.4 & 3.5) 151 36% 500 46% 182 33% 833 40%

Total 422 100% 1093 100% 544 100% 2059 100% Table 9: Classification il y a cleft vs. cleft lookalike according to traditional criteria

3.2 Il y a clefts

This section is dedicated to il y a… qui sentences that can easily be classified as eventive (section 3.2.1) or as specificational il y a clefts (section 3.2.2).

3.2.1 Eventive il y a clefts

As explained in section 1.3.1, il y a clefts come in two main subtypes: those that express events and those which are semantically specificational. The first subtype is illustrated in (95). All these instances introduce a discourse-new referent (une main ‘a

67 The cleft/cleft lookalike distinction in the il n’y a que… qui dataset will be discussed in Chapter 5.1. 68 These include both eventive and specificational il y a clefts (see chapter 1.3.4). In other words, even if some diagnostics that determine whether the sentence has eventive semantics yield negative results, I do not take this as an indication that the sentence is an il y a cleft lookalike or an unclear case if it is specificational (see further discussion of specificational clefts in section 3.2.2).

Chapter 3 | Application of criteria to the corpus data 71

hand’, un ouvrier ‘a worker’, quelqu’un ‘someone’) that is involved in some event (grabbing my throat, crying, asking for money).

(95) a. Je fais un rêve bizarre depuis quelques jours, vous avez une interprétation ? (…) subitement je vois une lumière blanche à la surface de l'eau et y'a une main qui me saisit par le cou et qui me relève.

‘I have this recurrent dream since a couple of days (…) Suddenly I see a white light above the water and there’s a hand that grabs my throat and that lifts me up.’ (YCCQA)

b. Le matin, on arrangeait ces cadavres avec de la chaux. Un peu de terre, un peu de chaux, jusque quand c'était plein. Un jour, il y a un ouvrier qui a crié : « Mais ça, c'est ma fille ! C'est ma fille ! »

‘In the morning, we arranged those corpses with chalk. A bit of soil, a bit of chalk, until it was full. One day, there’s a worker who cried: “But that’s my daughter! It’s my daughter!”’ (Le Monde)

c. le métro c'est c'est pas très joy- c'est pas très marrant + euh + + parce que on peut pas prendre une rame + sans qu'y ait euh sans qu'y ait quelqu'un euh qui vous demande de l'argent

‘the metro that’s that’s not very cheer- it’s not funny, uh, because you can’t get in without there being someone who asks you for money.’ (CFPP)

All these sentences can be declefted (see chapter 1.1 about the decleftability criterion). In other words, they can be transformed into an SV sentence that expresses the same proposition:

(96) a. Une main me saisit par le cou et me relève. (= (95a)) ‘A hand grabs my throat and lifts me up.’

b. Un ouvrier a crié: … (= (95b))

‘A worker cried: …’

c. … sans que quelqu’un vous demande de l’argent. (= (95c))

‘… without someone asking you for money.’

72 Chapter 3 | Application of criteria to the corpus data

Although constituency tests such as pronominalisation and dislocation yield grammatically acceptable results when applied to the clefts in (95) and hence indicate that the relative clause is a RRC, these reformulations clearly no longer express an event:

(97) a. Subitement, il y en a une, une main qui me saisit par le cou. ‘Suddenly, there’s one of them, a hand that lifts me up.’

b. Un jour, il y en a un, un ouvrier qui a crié :…

‘One day, there’s one of them, a worker who cried: …’

c. ?… sans qu’il y en ait un, quelqu’un qui vous demande de l’argent.

?‘… without there being one of them, someone who asks you for money.’

Note furthermore that all the examples in (95) have lexically eventive predicates. However, the corpus data confirm that this is not a requirement for an eventive (‘stage-level’) interpretation (see chapter 1.3.2). For instance, the VPs ‘be too dry’, ‘be too small’, ‘not being there’ and ‘not being round’ in (98) are lexically stative, but they are interpreted as stage-level predicates: the properties they express are temporary rather than permanent. The fact that these sentences are clefts is confirmed by the decleftability test in (99) and by the fact that several of the clefted NPs are proper nouns (Zizou, Riberi, Anelka), which are impossible antecedents for RRCs.

(98) Lexically stative predicate interpreted as stage-level – A: Les bleus sont ils la en touristes ? (…) qu'est ce qu'ils attendent pour marquer un but ????

‘Is the French soccer team there as a bunch of tourists? (…) What are they waiting for to score????’

– B: Déconnez pas ; La pelouse est trop séche y a zizou qu'y69 tient pas d'bout riberi qu'est trop p'tit Anelka qu'est pas là...... Le ballon qu'est pas rond

69 It is unclear whether the writer of this message misspelled the regular relative pronoun qui ‘that’ (a mistake which is not uncommon in the YCCQA corpus), or whether qu’y is intended as a contraction of the relative pronoun qui ‘that’ and the locative adverb y ‘there’.

Chapter 3 | Application of criteria to the corpus data 73

‘Don’t be stupid: the grass is too dry, there’s Zizou who can’t keep upright Riberi who is too small Anelka who isn’t there… the ball that isn’t round’ (YCCQA)

(99) Declefted SV equivalent La pelouse est trop sèche, Zizou ne tient pas debout, Riberi est trop petit, Anelka n’est pas là, le ballon n’est pas rond.

‘The grass is too dry, Zizou can’t keep upright, Riberi is too small, Anelka isn’t there, the ball isn’t round.’

3.2.2 Specificational il y a clefts

Recall that a few authors acknowledge the existence of a subtype of il y a clefts that has specificational rather than eventive semantics (see chapter 1.3.4). Such sentences can be considered clefts on the basis of the decleftability criterion and the type of relative criterion (the relative clause is not restrictive). Consider for instance the clefts in (100) - (102), which all specify a value (Lost, homeopathy, the Fnac Bastille) for a discourse-given variable (great TV shows, ways to eliminate stress, bankruptcies).

(100) – A: Quelles est votre meilleure série du moment ? ‘What’s your favorite TV show right now?’

– B: "How I Met Your Mother" c'est génial, y'a aussi "Lost" qui est bien.

‘”How I Met Your Mother” is great, there’s also “Lost” that is good.’

(YCCQA)

(101) – A: Je vais bientôt passer un examen, et j'ai le trac. pouvais vous me donner des conseils ?

‘I have an exam soon and I’m stressed out. Can you give me some advice?’

– B: détends-toi, fais aussi du sport, ça permet de bien éliminer le stress. Il y a aussi l'homéopathie qui fonctionne vraiment super bien !

‘Relax, do some sports, that enables you to get rid of the stress. There’s also homeopathy that works really well! (YCCQA)

74 Chapter 3 | Application of criteria to the corpus data

(102) – A: Y a pas d'faillite de choses comme ça ? ‘Aren’t there bankruptcies things like that?’

– B: non y a pas d' faillite ni rien -- ah si y a la Fnac Bastille qui va fermer.

‘no there are no bankruptcies no nothing – oh yes there’s the Fnac Bastille that’s going to close.’ (CFPP)

All three sentences can be declefted (103), and the presence of a proper noun in (100) and (102) also indicates that the relative clause is not a RRC, and hence, that those sentences are clefts. The fact that the relative clauses relate back to the preceding question indicates that they are cleft relative clauses rather than appositive ones (which express new information that is not at issue, see chapter 1.2.3).

(103) a. Lost est bien aussi. (= (100)) ‘Lost is also good.’

b. L’homéopathie fonctionne vraiment super bien. (= (101))

‘Homeopathy works very well.’

c. ah si, la Fnac Bastille va fermer. (= (102))

‘Oh yes, the Fnac Bastille is going to close.’

However, as already discussed in section 1.3.4, tests that determine whether the sentence has an eventive interpretation do not (always) apply to specificational il y a clefts, since they are a different subtype of il y a cleft that do not necessarily have eventive semantics (see Verwimp & Lahousse 2016). My corpus data further confirm that other criteria based on eventive semantics also do not hold. Firstly, there is no ban on individual-level predicates in specificational il y a clefts, as shown by (104). This example presents a value (le Magellan Maestro 3250) for a variable (des GPS à reconnaissance vocale ‘GPS systems with voice recognition’). The fact that the clefted NP is a proper noun and that the sentence can be declefted (105) indicate that it qualifies as a cleft. However, the relative clause clearly expresses an individual-level predicate (having speech recognition).70

70 See chapter 1.3.2 about the distinction between individual-level and stage-level predicates as a diagnostic for clefthood.

Chapter 3 | Application of criteria to the corpus data 75

(104) – A: Est-ce qu'il y a des GPS a reconnaissance vocal disponible au Québec ? ‘Are there GPS systems with speech recognition for sale in Quebec?’

– B: Oui il y a le Magellan Maestro 3250 qui a la reconnaissance vocale tel qu'expliquer sur le site de CAA Québec: www

‘Yes there’s the Magellan Maestro 3250 that has speech recognition as explained on the CAA Quebec website: www’ (YCCQA)

(105) Cleft + decleftability test Oui, le Magellan Maestro 3250 a la reconnaissance vocale.

‘Yes, the Magellan Maestro 3250 has speech recognition.’

Furthermore, it has been claimed that eventive il y a clefts cannot contain object relative clauses (see chapter 1.3.2). This ban does not hold for specificational il y a clefts, as shown in (106): the clefted element specifies a value (farting) for a variable (great pleasures you shouldn’t forget), and this variable is encoded by an object relative clause.

(106) Cleft + object relative clause – A: Faire caca est-il un des plus grands plaisirs à notre disposition sur cette

terre ?

‘Is pooping one of the greatest pleasures we have at our disposition on this earth?’

– B: mouai ... il y a aussi le pet qu'il ne faut pas oublier !

‘Well yeah… there’s also farting that you shouldn’t forget!’ (YCCQA)

Another test that supposedly holds for eventive il y a clefts is that you cannot omit the relative clause (see chapter 1.2.2.3). However, as already shown by Verwimp & Lahousse (2016) (see also Blanche-Benveniste 1990:61; Léard 1992:45; Charolles 2002:156 for this observation), the relative clause in specificational clefts can be omitted when it expresses discourse-given information. For instance, the relative clause ‘that has speech recognition’ in (104) literally repeats the preceding question, and can therefore be left out, as shown in (107).

76 Chapter 3 | Application of criteria to the corpus data

(107) Cleft + omission of the relative clause [Based on (104)] Oui, il y a le Magellan Maestro 3250.

‘Yes, there’s the Magellan Maestro 3250.’

To conclude, several previously presented diagnostics for clefthood do not apply to specificational il y a clefts. However, this does not necessarily question the validity of these diagnostics for il y a clefts of the eventive type. Furthermore, the fact that specificational il y a clefts can be declefted and do not contain RRCs clearly indicate that they are indeed clefts.

3.3 Il y a cleft lookalikes

This section illustrates clear cases of il y a cleft lookalikes. It is generally argued that il y a cleft lookalikes (i) cannot be declefted (chapter 1.1), (ii) contain RRCs or appositive relative clauses (chapter 1.2) and (iii) that they do not have eventive semantics (Chapter 1.3) (see e.g. Lambrecht 1986, 1988a). For the time being, I will assume this description of il y a cleft lookalikes. In what follows, I present clear cases of il y a cleft lookalikes with restrictive (section 3.3.1) or appositive (0) relative clauses.

3.3.1 Cleft lookalikes with RRCs

Many instances cannot be considered il y a clefts, because (i) they cannot be declefted, and (ii) the relative clause tests positive as restrictive. There are several possible reasons why a given sentence cannot be declefted. In a first group of il y a… qui sentences, il y a introduces a pronoun such as celui ‘that one’ or ceux ‘those’, as in (108) and (109).

(108) (…) les joueurs de chez nous ne sont pas hommes à croire aux sornettes. Certes, il y a celui qui joue toujours avec le même slip. Celui qui ne joue jamais sans la chaîne offerte par son grand-père (…).

Chapter 3 | Application of criteria to the corpus data 77

‘(…), our players don’t believe in nonsense. Sure, there’s the one who always wears the same underwear during a game. The one who never plays without the necklace his grandfather gave him (…).’ (Le Monde)

(109) « Face au vent du changement, a-t-il conclu, il y a ceux qui construisent des paravents, et il y a ceux qui construisent des moulins. » ‘“Confronted with change, he concluded, there are those who build windbreaks, and there are those who build windmills.”’ (Le Monde)

The pronouns celui ‘the one’ and ceux ‘those’ cannot function as subject in a declefted sentence (without a relative clause), as shown in the ungrammatical SV sentences in (110a) and (111a). Moreover, the relative clauses in (108) and (109) are restrictive: they can be replaced by the spatial clitics –ci ‘here’ or –là ‘there’ in (110b) and (111b).

(110) [Based on (108)] a. * Certes, celui joue toujours avec le même slip.

‘Sure, the one always wears the same underwear during the game.’

b. Certes, il y a celui-là.

‘Sure, there’s that one.’

(111) [Based on (109)] a. *… ceux construisent des paravents et ceux construisent des moulins.

* … ‘those build windbreaks and those build windmills.’

b. Il y a ceux-ci et il y a ceux-là.

‘There’s these here and there’s those there.’

There are also other reasons for which declefting is impossible, and hence these sentences cannot be classified as il y a clefts. Consider for instance the underlined il y a... qui sentence in (112).

78 Chapter 3 | Application of criteria to the corpus data

(112) Pour la majorité des Américains (…) il y a mensonge et mensonge. Il y a les mensonges formulés dans l'exercice de fonctions officielles et il y a les mensonges qui ont trait à la vie privée, et ces deux catégories ne sauraient être mises dans le même panier. ‘For the majority of Americans (…) there are lies and lies. There are lies (that are) formulated while performing official functions and there are lies that have to do with private life, and those two categories shouldn’t be put in the same basket.’ (Le Monde)

The sentence in (112) yields an infelicitous declefted equivalent in (113a): while the original sentence enumerates certain types of lies, the proposition expressed by the SV equivalent is a statement about lies in general. In other words, the proposition is no longer the same, meaning that the decleftability test failed. Moreover, the relative clause in (112) can be replaced by the present participle ayant ‘having’ (113b), indicating that the relative clause is restrictive, and, hence, that the sentence is not a cleft. This is also supported by the fact that the first-mentioned mensonges ‘lies’ in (112) is modified by the past participle formulés ‘formulated’.

(113) [Based on (112)] a. Il y a les mensonges formulés dans l'exercice de fonctions officielles et les

mensonges ont trait à la vie privée.

‘There are lies (that are) formulated while performing official functions and lies have to do with private life.’

b. … et il y a les mensonges ayant trait à la vie privée.

… ‘and there are lies having to do with private life.’

Another example is presented in (114). Here, the decleftability test yields an incomprehensible result (115a)71 and the relative clause can be replaced by the

71 The fact that this example contains an object relative clause seems to complicate the transition to a SV sentence. However, object relative clauses do not always block decleftability: specificational il y a clefts such as (i) do correspond to canonical equivalents (ii). (i) (Are there any TV shows about crime that you like?) – Oui, il y a The Wire que j’aime bien. ‘Yes, there’s The Wire that I like.’ (ii) Oui, j’aime bien The Wire. ‘Yes, I like The Wire.’

Chapter 3 | Application of criteria to the corpus data 79

adjective uniques ‘unique’ (115b), indicating that it is restrictive. Furthermore, the relative clause and its antecedent can be pronominalized and dislocated (115c).

(114) Chaque langue a sa richesse, sa spécificité, sa beauté. Dans chaque langue, il y a des notions et des subtilités que d'autres n'ont pas. ‘Every language has its richness, its specificity, its beauty. In every language, there are notions and subtleties that others don’t have. (YCCQA)

(115) a. * Dans chaque langue, d’autres n’ont pas des notions et des subtilités. * ‘In every language, others don’t have notions and subtleties.’

b. Dans chaque langue, il y a des notions et des subtilités uniques.

‘In every language, there are unique notions and subtleties.’

c. Chaque langue en a, des notions et des subtilités que d'autres n'ont pas.

‘Every language has them, notions and subtleties that others don’t have.’

3.3.2 Cleft lookalikes with appositive relative clauses

The second subcategory of il y a cleft lookalikes are il y a... qui sentences containing appositive relative clauses. These relative clauses can be preceded by a comma, as in (116). The presence of a comma and a proper noun indicate that the relative clause is appositive.

(116) – A: Vous rappelez vous de "télé chat" ? qui peut me décrire les différents personnages de cette émission et ce qu'ils diffusaient ?

‘Do you remember [the TV show] Téléchat? Who can describe the different characters of that show and what they were about?’

(…)

– B: Et il y avait aussi Léguman, qui sortait du frigo et avait une tête de citrouille si je me souviens bien.

‘And there was also Léguman, who appeared from the fridge and had a pumpkin head if I remember correctly.’ (YCCQA)

80 Chapter 3 | Application of criteria to the corpus data

As already reported by others, appositive relative clauses are not always preceded by a comma (see chapter 1.2.3), especially in informal written registers, where punctuation can sometimes be completely absent. Consider for instance the example in (117): the relative clause does not restrict the reference of ‘my father’s death certificate’, which is identifiable without knowing that it dates back to 1987. This information is presented as ‘extra information’. A cleft interpretation is also not warranted here, as shown by the awkward SV equivalent in (118), which is not compatible with the preceding context. The sentence in (117) states that a certain object (a death certificate, which happens to date back to 1987) is in a certain location (a drawer). It does not predicate a property (the date 1987) of the death certificate, as is the case in (118).

(117) – A: Qu'est-ce qui dort au fond d'un tiroir depuis une éternité, chez vous ? ‘What’s been lying in the back of a drawer at your place for an eternity?’

– B: Un agenda que j'ai tenu en 1973 et un porte feuille que j'avais offert à mon père dans ma jeunesse et je l'ai récupérer à sa mort et dedans il y a son annonce de décès qui est de 1987.

‘A diary I kept in 1973 and a wallet that I gave my father when I was young and I got it back when he died and inside there’s his death certificate which dates back to 1987’ (YCCQA)

(118) # et dedans son annonce de décès est de 1987. # ‘and inside, his death certificate dates from 1987’

To conclude, the diagnostics from previous works successfully identify the il y a… qui sentences discussed in this section as cleft lookalikes. Firstly, they cannot be declefted. Secondly, other properties and tests (e.g. substitution of relative clause by adjectival equivalent, presence of a comma) support the outcome of the decleftability criterion by showing that the il y a… qui sentences do not behave like clefts.

Chapter 3 | Application of criteria to the corpus data 81

3.4 Unclear case 1: type of relative clause underspecified

So far we have only looked at cases of clefts or cleft lookalikes. However, in section 3.1, I announced that in approximately 40% of all il y a… qui sentences in my dataset, the diagnostics from previous analyses of il y a clefts cannot be applied without problems (Table 9).72 In this section, I explain the source of the confusion in the first type of unclear case, in which the type of relative clause is underspecified.

Recall that it is generally assumed that clefts do not contain relative clauses of the restrictive type (RRCs). In order to determine whether a given relative clause is restrictive or not, one can apply constituency tests (see chapter 1.2.2). However, in what follows, I will show that the outcomes of several constituency tests depend on the aspectual interpretation of the sentence (stative or eventive interpretation). As a consequence, if the aspectual interpretation of the sentence is underspecified, it is not possible to evaluate whether the tests yield positive or negative results. Therefore, it is unclear whether the sentence should be considered a cleft or not.73

I first analyze several problematic il y a... qui sentences from the corpus data (section 3.4.1). I then propose that the unclear distinction between RRCs and cleft relative clauses comes as no surprise, given the blurry distinctions between other types of relative clauses in previous studies (section 3.4.2). I will argue there that my data support the idea presented in those studies that different relative clause types should not (always) be seen as discrete classes. In section 3.4.3, I present several hypotheses to account for the way such unclear il y a… qui sentences are processed by the speaker/hearer.

72 Interestingly, the fact that the distinction between clefts and cleft lookalikes can be difficult is also reported in corpus research on c’est clefts (e.g. Scappini 2006; Rouquier 2007, 2015). 73 In this sense, my data reflect the following observation by König & Lambrecht (1999): “As is often the case with constituency tests, the (…) evidence is not unequivocal” (König & Lambrecht 1999:4).

82 Chapter 3 | Application of criteria to the corpus data

3.4.1 Corpus data: aspectually underspecified il y a... qui sentences

Consider the example in (119), which appears without any prior context on the Yahoo forum. In this example, the VP qui posent des questions ‘who ask questions’ is in the present tense, which is underspecified for progressive or ‘ongoingness’ aspect in French (Copley & Roy 2015). The sentence can be interpreted as expressing (i) the existence of a certain type of people (‘stupid people’ = RRC reading), or (ii) as the observation of a process taking place at the moment of utterance (‘people asking stupid questions’ = cleft reading). This is reflected by the two English translations, corresponding either to a stative or an eventive reading.74

(119) Pourquoi y'a plein de gens qui posent des questions débiles au lieu de chercher sur Google ?? (YCCQA) – Interpretation 1. ‘Why is it that so many people are asking stupid questions instead of googling it?’ = cleft reading

– Interpretation 2. ‘Why are there so many people who ask stupid questions instead of googling it?’ = RRC reading

If we now try to apply the tests from the literature, the results are somewhat confusing. Firstly, it is possible to replace il y a by il existe ‘there exists’ (120), indicating that the sentence has stative rather than eventive semantics. Furthermore, the pronominalisation and dislocation tests can also be applied (121), indicating that the relative clause forms a single constituent with the antecedent.

(120) ‘Il existe’ reformulation = stative interpretation Pourquoi existe-t-il plein de gens qui posent des questions débiles ??

‘Why do there exist so many people who ask stupid question?’

74 Choi-Jonin & Lagae (2005:51) acknowledge that il y a… qui sentences can be ambiguous between a cleft reading and an existential reading (with a RRC) when a partitive indefinite NP combines with a relative clause that can be ‘specifying’, i.e. eventive (“Il convient de souligner ici que la distinction entre [cleft relative clauses] et [RRCs] est parfois difficile à établir, surtout dans le cas d’une relative avec prédicat pouvant être spécifiant et antécédent indéfini à lecture partitive”, Choi-Jonin & Lagae 2005:51, my emphasis).

Chapter 3 | Application of criteria to the corpus data 83

(121) Pronominalisation & dislocation = stative interpretation Pourquoi il y en a plein, des gens qui posent des questions débiles ??

‘Why are there so many of them, people who ask stupid questions?’

Importantly, the tests in (120) and (121) evoke a stative interpretation of the sentence, and, according to certain authors, indicate that the relative clause is restrictive. However, the sentence can also be declefted, as in (122), which would indicate that it is a cleft. Interestingly, the declefted sentence is more compatible with an eventive interpretation, perhaps because human referents in subject position are preferably interpreted as agents, which cause or initiate events.

(122) SV equivalent (i.e. declefted) = eventive interpretation Pourquoi plein de gens posent des questions débiles au lieu de chercher sur Google?

‘Why are so many people asking stupid questions instead of googling it?’

The il y a… qui sentence in (119) is thus problematic for classification purposes, because under a stative reading, it seems to be a cleft lookalike with a RRC, whereas under an eventive reading, the sentence expresses the same proposition as an SV sentence (i.e. it can be declefted) and is therefore presumably a cleft.75 In other words, applying the constituency tests tends to evoke the stative interpretation rather than the eventive one, whereas the opposite is true of the decleftability test. It thus seems impossible to classify (119), because there is nothing in the discourse context that favors either a stative or an eventive interpretation. Importantly, the different

75 Under a cleft lookalike reading, the relative clause presumably takes only the nominal gens ‘people’ as its antecedent (i), whereas under the cleft reading, the whole NP plein de gens ‘so many people’ is the antecedent of the relative clause (ii), as is the case in unambiguous clefts such as (iii) (see Davidse 2000 about the antecedent of cleft relative clauses in there clefts). (i) Cleft lookalike

Pourquoi y'a plein de [gens [qui posent des questions débiles au lieu de…]]COMPLEX NP

‘Why are there so many [people [who ask stupid questions instead of….]] COMPLEX NP (ii) Cleft Pourquoi y'a [plein de gens] [qui posent des questions débiles au lieu de….] ‘Why are [so many people] [asking stupid questions instead of…]’ (iii) Cleft Pourquoi y’a [ton frère] [qui pose des questions débiles?] ‘Why is your brother asking stupid questions?’

84 Chapter 3 | Application of criteria to the corpus data

formulations express more or less the same proposition: the differences are very slight.76

As a second illustration of an il y a… qui sentence allowing two readings, consider (123): the sentence can either be interpreted as (i) the description of a process (‘words losing their meaning’) which led to a certain final result, or as (ii) the description of a state (‘many banal words’), in which case the process of ‘banalisation’ is not deemed pertinent/is not evoked.

(123) – A: Pourquoi les q/ristes [= les internautes du forum ‘Questions et Réponses’] s'insultent les un les autres de nazi si facilement ? A force d'utiliser le mot nazi a tout bout de champs, il n'a plus de sens

‘Why do so many forum visitors call each other Nazis so quickly? By using the word Nazi all the time, it doesn’t have any meaning anymore.’

– B: il y a trop de mots qui sont devenus "banals" lorsqu'ils auraient du rester rares... nazi, guerre, mort, attentat, terrorisme, génocide, viol, sida, maladie, et j'en passe. (YCCQA)

Interpretation 1 - Too many words have become banal while they should have

remained rare. = cleft reading

Interpretation 2 -There are too many banal(ised) words that should have

remained rare. = RRC reading

The declefted equivalent in (124) and the first English translation in (123) are compatible with an eventive interpretation (in which the process is evoked). Furthermore, it is possible to add a temporal expression such as au cours des dernières années ‘during the past few years’ to the sentence, which also brings about a process reading. The tests in (125), on the other hand, evoke a stative interpretation.77

76 As stated in chapter 1.3.1, there is no general consensus in the linguistic literature about how to determine the distinction between states and events in corpus data. 77 Again, from the syntactic point of view, two parses seem possible, (ia) representing a cleft interpretation, while (ib) indicates that the relative clause is embedded inside a complex NP. (i) a. Il y a [trop de mots] [qui sont devenus "banals" lorsqu'ils…]

b. Il y a trop de [mots [qui sont devenus "banals" lorsqu'ils…]]COMPLEX NP

Chapter 3 | Application of criteria to the corpus data 85

(124) Interpretation: process reading Trop de mots sont devenus "banals" lorsqu'ils auraient du rester rares...

‘Too many words have become banal when they should have remained rare.’

(125) Interpretation: state (no process implied) a. Il existe trop de mots qui sont devenus "banals"…

‘There exist too many words that have become banal.’

b. Il y a trop de mots banalisés…

‘There are too many banalized words.’

c. Il y en a trop, des mots qui sont devenus "banals"…

‘There are too many of them, words that have become banal.’

To conclude, the cases discussed in this section show that so-called ‘formal’ tests such as pronominalisation cannot be applied blindly, but are sensitive to the aspectual interpretation of the sentence. However, when the interpretation is underspecified in the absence of contextual indications such as adverbs that favor either a stative or an eventive reading, it is unclear which tests apply and which don’t. According to Choi- Jonin & Lagae (2005), seemingly ambiguous il y a… qui sentences can be classified as clefts or as cleft lookalikes if one disposes of a sufficiently large discourse context. I’ve shown that even when considered in their original context, not all il y a... qui sentences can be classified as clefts or cleft lookalikes.

Importantly, aspectual underspecification has also been reported in analyses of other sentence types. For instance, it has been observed that “in certain cases the same situation can be described as a state or as an event” (Bhatt & Pancheva 2006:10, my emphasis). The authors present the sentences in (126), of which the first is stative while the second is eventive/dynamic according to them. The two sentences represent two readings or conceptualizations of the same situation.

(126) a. Maya is asleep. b. Maya is sleeping. (Bhatt & Pancheva 2006:10).

86 Chapter 3 | Application of criteria to the corpus data

Moreover, a single sentence can be interpreted as stative or as eventive, depending on context (Schlesinger 1995:121; Manninen 1997:2), as in (127): 78

(127) The car drives easily. (Schlesinger 1995:121)

According to Schlesinger (1995), “The interpretation of [(127)] as an Event or as a State may depend on the context: when driving the car, it may mean that it is now moving effortlessly, i.e., the sentence will have an Event predicate. In a different context, the sentence may refer to the car’s driving characteristics, that is, it may have a State predicate” (Schlesinger 1995:121). Crucially, without any additional contextual indications, the two interpretations remain possible: the sentence is aspectually underspecified.

In light of the fact that aspectual underspecification is also found in other contexts, it does not come as a surprise that it is sometimes difficult to distinguish between eventive and non-eventive il y a… qui sentences, as in (119) and (123). Importantly, however, this probably does not mean that such utterances are difficult to process for the hearer, since the difference in interpretation is so subtle.

To summarize, I showed that il y a… qui sentences can be aspectually underspecified and that this has repercussions for the application of constituency tests. Since in such cases, it is not clear whether the relative clause is restrictive or not, one cannot classify them as either clefts or cleft lookalikes with certainty. Furthermore, I showed that aspectual underspecification is not limited to il y a… qui sentences. In the next section, I present further support from previous studies for the fact that boundaries between relative clause types may be blurry.

78 See also Croft 1990 about event views; Olson (1994) and Apothéloz & Nowakowska (2010) about sentences that are aspectually underspecified for dynamicity or telicity and De Saussure (2014:2) about the telic/non-telic interpretation of NPs such as ‘construction’, as in the construction of the wall takes time vs. that’s a beautiful construction.

Chapter 3 | Application of criteria to the corpus data 87

3.4.2 About the non-discrete boundaries between relative clause types

RRCs are often taken to represent a well-established category in studies on clefts and this notion is still widely used in linguistic theory in general as well as in didactic language manuals. However, it has received strong criticism in the past few decades from scholars who try to distinguish RRCs from appositive relative clauses (e.g. Le Goffic 1979; Fuchs 1987a, see Gapany 2004 and Loock 2013 for detailed overviews). An example of such criticism is Gapany (2004), who traces the distinction between RRCs and non-RRCs back to the philosophy of logic, where it was picked up by grammarians, who eventually passed it on to linguistics. Gapany (2004) argues, however, that corpus research shows that the proposed semantic and morphosyntactic criteria are often empirically problematic (Gapany 2004; Loock 2013). Moreover, this distinction, which is often assumed to be universal according to Loock, may in fact not exist in all languages. Gapany (2004:16) and Loock (2013) also show that relative clauses with indefinite NPs as antecedents are more difficult to classify than those with definite antecedents, which are most often used in minimal pairs of restrictive and appositive relative clauses (see also Davidse 2000:1108).79

Another important insight comes from the literature on the L1/L2 acquisition of relative clauses across languages. Several researchers in this field (e.g. Jisa & Kern 1998; Aparici et al. 2016) acknowledge that relative clauses often have multiple semantic, pragmatic and text-structuring functions simultaneously (i.e. the ‘plurifunctionality of relative clauses in discourse’: Dasinger & Toupin 1994:460). For example, the relative clause in (128) allows to reidentify the bees that had been introduced earlier (restricting function), but at the same time, its narrative/pragmatic function is to ‘motivate a main-clause event’, i.e. to explain why the dog is limping. In other words, although this relative clause can be analyzed as restrictive, this does not exclude other higher-level (pragmatic or narrative-structuring) functions.

(128) The dog is limping… it looks like it got bit by a couple of the bees that were chasing him. (Dasinger & Toupin 1994:475)

79 Prosody may be a more reliable diagnostic (presence/absence of an intonational break). However, this diagnostic cannot be applied to RRCs/appositive relative clauses in written French.

88 Chapter 3 | Application of criteria to the corpus data

The observation that relative clauses may perform several functions at once is important, because relative clauses are often classified in discrete categories based on their ‘sole’ function (e.g. ‘restrictive’, ‘continuative/narrative-advancing’). What authors like Dasinger & Toupin (1994) show is that these functions are in fact not always incompatible with each other, and hence that discrete function-based categories may not be the best way to characterize relative clauses when classifying authentic language data.

Moreover, it is generally taken for granted that the NP-internal or NP-external position of the relative clause is a binary feature: it is either embedded inside the NP or it is not. However, an alternative road is pursued by Fox & Thompson (2007), who argue that the integration of the relative clause with respect to its antecedent is not a binary distinction, but a gradient phenomenon. They posit several factors that increase or diminish the degree of integration, and the more the relative clause and the main clause are integrated, the more they are monoclausal. This view of relative clause attachment is hardly compatible with regular views of clefts and the syntax of relative clauses in general, and I remain agnostic about its costs and benefits for the analysis of il y a clefts. However, I do believe it is important to acknowledge the existence of alternative analyses of relative clauses here.

To my knowledge, the empirical problems and alternative analyses concerning the type of relative clause described above are not often acknowledged in studies of clefts, which usually only present carefully selected, clear instances of cleft sentences. Although I do not question the fundamental distinction between the types of relative clauses in clear, prototypical examples, I did show in section 3.4.1 that my corpus data support the view that the boundaries between different types of relative clauses are not always clear-cut. My findings thus confirm previous studies about the difficulty to establish discrete categories of relative clauses in French.

Chapter 3 | Application of criteria to the corpus data 89

3.4.3 Hypotheses about processing il y a… qui sentences with ambiguous relative clauses

In section 1.2.2, I showed that the application of many so-called ‘formal’ tests such as dislocation, pronominalisation and omission must take into account the interpretation of the sentence in its original context. However, since the context can be underspecified or unclear, we do not always have the means to determine with absolute certainty whether the relative clause is of the restrictive, the appositive or the cleft type, as in (129).

(129) Pourquoi y'a plein de gens qui posent des questions débiles au lieu de chercher sur Google ?? (YCCQA) – Interpretation 1. ‘Why is it that so many people are asking stupid questions instead of googling it?’ = cleft reading

– Interpretation 2. ‘Why are there so many people who ask stupid questions instead of googling it?’ = RRC reading

The classification problems caused by instances such as (129) are in line with the empirical problems that have already been raised by authors working on the distinction between restrictive and appositive relative clauses (see section 3.4.2). In other words, these problems are not unique to cleft sentences. They also have consequences for the way in which relative clauses are processed. It is generally assumed that the distinction between RRCs and cleft relative clauses reflects the way in which speakers and hearers generate and parse relative clauses. This raises the question how speakers and hearers analyze the syntax of the relative clause in underspecified cases such as (129). Below, I present two possible hypotheses, although I do not exclude other possibilities.

Hypothesis 1.

Firstly, one could state that the position of the relative clause (NP-internal or NP- external) in cases such as (129) remains underspecified, in accordance with the fact that the aspectual-semantic distinction is underspecified (see section 3.4.1). This would be in line with proposals in psycholinguistics such as ‘Good enough parsing’ (e.g. Swets et al. 2008, see also Muskens 2001), and more specifically about relative

90 Chapter 3 | Application of criteria to the corpus data

clauses, ‘Construal theory’ (Frazier & Clifton 1996). Their general hypothesis is that people don’t always analyze all syntactic and semantic relations when they don’t need to do so for a proper comprehension of the sentence. In the case of ambiguous il y a... qui sentences, this could mean that the speaker/hearer establishes the antecedent of the relative clause, but not the exact interpretation of the relative clause (and, consequently, its NP-internal or NP-external position). If, however, a sentence is clearly eventive, the relative clause is processed as NP-external, in accordance with the fact that the standard constituency tests do not apply in eventive cases. The third option is that the sentence is clearly stative, in which case the relative clause is analyzed as NP-internal, because the constituency tests do apply in such instances.

Hypothesizing that the syntactic position of the relative clause may remain underspecified would also be in line with statements made by Piantadosi et al. (2012) about the benefits of underspecification in language and communicative systems in general. They argue that linguistic ambiguity is such a pervasive phenomenon because it allows for more efficient communication: it avoids redundantly specifying information and allows for the re-use of efficient linguistic units, thus “decreasing the overall effort needed to use a linguistic system” (Piantadosi et al. 2012:290).

Hypothesis 2.

One could also posit that relative clauses are by default NP-internal (and restrictive), unless there is clear evidence to the contrary. For instance, the presence of a proper noun indicates that the relative clause is not restrictive and NP-internal; and the presence of a comma or pause preceding the relative clause indicates that the relative clause is appositive instead of restrictive. Assuming this hypothesis of a default NP- internal relative clause, eventive il y a... qui sentences would have the same syntactic structure as stative il y a... qui sentences, i.e. the relative clause would form a single constituent with its antecedent just like regular RRCs (130):

(130) – A: Qu’est-ce qui se passe ? ‘What’s happening?’

– B: Il y a [des enfants qui jouent dans la cour]COMPLEX NP

There are [children who are playing in the courtyard]COMPLEX NP

Chapter 3 | Application of criteria to the corpus data 91

In this view, the fact that constituency tests such as pronominalisation and dislocation don’t apply to eventive il y a... qui sentences is not taken as evidence that the relative clause and its antecedent are two separate constituents.80

I believe there are no solid grounds for choosing one hypothesis rather than the other. The best shot at falsifying them probably lies in experimental psycholinguistic methods that investigate online language processing. Furthermore, recent psycholinguistic research may also provide insight in the analysis of ambiguous relative clauses in clefts, regardless of which aforementioned hypothesis one assumes: it has been acknowledged that factors such as level of attention or interest, personal expectations and noisiness of the channel affect parsing outcomes (see for instance Traxler 2014). Connecting these findings with the analysis of relative clauses in il y a clefts, it may be the case that different individuals assign different structures to the exact same sentence in the same context. If this is indeed the case, we must abandon the idea that each and every sentence in the corpora has only one ‘true’ interpretation (and syntactic structure) that we as linguists try to uncover. Instead, it is safer to acknowledge that there are several possible parses/interpretations, and that not all speakers and hearers necessarily produce the same one.

3.5 Unclear case 2: decleftable + RRC

In section 3.4, I presented a first type of unclear case: the il y a… qui sentence contains a relative clause that cannot unambiguously be classified as restrictive or not. The unclear cases presented here are of a different nature. In previous analyses (e.g. Lambrecht 1986), it is assumed that the decleftability criterion and the type of relative clause criterion yield the same results, as visualized in Figure 2. In other words, the idea is that il y a… qui sentences that can be declefted always contain cleft relative

80 Indeed, one could argue that these are contexts in which constituency tests such as pronominalisation and dislocation yield false negatives, as described by Tellier (2003:44) (see chapter 1.2.2). Recall that she warned against the fact that in some cases, a string of words may be a constituent despite the fact that standard constituency tests do not work, because there is some other, non-syntactic factor that blocks the test. In the case of eventive il y a... qui sentences that do not contain proper nouns, it is possible that the constituency tests are blocked because they cancel out the original eventive interpretation, as explained in chapter 1.2.2.

92 Chapter 3 | Application of criteria to the corpus data

clauses, and that il y a… qui sentences that contain restrictive (or appositive) relative clauses can never be declefted.

All il y a… qui sentences

Decleftable Not decleftable & Cleft relative clause & Restrictive relative clause

“Il y a cleft” “Il y a cleft lookalike”

Figure 2: Assumption in previous studies about the correlation between decleftability & type of relative clause in il y a clefts

However, in what follows, I will present instances of il y a… qui sentences that contain a RRC, but which can at the same time be declefted. In other words, there seems to be a conflict between the negative diagnostics for clefts (diagnostics that exclude clefthood by identifying a RRC) and the positive diagnostic of decleftability (which indicates that the sentence is a cleft). This group of unclear cases is referred to as “???” in Figure 3.

All il y a… qui sentences

Decleftable Not decleftable

Restrictive relative clause Cleft relative clause

??? “Il y a cleft” “Il y a cleft lookalike”

Figure 3: Possible combinations of decleftability & type of relative clause in the corpus data

Chapter 3 | Application of criteria to the corpus data 93

Consider for instance the il y a… qui sentence in (131), which contains a prototypical RRC. The restrictive status of the relative clause is confirmed by the tests in (132).81

(131) – A: Ce sont les enseignants qui du coup ne n'exigent pas assez des élèves ou c'est pas possible parce qu'il y avait des problèmes de discipline (…) ?

‘Is it the teachers who suddenly are not demanding enough towards the students or is it not possible because there are discipline problems (…)?’

– B: j'en veux pas trop aux enseignants (…) j'pense qu'ils pouvaient pas vraiment faire autrement enfin c'est vrai que y avait+ y avait beaucoup d'élèves qui étaient motivés mais y en avait aussi une bonne partie qui n'en avait rien à faire

‘I’m not really upset with the teachers (…) I think that they couldn’t really do anything differently well it’s true that there were there were many students who were motivated but there was also a large group that didn’t care.’ (CFPP)

(132) Tests for RRCs a. Il y avait beaucoup d’élèves motivés.

‘There were many motivated students.’

b. Il y en avait beaucoup, des élèves qui étaient motivés.

‘There were many of them, students who were motivated.’

However, the il y a… qui sentence in (131) can also be declefted, as shown in (133). Contrary to the underspecified examples discussed in section 3.4, there does not seem to be an aspectual difference between the sentences in (131), (132) and (133).

(133) Decleftability test Beaucoup d’élèves étaient motivés.

‘Many students were motivated.’

81 Moreover, this sentence expresses the fact that there are many instances of the type ‘motivated students’, and hence, the relative clause contributes to the ‘type specification’ of the antecedent (Davidse 2000:1112). According to Davidse (2000:1112), this is characteristic of RRCs.

94 Chapter 3 | Application of criteria to the corpus data

The results of the (negative and positive) diagnostics are thus contradictory: it is not clear whether such instances should be regarded as clefts.

Another example is given in (134): this il y a… qui sentence can be declefted (135a), indicating it is a cleft, but several tests that identify RRCs also yield positive results: e.g. an adjectival equivalent (135b) and pronominalisation (135c).82

(134) Antoine, vingt-cinq ans, chômeur au Mans, n'était pas dans le mouvement il y a trois semaines. « Je regardais la télé. Je me suis dit qu'il fallait aller voir. Dans ma cité, je sens qu'il y a plein de gens qui seraient prêts à nous rejoindre. (…) » “Antoine, 25 years old, unemployed and living in Mans, was not at the gathering three weeks ago. “I watched TV. I told myself I should see it. In my neighborhood, I feel that there are lots of people who would be willing to join us. (…)”’ (Le Monde)

(135) a. … je sens que plein de gens seraient prêts à nous rejoindre. ‘… I feel that lots of people would be willing to join us.’

b. … je sens qu’il y a plein de gens prêts à nous rejoindre.

‘… I feel that there are lots of people willing to join us.’

c. … je sens qu’il y en a plein, des gens qui seraient prêts à nous rejoindre.

‘… I feel that there are lots of them, people who’d be willing to join us.’

The same holds for the il y a… qui sentence in (136), which can be declefted (137a), but also passes the prepositional reformulation test (137b) and pronominalisation + dislocation (137c).

(136) À la halte suivante, un foyer d'urgence de vingt places qui reçoit, justement, beaucoup de ces jeunes en errance, et plus tard encore dans la journée, l'abbé Pierre reviendra sur le sujet, sa marotte, son obsession : il faut construire. « Quand, dans un pays, il y a plusieurs millions de familles qui n'ont pas à se loger, il n'y a pas trente-six solutions, dit-il, il faut bâtir ce qui manque ! »

82 Notice that the clefted element is again quantified (many students, many people), which will turn out to be an important factor, although I will show that not all decleftable il y a… qui sentences with RRCs contain quantified NPs (see Chapter 6).

Chapter 3 | Application of criteria to the corpus data 95

‘At the next stop, an emergency shelter with twenty places receives many wandering youngsters, and later in the day, l’abbé Pierre gets back to the topic, his pet subject, his obsession: we have to build. “When, in a country, there are several million families that don’t have a home, there aren’t thirty-six solutions, he says, we have to build what’s lacking!” (Le Monde)

(137) a. … plusieurs millions de familles n’ont pas à se loger, …. ‘… several millions of families don’t have a home.’

b. … Il y a plusieurs millions de familles sans abri, …

‘… there are several millions of homeless families.’

c. … il y en a plusieurs millions, des familles qui n’ont pas à se loger, …

‘… there are several millions of them, families that don’t have a home.’

To sum up, sentences such as (131), (134) and (136) defy the traditional views on il y a clefts, because they contain a RRC, but can simultaneously be declefted without a change in proposition.

According to Lambrecht (2002:189-192), il y a... qui sentences introducing quantified indefinite NPs such as beaucoup d’Américains ‘many Americans’ in (138a) are a subtype of cleft, because they have a declefted equivalent and a pragmatic motivation:

“Though grammatically well-formed, (…) canonical sentences [such as (138b)] are unnatural in spontaneous speech because they violate the generalized constraint against non-topical items in preverbal position (indefinite quantified NPs cannot be topics). The use of the bi-clausal cleft structure in [(138a)] is motivated by the need to prevent the indefinite NPs from occurring in initial subject position.” (Lambrecht 2002:191)83

83 See also Furukawa (1996a:71-85), who argues that the relative clause in il y a... qui sentences with quantified constituents such as (i) is restrictive, and hence not of the same type as the relative clauses found in il y a clefts, which he analyzes as predicative rather than restrictive. (i) Il y a beaucoup d’américains qui aiment l’opéra. ‘There are many Americans who like opera.’ (Furukawa 1996a:71, my translation)

96 Chapter 3 | Application of criteria to the corpus data

(138) a. Y a beaucoup d’Américains qui approuvent la violence. ‘There are many Americans who approve of violence.’

b. Beaucoup d’Américains approuvent la violence.

‘Many Americans approve of violence.’

(Lambrecht 2002:189-190)

However, Lambrecht (2002) doesn’t say anything about the fact that the relative clause seems to be restrictive, and he does not apply any semantic or syntactic diagnostics apart from the decleftability test. He apparently takes the decleftability test and the presence of a pragmatic motivation to be sufficient for considering (138a) as a cleft. This contradicts the fundamental assumption that il y a clefts cannot contain RRCs (Lambrecht 1986, 1988a, 2001; Choi-Jonin & Lagae 2005). In other words, the question is what status the two criteria (decleftability and type of relative clause) have: is one of them more important than the other? I will elaborate on this issue in the next section (3.6).

3.6 Possible ways of defining il y a clefts

Confronted with the types of il y a… qui sentences discussed in the previous section, several definitions of il y a clefts are possible. Consider first the two definitions below:

A. Broad definition

i. Decleftability is a necessary and a sufficient criterion for clefthood.

ii. The type of relative clause is not a necessary criterion: it can be restrictive.84

iii. Hence, the utterance is a cleft if it can be declefted, regardless of whether the relative clause is restrictive or not.

84 I found no instances of il y a… qui sentences with appositive relative clauses that can be declefted without a change in meaning. Therefore, both definitions A and B exclude il y a… qui sentences with appositive relative clauses as instances of clefts, since both definitions require that the sentence can be declefted.

Chapter 3 | Application of criteria to the corpus data 97

B. Narrow definition

i. Decleftability is a necessary but not a sufficient criterion for clefthood.

ii. The type of relative clause is a necessary but not a sufficient criterion: it must not be restrictive.

iii. Hence, the utterance is a cleft only in case it can be declefted AND the relative clause is not restrictive.

I believe it is not useful to state that one or the other of these two definitions is ‘right’ or ‘wrong’. They simply delimit slightly different phenomena. There is a large amount of overlap,85 because both definitions identify il y a... qui sentences such as (139) as a cleft: according to definition A, this sentence is a cleft because it can be declefted (140). According to definition B, this sentence is a cleft, because it fulfills the two necessary criteria: (i) the sentence can be declefted, and (ii) the relative clause is not a RRC (141).

(139) Il y a les chats qui miaulent. ‘The cats are meowing’ (Lit. ‘There are the cats that are meowing.’)

(140) Les chats miaulent. ‘The cats are meowing’

(141) # Il y en a, les chats qui miaulent # ‘There are some, the cats that meow.’

Furthermore, according to both definitions, (142) is a cleft lookalike: it cannot be declefted and contains a RRC (see section 3.3.1).

(142) « Face au vent du changement, a-t-il conclu, il y a ceux qui construisent des paravents, et il y a ceux qui construisent des moulins. » ‘”Confronted with change, he concluded, there are those who build windbreaks, and there are those who build windmills.”’ (Le Monde)

85 Recall that the diagnostics allow to classify 60% of the il y a… qui sentences as either clefts or cleft lookalikes, whereas 40% remain unclear (see section 3.1, Table 9).

98 Chapter 3 | Application of criteria to the corpus data

However, while definition A does qualify (143) as a cleft, because the sentence can be declefted (see (144)), definition B excludes sentences such as (143), because although they can be declefted, they contain a RRC (see tests in (145)).

(143) y avait beaucoup d'élèves qui étaient motivés mais y en avait aussi une bonne partie qui n'en avait rien à faire ‘There were many students who were motivated but there was also a large group that didn’t care.’ (CFPP)

(144) Decleftability test Beaucoup d’élèves étaient motivés.

‘Many students were motivated.’

(145) RRC tests a. Il y avait beaucoup d’élèves motivés.

‘There were many motivated students.’

b. Il y en avait beaucoup, des élèves qui étaient motivés.

‘There were many of them, students who were motivated.’

Yet another way of looking at il y a clefts is to see them as a gradient/prototype category (see e.g. Rosch 1978; Taylor 1995; Geeraerts 2006 about Prototype theory).

C. Prototype definition:

i. The most central instances of il y a clefts are characterized by the following three properties: (i) the sentence can be declefted, (ii) the relative clause is not restrictive, (iii) the utterance has either eventive or specificational (not stative, non-specificational) semantics.

ii. More peripheral instances of il y a clefts display one or two but not all three of these properties. For instance, if a sentence can be declefted, but contains a RRC (e.g. (143)), this il y a… qui sentence is a more peripheral instance of an il y a cleft.86

86 Given that eventive il y a clefts are more frequent than specificational ones, one could also argue that the eventive il y a clefts are more central instances than the specificational ones.

Chapter 3 | Application of criteria to the corpus data 99

This view on the il y a cleft sentence type may be a promising way to account for the data, especially given the intuition that some il y a… qui sentences can more easily be identified as clefts than others (see discussion in previous section). This is in line with one of the key properties often associated with prototype categories, namely the fact that there can be degrees of membership representativity (see e.g. Geeraerts 2006), or as Taylor (1995) puts it, “[t]he existence of intuitively clear cases alongside a number of not-so-clear cases strongly suggests that we are dealing with a prototype category.”(Taylor 1995:176). Moreover, seeing a particular phenomenon as a prototype also allows to attribute a more central role to certain criteria while seeing other criteria as more peripheral.

Interestingly, in a short remark, Lambrecht (1988a) also seems to hint at a gradient view of il y a clefts in a short remark: “(…) there is a logical link between the existentials and the presentational function, and (…) I prefer to see the two as endpoints on a continuum rather than as two discrete functions.” (Lambrecht 1988a:149) However, in all his other papers on il y a clefts, he assumes a strict distinction between clefts and non-clefts with restrictive relative clauses.

One complicating factor for a prototype approach to il y a cleft is the following. The well-known examples of prototype categories ‘bird’ and ‘red’ (e.g. Geeraerts 2006) correspond to cognitive concepts that are ‘tangible’ independently of any linguistic theory. Il y a clefts, on the other hand, are a syntactic structure that naïve speakers of French cannot identify without receiving any linguistic instructions. Therefore, I do not see how one can determine in a completely objective way which sentence should be seen as the most central exemplar of the prototype category, independently of previous observations about the properties of il y a clefts. However, this problem also occurs with the other definitions mentioned above (definitions A and B): whichever definition one assumes, it will be but one of several possible views to look at the phenomenon in question.

I will not discuss prototypes in more detail here, but it does seem possible to use a prototype account in order to describe the il y a cleft data.

100 Chapter 3 | Application of criteria to the corpus data

3.7 My definition of il y a clefts

I believe the safest – but admittedly also the least audacious – solution is to argue that there is no single definition of il y a clefts that is ideal or better than any other definitions in all respects. Instead, in order to avoid confusion, any analysis of il y a clefts should ideally state explicitly which criteria are used and in which way.

In the remainder of this dissertation, I will use the following terms, corresponding to the overview in Figure 4. Il y a… qui sentences are divided into two groups according to whether the sentence can be declefted, i.e. whether it expresses the same proposition as a canonical SV equivalent. If the il y a… qui sentence cannot be declefted, it is a cleft lookalike. If the il y a… qui sentence can be declefted and does not contain a restrictive relative clause, it is an il y a cleft. If the il y a… qui sentence can be declefted and contains a RRC, I refer to this sentence periphrastically as a ‘decleftable il y a… qui sentence with a RRC’ for the time being. In Chapter 6, I will explain in more detail which properties characterize this class of il y a… qui sentences and what allows for them to be declefted.87

All il y a… qui sentences

Decleftable Not decleftable

Restrictive relative clause Cleft relative clause (RRC) (not RRC)

“Decleftable il y a… qui “Il y a cleft” “Il y a cleft lookalike” sentence with RRC” (See (143)) (See (139)) (See (142))

Figure 4: Classification of different il y a... qui sentence types

87 It will become clear in Chapter 6 that the category of decleftable il y a… qui sentences does not only include instances with quantified NPs such as the ones discussed in section 3.5.

Chapter 3 | Application of criteria to the corpus data 101

In other words, for the label ‘il y a cleft’, both the decleftability criterion and the type of relative clause criterion (not a RRC) must be met: neither of these criteria is sufficient in and of itself. The motivation for this decision is the following: in any scientific field, it is desirable to use the same definition for the same label, in this case ‘il y a cleft’. Using the same criteria for a given term means that the results of different analyses remain comparable. All previous analyses of il y a clefts seem to exclude il y a… qui sentences that contain RRCs from the class of il y a clefts. If I were to refer to all decleftable il y a... qui sentences as il y a clefts, regardless of the type of relative clause, my hypotheses and findings could not easily be compared to other analyses of (il y a) clefts.

Note that the way I define il y a clefts is also compatible with a prototype approach: the instances that I refer to as ‘decleftable il y a… qui sentences with RRC’ could be seen as more peripheral instances of il y a clefts, whereas the sentences I label ‘il y a clefts’ are the central instances of the il y a cleft prototype category.

The largest part of the rest of this dissertation will focus on the instances defined as ‘il y a clefts’ in the manner described above, i.e. the 576 instances that I classified as clefts in section 3.1 (Table 9) and which are decleftable and contain a relative clause that is not restrictive. The only exception is Chapter 6, which considers the overarching group of ‘decleftable il y a... qui sentences’. This is because Chapter 6 aims to answer one of the main research questions raised in the General Introduction, which can only be addressed by analyzing all decleftable il y a… qui sentences (including those with a RRC): Why do people use extra structure (il y a… qui) when they could have expressed the same proposition in a ‘simpler’ way instead, i.e. by means of a sentence with canonical word order (i.e. the ‘declefted variant’)?

------

Part I: Conclusion

The main goal of this part of the thesis was to delimitate the phenomenon of il y a clefts in a bottom-up way, i.e. by analyzing all il y a… qui sentences without dismissing instances that at first sight seem unclear. In other words, rather than imposing a given definition of il y a clefts on the data, I disentangled the different criteria that have been proposed in previous studies and determined to what extent those criteria allow to delimit a uniform class of sentences.

In Chapter 1, I gave an overview of and commented on the main criteria for the il y a cleft definition that have been argued for in previous analyses. Chapter 2 presented the corpora I used and the extraction of the data. In Chapter 3, I confronted the diagnostics for il y a clefthood from previous studies with my corpus data in order to evaluate these diagnostics and to distinguish between il y a clefts and their lookalikes. In about two thirds of the extracted il y a… qui sentences, the application of those criteria leads to clear results, either indicating that the utterance is an il y a cleft (28%) or a cleft lookalike (32%). However, in 40% of the corpus data, it is not immediately clear whether the il y a… qui sentence should be regarded as an il y a cleft according to previous analyses. The two main sources for this confusion are:

i. The difficulty to establish the type of relative clauses (which is also reported as problematic when dealing with corpus data of il y a clefts in Jullien 2014 and is in line with classification problems of relative clauses in general).

ii. The fact that certain sentences correspond to an SV equivalent expressing the same content (i.e. they can be declefted, indicating they are clefts), but also contain a RRC (indicating they are not clefts, according to previous studies).

104 Part I | Conclusion

With respect to the unclear cases of type (i), I posited two hypotheses about the way in which speakers/hearers process ambiguous relative clauses: the type of relative clause either remains underspecified or is by default restrictive. Moreover, it may be the case that different individuals assign different analyses to the same utterance, and hence, that there is not a single ‘true’ analysis for each and every utterance.

As for the unclear cases of type (ii), I argued that there are several possible definitions of il y a clefts, depending on whether one sees decleftability as a sufficient criterion or not. I see no principled grounds for preferring one definition to the other. However, in this dissertation, I will take decleftability as a necessary, but not as a sufficient criterion, in order to increase the comparability of the research outcomes with respect to other studies of il y a clefts. In other words, I define il y a clefts as follows: they necessarily express the same proposition as a canonical SV equivalent (they can be declefted), AND their relative clause must not be restrictive (or appositive). Moreover, I argued that my account is compatible with a prototype approach, according to which there are central and more peripheral instances of the prototype category of il y a clefts.

Now that I have determined which il y a… qui sentences will be regarded as instances of il y a clefts, I will analyze the distribution, semantic and morphosyntactic properties of those clefts in the next part of the dissertation (Part II). The (discourse) functions of il y a clefts will be analyzed in Part III, and Part IV is dedicated to the Italian counterpart of the il y a clefts: the c’è cleft. Lastly, in Part V, I address the issue of the compositional/non-compositional nature of il y a clefts and their status as a ‘construction’.

Part II

Distribution, semantic and morphosyntactic properties

------

Part II: Introduction

In the previous Part of the dissertation, I explained how I identified the il y a… qui sentences that I consider instances of il y a clefts. In what follows, I present an analysis of the distribution and main semantic and morphosyntactic properties of il y a clefts of the eventive (146) and the specificational (147) subtype (Chapter 4).

(146) Eventive il y a cleft dans la rue oui j'entends qu'ils parlent pas français hein d'accord + donc en allant te promener (…) + ou quand y a Marta qui vient avec son mari elle parle roumain

‘in the street yes I hear that they don’t speak French huh right + so when you go for a walk (…) or when there’s Marta who comes over with her husband she speaks Romanian.’ (CFPP)

(147) Specificational il y a cleft – A: Pouvez-vous me conseiller un shampoing, un savon et gel douche bio?

‘Which organic shampoo, soap and shower gel would you recommend?’

– B: il ya douce nature qui est pas mal au niveau rapport qualité prix

‘There’s Douce Nature that’s not bad in terms of price/quality.’ (YCCQA)

Chapter 4 is followed by a presentation of the distribution and main semantic and morphosyntactic properties of il n’y a que clefts (148) (Chapter 5).

(148) Il n’y a que cleft y'a que toi qui comprend ou c'est moi qui pige rien à rien?

‘Are you the only one who understands or is it me who doesn’t understand a thing?’ (Lit. ‘Is there only you who understands…’) (YCCQA)

------

Chapter 4

Il y a clefts: two semantic types (eventive vs. specificational)

As explained in the General Introduction (section 3.2), il y a clefts can be divided into two semantic subtypes. Eventive il y a clefts such as (149) express an event (a worker crying), whereas specificational il y a clefts such as (150) specify a value (Lost) for a variable (good TV shows).

(149) Eventive il y a cleft Le matin, on arrangeait ces cadavres avec de la chaux. Un peu de terre, un peu de chaux, jusque quand c'était plein. Un jour, il y a un ouvrier qui a crié : « Mais ça, c'est ma fille! C'est ma fille! »

‘In the morning, we arranged those corpses with chalk. A bit of soil, a bit of chalk, until it was full. One day, a worker cried: “But that’s my daughter! It’s my daughter!”’ (Le Monde)

(150) Specificational il y a cleft – A: Quelles est votre meilleure série du moment?

‘What’s your favorite TV show right now?’

– B: "How I Met Your Mother" c'est génial, y'a aussi "Lost" qui est bien.

‘”How I Met Your Mother” is great, there’s also “Lost” that is good.’

(YCCQA) 110 Chapter 4 | Two types of il y a clefts

In what follows, I show that eventive and specificational il y a clefts display considerable differences when it comes to (i) their distribution throughout the corpora (section 4.1) and the type of constituent they introduce (section 4.2).88

4.1 Distribution

In this section, I present the distribution of all il y a clefts throughout the corpora (section 4.1.1), followed by a presentation of the frequency of eventive vs. specificational il y a clefts in the corpora (section 4.1.2).

4.1.1 Il y a clefts in general

Recall that it has been claimed that il y a clefts are a spoken rather than a written phenomenon (e.g. Lambrecht 1986; Blanche-Benveniste 1997:93; Katz 2000a; Choi- Jonin & Lagae 2005; Dufter 2008; Willems & Meulleman 2010). My corpus data empirically corroborate this claim. Table 10 shows that il y a clefts are 95.5 times more frequent in CFPP than in Le Monde (191/2=95.5).

Le Monde YCCQA CFPP Total

Corpus size in words 25700000 6100000 550.000 32.350.000

Total number of il y a clefts 81 285 210 576 Number of il y a clefts 2 23 191 9 per 500.000 words Table 10: Normalized frequency of il y a clefts

Interestingly, the fact that il y a clefts are more typical of spoken French than of written French is also reflected within the Le Monde corpus. I found that 83% of the il y a clefts in this corpus appear within a citation, while only 17% are not presented as ‘spoken’ French by the author of the newspaper article (Table 11).

88 Several other properties of il y a clefts (e.g. the distribution of subject/object relative clauses) can be found in the Appendix (Section 7). Furthermore, all properties reported in this chapter for il y a clefts have also been analyzed for the broader category of il y a… qui sentences (Appendix, Section 4).

Chapter 4 | Two types of il y a clefts 111

Le Monde

n % Occurs in citation 67 83% Does not occur in citation 14 17% Total 67 83%

Table 11: Il y a clefts appearing within a citation or not (Le Monde)

This is not to say that these cases should be seen as occurrences of ‘real’ spoken French, since journalists do not usually quote people verbatim. It is more likely that the writer uses the il y a cleft in order to mimic phenomena that are characteristic of real speech.89

Another factor that might influence the frequency of il y a clefts in the different registers has to do with the “epilinguistic level” that Deulofeu (2000:273)90 discusses in his analysis of the defining characteristics of genres: it may be the case that the il y a cleft is regarded as an “informal” use of language, which is therefore to be avoided by writers in more formal language, i.e. the journalistic register in the case of Le Monde.

I will return to possible underlying reasons for the difference in frequency of il y a clefts in spoken vs. written French in Chapter 6.

4.1.2 Eventive vs. specificational il y a clefts

After having looked at the distribution of all il y a clefts, let us consider the respective frequencies of eventive vs. specificational il y a clefts. As is evident from Table 12, eventive il y a clefts outnumber specificational il y a clefts in all three corpora, with an average of 84% eventive ones against 16% specificational ones. In this sense, it is not surprising that in the literature on il y a clefts, the eventive subtype has received much more attention than the specificational subtype.

89 See Burr (1998) and Dardano (1987) on the ‘stylization of spoken language’ in journalistic texts. 90 Deulofeu (2000:273) speaks of the respect of normative templates (le respect de tournures normatives) and the adoption of professional codes of language use (l’adoption de codes professionnels d’utilisation du langage).

112 Chapter 4 | Two types of il y a clefts

Le Monde YCCQA CFPP Total

n % n % n % n % Eventive 79 98% 226 79% 177 84% 482 84% il y a cleft Specificational 2 2% 59 21% 33 16% 94 16% il y a cleft Total 81 100% 285 100% 210 100% 576 100%

Table 12: Distribution of eventive vs. specificational il y a clefts

However, the number of specificational il y a clefts is also not negligible, especially in the YCCQA corpus, and to a lesser degree in the CFPP corpus. It is plausible that there is an influence of the ‘interactive setup’ of the corpora here: specificational il y a clefts are very often used in answer to a question/request, as in (151) (more examples of specificational il y a clefts used to answer a question are given in Table 14 and Table 16).

(151) – A: Cherche logiciel gratuit de création de musique ‘Looking for software to create music’

– B: il ya Audacity qui est totalement gratuit.

‘There’s Audacity that is completely for free.’ (YCCQA)

The journalistic corpus Le Monde only rarely contains question-answer pairs, whereas in the other two corpora questions are frequently asked and answered (by the interviewer/interviewee in the spoken corpus and by the visitors of the Yahoo online forum). As a consequence, specificational il y a clefts are more frequent in CFPP and YCCQA than in Le Monde. This can be seen as a confirmation of the importance of investigating not only the opposition between spoken and written language, but also different types of corpora within spoken/written French, since some grammatical forms may be much more frequent in certain situations than in others (Cappeau 2008:75).

Chapter 4 | Two types of il y a clefts 113

4.2 Type of clefted constituent

Apart from their relative frequencies, eventive and specificational il y a clefts also show clear differences with respect to the clefted constituent, and more specifically, (i) the number of proper nouns they introduce and, relatedly, (ii) the distinction between definite and indefinite constituents.

4.2.1 Grammatical category & frequency of proper nouns

Many examples in the literature on eventive il y a clefts contain proper nouns such as Jean (e.g. Il y a Jean qui a téléphoné ‘Jean called’, Lambrecht 1988a:136). However, it turns out that only 5% of all eventive il y a clefts introduce proper nouns.91 The number of proper nouns in specificational il y a clefts, on the other hand, is higher: 31% (see Table 13).

Specificational Eventive clefts Total clefts n % n % n % Noun phrase 419 87% 63 67% 482 84% Proper noun 22 5% 29 31% 51 9% Pronoun 41 9% 2 2% 43 7% Total 482 100% 94 100% 576 100%

Table 13: Grammatical category of the clefted constituent (il y a clefts)

The fact that proper nouns are more frequent in specificational il y a clefts than in eventive il y a clefts may have to do with the types of interactions in the corpora. In the typical examples used by Lambrecht and others (such as Il y a Jean qui a téléphoné ‘Jean called’), the use of Jean presupposes that both interlocutors know the person

91 The reason why proper nouns are relatively frequent in made-up examples of il y a clefts (despite their relative infrequency in the corpus data) is likely due to the fact that proper nouns cannot be the antecedent of a restrictive relative clause. Hence, sentences such as Y’a Jean qui a téléphoné are indisputably clear examples of clefts, despite the fact that il y a clefts are much more often used to introduce indefinite NPs (see next section, Table 15) rather than proper nouns. See also chapter 6.5.

114 Chapter 4 | Two types of il y a clefts

Jean refers to before the sentence is uttered.92 However, in all three corpora, the speaker/writer and his/her interlocutor share little or no knowledge. Therefore, if the speaker/writer wishes to indicate, for instance, that her sister called Sarah is knocking on the door, she will be more likely to use the NP ‘my sister’ than the proper noun ‘Sarah’.93

As for the higher number number of proper nouns in specificational il y a clefts, this is likely due to the types of questions that those clefts answer in the YCCQA and CFPP corpora. In these corpora, specificational il y a clefts often appear as answers to questions that inquire about products or restaurants, for instance, and which therefore prompt an answer with a proper noun that specifies a value for those variables. A few of these questions are enumerated in Table 14. For the sake of brevity, I only present the proper noun that functions as clefted constituent in the specificational il y a cleft. The whole cleft can be found in the Appendix (Section 7.4).

The clefted element Question in the answer

Est-ce qu'il y a des GPS a reconnaissance vocal disponible au Québec ? le Magellan Maestro ‘Are there GPS systems with speech recognition in Quebec?’ 3250

Pouvez-vous me conseiller un shampoing, un savon et gel douche bio ? Douce Nature ‘Which organic shampoo, soap and shower gel would you recommend?’

Je cherche une bonne crème de jour hydratante pour le visage. Uriage ‘I’m looking for a good hydrating facial cream.’

Cherche logiciel gratuit de création de musique Audacity ‘Looking for software to create music’

C'est quoi le mieux, firefox ou internet explorer ? Google Chrome ‘Which one is better, Firefox or Internet Explorer?’

Quelles est votre meilleure série du moment ? Lost ‘What’s your favorite TV show at the moment?’

Table 14: Questions leading to specificational il y a clefts with proper nouns

92 This is not to say that Jean is discourse-given when the cleft is uttered. 93 One could also speculate that in general, people speak more about events involving several people (encoded by a plural NP) rather than a single person (encoded by a proper noun.

Chapter 4 | Two types of il y a clefts 115

4.2.2 (In)definiteness of the clefted constituent

Another clear difference between eventive and specificational il y a clefts, which is related to the frequency of proper nouns, is the proportion of indefinite vs. definite constituents. As shown in Table 15, eventive il y a clefts almost always introduce indefinite constituents (89%), whereas only 41% of specificational il y a clefts contain indefinite clefted constituents. This is also in line with the difference in frequency of proper nouns in eventive and specificational il y a clefts (Table 13).

Specificational Eventive clefts Total clefts

n % n % n %

Indefinite 427 89% 39 41% 466 81%

Definite 55 11% 55 59% 110 19%

Total 482 100% 94 100% 576 100%

Table 15: (In)definiteness of the clefted constituent in il y a clefts

Again, I argue that the function of eventive and specificational sentences has an influence. Due to the prior context, many specificational il y a clefts present a specific value encoded by a proper noun (see examples in Table 14) or by a definite NP (see examples in Table 16, full examples can be found in the Appendix, Section 7.4).

The clefted element Question in the answer Je vais bientôt passer un examen, et j'ai le trac. pouvais vous me donner des conseils ? l'homéopathie ‘I’ve got exams coming up and I’m really nervous. Can you give me ‘homeopathy’ some advice?’ leur apparente Aimez-vous les chats ?? Si oui, pourquoi ? indépendance ‘Do you like cats?? If so, why?’ ‘their apparent independence

116 Chapter 4 | Two types of il y a clefts

J'ai des cheveux qui régressent trop vite, je suis obligée de les laver tous les jours avez vous des conseils ? les shampoings secs ‘My hair becomes greasy very quickly, I have to wash it every day, do ‘dry shampoos’ you have any advice?’ Comment on fait pour trouver l'amour ? je suis tout seul depuis la nouvelle longtmps je comprends pas, je m'intéresse aux gens mais ça donne rien technologie ‘How do people find love? I’ve been all alone for a long time, I don’t ‘new technology’ understand, I’m interested in people but it doesn’t work’ Pourqoui kurt cobain s'est t'il suicidé ? la drogue ‘Why did Kurt Cobain kill himself?’ ‘drugs’

Table 16: Questions leading to specificational il y a clefts with definite NPs

Furthermore, if we zoom in on the indefinite constituents introduced by specificational il y a clefts, we find that more than half of them are quantified expressions, as shown in Table 17. In contrast, most eventive il y a clefts contain indefinite constituents such as des gens ‘people’ and des étudiants ‘students’. The indefinite constituents in eventive il y a clefts are not as often quantified (21%)94 as indefinites in specificational il y a clefts (54%) (see Table 17).

Specificational Eventive clefts Total clefts

n % n % n % Quantified 89 21% 21 54% 110 24%

Non-quantified 338 79% 18 46% 356 76%

Total 427 100% 39 100% 466 100%

Table 17: Quantified vs. non-quantified indefinite constituents in il y a clefts

94 An example of an eventive il y a clefts with a quantified NP is given in (i): (i) Ce que je vois par la fenêtre de mon salon est 2 enfants qui jouent avec leur chiot, (…) y a aussi une dizaine d'enfants âgés d'environ 13 ans qui jouent au football. ‘What I see from my living room window is 2 children playing with their puppy, (…) there’s also a dozen or so children of about 13 years old who are playing soccer.’ (YCCQA)

Chapter 4 | Two types of il y a clefts 117

It is worth zooming in the relatively high number of quantified constituents in specificational il y a clefts. 95 These clefts can in fact be analyzed as the French equivalents of the so-called “quantifying” there cleft, which is a subtype of specificational there clefts in English.96 While “enumerative” specificational there clefts present an instance corresponding to a variable, “quantifying” specificational there clefts specify a quantity of instantiation of the variable. For instance, the variable ‘water fountains that are turned off’ is presented in the context in (152) and picked up by the cleft relative clause that do not work. The there cleft quantifies this variable: ‘at least three’. This type of there cleft is on a par with specificational it clefts of the quantifying type such as (153).

(152) Specificational there cleft of the “quantifying” type Does anyone know why all of the water fountains are turned off, there's at least three that do not work. (Njende 2016:29)

(153) Specificational it cleft of the “quantifying” type He can inform you as to the decrease in the number of airlines out here. It's at least three that are not here. (Njende 2016:5)

This analysis also applies to specificational il y a clefts in French: the il y a clefts referred to in Table 14 and Table 16 are of the “enumerative” type, providing an instance of a variable (e.g. variable: good facial creams, instance: Uriage). Specificational il y a clefts such as (154) and (155), on the other hand, do not specify an instance, but a quantity: in (154), the variable is ‘nice girls who…’, and the quantity is ‘about 999.999’, and in (155), the variable is ‘organizing countries that won’, for which the cleft specifies the quantity ‘six’.

95 An example of a specificational il y a cleft with a non-quantified indefinite is given in (i): (i) A: Enceinte, plaquee et suicidaire? ‘Pregnant, abandoned and suicidal?’ B: Je te conseille de te faire aider. (…). Vas voir un psychiatre, il y a aussi des numéros d'urgence qui existent. ‘I advise you to look for help. Go see a psychiatrist, there are also emergency numbers that exist.’ (YCCQA) 96 See Huddleston (1971); Davidse (1999, 2000, 2014); Davidse & Kimps (2016) and Njende (2016) about this subtype of there cleft.

118 Chapter 4 | Two types of il y a clefts

(154) Specificational il y a cleft of the “quantifying” type – A: Y a t'il 1 fille sympa qui cherche 1 mec pas chiant ?

‘Is there a/one nice girl who’s looking for a guy who’s not a pain in the ass?’

– B: Il y a environ 999'999 de filles qui cherche un mec pas chiant

‘There are about 999.999 girls who are looking for a guy who’s not a pain in the ass.’ (YCCQA)

(155) Specificational il y a cleft of the “quantifying” type – A: Combien de pays organisante de la coup du monde gagne cette coup ?

‘How many countries that organized the World Cup won that cup?’

– B: Il y a eu 6 pays organisateurs qui ont gagné le Mondial (Uruguay 1930, Italie 1934, Angleterre1966, Allemagne 1974, Argentine 1978 et France 98).

‘There have been 6 organizing countries that won the World Cup (Uruguay 1930, 1934, Great-Britain 1966, Germany 1974, Argentina 1978 and France 1998)’ (YCCQA)

To conclude, eventive and specificational il y a clefts display clear differences when it comes to (i) the frequency of proper nouns and (ii) the indefiniteness and quantification of the clefted constituent. I explained these differences in terms of the discourse functions of the two types of il y a clefts and the influence of the preceding discourse context. Furthermore, I provided evidence in support of the distinction between “enumerative” and “quantifying” specificational there clefts also holds for French il y a clefts.

------

Chapter 5 Il n’y a que clefts

In the previous chapter, I presented the main semantic and morphosyntactic properties of the il y a clefts in my dataset. As explained in chapter 2.3, I used a separate query to extract il y a… qui sentences that contain the exclusive particle ne… que ‘only’, which will henceforth be referred to as il n’y a que… qui sentences. If the sentence is a cleft, as in (156), it is labeled an il n’y a que cleft.

(156) Il n’y a que cleft y'a que toi qui comprend ou c'est moi qui pige rien à rien?

‘Are you the only one who understands or is it me who doesn’t understand a thing?’ (Lit. ‘Is there only you who understands…’) (YCCQA)

In what follows, I present the most relevant findings of the analysis of the il n’y a que clefts that were extracted from Le Monde and YCCQA. Section 5.1 comments on the distinction between il n’y a que clefts and their lookalikes and section 5.2 presents the distribution of il n’y a que clefts in the two corpora. The main semantic and morphosyntactic properties of the clefted constituent are discussed in section 5.3.97

97 The information structure of il y a clefts and il n’y a que clefts is presented in Chapter 7. 120 Chapter 5 | Il n’y a que clefts

5.1 Il n’y a que: clefts vs. cleft lookalikes

When it comes to the distinction between clefts and cleft lookalikes, il y a... qui sentences with the ne… que ‘only’ particle behave very differently from il y a... qui sentences without this particle. Firstly, the distinction between types of relative clauses (restrictive, appositive, other) is much clearer (I found no ambiguous cases). Secondly, the corpus data contain only very few restrictive and appositive relative clauses (Table 18), despite the fact that the query did not impose any restrictions on what follows the word sequence il n’y a que (see chapter 2.3).

Le Monde YCCQA Total

n % n % n %

Cleft 39 95% 173 97% 212 96%

Cleft lookalike 2 5% 6 3% 8 4%

Total 41 100% 179 100% 220 100%

Table 18: Il n’y a que clefts vs. cleft lookalikes with qui/que clauses

Of the 8 cleft lookalikes, 3 cases contain a relative clause that is clearly appositive (157). In the other cases, the relative clause is restrictive: it forms a single constituent with its antecedent, whose reference it restricts. This is particularly clear in cases such as (158) and (159), where the sequence ‘noun + RRC’ is contrasted with another constituent (étrangers ‘strangers’ in (158), femmes laides ‘ugly women’ in (159)).

(157) Il n’y a que + NP + appositive relative clause (≠ cleft) Lorsque Claude Vinci, déserteur, raconte l'attaque d'un dohar au lance-flammes par l'unité de l'armée française à laquelle il appartenait (il n'y avait que des femmes et des enfants, qui ont brûlé sous mes yeux), on comprend que la télévision avance ici sur un terrain excessivement sensible.

‘When Claude Vinci, deserter, speaks about the attack of a dohar by means of a flamethrower by the French army unit that he belonged to (there were only women and children, who burned before my eyes), one understands that television is walking on extremely sensitive ground here.’ (Le Monde)

Chapter 5 | Il n’y a que clefts 121

(158) Il n’y a que + NP + RRC (≠ cleft) Il n'y a pas d'étrangers sur terre. Il n'y a que des gens que l'on ne connaît pas (encore !).

‘There are no strangers on Earth. There are only people that you don’t know (yet!).’ (YCCQA)

(159) Il n’y a que + NP + RRC (≠ cleft) Il n'y a pas de femmes laides il n'y a que des femmes qui s'enlaidissent !

‘There are no ugly women there are only women who become ugly!’ (YCCQA)

Moreover, none of these sentences can be transformed into a SV counterpart without altering the proposition expressed by the original utterance, as shown in (160). Hence, they do not meet the decleftability criterion, because the meaning changes.

(160) a. # Seuls des femmes et des enfants ont brûlé sous mes yeux. # ‘Only women and children burned before my eyes.’

b. * On ne connaît pas (encore !) que des gens.

* ‘You only don’t know people (yet!).’

c. # Seules des femmes s’enlaidissent.

# ‘Only women become ugly.’

The fact that only very few instances of il n’y a que… qui sentences are not clefts may indicate that the cleft function has become so predominant that speakers avoid using this sequence for a different function (i.e. in combination with a RRC). This would be in contrast with il y a… qui sentences without the ne… que ‘only’ particle, which are used as cleft lookalikes more than two thirds of the time in my dataset (see chapter 3.1, Table 9). However, this hypothesis cannot be tested by means of corpus data alone.98

98 One could, for instance, design a self-paced reading experiment with instances of il n’y a que clefts and il n’y a que cleft lookalikes and measure reaction times. The prediction would be that all other things being equal, il n’y a que cleft lookalikes take more time to process than il n’y a que clefts. This prediction would be based on the idea that it’s easier to process a sentence with a frequently occurring interpretation (cleft interpretation) than the same string of words with an infrequent interpretation (cleft lookalike interpretation).

122 Chapter 5 | Il n’y a que clefts

5.2 Distribution of il n’y a que clefts

As shown in Table 19, il n’y a que clefts are 14 times more frequent in the YCCQA corpus than in Le Monde.

Le Monde YCCQA Total

Corpus size in words 25700000 6100000 31.800.000

Total number of il n'y a que... qui clefts 39 173 212

Number of il n'y a que... qui/que clefts 1 14 3 per 500.000 words

Table 19: Normalized frequency of il n’y a que clefts

This reflects the results found for il y a clefts (11.5 times more frequent, chapter 4.1.1, Table 10). Furthermore, a considerable number of il n’y a que clefts in Le Monde appear within a citation: 54% (Table 20). This is illustrated by (161).

Le Monde

n % Occurs in a citation 21 54% Does not occur in a citation 18 46% Total 39 100% Table 20: Il n’y a que clefts appearing within a citation or not (Le Monde)

(161) « Avant il n'y avait que les Noirs qui me reconnaissaient dans la rue. Aujourd'hui même des Blancs m'arrêtent pour me saluer (…)», affirme-t-il. “Before, there were only black people who recognized me in the streets. Now even white people stop me to say hello (…)”, he stated. (Le Monde)

The percentage of clefts appearing within citations is not as high as the one found for il y a clefts without ne… que ‘only’ (of which 83% appear in citations, see chapter 4.1.1, Table 11). However, given the overall limited number of il n’y a que clefts found in the Le Monde corpus, it would be imprudent to draw hard conclusions from this difference.

Chapter 5 | Il n’y a que clefts 123

5.3 Properties of the constituent introduced by il n’y a que

In what follows, I present (i) the grammatical category and (ii) the (in)definiteness of the constituents introduced by il n’y a que clefts.

5.3.1 Grammatical category

The findings in Table 21 show that most constituents introduced by il n’y a que are NPs (an average of 66%) and pronouns (17%). These categories are followed by prepositional phrases (11%, see (162)), proper nouns (5%), embedded clauses (1%, see (163)) and one case with the adverb là-bas ‘there’ (see (164)).

Le Monde YCCQA Total

n % n % n %

Noun phrase 30 77% 109 63% 139 66%

Pronoun 3 8% 32 18% 35 17%

Prepositional phrase 6 15% 17 10% 23 11%

Proper noun 0 0% 11 6% 11 5% Temporal / 0 0% 3 2% 3 1% conditional clause Adverb 0 0% 1 1% 1 0%

Total 39 100% 173 100% 212 100%

Table 21: Type of constituent behind il n’y a que

(162) Il n’y a que + prepositional phrase Il n'y a que comme cela que le couple peut durer!

‘Only like that can the couple last!’

(Lit. ‘There is only like that that the couple can last!’) (YCCQA)

124 Chapter 5 | Il n’y a que clefts

(163) Il n’y a que + conditional clause y'a que si tu es handicapé que tu ne peux pas franchir les portes

‘Only if you are handicapped are you unable to go through doors.’

(Lit. ‘There is only if you are handicapped that you cannot…’) (YCCQA)

(164) Il n’y a que + adverb y a que làba que tu auras une réponse fiable!

‘Only there will you find a reliable answer!’

(Lit. ‘There is only there that you will find a reliable answer!’) (YCCQA)

Remember that il y a sentences in general are unable to introduce a non-nominal constituent (i.e. a constituent that is not a NP, pronoun or proper noun) unless the particle ne… que ‘only’ is present (see chapter 2.3). Table 22 contrasts nominal constituents and non-nominal constituents, showing that almost 90% of all constituents introduced by il n’y a que are nominal. In other words, although il n’y a que clefts, unlike il y a clefts, are able to introduce non-nominal constituents, they do not do so very frequently.

Le Monde YCCQA Total n % n % n % Nominal (NP, pronoun, proper noun) 33 85% 152 88% 185 87%

Other (prepositional phrase, adverb, 6 15% 21 12% 27 13% temporal/conditional claus) Total 39 100% 173 100% 212 100%

Table 22: Nominal vs. non-nominal constituents behind il n’y a que

Chapter 5 | Il n’y a que clefts 125

5.3.2 (In)definiteness

Zooming in on nominal constituents (i.e. noun phrases, pronouns and proper nouns) introduced by il n’y a que clefts, it turns out that 89% of them are definite, while only 11% are indefinite expressions (see Table 23):

Le Monde YCCQA Total

n % n % n % Definite 17 52% 147 97% 164 89% Indefinite 16 48% 5 3% 21 11% Total 33 100% 152 100% 185 100%

Table 23: (In)definiteness of nominal constituents behind il n’y a que

In this sense, il n’y a que clefts pattern more like specificational il y a clefts (59% definites, 41% indefinites) than like eventive il y a clefts (89% indefinites, 11% definites) (see section 4.2.2, Table 15). This is no surprise, given that most il n’y a que clefts are specificational. As an illustration, consider (165):99 the relative clause ‘that can change things’ refers to the question raised by the first writer, i.e. ‘ways in which I can make my boyfriend be less interested in soccer’. The value that the il n’y a que cleft presents for this variable is ‘only time’.

(165) Specificational il n’y a que sentence – A: Comment faire pour que mon copain s'interresse moins au foot?

‘How can I make my boyfriend become less interested in soccer?’

– B: Mon chéri doit etre pareil que ton copain mais je te dis juste attends il n'y a que le temsp qui pourra changer les choses... Il faut etre patiente....

‘My honey must be like your boyfriend but let me just tell you to wait, only time can change things… You have to be patient.’

(Lit. ‘There is only time that can change things’) (YCCQA)

99 Not all instances of il n’y a que clefts are clearly specificational. I will return to this in 7.3.5.1.

------

Part II: Conclusion

In this Part of the dissertation, I have discussed the distribution of il y a clefts and il n’y a que clefts throughout the corpora, as well as their main semantic and morphosyntactic properties, in order to get a better understanding of the characteristics of such clefts as they are used in authentic language data (as opposed to analyses based on only few or made-up examples). The main findings are the following. As for il y a clefts without the ne… que ‘only’ particle, it turns out that they are 95.5 times more frequent in the spoken corpus CFPP than in the written corpus Le Monde. This difference is even more pronounced if one takes into consideration the fact that 83% of the il y a clefts in Le Monde occur within citations. Thus, the corpus data empirically corroborate the claim in previous studies that il y a clefts are a spoken rather than a written phenomenon. I then turned my attention to the differences and similarities between eventive and specificational il y a clefts. Although specificational il y a clefts have only been mentioned a few times in prior studies, they are not marginal in my corpus data: 16% of all il y a clefts are specificational. The two semantic subtypes of il y a clefts display several clear differences, such as the number of indefinite clefted constituents (89% of eventive clefts, 41% of specificational clefts) and the number of proper nouns (5% of eventive clefts, 31% of specificational clefts). I explained these differences in terms of their discourse function. Furthermore, within the class of specificational il y a clefts, I identified a number of “quantifying” il y a clefts, which are similar to quantifying specificational there clefts.

I then analyzed il n’y a que clefts. The corpus data contain considerably less cleft lookalikes than the il y a… qui dataset without the ne… que ‘only’ particle. This may indicate that the il n’y a que… qui sentence type is specialized to be used as a cleft, contrary to the il y a… qui sentence type. I then showed that almost 90% of all clefted elements are nominal (i.e. NP, pronoun or proper noun), and that the majority of those nominal clefted elements are definite, paralleling specificational il y a clefts without ne… que ‘only’.

Part III

Functions of il y a clefts

“No French speaker could do without it”

(Lambrecht 1988a:153)

------

Part III: Introduction

The main goal of this part of the dissertation is to examine the functions of il y a clefts by means of a corpus analysis, in order to explain what motivates speakers to use such clefts rather than their SV counterpart.

One first line of reasoning is that il y a clefts (and il y a sentences more generally) are used in order to avoid certain types of nominal expressions, such as indefinite NPs, from appearing in preverbal subject position, because those expressions would be ‘bad’ or ‘suboptimal’ preverbal subjects (e.g. Lambrecht 1986; Cappeau & Deulofeu 2001). However, what exactly constitutes ‘suboptimality’ or ‘bad subjects’ in il y a… qui sentences has not yet been analyzed in detail, and it is not yet clear whether this function holds for il y a clefts or for il y a sentences in general. Moreover, the hypothesis that il y a clefts are used to avoid ‘bad subjects’ has not yet been corroborated by means of systematic corpus analysis. Therefore, I will test this hypothesis and present a more precise analysis of the factors that make preverbal (indefinite) subjects ‘optimal’ or ‘suboptimal’ or something in between (Chapter 6).

Besides the function of introducing indefinites, il y a clefts and other cleft types have often been connected to the study of ‘information structure’ (see General Introduction). In this respect, there are two common assumptions in the literature on clefts in general that I will address: (i) all clefts (also c’est clefts etc.) are used to avoid a topic-comment interpretation of the sentence. (ii) Clefts are only motivated by the function of marking focality and/or non-topicality (see also Herslund 2005; Blanche- Benveniste 2006 and Dufter 2009b among others for criticism of this assumption in the study of c’est clefts)

132 Part III | Introduction

The hypothesis that clefts are used to avoid a topic-comment interpretation of the sentence is presented by several authors (e.g. Lambrecht 1988a; 1994:22; 2001; Erteschik-Shir 2007:121). However, this view has also been criticized by researchers who show that clefts introduced by it/c’est are able to express a topic- comment interpretation (e.g. Prince 1978; Doetjes et al. 2004; Gundel 2006; Hedberg 2006; Hedberg & Fadden 2007; Dufter 2008; Hedberg 2013). This raises the question whether the hypothesis does (still) hold for il y a clefts, or whether it should be refuted, as in the case of it/c’est clefts. I will argue that contrary to it/c’est clefts, il y a clefts never introduce regular topics (although they do sporadically introduce marked topics), and that in this sense, the hypothesis according to which il y a clefts are used to avoid a topic-comment interpretation is correct. However, I will also nuance this hypothesis by arguing that the origin of this function does not lie in the cleft format (as argued by Lambrecht 2001 for instance), but stems from the existential expression ‘il y a’ only. Therefore, avoiding a topic-comment interpretation is not a motivation that is unique to il y a clefts, but characterizes the category of ‘il y a sentences’ more generally.

The second common assumption, i.e. that clefts are solely used as markers of focality and/or non-topicality, has also received an increasing amount of criticism. As stated by Dufter (2009b), “many syntactic analyses do not even feel the need to elaborate on the assumption that universally, clefts are special devices employed for highlighting purposes” (Dufter 2009b:85). In contrast, researchers such as Delin & Oberlander (1995) and Dufter (2006, 2009b) show that clefts introduced by c’est/it can also have other underlying motivations, making Delin & Oberlander (2005) conclude that “it is hard to uphold any version of the view that [clefts] are ‘focusing’ constructions” (Delin & Oberlander 2005:5). I will support this line of research by showing that il y a clefts also perform functions that cannot be explained by referring to the notions ‘focus’ and ‘topic’.100

100 In this sense, my findings are in line with Jullien (2014), who shows that il y a clefts and related structures can be used for interactional purposes such as negotiating turn-taking and ‘try-marking’. I will not focus on the role of il y a clefts from an interactional point of view and refer the reader to Jullien’s (2014) dissertation for an in-depth analysis of il y a sentences in this respect.

Part III | Introduction 133

I will also compare the information structure properties of il y a clefts as they appear in the corpus data with what is reported about the information structure of c’est clefts in previous (corpus-based) studies. As it turns out, the two cleft types have the same potential, but the distribution of the different information structure interpretations is very different. I will present different factors that (partially) explain this difference.

The structure of this part of the dissertation is as follows. In Chapter 6, I investigate the hypothesis according to which il y a clefts are used in order to avoid indefinites and other types of nominal expressions from appearing in preverbal subject position. Chapter 7 addresses the assumption that clefts are used to avoid a topic-comment interpretation, by presenting an in-depth analysis of the information structure properties of il y a clefts. In Chapter 8, I compare the information structure properties of c’est clefts and il y a clefts. Chapter 9 provides a description of ‘additional’ pragmatic functions of il y a clefts that are not directly linked to the notion of ‘topic’ and ‘focus’.

------

Chapter 6

The function of avoiding ‘bad’ preverbal subjects

The goal of this chapter is to investigate the first hypothesis about the speaker’s motivations for the use of an il y a cleft. In section 6.1, I will present previous hypotheses about il y a clefts, il y a… qui sentences more generally and existential sentences in the cross-linguistic literature. I then introduce the hypothesis that I will evaluate in this chapter, i.e. that il y a… qui sentences are used in order to avoid ‘bad preverbal subjects’ (section 6.2). There is a considerable amount of literature about the different factors that influence/determine whether a given preverbal subject is ‘good’ or ‘bad’. I present an overview of those factors and their underlying rationale in section 6.3. I then test the hypothesis presented in section 6.2 by analyzing to what extent the constituents introduced by il y a make ‘good’ or ‘bad’ preverbal subjects in the corresponding SV sentence (section 6.4). In order to do this, I systematically determine for all il y a… qui sentences whether or not they can be declefted and how the factors influencing subject acceptability (discussed in section 6.3) do or do not map onto the corpus data. In section 6.5, I argue that although the hypothesis is correct, the corpus data show that it should also be nuanced, because there are different degrees to which expressions are ‘optimal’ or ‘suboptimal’ in preverbal subject position. Those degrees of subject acceptability, I will argue, influence the ease with which a given il y a… qui sentence can be declefted. Moreover, I will argue that the reason why spoken French resorts to il y a… qui sentences more frequently than written French is that the conditions that a preverbal subject must fulfill are 136 Chapter 6 | The function of avoiding ‘bad’ preverbal subjects

stricter in spoken French. In other words, since the threshold for acceptable preverbal subjects is higher in spoken French than in written French (presumably due to a higher need to facilitate online language processing for the interlocutor), speakers use il y a… qui sentences more often in spoken than in written French. I then discuss the consequences of the findings for the validity of the decleftability criterion (section 6.6). This section is followed by a more general conclusion of the chapter.

6.1 Previous studies: motivations of il y a clefts/il y a… qui sentences

As I briefly mentioned in the General Introduction (section 3.2), it has been argued by several authors that il y a clefts are used in order to introduce a discourse-new constituent (Lambrecht 1986; Giry-Schneider 1988; Lambrecht 1988a, 1994; Ashby 1995, 1999; Lambrecht 2000a, 2000b, 2001, 2002; Jullien 2007; Pitavy 2009; Secova 2010). This hypothesis, which is defended most explicitly by Lambrecht, is based on the idea that there is a general preference for presenting discourse-given information before discourse-new information, because this is easier to process (e.g. Clark & Clark 1977; Gundel 1988). In other words, il y a clefts such as (166a) are supposedly used to avoid discourse-new referents from appearing in preverbal subject position, as in (166b), the latter being more difficult to process.101

(166) a. Y’a Jean qui s’est cassé la jambe. ‘Jean broke his leg.’

b. Jean s’est cassé la jambe.

‘Jean broke his leg.’ (Lambrecht 1986:115)

This line of reasoning shows similarities with related claims made in research on preverbal indefinite subjects in French. It is well known that there are restrictions on the appearance of indefinite expressions in preverbal subject position, especially in spoken French. For instance, it is argued that indefinite preverbal subjects such as un ami ‘a friend’ in (167) are problematic and infrequent in spoken French.

101 However, since Lambrecht does not present any context for this example, it’s difficult to judge how it would be processed in naturally occurring speech.

Chapter 6 | The function of avoiding ‘bad’ preverbal subjects 137

(167) ? Un ami m’a apporté ce livre. ‘A friend brought me that book.’ (Cappeau & Deulofeu 2001:4, my translation)

Several authors note that il y a… qui sentences such as (168) can be seen as a way to avoid indefinite nominal expressions from functioning as preverbal subjects (as in (167)) (e.g. Jeanjean 1979; Léard 1992: section 3.3.2; Van De Velde 1995:28; Blanche- Benveniste 1997:92-93; Cappeau & Deulofeu 2001). The link with the hypothesis presented by Lambrecht and associates is clear: indefinite expressions usually encode discourse-new referents.

(168) Il y a un ami qui m’a téléphoné. ‘A friend called me.’ (Lit. ‘There’s a friend that called me.’)

(Cappeau & Deulofeu 2001:4, my translation)

Another line of research in linguistics that has produced a similar hypothesis is the analysis of ‘existential sentences’ introduced by existential expressions such as il y a, there is and other cross-linguistic equivalents. Most of the existential sentences that are analyzed in those studies do not contain a relative clause. Instead, the NP introduced by the existential expression is followed by a locative prepositional phrase (169), or by nothing at all (170).

(169) C’è un serpente in cucina! ‘There is a snake in the kitchen!’ (Leonetti 2016a:92)

(170) Il y a beaucoup de problèmes. ‘There are many problems.’ (Bentley 2013:677)

Many researchers have observed that existential sentences can only introduce discourse-new referents (e.g. Breivik 1981; Abbott 1993; Ward & Birner 1995; McNally 1997; Francez 2007; Breivik & Martínez-Insua 2008; Leonetti 2008; Bentley 2013; Bentley et al. 2015; Leonetti 2016a). More recently, it has been claimed that existential sentences are used to avoid so-called ‘bad’ or ‘non-canonical’ constituents from appearing in preverbal subject position (Beaver et al. 2006; Bentley 2013; Bentley et al. 2015). This hypothesis encompasses the two hypotheses presented above,

138 Chapter 6 | The function of avoiding ‘bad’ preverbal subjects

because both discourse-newness and indefiniteness are not typical of preverbal subjects.

Summing up so far, we can state that all the above-mentioned researchers present similar claims. However, it is important to realize that the types of sentences they analyze are not exactly the same. While Lambrecht only makes claims about what he calls ‘il y a clefts’, authors such as Van De Velde (1995) speak of ‘il y a… qui sentences’ more generally, and the hypothesis about existential sentences covers even more ground, because it includes il y a sentences without relative clauses as well.

For the sake of clarity, the different hypotheses are presented in Table 24:

Hypotheses in previous literature discourse-new A. Il y a clefts102 are used referents and/or foci in preverbal

B. Il y a… qui sentences103 to avoid indefinite expressions subject position

C. Existential sentences104 ‘bad subjects'

Table 24: Overview of hypotheses about the motivation of il y a sentences

It should be noted that the hypotheses in Table 24 are not verified by means of systematic corpus research of French il y a sentences.105 In the next section, I explain which version of the hypotheses in Table 24 I will consider in this chapter.

102 See Lambrecht 1986; Giry-Schneider 1988; Lambrecht 1988a, 1994; Ashby 1995, 1999; Lambrecht 2000a, 2000b, 2001, 2002; Jullien 2007; Pitavy 2009; Secova 2010. 103 See Jeanjean 1979; Van De Velde 1995:28; Cappeau & Deulofeu 2001. 104 See most notably Beaver et al. (2006); Bentley (2013). Existential sentences are discussed in more detail in chapter 7.5. 105 Cappeau & Deulofeu (2001); Cappeau (2005); Cappeau & Deulofeu (2006b) and Cappeau (2008) do consider corpus data of (mainly spoken) French. However, these analyses are based on targeted queries of specific types of constituents (e.g. indefinite pronouns such as personne ‘nobody’ and indefinite NPs such as des gens ‘people’). In contrast, my data contain all types of possible constituents behind il y a, without restrictions on (in)definiteness or syntactic category (e.g. pronoun or NP). Furthermore, Cappeau and Deulofeu mainly concentrate on the il y en a ‘there are some’ sequence rather than the il y a… qui sequence. Other studies do consider corpus data, but do not explicitly use them to address the hypothesis in (171) (e.g. Choi-Jonin & Lagae 2005; Jullien 2007; Secova 2010).

Chapter 6 | The function of avoiding ‘bad’ preverbal subjects 139

6.2 The hypothesis evaluated in this chapter

Synthesizing the hypotheses in Table 24, I will investigate the following hypothesis:

(171) Hypothesis evaluated in the rest of this chapter Il y a… qui sentences are used to avoid ‘bad subjects’ in preverbal subject position.

The reasons for which I choose the variants ‘il y a... qui sentences’ (rather than ‘il y a clefts’) and ‘bad subjects’ (rather than ‘discourse-new referents’, ‘foci’ or ‘indefinite expressions’) are the following. Firstly, it may be the case that the motivation holding for il y a clefts also holds for il y a… qui sentences more generally, in which case it is more accurate to acknowledge this. Secondly, since my data only contain il y a sentences with relative clauses, it would be unwarranted to make claims about all existential sentences. Therefore, the hypothesis concerns ‘il y a… qui sentences’ rather than ‘existential sentences’. Thirdly, I will consider several factors that may explain why an il y a… qui sentence is used instead of its SV equivalent (discourse-newness, (in)definiteness, focality and other factors), rather than focusing on just one of those. Therefore, the hypothesis contains the more general notion ‘bad subject’, which will be explained in the next section (section 6.3).106

The main goal of the rest of this chapter is to verify whether the hypothesis in (171) is confirmed by my corpus data. Furthermore, if the hypothesis is indeed correct, the next step is to find out for which types of il y a... qui sentences it holds: all il y a… qui sentences in the corpus data, only the instances that are traditionally seen as il y a clefts, or yet a different subgroup of the corpus data. In order to verify the hypothesis that il y a… qui sentences are motivated by the need to avoid ‘bad preverbal subjects’, I will address the following sub-questions:

106 Note that this hypothesis does not hold for il y a… qui sentences with object relative clauses, since the constituent introduced by il y a in such instances does not appear in preverbal subject position when the sentence is transformed into the canonical equivalent. Object relative clauses are relatively infrequent in il y a… qui sentences (11%, see Appendix, section 4.1, Table 36). I leave a more in-depth analysis of the motivation of those instances for future research.

140 Chapter 6 | The function of avoiding ‘bad’ preverbal subjects

I. Which factors determine whether a given constituent is a ‘good’ or a ‘bad’ preverbal subject? (Section 6.3)

II. What kinds of constituents do il y a... qui sentences introduce? Do they indeed display properties of ‘bad’ subjects? (Section 6.4)

Question I is addressed in the next section, which presents an overview of the factors that determine whether a subject is ‘good’ or ‘bad’ according to previous accounts (section 6.3). I then address question II in section 6.4.

6.3 Some subjects are better than others

There is a general consensus that cross-linguistically, certain types of NPs and pronouns are more suitable as preverbal subjects than others.107 In section 6.3.1, I give an overview of the factors that are claimed to influence subject acceptability.108 The underlying rationale for why those factors influence subject acceptability is presented in section 6.3.2.

6.3.1 Factors that influence subject acceptability

One of the clearest distinctions between ‘good’ and ‘bad’ subjects is the one between definite and indefinite referents: definite NPs and pronouns are ‘good subjects’, while most indefinite ones are ‘bad subjects’ (e.g. Givón 1978; Leonetti 1998). Furthermore, discourse-given NPs are better candidates for subject position than discourse-new NPs (e.g. Lambrecht 1994). Other internal (inherent) and external (contextual) factors that influence subject acceptability (especially if the NP or pronoun is indefinite) include the stage-level/individual-level distinction, specificity, quantification, partitivity and genericity. The most important factors that influence subject acceptability are cited in Table 25.

107 The analyses presented below all concern languages with SV word order. 108 The term ‘subject acceptability’ will be used to indicate the extent to which a given constituent is acceptable in preverbal subject position.

Chapter 6 | The function of avoiding ‘bad’ preverbal subjects 141

Positive factors Negative factors

Definite NP Indefinite NP Discourse-given NP Discourse-new NP Human NP109 Non-human NP

Especially relevant if indefinite: Stage-level (eventive) Individual-level (stative) predicate110 predicate Specific NP111 Non-specific NP Quantified NP112 Non-quantified NP (Explicitly) partitive NP113 Non-partitive NP

Table 25: Properties that influence subject acceptability

All these properties can be seen as cumulative: the more ‘positive’ factors a preverbal subject displays, the higher the constituent is ranked on the gradient subject acceptability scale. Inversely, fewer positive factors equal a lower subject acceptability score. For instance, the subject il ‘he’ in (172) is (i) discourse-given, (ii) specific, (iii) definite and it combines with a stage-level predicate (preparing a budget). All these factors make il ‘he’ a very good subject.

(172) Que fait Klaus ? – Il prépare le budget de 2017-2019 en ce moment. ‘What’s Klaus doing? – He’s preparing the budget for 2017-2019 right now.’

Conversely, when presented without any context, the subject une linguiste ‘a linguist’ in (173) is interpreted as (i) discourse-new, (ii) non-specific, (iii) indefinite and (iv) it combines with an individual-level predicate (liking something). The combination of all these factors gives rise to a questionable result.

109 See for instance Cappeau & Deulofeu (2006b:129) and Hung & Schumacher (2014:37). 110 See Danon-Boileau (1989); Van De Velde (1995:19); Kleiber (2001); Dobrovie-Sorin & Beyssade (2004:50). 111 See e.g. Leonetti (1998:26); Beaver et al. (2006). 112 See Van De Velde (1995:22, 26). See also the notion of ‘proportional NP’ in Beaver et al. (2006). 113 See Leonetti (1998:141); Dobrovie-Sorin & Beyssade (2004:31); Van De Velde (2005:38).

142 Chapter 6 | The function of avoiding ‘bad’ preverbal subjects

(173) ? Une linguiste aime The Knife. ? ‘A linguist likes The Knife.’

Cappeau & Deulofeu (2001:3) note that there is no absolute restriction on indefinite preverbal subjects. Furthermore, Cappeau (2008) shows that certain types of indefinite preverbal subjects are more frequent in written and/or more formal registers than in spontaneous speech. For instance, Table 26 presents the total number of some indefinite pronouns used in preverbal subject position in his spoken and written corpora of equal size (a little less than 2 million words).

Spoken corpus Written corpus (press) certain(e)s 'some' 51 80 d'autres 'others' 17 51 beaucoup 'many' 10 33 plusieurs 'several' 0 3 Table 26: Indefinite pronouns in preverbal position in Cappeau (2008:75)114

The type of register can thus be seen as a contextual factor that influences subject acceptability: the same utterance with, say, an indefinite preverbal subject may be acceptable in a newspaper, but awkward in everyday conversation. This is important in light of the finding that il y a clefts (and il y a… qui sentences in general) were found to be much more frequent in the spoken CFPP corpus than in the written YCCQA and Le Monde corpora (see chapter 4.1.1 about the frequency of il y a clefts and chapter 2.2.5 about the frequency of il y a… qui sentences). I will return to this in section 6.5.

114 My translation. I omitted results from a 19th century literary corpus.

Chapter 6 | The function of avoiding ‘bad’ preverbal subjects 143

6.3.2 Underlying reasons for (in)acceptability

As stated in the previous section, there is a general consensus that some constituents function more easily as preverbal subject than others. The underlying reasons why a given referent is (un)acceptable as a preverbal subject are often described in slightly different terms, but most authors converge on the idea that subjects must be referentially ‘autonomous’ (Van De Velde 1995:28), i.e. they must be ‘identifiable’ so that something can be predicated of them (Cappeau & Deulofeu 2001:4).115 This autonomy is sometimes equated with the fact that subjects tend to be specific (Beaver et al. 2006:26). Furthermore, it has been argued that subjects in French are preferably the topic of the sentence, which are necessarily identifiable (Lambrecht 1994).116 All these characterizations are in line with the general consensus that discourse-given, definite expressions make the best subjects. However, this does not mean that discourse-new, indefinite expressions cannot function as preverbal subjects at all. Given that the large majority of all il y a... qui sentences introduce indefinite expressions (84%, see Appendix, section 4.3), factors that license indefinite preverbal subjects are particularly relevant for the present discussion.

One account of (dis)allowed types of indefinite subjects that will be very useful for gaining insight in the functions of il y a... qui sentences is the analysis presented in Leonetti (1998). He presents a Relevance-theoretic analysis of indefinite preverbal subjects in combination with stage-level and individual-level predicates. According to Leonetti, one main requirement that all utterances must fulfill is that they be sufficiently relevant, i.e. intuitively speaking, they must generate enough content for the amount of processing that they impose on the hearer. Leonetti (1998) argues that indefinite preverbal subjects can combine with stage-level predicates, as in (174), because the predicate indicates the spatiotemporal parameters of the proposition,

115 See Bentley (2013:685-687) for more references. 116 In French (and other languages), ‘subject acceptability’ is closely related to the notion of ‘topic- worthiness’, which describes the likelihood that a given referent can be used as a topic (Givón 1976, 1992; Mak et al. 2008; Hung & Schumacher 2014). There is a vast literature about which types of referents make good topics and which do not. See for instance the hierarchies proposed by Givón (1983) and Ariel (1990), which indicate in decreasing order the forms that are most likely to be used as topics (e.g. zero anaphors > pronouns > definite NPs > indefinite NPs). See also Bock & Warren (1985); Prat-Sala & Branigan (2000); Branigan et al. (2008). I will present the notion of ‘topic’ in more detail in chapter 7.1.4.

144 Chapter 6 | The function of avoiding ‘bad’ preverbal subjects

and therefore allows for the subject to be grounded (see also Kleiber 2001; Choi-Jonin & Lagae 2005:42). Since the proposition is grounded, the hearer can process it.

(174) Indefinite subject + stage-level predicate Someone has been following me. (Leonetti 1998:143)

Individual-level predicates, on the other hand, do not refer to any spatiotemporal settings, and therefore need to be grounded by their subject. Consequently, individual-level predicates generally cannot combine with indefinite subjects, because in this case “neither the predicate nor the subject are able to create an appropriate context for each other. This causes an increase in processing effort, as the recovery of the proposition expressed is not adequately constrained” (Leonetti 1998:158, see also Herburger 1994:527; Van De Velde 1995:19). Thus, Leonetti argues, sentences such as (175) are costly in terms of processing effort (due to the lack of grounding) and relatively uninformative, and therefore violate the Principle of Relevance.

(175) Indefinite subject + individual-level predicate ? A student is tall. (Leonetti 1998:142)

Interestingly, however, indefinite preverbal subjects are not completely ruled out from sentences with an individual-level predicate, as long as “some kind of enrichment” indicates that the speaker wishes to increase the informativeness of the indefinite. Leonetti mentions several ways in which an indefinite preverbal subject can be ‘enriched’ so that it can combine with an individual-level predicate. Firstly, if an indefinite NP such as a student is further specified (e.g. a student of mine/I know), this indicates that the speaker has some referent in mind, making the indefinite specific, and thereby grounding the subject. This is why (176) is more acceptable than (175).

(176) A student of mine is tall. (Leonetti 1998:156)

Secondly, if the indefinite is quantified, this also increases informativeness according to Leonetti. For instance, the NP some linguists in (177) can combine with the individual-level predicate like Metallica, because the sentence expresses the informative content “the set of linguists who like Metallica is not an empty set (there

Chapter 6 | The function of avoiding ‘bad’ preverbal subjects 145

are some of them)” (Leonetti 1998:158). The author argues that ‘there sentences’ such as (178) allow to draw the same inference of a non-empty set.

(177) Quantified indefinite subject + individual-level predicate Some linguists like Metallica. (Leonetti 1998:153)

(178) There are intelligent violists. (Leonetti 1998:158)

The same contrasts in Leonetti’s English examples have been reported for French. Firstly, several authors note that indefinite preverbal subjects can combine more easily with eventive/stage-level predicates than with stative/individual-level predicates, as illustrated in (179) (Danon-Boileau 1989; Van De Velde 1995:19; Kleiber 2001; Dobrovie-Sorin & Beyssade 2004:50).

(179) a. Du beurre était en train de fondre sur la table. ‘Butter was melting on the table.’

b. * Du beurre était savoureux. * ‘Butter was delicious.’

(Dobrovie-Sorin & Beyssade 2004:50, my translation)

Secondly, it has been acknowledged that quantification makes indefinite NPs better candidates for the preverbal subject slot: Van De Velde (1995:22&26) argues that the indefinite NP de la neige ‘snow’ in (180) becomes much more acceptable as a subject when neige ‘snow’ is modified by the quantifying expression une épaisse couche de ‘a thick layer of’ in (181) (see also Cappeau & Deulofeu 2001:3 and Cappeau & Deulofeu 2006a:434).

(180) * De la neige a effacé leurs traces. ‘Snow erased their traces.’117 (Van De Velde 1995:22, my translation)

(181) Une épaisse couche de neige a effacé leurs traces. ‘A thick layer of snow erased their traces.’

(Van De Velde 1995:23, my translation and emphasis)

117 Note that although the English translation is felicitous, the French original is not.

146 Chapter 6 | The function of avoiding ‘bad’ preverbal subjects

Furthermore, some researchers note that explicit partitivity improves the acceptability of preverbal indefinite subjects in French (Dobrovie-Sorin & Beyssade 2004:31; Van De Velde 2005:38). For instance, des livres ‘some books’ in (182) is partitive because the context evokes a set that some books belong to (in contrast to other books), and therefore this indefinite subject is more acceptable than if such a part-whole relationship were not evoked.

(182) Des livres étaient sur la table (d’autres par terre).

PART books were on the table PART.others on ground

‘Some books were on the table (some others on the floor).’

(Van De Velde 2005:38, my translation)

Partitivity can be included in Leonetti’s category of ‘enrichment by means of quantification’. In slightly different terms, one could state that the reason for the improved acceptability of explicitly partitive indefinites such as (182) is that the ‘whole’ they belong to can function as a ‘hypertheme’ that anchors the referent (see Daneš 1974; Dupont et al. 1994; Erteschik-Shir 2007 about hyperthemes). Dobrovie- Sorin & Beyssade (2004) explain the improved acceptability of partitive indefinite subjects as follows: “(implicitly or explicitly) partitive indefinites presuppose the existence of a collection of individuals that is contextually determined, which is why they are D(iscourse)-linked (…)” (Dobrovie-Sorin & Beyssade 2004:31, my translation, see also Choi-Jonin & Lagae 2005:56). To the extent that these accounts refer to the ‘anchoring’ function of partitivity, they are compatible with Leonetti’s analysis: partitive indefinites are ‘strong enough’ to ground the assertion, and are therefore able to combine with individual-level predicates.

To conclude, there are different degrees to which a constituent is a good or a bad preverbal subject, depending on different morphosyntactic, semantic and contextual factors that determine (i) how easily the proposition expressed by the utterance can be retrieved and (ii) how informative it is.

Chapter 6 | The function of avoiding ‘bad’ preverbal subjects 147

6.4 Corpus data

In the previous section, I explained which are the main factors that influence whether a given constituent is a good or a bad preverbal subject, and why. This section sets out to determine to what extent the constituents introduced by il y a in my corpus data would be ‘good’ or ‘bad’ preverbal subjects in the sense explained in section 6.3. In order to determine this, I will first analyze which il y a... qui sentences in the corpus data can be declefted, i.e. transformed into an SV sentence expressing the same proposition.118 I will show that decleftability is not a black-and-white phenomenon: some sentences can be declefted ‘easily’, while others require some adjustments. I will then analyze the factors that allow for il y a… qui sentences to be declefted by looking at the properties that have been shown to influence subject acceptability (e.g. eventive predicate, quantification, partitivity, see previous section).

In section 6.4.1, I comment on some basic requirements for the possibility to decleft a sentence. I then present the distribution of the different classes of decleftability in the corpus data (section 6.4.2). In section 6.4.3, I present examples of the class of il y a… qui sentences that can be declefted most easily, which, not coincidentally, correspond to il y a clefts. In section 6.4.4, I explain why other il y a… qui sentences, which do not have eventive or specificational semantics, can also be declefted. Finally, I present instances of il y a… qui sentences that do correspond to an SV equivalent, but which require a lexical adjustment in order to be declefted (section 6.4.5). In other words, declefting is less ‘easy’ in such instances.

6.4.1 Basic requirements for successful declefting

Before presenting the corpus data, I wish to emphasize an important aspect of the decleftability test that, as far as I know, has not been explicitly discussed before. The process of declefting an il y a... qui sentence with a subject relative clause necessarily puts the constituent originally introduced by il y a in preverbal subject position:

(183) Il y a X qui VP à DECLEFTING à XSUBJECT VP

118 See chapter 1.1 about decleftability.

148 Chapter 6 | The function of avoiding ‘bad’ preverbal subjects

Hence, constituent ‘X’ must be able to function as a preverbal subject, otherwise the transformation cannot yield an acceptable result, as in (184) - (185): ceux ‘those’ can never function as a full-fledged nominal constituent without the presence of a relative clause or the particles –ci ‘here’ or –là ‘there’.

(184) Il y a ceux qui – A: Aimer vous la boxe ?

‘Do you like boxing?’

– B: Mais c’est comme tout, il y a ceux qui aiment et ceux qui n’aiment pas.

‘But it’s like everything, there are those who like it and those who don’t like it.’ (YCCQA)

(185) Not decleftable * Ceux aiment et ceux n’aiment pas.

‘Those like it and those don’t like it.’119

In other words, one of the main requirements for successful declefting is that the constituent introduced by il y a can function as a preverbal subject.120 This may seem self-evident, but it gives us an important first insight in the degree of subject acceptability of the constituents introduced by decleftable il y a… qui sentences: even if the hypothesis that they introduce ‘bad subjects’ turns out to be true, the constituents they introduce are not completely ruled out as preverbal subjects, otherwise the sentence could not be declefted.121

119 Note that the English translation, contrary to the French original, is grammatically correct. However, it does not convey the same proposition as the English translation of the il y a sentence. 120 The fact that (184) has an existential interpretation, as evidenced by the fact that il y a can be replaced by il existe ‘there exist’, is another indication that this sentence is not a cleft. 121 A second fact that follows from declefting, as indicated in (183), is that the introducing expression il y a (or, in other cleft types, c’est ‘it’s’, j’ai ‘I have’ etc.) must be sufficiently ‘light’ on a semantic level, because otherwise it cannot be omitted. The fact that il y a is semantically light in cleft instances and in other uses has been pointed out by several authors (e.g. Lambrecht 1986; Choi-Jonin & Lagae 2005; Pekarek Doehler 2015:177).

Chapter 6 | The function of avoiding ‘bad’ preverbal subjects 149

6.4.2 Corpus data: degrees of decleftability

When determining whether or not the il y a… qui sentences in my corpus data could be declefted or not, I found that there is no discrete distinction between ‘decleftable’ and ‘not decleftable’. Instead, among ‘decleftable il y a... qui sentences’, there are degrees: some il y a… qui sentences can be declefted much more easily than others, for various reasons. I coded the decleftable il y a… qui sentences using the following classes: (i) declefting is possible in combination with eventive/specificational semantics (i.e. sentences that are traditionally analyzed as clefts); (ii) declefting is possible in combination with other properties; (iii) declefting is possible, but requires a lexical adjustment and (iv) declefting is not possible.122 I will present the different classes in sections 6.4.3 (il y a clefts), 6.4.4 (other decleftable il y a... qui sentences) and 6.4.5 (lexical adjustment required).123 Table 27 already presents a quantitative overview of the distribution of the different classes of decleftability.124

Total n % Declefting is possible in combination with 576 28% eventive/specificational semantics (i.e. il y a clefts) Declefting is possible in combination with other properties 387 19%

Declefting is possible, but requires a lexical adjustment 446 22%

Not decleftable 650 32%

Total 2059 100%

Table 27: All il y a… qui sentences in the corpus data: classes of decleftability

What is important to notice in Table 27 is that the third class (declefting requires a lexical adjustment: 22%) is by no means a marginal phenomenon, despite the fact that

122 I realize that the use of the term ‘declefting’ and ‘decleftability’ in combination with il y a… qui sentences that are not traditionally seen as clefts is unfortunate. A more neutral expression would perhaps be the periphrastic ‘transforming an il y a… qui sentence into its canonical equivalent’ and ‘the possibility to transform an il y a… qui sentence into its canonical equivalent’. I will nonetheless use the terms ‘declefting’/ ‘decleftability’ for the sake of brevity. 123 About sentences that cannot be declefted, see chapter 3.3. 124 The different classes of decleftability per corpus can be found in the Appendix, Table 42.

150 Chapter 6 | The function of avoiding ‘bad’ preverbal subjects

such instances have not been analyzed in previous studies of il y a clefts. Hence, I will analyze them more thoroughly in the next sections. I will show that there is a close connection between (i) ease of decleftability and (ii) degree of subject acceptability of the constituent introduced by il y a, as shown in Figure 5:

Declefting is easy Constituent is highly acceptable as a preverbal subject

Declefting is relatively easy Constituent is relatively acceptable as a preverbal subject

Declefting is difficult Constituent is hardly acceptable as a preverbal subject

Figure 5: Link between decleftability and subject acceptability

What I will argue for is that the ease of declefting depends in part on the extent to which the constituent introduced by il y a would be acceptable in preverbal subject position: the less the constituent is acceptable in preverbal subject position, the more difficult it becomes to decleft the il y a… qui sentence.

6.4.3 Declefting is possible: eventive/specificational semantics

Most of the il y a… qui sentences that are the easiest to decleft are those that closely resemble the typical examples in previous studies on il y a clefts, i.e. with a definite constituent and a predicate that expresses an observable event, such as (186).

(186) Observable event + definite NP (proper noun) bah si j'passe à côté d'eux dans la rue oui j'entends qu'ils parlent pas français (…) ou quand y a Marta qui vient avec son mari elle parle roumain

‘Well when I walk next to them in the street, yes, I hear that they don’t speak French, or when Marta comes over with her husband she speaks Romanian.’

(Lit. ‘When there’s Marta who comes over with her husband’) (CFPP)

Chapter 6 | The function of avoiding ‘bad’ preverbal subjects 151

(187) … ou quand Marta vient avec son mari, elle parle roumain. ‘… or when Marta comes over with her husband, she speaks Romanian.’

Predicates that express observable events are also ‘strong’ enough to license an indefinite constituent in preverbal subject position, especially if this indefinite is further grounded in some way (see section 6.3.2). For instance, the utterance in (188) expresses an observable event, and the modification à moi ‘of mine’ in une copine à moi ‘a friend of mine’ indicates that the speaker has a specific referent in mind.

(188) Observable event + ‘grounded’ indefinite NP c' qui était rigolo c'est que la même semaine y a une copine à moi qui m' dit « oh j'ai re-rencontré un mec avec qui j'étais super amie au collège qui est génial écoute c'est extraordinaire comme histoire il vient d'ouvrir un bar bla bla »

‘what was funny is that the same week there’s a friend of mine who tells me “oh I met a guy I was really close friends with in high school who’s great listen it’s an extraordinary story he just opened this bar bla bla”’ (CFPP)

(189) [Same context as (188)] Une copine à moi m’dit « oh… »

‘A friend of mine tells me “oh…”’

Specificational il y a... qui sentences introducing a definite expression, such as (190), can also be declefted (see (191)), thanks to the specificity/definiteness of the NP.

(190) Specificational semantics + definite NP – A: Comment on fait pour trouver l'amour? je suis tout seul depuis longtmps je comprends pas, je m'intéresse aux gens mais ça donne rien

‘How do people find love? I’ve been all alone for a long time I don’t understand, I’m interested in people but it doesn’t lead to anything.’

– B: ça tombe parfois par surprise mais maintenant il y a aussi la nouvelle technologie qui aide, Internet peut t'aider à trouver l'amour

‘Sometimes it happens by surprise but now there’s also new technologies that help, the Internet can help you find love.’ (YCCQA)

152 Chapter 6 | The function of avoiding ‘bad’ preverbal subjects

(191) La nouvelle technologie aide aussi.125 ‘New technologies also help.’

In sum, the il y a… qui sentences that are the easiest to decleft have one or several of the following properties, which have been shown to increase the acceptability of preverbal subjects: (i) definite/specific NP, (ii) expression of an observable event, (iii) ‘grounded’ indefinite NP.

6.4.4 Declefting is possible due to other properties

As I explained in chapter 3.5, clear instances of il y a clefts (having eventive or specificational semantics) are not the only types of il y a… qui sentences that correspond to an SV counterpart expressing the same proposition. There is a considerable variety of properties that characterize the class of decleftable il y a… qui sentences that are not semantically eventive or specificational. Below, I discuss the 3 predominant properties that allow for these il y a… qui sentences to be declefted: (i) quantification, (ii) embedding and (iii) explicit partitivity. As was the case in the previous section, it is easier to decleft a given il y a… qui sentence if it combines several of these properties.

(i) Quantification and (ii) embedding

As explained in section 6.3.2, quantification can improve the subject acceptability of an indefinite NP. It thus comes as no surprise that many of the il y a… qui sentences in the corpus data that can be declefted but which are not eventive or specificational (and hence, which would not traditionally be considered as il y a clefts), contain some form of quantification, such as beaucoup de gays ‘many homosexuals’ (192), or 1 million de Français ‘one million Frenchmen’ in (194).126 Furthermore, if the il y a… qui

125 The position of aussi ‘too’ is different in the SV counterpart. Such small adjustments are generally not regarded as problematic for the decleftability criterion, as long as the proposition doesn’t change (see chapter 1.1 and section 6.4.5). 126 The fact that ‘il y a beaucoup de N qui VP’ (‘There are many N who VP’) sentences correspond to ‘Beaucoup de N VP’ (‘Many N VP’) sentences has also been noted by Choi-Jonin & Lagae (2005:64).

Chapter 6 | The function of avoiding ‘bad’ preverbal subjects 153

sentence is embedded, as in (192), this further facilitates declefting, because this is a way of ‘introducing’ the constituent in an ongoing flow of information.

(192) Quantified NP + embedding Y-a-t-il ici des fans des gays nounours? j'entends par gays nounours, les bears c'est dire les types qui se réclament ressembler à des ours (enveloppés et velus) je sais qu'il y a beaucoup de gays qui apprécient ce genre d'hommes.

‘Are there fans of teddy bear homosexuals? By teddy bear homosexual I mean ‘bears’, the guys who claim they look like bears (big and hairy) I know that there are a lot of gay people who like that type of men.’ (YCCQA)

(193) Je sais que beaucoup de gays apprécient ce genre d’hommes. ‘I know that a lot of gay people like that type of men.’

(194) – A: Pensez vous que Degaulle est le dernier des grands president qu'a eu la France et l'Algérie?

‘Do you think De Gaulle was the last great president of France and Algeria?

– B: Il ya 1 millions de francais qui pensent le contraire: les pieds noires et les harkis.

‘There are one million Frenchmen who think the opposite: the pieds noirs and the harkis.’ 127 (YCCQA)

(195) 1 million de français pensent le contraire: les pieds noirs et les harkis. ‘One million Frenchmen think the opposite: the pieds noirs and the harkis.’

(iii) Explicit partitivity

As explained in section 6.3.1, explicit partitivity can license indefinite preverbal subjects (e.g. Van De Velde 1995). Partitivity can be seen as another way of ‘enriching’

127 Les pieds noirs refers to Algerian-born Europeans with the French nationality. Les Harkis refers to Muslim Algerians who served as auxiliaries in the French Army during the Algerian War (1954-1962).

154 Chapter 6 | The function of avoiding ‘bad’ preverbal subjects

the indefinite so as to make it informative enough (Leonetti 1998). The corpus data indeed contain il y a… qui sentences with indefinite NPs that can be declefted, despite the fact that they are not eventive or specificational. For instance, certains hommes ‘some men’ in (196) and un de mes co-internes ‘one of my fellow interns’ in (198) are able to function as indefinite preverbal subjects in (197) and (199) respectively, even though the relative clauses (who really feel the need to be loved, live in the rue du Chemin-Vert) are not stage-level predicates.

(196) – A: Pourquoi nous les hommes, on as peur que les femmes nous aiment pour de vrai ??

‘Why are we, men, scared that women really love us?’

– B: Il ne faut pas généraliser. Il y a certains hommes qui ont vraiment besoin de sentir aimés et/ou qui se complaisent dans cet état car ils peuvent en tirer certains avantages.

‘You shouldn’t generalize. There are some/certain men who really feel the need to be loved and/or who enjoy that state because they can obtain certain advantages from it.’ (YCCQA)

(197) Certains hommes ont vraiment besoin de se sentir aimés. ‘Some men really feel the need to be loved.’

(198) j'étais donc à Ivry en stage y a un d'mes co-internes qui habite rue du Chemin- Vert comme moi, j'l'ai vu un jour sur une liste ‘So I was at Ivry for an internship there’s one of my fellow interns who lives in the rue du Chemin-Vert like me, I saw him on a list (CFPP)

(199) un d'mes co-internes habite rue du Chemin-Vert comme moi ‘one of my fellow interns lives in the rue du Chemin-Vert like me’

To conclude, I’ve shown that certain il y a… qui sentences without clear eventive/specificational semantics can be declefted if they display one or several properties (e.g. quantification) that make the constituent they introduce an acceptable preverbal subject.

Chapter 6 | The function of avoiding ‘bad’ preverbal subjects 155

6.4.5 Declefting is possible, but requires a lexical adjustment

In a large group of the corpus data, declefting is possible, but only if a lexical adjustment is made. This in itself does not necessarily mean that the decleftability test does not apply: Lambrecht (2001:499); Dufter (2006:38) and Dufter (2008:38) observe that the transformation of clefts into their canonical counterparts sometimes requires lexical adjustments. As an illustration, Lambrecht (2001) cites an example from Collins (1991), namely a subtype of a so-called wh-cleft (200), and Dufter (2006) shows the same for a French c’est cleft (201).

(200) a. All this machine does is swim and eat. b. This machine only swims and eats. (Lambrecht 2001:500)

(201) a. C’est moi qui l’ai dit. ‘It’s me who said it.’

b. *Moi l’ai dit.

*‘Me said it.’

c. (Moi) je l’ai dit.

‘(Me) I said it.’ (Dufter 2006:38, my translation)

This of course raises the question how far one can go with these lexical adjustments and still state that the sentence can be declefted. The crucial requirement is that both sentences express the same proposition, i.e. their semantic representation must remain the same (Lambrecht 1986, 1994, 2001). Assuming this line of reasoning, the examples in (200) and (201) indeed pass the test: the propositions remain the same, only in (200b) can be seen as a lexical equivalent of all in (200a) in a different syntactic context, and the same holds for the full pronoun moi ‘me’ and the clitic pronoun je ‘I’ in (201), which are lexically equivalent.

As for the il y a... qui sentences that require a lexical adjustment, the large majority of them contain plural indefinite NPs such as des gens ‘people’. Without lexical adjustments, such sentences do not immediately yield an acceptable declefted equivalent, as illustrated in (202) - (203).

156 Chapter 6 | The function of avoiding ‘bad’ preverbal subjects

(202) – A: Je voudrai acheter une caméra numérique pas trop chère et bien? que me conseillez vous?

‘I’d like to buy a digital camera that’s not too expensive and good? What do you recommend?’

– B: Y'a des appareils photos qui filment très bien avec un bon son en plus...

‘There are photo cameras that film very well and also with good sound…’

(YCCQA)

(203) ?? Des appareils photos filment très bien avec un bon son en plus. ?? ‘Photo cameras film very well and also with good sound.’

In the il y a... qui sentence in (202), des appareils photos ‘photo cameras’ can be interpreted partitively, i.e. as a subgroup of all possible photo cameras (some cameras film very well). The possibility to interpret this NP partitively is also clear from the fact that one can add ‘but other cameras do not film well’ (this diagnostic for partitivity is presented in Choi-Jonin & Lagae 2005:44; Riegel et al. 2011:293-294, see also Kleiber 2001128). The problem that occurs when the sentence is declefted is that the partitive reading seems to be unavailable when the NP appears in subject position. In the absence of other factors that increase subject acceptability (e.g. an eventive predicate, specificity or quantification), NPs such as des appareils photo ‘video cameras’ are not licensed in subject position. However, their acceptability can be improved considerably by replacing des by the explicitly partitive determiner certains ‘some/certain’, as in (204).129

128 In sentences with a non-partitive, purely ‘existential’ NP, other instances of the noun are not evoked. For instance, Kleiber (2001) notes that the NP des inconnus ‘(some) strangers’ does not have a partitive interpretation that refers to certain strangers having burgled Lea’s house, whereas other strangers have not burgled Lea’s house. (i) Des inconnus ont cambriolé la maison de Léa. ‘(Some) strangers burgled Lea’s house’ (Kleiber 2001:57, my translation) (“[i] ne saurait avoir une interprétation partitive qui extrait ou prélève sur un ensemble d’inconnus ceux qui ont cambriolé la maison de Léa et qui donnerait donc à penser que tous n’ont pas cambriolé la maison de Léa (ou que seulement certains l’ont fait).” Kleiber 2001:57) 129 The fact that il y a... qui sentences introducing plural indefinite NPs correspond to SV equivalents with the determiner certains ‘some/certain’ has also been noted by Willems & Meulleman (2010:16). According to Blanche-Benveniste (1997:94), the expression ‘il y en a qui’ is a morphological variant of ‘certains’ in preverbal subject position (see, however, Choi-Jonin & Lagae 2005:50 for criticism).

Chapter 6 | The function of avoiding ‘bad’ preverbal subjects 157

(204) Certains appareils photos filment très bien avec un bon son en plus. ‘Certain/some photo cameras film very well and also with good sound.’

Two other examples are given in (205) and (206), which introduce the NPs des vétos ‘vets’ and des gens ‘people’ respectively.

(205) – A: J'ai 4 chatons que je n'arrive pas à donner. qu'est-ce-que je peux en faire? ‘I have four kittens that nobody wants to have. What I can I do?’

– B: Il ya des vétos qui les prennent pour ensuite les donner aux gens qui viennent chez eux.

‘There are vets who take them and then give them to people who visit them.’ (YCCQA)

(206) – A: Comment faire pour avoir le plus de chance d'avoir un garçon? ‘What can I do to improve my chances of having a boy?

– B: ma cheri c'est dieu qui nous donne remercie a dieu que tu a une ou des filles. il ya des gens qui ne gagne jamais.

‘Honey, it’s God who gives, thank the Lord that you have one or several girls. There are people who never win.’ (YCCQA)

These sentences again yield awkward declefted equivalents in (207a) and (208a), and are much more natural when the determiner des is replaced by certain(e)s ‘some’ in (207b) and (208b).130

(207) a. ?? Des vétos les prennent pour ensuite les donner aux gens qui viennent… ‘Vets take them and then give them to people who visit them.’

b. Certains vétos les prennent pour ensuite les donner aux gens qui viennent…

‘Some vets take them and then give them to people who visit them.’

(208) a. ?? Des gens ne gagnent jamais. ‘People never win.’

130 Note that a similar problem arises in the English translations with bare plurals, which are interpreted generically in this position, contrary to the original ‘there are NP’ sentences.

158 Chapter 6 | The function of avoiding ‘bad’ preverbal subjects

b. Certaines personnes ne gagnent jamais.

‘Some people never win.’

In other words, the NPs of the type ‘des N’ introduced by il y a can have a partitive interpretation, but the preverbal subject position does not easily allow for such a reading when the subject is indefinite. Consequently, the NP is not licensed as a preverbal subject, and the utterance requires a lexical adjustment (which does not alter the original proposition) in order to be acceptable.131 The crucial point is that in all these cases, the lexical adjustment is required because otherwise, the NP cannot function as a preverbal subject.

To the extent that the il y a… qui sentences discussed in this section can be declefted, they are similar to the more typical il y a... qui sentences with eventive or specificational semantics. Yet one cannot deny that typical cases such as Y’a Jean qui a téléphoné ‘Jean phoned’ are in a way ‘easier’ to decleft than the examples discussed here. One way of looking at this difference is to argue that decleftability is a gradient phenomenon (see discussion about il y a clefts as a prototype category in chapter 3.6): there are instances in which decleftability is immediately obvious, and other cases (requiring lexical adjustments), in which declefting is less straightforward but nonetheless possible. The view that decleftability is a gradient rather than a discrete phenomenon comes as no surprise if one accepts, as I posited in section 6.4.2, that decleftability depends in part on the constituent’s degree of subject acceptability: subject acceptability has independently been shown to be a gradient phenomenon itself, influenced by semantic, morphosyntactic and contextual factors (section 6.3).

131 Alternatively, one could argue that the NPs introduced by the il y a… qui sentences in this section do not have a clear partitive reading, but should be interpreted purely existentially (i.e. non- partitively). One could then 0bject that adding certains ‘some’ to the SV counterpart changes the semantics of the NP from non-partitive to partitive. However, this argument is not convincing, since it has been shown that NPs with determiners such as certains ‘some’ and du can have a non- partitive reading as well (e.g. Kleiber 2001: section 1). In other words, there is no change in proposition between (206) and (208b), regardless of whether one interprets the NP des gens as partitive or as non-partitive.

Chapter 6 | The function of avoiding ‘bad’ preverbal subjects 159

6.5 Confirmation of the hypothesis

In the previous section, it became clear that there is a link between (i) whether or not an il y a… qui sentence can be declefted and (ii) to what extent the constituent introduced by il y a makes a good or bad preverbal subject in the corresponding SV sentence. This was visualized in Figure 6, repeated below:

Declefting is easy Constituent is highly acceptable as a preverbal subject

Declefting is relatively easy Constituent is relatively acceptable as a preverbal subject

Declefting is difficult Constituent is hardly acceptable as a preverbal subject

Figure 6: Link between decleftability and subject acceptability

Some combinations of properties (e.g. a definite NP + a relative clause that expresses an observable event) facilitate declefting more than other combinations (e.g. a partitive indefinite + an individual-level predicate). In general, if several ‘positive’ factors are combined, the constituent is a better preverbal subject (and, hence, can be declefted more easily) than if there is only one positive factor. In other words, there is no binary distinction between ‘good’ and ‘bad’ subjects, but rather a scale ranging from ‘downright unacceptable’ to ‘optimal’ subjects, with other values in between those extremes. This scale has an effect on the ease with which a sentence can be declefted. For instance, it is possible to decleft (209) into (210), because the NP is partitive, but this sentence is still considered suboptimal, since the preverbal subject un de mes co-internes ‘one of my fellow interns’ is indefinite and combines with an individual-level predicate.

160 Chapter 6 | The function of avoiding ‘bad’ preverbal subjects

(209) j'étais donc à Ivry en stage y a un d' mes co-internes qui habite rue du Chemin- Vert comme moi (…) ‘So I was at Ivry for an internship there’s one of my fellow interns who lives on Chemin-Vert Street like me (…)’ (CFPP)

(210) un d' mes co-internes habite rue du Chemin-Vert comme moi ‘one of my fellow interns lives on Chemin-Vert Street like me’

Interestingly, the il y a… qui sentences that are traditionally regarded as typical instances of il y a clefts, such as (211a), generally introduce constituents that are relatively high on the subject acceptability scale, due to their eventive semantics and the definiteness of the introduced constituent. Hence, they can be declefted (see (211b) most easily with respect to other types of il y a… qui sentences.

(211) a. Y’a Jean qu’a téléphoné. ‘Jean called.’ (Lambrecht 1988a:136)

b. Jean a téléphoné.

‘Jean called.’

If it is so ‘easy’ to decleft sentences such as (211a), one could wonder whether we should still consider that such il y a… qui sentences are used to avoid suboptimal preverbal subjects, as hypothesized in section 6.1. I argue that this hypothesis is indeed correct, because the constituents introduced by the il y a… qui sentences in my corpus data all have at least one aspect in common that would make them suboptimal subjects if they appeared in preverbal subject position and which I have not yet discussed so far: they are discourse-new. In other words, even though the constituents un de mes co-internes ‘one of my fellow interns’ in (210) and Jean in (211b) are relatively acceptable preverbal subjects, they are not optimal, because they are discourse-new. In the case of (210), the factors that make the NP a suboptimal subject are: (i) indefinite NP, (ii) discourse-new NP, (iii) individual-level predicate, while in (211b), only the discourse-newness of the proper noun makes Jean a suboptimal preverbal subject.

Chapter 6 | The function of avoiding ‘bad’ preverbal subjects 161

Hence, in (211b), there is only one factor that makes the constituent a suboptimal subject (discourse-newness), while in (210), there are several such factors. I believe this explains why sentences such as (211a) (but not (209)) are often seen as ‘typical’ instances of il y a clefts and have received most attention. Given that (211a) can be declefted quite easily, and that Jean is a relatively acceptable preverbal subject, the fact that a speaker would use the clefted version rather than the declefted one is in a way more surprising than in the case of (209). Put differently, the il y a… qui structure seems less necessary in the il y a clefts in (211a) and therefore the question why a speaker would use this structure seems more acute. I nonetheless argue that both (210) and (211a) are motivated by the need to avoid a suboptimal preverbal subject. However, in some cases, this suboptimality is due only to discourse-newness, while in other cases, factors such as indefiniteness and individual-level predicates make the subject even more suboptimal.

The present line of reasoning may also help explain the frequency differences of il y a clefts throughout the corpora (chapter 4.1.1). Recall that il y a clefts are 95.5 times more frequent in CFPP than in Le Monde. Moreover, 83% of the il y a clefts in Le Monde were found within citations (chapter 4.1.1).132 As noted by Cappeau (2008), written French is more tolerant towards indefinite preverbal subjects than spoken French. I believe we can generalize this statement as follows: the threshold for acceptable preverbal subjects is higher in spoken French than in written French, because in the spoken modality, there is a higher need to facilitate online language processing for the interlocutor than in writing. In other words, spoken French will resort to il y a… qui sentences more frequently than written French, because the conditions that a preverbal subject must fulfill are stricter in spoken French.

132 Similarly, the broader class of decleftable il y a… qui sentences is almost 53 times more frequent in the spoken corpus than in the formal written corpus (Appendix, Section 5, Table 43). Moreover, 77% of all decleftable il y a... qui sentences in Le Monde appear in a citation (Appendix, Table 44).

162 Chapter 6 | The function of avoiding ‘bad’ preverbal subjects

6.6 Demystifying decleftability

In previous studies, it is sometimes stated (and often seems to be assumed) that the fact that il y a clefts can be declefted is a unique property of those clefts, in contrast with other types of sentences, such as cleft lookalikes (e.g. Lambrecht 1986).133 The findings from Chapter 3 and those presented in this chapter ‘demystify’ this assumption: I first showed that il y a... qui sentences with RRCs, which are traditionally not analyzed as clefts, can also be declefted in certain cases (chapter 3.5). In other words, this finding constitutes a first indication that the availability of an SV variant is not solely reserved for il y a... qui sentences displaying properties that are typically associated with il y a clefts (i.e. a certain type of relative clause and eventive or specificational semantics). In this chapter I showed that whether or not an il y a... qui sentence can be declefted depends to a large extent on whether or not the constituent it introduces can function as a preverbal subject. Since the typical eventive/specificational il y a clefts are not the only types of il y a... qui sentences that display the necessary properties for subject acceptability, they are also not the only types of il y a... qui sentences that can be declefted.134

6.7 Conclusion

The goal of this chapter was to investigate the hypothesis according to which il y a… qui sentences are used to avoid ‘bad’ preverbal subjects. In so doing, I also provided more insight in the phenomenon of ‘decleftability’.

Confirmation of the hypothesis

A close scrutinization of the corpus data led me to conclude that the hypothesis according to which il y a… qui sentences avoid ‘bad subjects’ from appearing in preverbal subject position is fundamentally correct, although it also needs to be nuanced. The hypothesis is correct, because all il y a… qui sentences introduce

133 Davidse (2000) also argues that the decleftability of it clefts and there clefts is not as mystical as is often assumed, although she gives different arguments for this than the ones I present above. 134 Consequently, the term “de-cleft-ing”/”de-cleft-ability” may not be ideal, unless one adopts the position that all decleftable sentences should be seen as clefts, see footnote 122 and chapter 3.6.

Chapter 6 | The function of avoiding ‘bad’ preverbal subjects 163

constituents that are not perfect candidates for the preverbal subject slot. However, if the sentence can be declefted, this entails that the constituent is at least minimally acceptable as a preverbal subject, otherwise declefting would not be possible.

I also showed that there is a variety of interacting factors that determine to what extent a constituent is acceptable as a preverbal subject. Consequently, there are varying degrees of subject acceptability of the constituents introduced by il y a. This constitutes a more nuanced picture than previous accounts, which often mainly focus on one factor that characterizes the constituents introduced by il y a… qui sentences (e.g. discourse-newness: Lambrecht 1986, indefiniteness: Cappeau & Deulofeu 2001), rather than analyzing the bundle of factors that determine the constituent’s subject acceptability.

Moreover, in order to account for the fact that il y a clefts occur much more frequently in spoken French than in written French, I hypothesized that the threshold for acceptable preverbal subjects is higher in spoken French than in written French, because in the spoken modality, there is a higher need to facilitate online language processing for the interlocutor than in writing. In other words, the conditions that a preverbal subject must fulfill are stricter.

Demystifying decleftability

By determining for each il y a… qui sentence in the corpus data whether it can be declefted or not and why this is possible or impossible, I was able to demystify the phenomenon of decleftability: I argued that the possibility of declefting depends to a large extent on whether or not the introduced constituent would be acceptable in preverbal subject position (i.e. subject acceptability). Since subject acceptability has independently been shown to be a gradient phenomenon, it follows that some sentences can be declefted more easily (relatively high acceptability) than others (low acceptability). This allowed me to explain why declefting is possible in a large number of cases that are not commonly analyzed as il y a clefts.

------

Chapter 7

The information structure of il y a clefts

Cleft sentences are often mentioned as a textbook example of a syntactic construction that is motivated by information structure, or, put slightly differently, they are seen as special constructions that overtly mark information structure properties (e.g. Lambrecht 2001; Erteschik-Shir 2007:121). Two common hypotheses about the link between clefts and information structure are the following: (i) clefts are focus markers, i.e. they mark a constituent (or a whole sentence) as focal,135 and (ii) clefts are used to prevent a topic-comment interpretation of the sentence. These two hypotheses have received an increasing amount of attention, especially in (corpus) research on c’est clefts and their cross-linguistic equivalents (see Chapter 8 about the information structure of c’est clefts). However, the exact information structure properties of il y a clefts have not yet been investigated by means of an in-depth corpus analysis. In this chapter, I will fill this descriptive and theoretical gap.136

Two main results were obtained. Firstly, my corpus data display a wider variety of information structure properties than previously assumed. Secondly, I will argue that the data support the hypothesis that il y a clefts mark the constituent they introduce as ‘non-topical’, but I will also argue that this function is not due to the cleft format, but to the existential expression il y a. Hence, il y a clefts share this function with existential sentences more generally. This chapter also has methodological relevance by showing that information structure notions cannot always be

135 As Dufter (2006:43) notes, the idea that clefts are focus-markers is so pervasive that some authors even use the term ‘focus phrase’ to refer to the clefted constituent (e.g. Smits 1989a; Lambrecht 2001) rather than a pragmatically neutral term such as ‘clefted constituent’. 136 Parts of this chapter are/will be published in Karssenberg (2016a, 2016b); Karssenberg & Lahousse (forthcoming-a). 166 Chapter 7 | The information structure of il y a clefts

operationalized without problems when confronting them with authentic corpus data, as has already been pointed out by others (e.g. Ritz et al. 2008; Lahousse 2011; Lahousse & Borremans 2014; Stede & Mamprin 2014:1722).

The chapter is structured as follows: in section 7.1, I present and define the main information structure notions that will be used to analyze the corpus data. I then briefly present the information structure properties that have been attributed to il y a clefts and Lambrecht’s hypothesis that they function as markers of non-topicality (the ‘non-topic marking hypothesis’) (section 7.2). Section 7.3 is devoted to a detailed analysis of the information structure properties of the il y a clefts in the corpus data. Section 7.4 evaluates the non-topic marking hypothesis in light of the corpus findings. In section 7.5, I compare the articulations of il y a clefts with different types of existential sentences, revealing significant parallels. The hypothesis about non- topicality marking is then shown to be closely related to the hypothesis that il y a clefts avoid suboptimal preverbal subjects, as put forward in Chapter 6 (section 7.6). Finally, I present psycholinguistic studies of expressions that are similar to il y a in support of my hypothesis that il y a is a processing cue (section 7.6).

7.1 Background: information structure terminology

“a terminological minefield” (Vallduví & Engdahl 1996:466)

The field of information structure is notorious for its diversity of notions that are used (e.g. topic, focus, theme, rheme, background, emphasis, salience, contrast) and the different ways in which all of these notions are defined. In some cases, two terms display overlap with each other (e.g. rheme and comment), in other cases, a single notion is used in fundamentally different ways (for instance topic, which Vallduví & Engdahl 1996:466 refer to as a terminological minefield). Therefore, it is crucial to state which notions and definitions I will use in this dissertation.

The notions ‘information structure’, ‘Common Ground’ and ‘Question Under Discussion (QUD)’ are introduced in section 7.1.1. In section 7.1.2, I explain the important distinction between the ‘referential’ and the ‘relational’ dimension of

Chapter 7 | The information structure of il y a clefts 167

information structure. I then turn my attention to the notions ‘focus’ and ‘background’ (section 7.1.3) and ‘topic’ and ‘comment’ (section 7.1.4). Section 7.1.5 is devoted to the information structure of ‘thetic sentences’.

7.1.1 Information Structure, Common Ground and QUDs

Information structure analysis can be seen as the study of the manner in which speakers ‘package’ their message so that it fits the hearer’s current knowledge and expectations. In this way, the speaker can make it easier for the hearer to process the information expressed by the utterance (e.g. Chafe 1976; Krifka 2008). For instance, while the sentences in (212) express the same proposition, they structure the information differently.

(212) a. Weightlifting really clears my mind. b. What really clears my mind is weightlifting.

c. It’s weightlifting that really clears my mind.

Depending on the discourse context, one of these forms may be preferred over the other in order to optimize the message. For instance, in the context of (213), answer A is more appropriate than answer B.137

(213) – Q: Why do you like weightlifting so much? – A: Weightlifting really clears my mind.

– B: # What really clears my mind is weightlifting.

One notion that is of particular relevance in this respect is the Common Ground (Karttunen 1974; Stalnaker 1974; Lewis 1979), which refers to the information that it mutually known to be shared by the speaker and the hearer, and which is constantly being modified/updated during communication (Krifka 2008:245). The way information is packaged (see (212)) should be in optimal accordance with the current state of the Common Ground.

137 See e.g. Prince 1978 about the difference in interpretation between wh-clefts (212a), regular SV sentences (212b) and it clefts (212c).

168 Chapter 7 | The information structure of il y a clefts

It has been argued that so-called ‘Questions Under Discussion’ (QUDs) influence the way in which the Common Ground is constantly being modified (see e.g. Van Kuppevelt 1995; Roberts 1996; Kehler & Rohde 2016). Under this view, communicative interactions are structured by question-answer pairs. For instance, in (214), the QUD “Are you going anywhere this summer?” is explicitly raised by the question. Speaker A then answers this QUD.

(214) – Q: Are you going anywhere this summer? – A: Yeah, we’re taking the van to go camping in Wales.

However, in many cases, the QUD is implicit. For example, the second sentence in (215) is normally interpreted as an answer to the implicit question ‘Why?’ (Kehler & Rohde 2016:1). By inferring this implied QUD, the hearer is able to understand the coherence link between the first and the second sentence.

(215) We should write a paper. There’s a special issue of Discourse Processes coming up that will be dedicated to QUDs. (Kehler & Rohde 2016:1)

7.1.2 The ‘referential’ and the ‘relational’ dimension of information structure

There is an important distinction between two dimensions of information structure, often referred to as the ‘referential’ and the ‘relational’ dimension (see e.g. Prince 1981; Lambrecht 1988a, 1994; Gundel & Fretheim 2003; Cruschina 2015a:45).

Put simply, the referential dimension refers to the fact that in a given discourse context, some referents may already have been evoked while other referents are newly introduced into the discourse, i.e. the degree of activation of a given referent (e.g. Givón 1978; Prince 1981; Givón 1983; Ariel 1990; Zimmerman & Féry 2009). The speaker adapts his/her utterances according to whether the hearer is already ‘familiar’ with a referent. Thus, referents that are discourse-new are generally encoded differently than referents that are discourse-given. For instance, in (216), the noun phrase ‘the instructor’ denotes the referent ‘the instructor of my weightlifting class’, which can consequently be encoded by the pronoun he in the next sentence. In

Chapter 7 | The information structure of il y a clefts 169

this context, it would be odd to encode the referent by another full NP, as in (217).138 This is because the speaker is aware that the hearer already knows which referent is intended. Therefore, repeating the full NP would be superfluous, and hence, unexpected for the hearer.

(216) I went to my weightlifting class last night and the instructor was wearing a Superman T-shirt with matching sweatbands. He looked fabulous!

(217) I went to my weightlifting class last night and the instructor was wearing a Superman T-shirt with matching sweatbands. # The instructor of my weightlifting class looked fabulous!

There are varying degrees of discourse-givenness and discourse-newness. For instance, Prince (1981) makes a distinction between new referents, inferrable referents and evoked (i.e. discourse-given) referents. 139 Thus, the NP the instructor in (216) denotes an inferrable referent: the hearer is assumed to infer from the previously mentioned ‘weightlifting class’ that an instructor was present, due to world knowledge (classes are generally led by teachers/instructors). Other taxonomies of familiarity/discourse-givenness include the ‘activation states’ in Chafe (1987), the ‘Accessibility scale’ in Ariel (1990) and the ‘Givenness Hierarchy’ in Gundel et al. (1993).

The relational dimension, on the other hand, concerns the relation between the constituents of an utterance (see e.g. Lambrecht 1994; Jacobs 2001; Gundel & Fretheim 2003; Krifka 2008). This will become clearer when four particularly important notions are defined: ‘focus’ and ‘background’ (next section), and ‘topic’ and ‘comment’ (section 7.1.4).

138 I will indicate pragmatic awkwardness by means of ‘#’. 139 Prince (1981) also makes more fine-grained distinctions within those three main classes: new referents can be brand-new/unused, inferrable referents can be noncontaining/containing and evoked referents can be evoked textually or situationally.

170 Chapter 7 | The information structure of il y a clefts

7.1.3 Focus & background

The notion ‘focus’ (plural: ‘foci’) has been defined in different ways. One major line of reasoning is that ‘focus’ refers to the most ‘salient’, ‘highlighted’ or ‘important’ part of a sentence (e.g. Lambrecht 1994; De Cesare et al. 2016:41), i.e. the part to which the hearer’s attention is drawn (Erteschik-Shir 2007:39). If a specific part of the sentence is salient, while another part of the sentence is not, the non-salient part is often referred to as the ‘background’. For instance, in the context of (218), my cat is the element that the speaker wishes to emphasize (i.e. the focal constituent), while the VP can be considered as the ‘background’. With respect to the referential dimension, my cat is discourse-new, whereas the VP expresses discourse-given information.

(218) Relational: focus-background, Referential: new-given – A: Who is hungry?

– B: [My cat]FOCUS [is hungry]BACKGROUND

It should be noted that the account of focus as ‘salience’, ‘importance’ or ‘relevance’ has been criticized, because those notions, although intuitively appealing, are difficult to define (e.g. Krifka 2008:255).140 I will nonetheless implement it, and address its shortcomings when problems arise.

It’s important to realize that foci are not necessarily discourse-new on the referential level (e.g. Dufter 2006:43). For instance, in (219), the salient constituent in speaker B’s answer is Felix, which was mentioned in the preceding question and is therefore discourse-given. In this example, Felix is contrasted with respect to Curly. Therefore, Felix is an instance of contrastive focus.141

140 For instance, Doherty (2005) notes that ‘relevance’ must have relative values: “more relevant as opposed to less relevant information. This may even result in a hierarchy of informational values from highest to lower and lowest relevant” (Doherty 2005:189-190). It is unclear how such gradience could be operationalized in corpus research. See also Matić & Wedgwood (2012); Van Der Wal (2016) for criticism about the notion ‘focus’. 141 In this dissertation, I will analyze constituents as contrastive if they are explicitly contrasted with respect to another constituent in context. See Lahousse & Borremans (2014:798-799) about the danger of ‘eroding the concept of contrastiveness’ in studies of the information structure of clefts.

Chapter 7 | The information structure of il y a clefts 171

(219) Relational: focus-background, Referential: Felix = discourse-given – A: Which one of them ate all the cheese? Felix or Curly?

– B: [Felix]FOCUS [ate all of it]BACKGROUND

In languages such as English, it is often stated that foci correspond to a specific prosodic contour, and some authors therefore use prosody to identify foci. However, in French, prosody less consistently maps onto discourse-new/discourse-given/salient information (e.g. Lambrecht 1994). Moreover, whichever the language under consideration, not all communication is spoken, and addressees presumably also assign information structure properties to written language when they process utterances. Therefore, we cannot assume that addressees always disambiguate between focal and non-focal constituents by means of prosody.142

7.1.4 Topic & comment

Topics are often defined as ‘that which the sentence is about’, (Strawson 1964; Gundel 1974; Reinhart 1982; Lambrecht 1994:118; Erteschik-Shir 2007; Krifka 2008). They are regarded as mental locations at which the ‘comment’ (i.e. the information expressed about the topic) is stored by the hearer. In terms of the so-called ‘file-card metaphor’, a topic represents a card in a file card system on which the hearer writes the new information that he/she has learned about the particular referent at hand (see e.g. Reinhart 1982; Vallduví & Engdahl 1996; Erteschik-Shir 2007; Krifka 2008). Since all new information must be stored somewhere, some argue that all sentences must contain a topic (e.g. Erteschik-Shir 2007). On the referential level, topics are usually (though not always) discourse-given. In the context of (220), for instance, my cat constitutes that which the sentence is about, i.e. it functions as the topic. The information that is expressed about the topic (is hungry) is the comment. As for the referential dimension, my cat is discourse-given (because it is mentioned in the preceding question), and the VP expresses discourse-new information.

142 Of course, it is possible that readers attribute a certain prosodic profile to the sentence in their head when they read it, but this cannot be taken into account when analyzing written corpus data.

172 Chapter 7 | The information structure of il y a clefts

(220) Relational: topic-comment, Referential: given-new – A: How’s your cat doing today?

– B: [My cat]TOPIC [is hungry]COMMENT

In order to test whether a given constituent is topical or not, one can apply the so- called ‘topic tests’ (e.g. Reinhart 1982; Lambrecht 1994; Erteschik-Shir 2007; Hedberg 2013). These tests consist in asking a question about the constituent in the form of ‘What about X?’ (221a) or ‘Tell me about X’ (221b). Alternatively, one can introduce the sentence by ‘I say about X that’ (221c). If the result is coherent, X is indeed a topic.

(221) a. What about your cat? – My cat is hungry. b. Tell me about your cat. – My cat is hungry.

c. I say about my cat that it’s hungry.

Importantly, as Krifka (2008) points out, the notions ‘topic’ vs. ‘comment’ are sometimes equated with ‘background’ vs. ‘focus’ respectively. However, several authors show that this is incorrect, because (i) topics may contain focal elements, and (ii) the comment is not necessarily identical with the focal element of the utterance. The fact that constituents may simultaneously function as topics and contain focal elements is especially relevant for the analysis of contrastive topics.143 For instance, the constituents my oldest sister and my other sister in (222) are topical, as indicated by the fact that the preceding question inquires about them. However, the adjectives oldest and other are focal because they are contrasted with respect to each other. Hence, these adjectives are analyzed as focal elements within the topic constituents.

(222) Contrastive topics containing focal elements (based on Krifka 2008) – A: What do your sisters do for a living?

– B: [My [OLdest]FOCUS sister]TOPIC [is a fashion designer]COMMENT ,

and [my [OTHer]FOCUS sister]TOPIC [is a physiotherapist]COMMENT

143 There is an increasing body of work on contrastive topics (e.g. Jäger 2001; Erteschik-Shir 2007; Krifka 2008; Constant 2012; Leonetti 2013; Constant 2014; Büring 2016).

Chapter 7 | The information structure of il y a clefts 173

Moreover, although the comment often coincides with focal information (as in (222)), this is not necessarily the case in all contexts. For instance, Krifka (2008) shows that the comment in (223B) is married her in 1968, but only the prepositional phrase in 1968 is focal, because married her is already evoked in the preceding question.

(223) Comment containing discourse-given information

– A: When did [Aristotle Onassis]TOPIC marry Jacqueline Kennedy?

– B: [He]TOPIC [married her [in 1968]FOCUS]]COMMENT

(Krifka 2008:266)

Thus, as shown by the notation in (222) and (223), ‘topic’ and ‘background’ do not always mark exactly the same constituents, and neither do ‘focus’ and ‘comment’.

Several types of topics have been proposed in the literature. A fundamental difference can be drawn between two types: aboutness-topics, which usually designate referents (as in my oldest/other sister in (222)), and spatiotemporal topics, which designate a spatiotemporal frame such as the here-and-now.144 Spatiotemporal topics (or ‘stage’/‘scene-setting’ topics, cf. Kuno 1972; Chafe 1976; Prince 1978; Jacobs 2001; Lahousse 2003; Charolles & Vigier 2005; Erteschik-Shir 2007; Lahousse 2008) introduce “a spatial, temporal or individual framework within which the main predication holds” (Chafe 1976:50). For instance, Erteschik-Shir (2007:17) analyzes the fronted prepositional phrase in (224) as a spatiotemporal topic that specifies the location with respect to which the sentence is to be evaluated. Similarly, Jacobs (2001) sees the expression in meinem Traum ‘in my dream’ as the situation “for which the whole sentence is asserted to hold” (Jacobs 2001:657).

(224) Outside the door, there’s a cat. (Erteschik-Shir 2007:17)

(225) In meinem Traum war Peter ein Krokodil. ‘In my dream, Peter was a crocodile.’ (Jacobs 2001:657)

144 See also Frascarelli & Hinterhölzl (2007) for an even more fine-grained distinction between different types of topics in Italian and German.

174 Chapter 7 | The information structure of il y a clefts

At this point, it is important to note that information structure interpretations arise in context: the context determines which information is already or not yet in the shared Common Ground (referential dimension) and which information is focal or topical (relational dimension). Put differently, out of context, the utterance ‘My cat is hungry’ is underspecified with respect to the information structure articulations it can express; it does not have an ‘inherent’ information structure interpretation. The importance of context has implications for the use of topic tests. Question-answer pairs (‘What about X?’) and the ‘I say about X that’ test are very useful to illustrate topics, but they cannot always be used without problems when investigating existing corpus data. The goal of analyzing authentic data is to determine whether a constituent is topical in its original context. When the original context is replaced by a question such as ‘Tell me about X’, what you uncover is whether the constituent has the potential to be a topic, but not whether it really is one in its original context. The topic test must therefore be used with caution. Consider for instance the utterance in (226), which is the first sentence in the famous novel by Proust. If one wants to determine which expression in the sentence is the topic, it is possible to apply the topic test as in (227). In this case, the first person pronoun je ‘I’ seems to encode the topic of the sentence.

(226) Longtemps, je me suis couché de bonne heure. ‘For a long time, I went to bed early.’

(Marcel Proust, Du côté de chez Swann, 2008 [1913], p.3).

(227) Topic test à topic-comment articulation Je dis à propos de moi-même que longtemps, je me suis couché de bonne heure.

‘I say about myself that for a long time, I went to bed early.’

However, one could also argue that this utterance is a thetic sentence in which the proposition ‘I went to bed early’ is predicated of the spatiotemporal topic encoding the settings ‘at that time, for a long time’ (see next section about thetic/all-focus sentences). This is illustrated in (228).

Chapter 7 | The information structure of il y a clefts 175

(228) Topic test à all-focus articulation Je dis à propos de cette période qu’à cette époque, pendant longtemps, je me suis couché de bonne heure.

‘I say about that period that during that time, for a long time, I went to bed early.’

The difficulty of establishing what the sentence is about is related to the fact that it is not straightforward to pinpoint the exact implicit QUD that the sentence in (226) answers. Given that there is no prior context, one could argue that the QUD is the most general one that is possible: “What is the way things are?” (called ‘the Big Question’ in Roberts 1996). However, another possibility is to state that upon reading the sentence, the reader infers that the writer intends to answer a more specific QUD, such as ‘What was I/my childhood like?’ or ‘How were my sleeping habits during my childhood?’. In other words, a more specific QUD can be accommodated in the Common Ground by the reader (Krifka 2008:251). The fact that it is difficult to establish the exact implicit QUD in authentic data has consequences for determining which constituents are topical/non-topical and which are focal/non-focal. One option would be to state that different readers may infer different QUDs when reading (226). Consequently, there would not be one ‘true’ information structure reading of the utterance, but several possible readings. I will return to this issue in section 7.3.4.

7.1.5 Thetic/‘all-focus’/‘all-comment’ sentences

A distinction is often made between three main information structure articulations of the whole sentence: ‘topic-comment’, ‘focus-background’ and ‘all-focus’. These are also referred to as ‘predicate-focus’, ‘argument-focus’ and ‘sentence-focus’ respectively by Lambrecht (1994). The all-focus articulation is sometimes equated with the term ‘thetic sentence’ (Kuroda 1972; Sasse 1996; Rosengren 1997; Choi-Jonin & Lagae 2005:54). There are two properties that are commonly associated with thetic sentences: (i) the grammatical subject is not the topic of the sentence, (ii) all the information expressed by the sentence is focal.145

145 For instance, Rosengren (1997) claims that thetic sentences are ‘all-focus’ and ‘all-comment’.

176 Chapter 7 | The information structure of il y a clefts

Consider for instance (229). In this context, all the information expressed by the answer is salient, and hence focal. Moreover, my cat is not analyzed as ‘that which the sentence is about’, witness the fact that it is inappropriate to add ‘I say about…’ in (230) or ‘as for my cat’ in (231). As for the referential dimension, all the information in B’s answer in (229) is discourse-new. This is typical of examples that are used to illustrate all-focus/thetic sentences.

(229) Relational: all-focus, Referential: all-new – A: What’s that weird noise?

– B: [My cat is hungry]FOCUS

(230) – A: What’s that weird noise? – B: # I say about my cat that it’s hungry

(231) – A: What’s that weird noise? – B: # As for my cat, it’s hungry

Even though my cat is not topical in (229), this doesn’t mean that the sentence is about nothing (e.g. Erteschik-Shir 2007; Krifka 2008): “(...) thetic sentences are viewed as having implicit ‘stage’ topics indicating the spatiotemporal parameters of the sentence (the here-and-now of the discourse). These are contextually defined.” (Erteschik-Shir 2007:16146). In other words, the new information expressed by the thetic sentence in (229) is stored under the “time and space information that is either mentioned in the preceding discourse (…) or may be inferred from the contextual setting of the conversation” (Vallduví & Engdahl 1996:471).

Now that the main information structure notions have been defined, let us move on to previous analyses of the information structure of il y a clefts.

146 See also Vallduví & Engdahl (1996: 471); Pinto (1997); Tortora (1997); De Cat (2007); Krifka (2008).

Chapter 7 | The information structure of il y a clefts 177

7.2 Previous studies: the information structure of il y a clefts

In section 7.2.1, I illustrate the information structure articulations that have been associated with il y a clefts. I then present the hypothesis according to which (il y a/c’est) clefts are used to mark the constituent they introduce as non-topical and/or as focal (section 7.2.2).

7.2.1 Information structure articulations of il y a clefts

The articulation that is usually attributed to il y a clefts is all-focus (e.g. Lambrecht 1994, 2000a, 2001), i.e. il y a clefts are usually analyzed as thetic sentences. An example of an all-focus il y a cleft is given in (232): this sentence, Lambrecht argues, is uttered in an out-of-the-blue context (i.e. Jean is not mentioned in the preceding discourse), hence all the information is discourse-new and focal.147

(232) All-focus il y a cleft Y’a Jean qu’a téléphoné.

‘Jean called’ (Lambrecht 1988a:136)

147 Lambrecht (1988a) argues that within the class of all-focus il y a clefts, a distinction should be made between the ‘presentational function proper’ and the ‘event-reporting’ function of il y a clefts. Truly presentational il y a clefts (232), he says, present a new referent (Jean) and then predicate something about it (he called), i.e. they are ‘entity-introducing’. In event-reporting clefts (i), on the other hand, the clefted element is just a participant in the event that is introduced as a whole in the discourse. (i) Y’a le téléphone qui sonne! ‘The phone’s ringing!’ (Lambrecht 1988a:137) However, Lambrecht himself admits that it’s unclear when a referent is ‘important’ enough to be considered an aboutness-topic, and he therefore states that the difference between sentences such as (232) and (i) is not categorical (Lambrecht 1988a:149; 2001:507). Indeed, it is possible to see even (232) as an event-reporting cleft if Jean is “mentioned only as a necessary participant in the reported event, i.e. the phone call” (Lambrecht 1988a:160, see also the discussion about the two readings of Il y a Paul qui arrive ‘Paul is coming’ in Choi-Jonin & Lagae 2005:63). I do not maintain the distinction between entity-introducing and event-reporting il y a clefts in the corpus analysis, because it can hardly be operationalized. Davidse & Kimps (2016) don’t maintain this distinction for English there clefts either, since “Lambrecht himself provides no criteria for the distinction” (Davidse & Kimps 2016:128).

178 Chapter 7 | The information structure of il y a clefts

Apart from il y a clefts with an all-focus articulation such as (232), which are semantically eventive, there are also semantically specificational il y a clefts (Lambrecht 2001; Verwimp & Lahousse 2016). Lambrecht associates these with a different information structure, namely focus-background. He states that in such cases, the clefted constituent is focal and the relative clause expresses discourse-given background information. For instance, the author states that the underlined cleft in (233) was uttered in a context where people discussed who would like to go to the theater. Therefore, Beth is focal and the relative clause is the background.

(233) Focus-background il y a cleft Y’a Beth qui veut y aller, euh, y’a y’a Jean-Marc, y’a moi, bon.

‘There’s Beth who wants to go, uh, there’s there’s Jean-Marc, there’s me, ok.’

(Lambrecht 1988a:154)

One thing all-focus and focus-background il y a clefts have in common is that the clefted constituent is discourse-new and encodes a focal rather than a topical referent. This generalization has led to the hypothesis that il y a clefts, like other types of clefts, are used to signal that a focal, non-topical constituent is coming, which I will now explain in more detail.

7.2.2 Clefts in general: markers of non-topicality and/or focality

It has been claimed that the underlying motivation of clefts in general is to prevent the hearer from interpreting the sentence as having a topic-comment articulation (e.g. Lambrecht 1988a, 1994, 2001; Gundel 2006; Erteschik-Shir 2007). The argument is that cross-linguistically (in SVO languages), preverbal subjects (e.g. Felix in (234a)) tend to coincide with the topic of the sentence. Hence, if the speaker wishes to express a proposition in which Felix is not to be interpreted as a topic by the hearer, the speaker can use a cleft sentence to signal this explicitly, as in (234b). In other words, under this view, clefts are seen as markers of non-topicality.

(234) a. Felix ate all the cheese. b. It’s Felix who ate all the cheese.

Chapter 7 | The information structure of il y a clefts 179

A different but closely related hypothesis is that clefts are used to mark the constituent they introduce as focal, i.e. clefts are focus markers (e.g. Smits 1989a; Lambrecht 2001; Grevisse & Goosse 2008:575; Riegel et al. 2011:726). This function is supposedly especially important in languages that cannot easily mark sentence-initial constituents as focal by means of prosody, such as French (as opposed to English148) (see e.g. Lambrecht 1994). The terms ‘non-topical’ and ‘focal’ are sometimes equated, but we will see shortly that the distinction between the two is important to maintain.

Lambrecht (1994) argues that both hypotheses (markers of non-topicality and of focality) also hold for il y a clefts: “due to a powerful grammatical constraint against the co-mapping of the pragmatic relation focus and the grammatical relation subject (…), spoken French makes abundant use of [cleft constructions] to avoid focus-initial SV structures.” (Lambrecht 1994:22). He relates this constraint to what he calls the “Principle of the Separation of Reference and Relation”: “Do not introduce a referent and talk about it in the same clause” (Lambrecht 1994:185; 2000a). For instance, in a context where the subject Jean is focal, the SV sentence in (235b) is supposedly pragmatically less felicitous than the cleft in (235a), because in (235b), Jean and the predication about Jean are presented in one clause.

(235) a. Y’a Jean qui s’est cassé la jambe. ‘Jean broke his leg.’

b. Jean s’est cassé la jambe.

‘Jean broke his leg.’ (Lambrecht 1986:115, glosses omitted)

However, it is important to note that the hypothesis that clefts are markers of non- topicality has been discredited with respect to it clefts, c’est clefts and wh-clefts, because these clefts can also express a topic-comment articulation, as in (236) (see for instance Prince 1978; Declerck 1984, 1988; Doetjes et al. 2004; Hedberg 2006; Hedberg & Fadden 2007; Dufter 2008; Hedberg 2013; Lahousse & Borremans 2014).149

148 English does regularly use prosodic stress in order to indicate that a preverbal subject is focal, as in (i) (capitalization indicates prosodic stress). This is uncommon in French. (i) – A: Who ate the cheese? – B: FELIX ate it. 149 Relatedly, clefts used to be associated with less variety in terms of discourse-given/discourse- new mapping onto the clefted element and the relative clause. As stated by Delin & Oberlander (2005), "in the early literature on clefts, it was often assumed that clefts conveyed ‘Old’ or ‘Given’

180 Chapter 7 | The information structure of il y a clefts

In example (236), speaker A asks for more information about ‘the topic’, and speaker B uses a cleft in which the relative clause provides new information about ‘the topic’. In other words, the clefted element is an aboutness-topic, and the relative clause is the comment.

(236) – A: But why is the topic so important? – B: Apparently, it is the topic that enables the listener to compute the intended antecedents of each sentence in the paragraph.

(Prince 1978:902)

Similar c’est clefts have been found in French, as shown in (237) (see e.g. Doetjes et al. 2004; Blanche-Benveniste 2006; Dufter 2008, 2009b; Mertens 2012; Lahousse & Borremans 2014). In this example, the clefted element elle ‘her’ is discourse-given and the relative clause presents new information about the clefted element.

(237) Sophie tu la connais depuis longtemps, enfin c’est elle que tu connaissais d’abord ‘Sophie you’ve known her for a long time, indeed it was her that you knew first.’

(Dufter 2008:49, translation slightly adapted)

What is somewhat confusing about examples such as (236) and (237) is that although the clefted constituent is indeed what the sentence is about (i.e. topical), it simultaneously seems to be focal, because it seems more salient than if it appears in preverbal subject position. Thus, this NP is both topical and focal at once, which is possible under a number of accounts: it is a contrastive topic (see section 7.1.4). Hence, sentences such as (236) do not refute the hypothesis that clefts mark their clefted constituent as focal (if focality is defined as the salient part of the sentence), but they are problematic for the hypothesis that clefts signal that their clefted element is non-topical. I will present the different information structure articulations that c’est clefts can express in chapter 8.1, but I will first analyze il y a clefts.

information in the [relative clause], and ‘New’ information in the clefted constituent (e.g. Akmajian 1970; Keenan 1971)" (Delin & Oberlander 2005:3). This assumption was also refuted by examples such as (236): the clefted element is discourse-given and the relative clause is discourse-new, instead of the other way around.

Chapter 7 | The information structure of il y a clefts 181

The discussion about topic-comment it clefts and c’est clefts raises the question whether il y a clefts pattern like c’est clefts: can il y a clefts also express the topic-comment articulation and, hence, should the hypothesis that they are markers of non-topicality be refuted?150 I will address this question by means of a qualitative analysis of the corpus data.

7.3 Corpus data: different information structure articulations

In what follows, I will show that the il y a clefts in my corpus data are able to express a wider variety of information structure articulations than previously assumed. Section 7.3.1 presents a quantitative overview of the different articulations throughout the corpora, which are then presented in more detail: all-focus (with or without a ‘light’ relative clause) (section 7.3.2), focus-background (section 7.3.3), il y a clefts that are ambiguous between multiple focus and focus-background (7.3.4) and contrastive topic-comment (section 7.3.5). This array of interpretations contradicts the common assumption that il y a clefts only express all-focus or focus-background: il y a clefts are not as uniform as they are sometimes portrayed when it comes to their information structure properties. Moreover, the ambiguous il y a clefts support the general insight that information structure properties are often difficult to pinpoint when investigating corpus data, and indicate that the classes themselves may be gradient. The consequences of the existence of topic-comment clefts for the non-topic marking hypothesis will be discussed in section 7.4.

150 See Piotrowski (2009) for an analysis of the information structure of there clefts, it clefts and reverse wh-clefts in light of the non-topic marking hypothesis.

182 Chapter 7 | The information structure of il y a clefts

7.3.1 Quantitative overview

Table 28 gives an overview of the articulations found in the corpus data of il y a clefts.151 The information structure articulations of the il n’y a que clefts extracted from Le Monde and YCCQA are presented in Table 29.

Information structure: il y a clefts n %

Section 7.3.2 All-focus 447 78%

Section 7.3.2 All-focus with ‘light’ relative clause 28 5%

Section 7.3.3 Focus + background 64 11%

Section 7.3.4 Multiple focus?/Ambiguous 30 5%

Section 7.3.5.2 Contrastive aboutness-topic + comment 7 1%

Total 576 100%

Table 28: Information structure articulations of il y a clefts

Information structure: il n'y a que clefts n %

Section 7.3.3 Focus + background 116 55%

Section 7.3.4 Ambiguous/multiple focus? 95 45%

Section 7.3.5.1 Spatiotemporal topic + comment 1 0%

Total 212 100%

Table 29: Information structure articulations of il n'y a que clefts

7.3.2 All-focus (with/without light relative clause)

The most frequent articulation in the data is all-focus (also referred to as ‘wide focus’, ‘thetic sentence’, ‘all-comment’, see section 7.2.1), which may explain why this articulation is mentioned most often in studies of il y a clefts. In all-focus sentences, the subject and the predicate do not correspond to the topic and the comment

151 The quantitative findings reported here are based on only cases of clear il y a clefts that do not contain restrictive relative clauses (see discussion in chapter 3.6).

Chapter 7 | The information structure of il y a clefts 183

respectively (see section 7.1.3). In (238) – (240), for instance, the NPs trois personnes ‘three people’; une voiture ‘a car’ and une personnalité importante ‘someone important’ do not function as the topic about which the relative clause expresses a comment.152 Rather, the whole event expressed by the il y a cleft is presented in one block, and the whole sentence can be seen as focal. Furthermore, all the information expressed by the cleft is referentially discourse-new.

(238) Ceci dit, au moindre doute, je passe un coup de fil, et il y a trois personnes qui descendent dans la seconde. ‘With that said, if there is any doubt I make a call, and three people come down within a second.’ (Le Monde)

(239) On devrait ouvrir un site (...) si il y a quand il y a une personnalité importante qui meurt. ‘They should make a website (...) if there’s when there’s someone important who dies.’ (CFPP)

(240) – Toto, pourquoi n'irais-tu pas sur le balcon. Comme ça tu nous raconteras tout ce qui se passe en bas! (…)

‘Toto, why don’t you go to the balcony. Then you can tell us about everything that’s happening below!’

– Toto: Il y a une voiture qui se fait remorquer par une dépanneuse dans le parking...

Toto: ‘There’s a car that’s getting towed by a tow truck in the parking lot.’

(YCCQA)

152 One could argue that the clefted elements are not topical because they are indefinite. However, indefinite expressions can function as topics in other contexts (see e.g. Erteschik-Shir 2007:52 about indefinite topics). Note also that all-focus il y a clefts also introduce definite constituents, as illustrated by (i). In this example, neither the clefted element (that famous flashback) nor the information expressed by the relative clause (it passes before our eyes) is discourse-given, and the whole sentence can be seen as salient. (i) (…) Le paradis est pour moi, l'instant de notre mort, où l'on repense à notre vie et/ou(ou pas) il y a ce fameux flash-back de notre vie qui défile devant nos yeux. ‘(…) For me, paradise is the moment of our death, when we look back at our life and/or (or not) there is this famous flash-back of our life that passes before our eyes.’ (YCCQA)

184 Chapter 7 | The information structure of il y a clefts

The information structure of sentences such as (240) can be visualized as in (241): the NP une voiture ‘a car’ is not the topic of the sentence, but rather, the whole sentence is predicated of an implicit stage topic that specifies the spatiotemporal settings of the utterance (in this case Toto’s ‘here-and-now’). Furthermore, the whole sentence is focal, indicating that all the information is salient.

(241) … IMPL. STAGE TOPIC [il y a une voiture qui se fait…]FOCUS

… IMPL. STAGE TOPIC [there’s a a car that’s getting…]FOCUS

As explained in section 7.1.3, thetic sentences are usually associated with three properties: (i) the subject and the predicate do not have a topic-comment relationship, (ii) all the information is focal (i.e. salient) and (iii) all the information is referentially discourse-new. The sentences in (238) - (240) combine these three properties. However, the corpus data contain a group of il y a clefts that do display properties (i) and (iii), but in which the focal status of the relative clause is not exactly clear. In those cases, the relative clause contains a semantically light verbal phrase, such as qui s’opère ‘that takes place’ in (242).

(242) – A: Quels mélanges ne faut-il pas effectuer avec du Baileys? Pourquoi? ‘What shouldn’t you mix with Baileys? Why?’

– B: C'est surtout le bayleys et le schweppes, y a une reaction chimique qui s’opere et le melange devient solide

‘It’s mostly Baileys and Schweppes, there’s a chemical reaction that takes place and the mix becomes solid.’ (YCCQA)

Recall that ‘focus’ is defined in terms of what is most salient or what the speaker wishes to draw the hearer’s attention to (e.g. Erteschik-Shir 2007). In this sense, the relative clause in (242) can be considered as backgrounded rather than focal. In this example, the NP une réaction chimique ‘a chemical reaction’ by itself denotes an event. Therefore, the added relative clause expressing that the event takes place is in a way superfluous: stating that ‘there is a chemical reaction’ already suffices to indicate that that reaction takes place. Consequently, the relative clause does not seem very salient: it’s not what the speaker wishes to draw the hearer’s attention to. Therefore, it is

Chapter 7 | The information structure of il y a clefts 185

unclear whether the relative clause should be analyzed as focal, as indicated by (243), or as non-focal, in which case the sentence has a focus-background articulation (244).

(243) Analysis 1 (focal relative clause)

… IMPL. STAGE TOPIC [il y a une réaction chimique qui s’opère]FOCUS

… IMPL. STAGE TOPIC [there’s a chemical reaction that takes place]FOCUS

(244) Analysis 2 (non-focal relative clause)

… IMPL. ST. TOPIC [[il y a une réaction chimique]FOCUS [qui s’opère]BACKGR.

… IMPL. ST. TOPIC [[there’s a chemical reaction]FOCUS [that… ]BACKGR.

Another example of an eventive il y a cleft with a semantically light relative clause is (245): the fact that the error message appears is obvious in this particular context. The relative clause hardly contributes to the proposition expressed by the sentence.

(245) Lorsque je veux transmettre un courrier ou répondre à un mail, il y a un message d’erreur qui apparaît. ‘When I want to send a message or answer an email, there’s an error message that appears.’ (YCCQA)

I’m not aware of any principled reason for preferring either one of the analyses in (243) and (244). Instances such as (242) and (245) seem at the verge of the all-focus and the focus-background articulations. 153 However, this difference is presumably not problematic for the hearer or reader: it does not seem to matter whether one interprets the sentence as having an all-focus or a focus-background articulation, both interpretations are logical in the discourse context. I will return to sentences that seem to allow two information structure articulations in section 7.3.4.

153 Note also that if one accepts that sentences such as (242) and (245) can have a focus- background interpretation, then one must acknowledge that not all focus-background il y a clefts have specificational semantics, as Lambrecht (2001) seems to imply.

186 Chapter 7 | The information structure of il y a clefts

7.3.3 Focus + background

Previous studies of il y a clefts do not often mention the focus-background articulation or semantically specificational il y a clefts (but see Lambrecht 2001; Verwimp & Lahousse 2016). However, I already showed in chapter 4.1.2 that the corpus data contain a non-negligible number of specificational il y a clefts. These clefts are predominantly (although not always: see next section154) associated with the focus- background articulation, in which (part of) the clefted element is focal, whereas the relative clause is backgrounded. In other words, these sentences display narrow focus on the clefted element. For example, the relative clause in (246) expresses a variable – cars cheaper than €10.000 – which is given in the preceding question. The clefted element is the focus of the sentence (it is the most salient part) and specifies a value for this variable – the Citroen c1.

(246) – A: Je recherche des modèles de voiture à acheter neuve moins de 10 000 euros, ou aller!?

‘I’m looking for new car models that cost less than €10.000, where should I go?’

– B: bonjours. il y a la citroen c1 qui est a moins de 10 000 euros.

‘Hello. There’s the Citroën C1 that costs less than €10.000.’ (YCCQA)

Similarly, in (247), the relative clause expresses the backgrounded variable free software programs that can open mp4 files, and the clefted element specifies the salient/focal value VLC. In (248), the backgrounded variable expressed by the relative clause is (old) people who come to neighborhood meetings, and the clefted element provides the salient/focal value Mrs. Blumenthal.155

154 English it clefts with specificational semantics can also correspond to different information structure articulations, see e.g. Declerck (1984); Bicler & Davidse (2008). However, the “specificational function works best when the variable [expressed by the relative clause] is given and can be informationally presupposed” (Bicler & Davidse 2008:265). 155 In il y a clefts such as (246) - (248), the variable can be seen as the topic of the sentence, while the value is the comment about this topic. For instance, speaker A in (246) raises the question “which car is less than €10.000?” as a topic, and speaker B presents the value the Citroën C1 as a comment about this topic (see Furukawa 1996b; Lambrecht & Michaelis 1998; Hartmann 2006;

Francez 2007; Lahousse & Borremans 2014:829 for similar analyses of related constructions).

Chapter 7 | The information structure of il y a clefts 187

(247) – A: Avec quels logiciels libres et gratuits puis-je lire des fichiers vidéos encodés en MP4?

‘With which freeware programs can I open video files encoded in MP4?’

– B: La meilleure sources de logiciel libre est [www]. En ce qui concerne la lecture de mp4 il y a VLC qui les ouvrent très bien.

‘The best source for freeware programs is [www]. As for opening mp4 files, there’s VLC that opens them very well. (YCCQA)

(248) C’est pas les jeunes les gens jeunes qui du quartier qui viennent au conseil, c’est toutes les personnes âgées (...) c’est les personnes âgées alors comme y a madame Blumenthal qui vient. ‘It’s not the young the young people who from the neighborhood who go to the neighborhood meeting, it’s all the old people (...) it’s the old people so like there’s Mrs. Blumenthal who goes there. (CFPP)

Note that the focal referents introduced by il y a in the corpora are not necessarily referentially new, in accordance with previous studies showing that foci (defined as the salient part of the sentence) can be discourse-given (see section 7.1.3). For instance, in (249), lui ‘him’ refers to mon fils Stewan ‘my son Stewan’, mentioned in the preceding discourse. The relative clause also expresses discourse-given information: we already know that the speaker’s son is called ‘Stewan’. What is salient and new information in this cleft is the fact that the speaker has the impression that only he has that name. This analysis is represented in (250).

(249) Connaissez-vous la signification, l'origine du prénom de mon fils STEWAN (prononcer STIVANE), Merci? j'ai l'impression qu'il n’y a que lui qui porte ce prénom. ‘Do you know the meaning, the origin of my son’s first name ‘Stewan’ (pronounce Steevahn), thank you? I have the impression that only he has that name.’ (Lit. ‘I have the impression that there’s only he who has that name’) (YCCQA)

188 Chapter 7 | The information structure of il y a clefts

(250) Il n’y a [que lui]FOCUS [qui porte ce prénom]BACKGROUND

There’s [only him]FOCUS [who has that name]BACKGROUND

Besides examples in which the whole clefted element constitutes the focus, there are also instances of il y a clefts in which only part of the clefted element is focal. These are instances that I referred to as specificational il y a clefts of the “quantifying type” in chapter 4.2.2, on a par with the same type of there cleft in English (see among others Huddleston 1971 and Davidse 2000). In such cases, it’s the quantification within the clefted element that is focal. For instance, what is focal in (251) is the numeral 999.999, which is contrasted with respect to the determiner ‘1’ in 1 fille ‘a/one girl’ in the preceding question, i.e. ‘1’.156 Hence, the information structure of (251) can be visualized as in (252).157

(251) Specificational il y a cleft of the “quantifying” type – A: Y a t'il 1 fille sympa qui cherche 1 mec pas chiant?

‘Is there a nice girl who’s looking for a guy who’s not a pain in the ass?’

– B: Il y a environ 999'999 de filles qui cherche un mec pas chiant

‘There are about 999.999 girls who are looking for a guy who’s not a pain in the ass.’ (YCCQA)

(252) Il y a [environ 999.999]FOCUS [de filles qui cherchent…]BACKGROUND

There are [about 999.999]FOCUS [girls who are…]BACKGROUND

156 The digit ‘1’ is regularly used in informal written French as an equivalent of the indefinite determiner ‘un’ (the numeral ‘one’ and the indefinite determiner ‘a’ are both encoded by un in French). It is very unlikely that the first interlocutor in (251) intends ‘1’ to be interpreted as a numeral, as in ‘is there one girl who…?’. 157 This example can be seen as an instance of corrective focus (not 1, but rather 999.999).

Chapter 7 | The information structure of il y a clefts 189

7.3.4 Ambiguous focus (multiple focus?)

In what follows I discuss a few il y a clefts that do not easily fall within one information structure class or another. Information structure ambiguity was already shown to be relevant for the analysis of il y a... qui sentences with semantically light relative clauses (section 7.3.2), but it also has consequences for the distinction between focus- background and multiple focus il y a clefts, as we will see shortly. 158

In a considerable number of instances, the relative clause contains an element that can be interpreted as contrastive with respect to another element, but it is unclear whether it should be interpreted as contrastive. Given that contrast is associated with focus (see section 7.1.4), it is unclear whether the relative clause contains a focal element or not. Consider for instance the underlined il y a cleft in (253). The adjective originales ‘original’ can be seen as explicitly contrastive with respect to the preceding remark that nothing happens in the movie ‘Bandits’.

(253) – A: Quelle est le pire film qui a été fait? ‘What is the worst movie ever made?’

– B: Bandits avec Bruce Willis. C’es tellement c-on ce film ! Et y a rien qui se passe. Y a que les dernières cinq minutes qui sont originales…

‘Bandits with Bruce Willis. That movie is soo bad! And nothing happens. Only the last five minutes are original.’

(Lit. ‘There’s only the last five minutes that are original.’) (YCCQA)

Slightly reformulating the prior context, the contrastive reading corresponds to the representation in (254). The underlined constituents form a first contrast, and the constituents in boldface form a second contrast. In this case, the adjective originales ‘original’ could receive a rising contour if it were uttered in spoken French/English.

158 It has been acknowledged that it clefts can have a ‘multiple focus’ articulation with a first focus on the clefted element and second one on the relative (Carlson 1983; Huber 2002, 2006; Dufter 2009b; Hedberg 2013; Büring 2016; De Cesare et al. 2016:44). See chapter 8.1.3 about such clefts.

190 Chapter 7 | The information structure of il y a clefts

(254) Presque TOUT LE TEMPS, y a rien qui se passe, y a [que LES DERNIÈRES CINQ

MINUTES]FOCUS qui sont [ORIGINALES]FOCUS ‘Almost THE WHOLE TIME, NOTHING HAPPENS, [only the LAST FIVE

MINUTES]FOCUS are [ORIGINAL]FOCUS’

This kind of interpretation resembled what Krifka (2008) refers to as ‘foci that highlight parallels’, illustrated in (255).

(255) MAry stole the COOkie and PEter stole the CHOcolate. (Krifka 2008:252)

However, it is also possible that the adjective originales ‘original’ is not intended to be interpreted as explicitly contrastive. Under this reading, the relative clause is not focal, and if uttered in a spoken context, the adjective wouldn’t receive prosodic stress. This is visualized in (256):

(256) Presque TOUT LE TEMPS, y a rien qui se passe, y a [que LES DERNIÈRES CINQ

MINUTES]FOCUS [qui sont originales]BACKGROUND ‘Almost THE WHOLE TIME, nothing happens, [only the LAST FIVE

MINUTES]FOCUS [are original]BACKGROUND’

Since the original example in (253) was extracted from a written corpus and there are no other contextual clues to indicate which interpretation is more likely, the two analyses are possible: multiple focus (254) or focus-background (256).159

Summing up, examples such as (253) show once again that there are limits to the application of information structure criteria to corpus data (see e.g. Ritz et al. 2008; Marzo et al. 2012; Stede & Mamprin 2014:1722). I believe there are several ways of looking at such instances that are ambiguous between a multiple focus and a focus- background interpretation, as I also argued for the analysis of aspectually underspecified relative clauses (see chapter 3.4.3). Primo, it may be the case that different interlocutors assign different properties to the exact same sentence appearing in the same context, according to factors that vary from individual to individual. This cannot be investigated by means of corpus data. Secundo, it is possible

159 It may be the case that such ambiguities are less likely to occur in spoken French, because contrast is often accompanied by prosodic stress.

Chapter 7 | The information structure of il y a clefts 191

that people don’t assign a specific information structure value to each and every constituent of a given sentence when they hear or read it: certain parts may remain underspecified provided they do not have any influence on the correct (semantic and pragmatic) interpretation of the utterance. This could be the case in the examples discussed above: the difference between the articulations (multiple focus or focus- background) does not seem crucial for a correct understanding of the utterance in its context. In other words, the different analyses of the utterance do not lead to fundamentally different updates of the Common Ground, and hence they don’t lead to communicative misunderstandings. I speculate that the assignment of information structure properties to a sentence is indeed not as strict as we might sometimes believe, i.e. that not all interlocutors process the same sentence in the same way, and that underspecified analyses are possible. I leave this issue for future research.

7.3.5 Contrastive topic + comment

The corpus data also contain several instances in which il y a introduces a contrastive topic followed by a comment. These topics may either be contrastive spatiotemporal topics (section 7.3.5.1) or contrastive aboutness-topics followed by a contrastive relative clause (section 7.3.5.2).

7.3.5.1 Contrastive spatiotemporal topic + comment

In the il n’y a que cleft in (257), the clefted element has already been evoked: ‘Turkey’ is the subject of the whole newspaper article and is explicitly mentioned in the context preceding the cleft (son pays ‘his country’; the pronoun y ‘there’). The relative clause, however, does not express information that can be inferred from the previous context: it conveys discourse-new information. In other words, on the referential level, this cleft has a given-new interpretation.

192 Chapter 7 | The information structure of il y a clefts

(257) [Context: newspaper article about Turkey] Mais l'essentiel de la marque qu'Atatürk a imposée à son pays semble indélébile. Personne d'ailleurs n'y parle sérieusement de remettre son œuvre en question. Et il n'y a qu'en Turquie que le dirigeant du parti intégriste a pu faire alliance, sans que personne parmi ses partisans s'en choque, avec une femme aussi visiblement occidentalisée que Tansu Çiller, économiste made in USA.

‘But the essence of the mark left by Atatürk on his country seems inerasable. Incidentally, nobody there speaks seriously of questioning his work. And there’s only in Turkey that the leader of the integrist party can start an alliance, without any of his partisans being shocked, with a woman who is as clearly westernized as Tansu Çiller, economist made in the USA.’ (Le Monde)

Let us now consider the relational dimension of information structure. Given the fact that the clefted element in (257) presents a frame for the predication expressed by the relative clause, this il y a cleft can be analyzed as an instance of the articulation ‘spatiotemporal topic + comment’. Moreover, since the topic is modified by ne… que ‘only’, it is a contrastive topic: only in Turkey indicates a contrast with respect to all other possible countries (see chapter 8.1.4 about similar c’est clefts).160 Hence, the articulation expressed by (257) can be visualized as in (258).161

(258) Il n’y a [qu’en Turquie]CONTR. TOPIC [que le dirigeant a pu…]COMMENT

There’s [only in Turkey]CONTR. TOPIC [that the leader can…]COMMENT

160 The same type of cleft can also be found in English (Kristin Davidse, p.c.): (i) (…) when I arrived in Paris I lived in a 11m2 flat in Montmartre (…) and once a week, a knife grinder passed in the street with a bell, shouting “grinder grinder!!” (…) there’s only in Paris that you can still see this kind of scene. 161 The il y a… où sentence in (i) also displays a spatiotemporal topic + comment articulation. The relative clause expresses new information about the evoked clefted element in France. (i) Toute la difficulté de l'exercice annuel de l'Apec tient cependant à la mise au point d'une définition de la notion de cadre qui soit commune aux différents pays européens. Cette notion est, on le sait, extrêmement hexagonale: il n’y a qu’en France où elle correspond à la fois à un statut (…), à un mode de couverture sociale (…), mais aussi à un certain prestige social. ‘The whole difficulty of the APEC’s annual exercise is, however, the establishment of a definition of the notion ‘cadre’ that applies to the different European countries. This notion is, as is well known, extremely French: only in France does it correspond simultaneously to a statute (…), to a type of social support (…), but also to a certain social prestige.’ (Lit. There’s only in France where it corresponds to…’) (Le Monde)

Chapter 7 | The information structure of il y a clefts 193

Recall that there is a syntactic constraint on il y a... qui sentences that makes them unable to introduce non-(pro)nominal constituents, unless the particle ne… que ‘only’ is present (chapter 5.3.1). It follows that if an il n’y a que cleft introduces a spatiotemporal topic, this topic will always be contrastive, because it is necessarily modified by ne… que ‘only’. In other words, when they appear as clefted elements, spatiotemporal topics in il n’y a que clefts will always be topical and focal simultaneously.162

7.3.5.2 Contrastive aboutness-topic + contrastive comment

In a few il y a clefts, both the clefted element and the relative clause are explicitly contrastive with respect to two other linguistic expressions present in the context. I will argue that the clefted element in such instances is a contrastive aboutness-topic (see section 7.1.4 about such topics).

Consider for instance (259). A first contrast is evoked between on ‘we’ on the one hand and plein de petites jeunes ‘plenty of youngsters’ on the other. This contrast is further emphasized by the dislocated nous ‘us’. Furthermore, there is a second contrast between avoir de la chance, avoir des employeurs corrects ‘being lucky, having good employers’ on the one hand and se faire avoir ‘getting ripped off’ on the other.

(259) « Nous, on a de la chance, on a des employeurs corrects. Mais il y a plein de petites jeunes qui se font avoir », poursuit Lina. ‘“Us, we’re lucky, we have good employers. But there are plenty of youngsters who are getting ripped off”, continues Lina.’ (Le Monde)

162 Not all spatiotemporal clefted elements in il y a clefts are topics, as shown in (i). In this example, the spatial expression ‘(only) at home’ is focal and specifies a value for the variable ‘places where I sit down on the toilet seat’. (i) – A: Dans les toilettes publiques es ce que vous faites pipi assises sur la lunettes ou pas? ‘Do you sit down on the toilet seat in public places or not?’ – B: y a que chez moi que je m'assois ‘I only sit down at home .’ (YCCQA)

194 Chapter 7 | The information structure of il y a clefts

Similarly, in the underlined cleft sentence in (260), toi ‘you’ is contrasted with un pilote ‘a pilot’, and the relative clause qui n’arrive même pas… ‘who can’t even…’ is contrasted with the relative clause qui arrive à poser… ‘who manages to land…’.

(260) – A: Atterissage de l'airbus à New-York? [video link] Le co-pilote n'etait t-il pas un arabe (lol) et traiter les arabes de terroriste.

‘Landing of the airbus in NY? [link] Wasn’t the co-pilot an Arab (lol), and to treat Arabs like terrorists!’

– B: il y'a un pilote qui arrive a poser un avion sur l'eau et évite la perte de 160 personne, et d'un autre coté il y'a toi qui n'arrive même pas a faire une phrase compréhensible!!!

‘There’s a pilot who manages to land a plane on the water and avoid the deaths of 160 people, and on the other hand there’s you who can’t even write a comprehensible sentence!!!’ (YCCQA)

As a final illustration, consider (261). Here, le peuple ‘the people’ is contrasted with l’establishment ‘the establishment’ and tous ces salopards ‘all those assholes’. Moreover, the relative clause qui a toujours été… ‘who have always been…’ is contrasted with the two immediately preceding relative clauses (qui me déteste ‘that hates me’ and qui pensent… ‘who think…’). The contrast is reinforced by the stressed pronoun lui ‘him’.

(261) [Salman Rushdy and the writer of the article are walking down the street. A man recognizes Salman and says something encouraging to them.] Et Salman : "Je suis content que vous voyez ça... Il y a l'establishment qui me déteste. Il y a tous ces salopards qui pensent que je me suis fait de la pub avec la fatwa. Mais il y a le peuple qui, lui, a toujours été formidable avec moi.”

‘And Salman says: “I’m happy that you see that [the stranger being friendly to Salman]... There’s the establishment that hates me. There’s all those assholes who think I just made publicity with the fatwa. But there’s the people who, them, have always been fantastic to me.’ (Le Monde)

Chapter 7 | The information structure of il y a clefts 195

I argue that the contrastive clefted elements in (259) - (261) constitute contrastive aboutness-topics and that the relative clauses represent the (contrastive) comment. Such il y a clefts are on a par with c’est clefts such as (262), which Dufter (2006) analyses as ‘contrastive topic + comment’. 163 This analysis is intuitively plausible, since the relative clauses seem to express properties about the referents.164

(262) Contrastive topic + comment c’est cleft Au fond, c’est toi qui fais les conneries, et c’est moi qui suis puni.

‘Basically, it’s you who’s doing stupid things, and it’s me who’s getting punished.’

[Clavel, La Maison des autres (1962), Dufter 2006:51, my translation)

The fact that there are only 7 instances (on a total of 576 il y a clefts, i.e. 1%, see Table 28) that display a double contrast indicates that this is a marginal phenomenon. This low frequency, together with the fact that the topics are contrastive and hence marked, indicate that these few examples should not be taken as evidence against the hypothesis that il y a clefts are used to avoid a topic-comment interpretation (section 7.2.2).165

I believe that the contrastive nature of these constituents is in fact the licensing factor that enables them to appear in il y a clefts, which normally do not contain topics. In Leonetti’s (2013) terminology, contrast can be seen as a ‘survival strategy’ for weak topics in a ‘hostile environment’. Leonetti argues that contrast enables several types of constituents (e.g. bare plurals and indefinites) to appear in positions that they are normally banned from (in his analysis: preverbal subject position and topic positions respectively).166 He argues that “contrast operates as a last resort interpretative mechanism that is able to “rescue” a nominal placed in a “hostile” environment, i.e. an environment where the nominal does not fit in

163 These il y a clefts also resemble Hedberg’s (2013) ‘emphatic repetition cleft’, illustrated in (i). (i) The women who went were almost all married. But it was husbands who were captured by the glowing descriptions of the West, wives who were skeptical. (…) (Hedberg 2013:242) 164 In terms of the file-card metaphor, the clefted element (e.g. toi ‘you’ in (260)) would be the topic card on which new information is added (i.e. the location for information update). 165 One could at best state that these examples show that the hypothesis that il y a clefts never introduce topics characterizes a strong quantitative tendency, rather than an absolute rule. 166 See Leonetti (2013, 2016b) for more relevant examples of (pro)nominal expressions in hostile environments that are ‘rescued’ by contrast.

196 Chapter 7 | The information structure of il y a clefts

adequately for some reason” (Leonetti 2013:99). Leonetti speculates that the reason why contrast is able to increase the acceptability of certain nominal expressions in specific positions can be captured in Relevance-theoretic terms. For instance, placing an indefinite in a position that is normally associated with topics leads to more processing effort, because indefinites do not indicate the location of information update as clearly as regular (definite) topics do. Hence, the hearer infers that there must be some additional motivation why the speaker uses an indefinite in this hostile environment: contrast. In other words, contrast is the “reward” for the extra processing efforts, because it enriches the interpretation of the sentence (Leonetti 2013:114).167

In my view, the same line of reasoning applies to the il y a clefts discussed above: il y a clefts are a hostile environment for topics, therefore the few topics that do appear in il y a clefts are contrastive.

To conclude, in only very few cases, the clefted element of the il y a cleft can be analyzed as topical in terms of ‘aboutness’. All of those cases display a double contrast between (i) the clefted element and some salient referent in the context and (ii) the property expressed by the relative clause and some other contextually available property. In line with Leonetti (2013), I argued that contrast (which combines topical and focal properties) is the survival strategy for these topics in an environment that is normally hostile with respect to aboutness-topics, namely the il y a cleft. 168

167 See also Lahousse et al. (2014), who argue on the basis of corpus data that contrast licenses it clefts and c’est clefts in temporal clauses. 168 I would also like to argue that the cases discussed in this section reveal a problematic aspect of the definition of ‘thetic sentences’. The crucial thing to realize is that ‘thetic sentences’ can be defined in two ways that often seem to be conflated (see section 7.1.5). The first definition relates to the focality of the constituents in the utterance, whereas the second definition is about the non- topicality of the grammatical subject. However, to the extent that contrastive aboutness-topics are both focal and topical (e.g. Krifka 2008; Büring 2016), the two definitions do not delimit exactly the same phenomenon: all the constituents in the examples in (259) - (261) are focal, despite the fact that the clefted elements are simultaneously topical. This is illustrated in (i). (i) Topic + comment & all-focus simultaneously (= (260))

il y a [toi]TOPIC+FOCUS [qui n'arrive même pas à….]COMMENT+FOCUS

there’s [you]TOPIC+FOCUS [who can’t even….]COMMENT+FOCUS Therefore, (i) does have an all-focus articulation, but not an all-comment articulation. Consequently, it is unclear whether (i) should be seen as a thetic sentence or not.

Chapter 7 | The information structure of il y a clefts 197

7.4 The non-topic marking hypothesis: evaluation

Recall that according to the non-topic marking hypothesis (section 7.2.2), clefts in general are used to prevent the hearer from construing a topic-comment interpretation. This hypothesis has received criticism from researchers working on it clefts and their cross-linguistic equivalents. The question I tried to answer is whether the hypothesis does or does not hold in the case of il y a clefts.

I argue that in general, the corpus data support the hypothesis, although it should be nuanced in some respects. The main piece of evidence in favor of the hypothesis is that 99% of the clefted elements in il y a clefts are non-topical. Secondly, the few instances in which the clefted element does seem to be topical are very marked: I argued that they require an explicitly contrastive context in order to be licensed. Hence, the fact that they need to be contrastive actually supports the view that il y a clefts are hostile environments for topical constituents.169 Another piece of evidence is the fact that it is very difficult, if not impossible, to construe an il y a cleft with a non-contrastive aboutness topic (263):

(263) – A: Que fait le requin ? / Qu’en est-il du requin ? ‘What is the shark doing? / What about the shark?’

– B: # Il y a le requin/lui qui mange un caniche.

‘There’s the shark/it that’s eating a poodle.’

In sum, everything seems to point in the direction that il y a clefts function as a processing cue that announces to the hearer that a non-topical (and focal) constituent is coming. However, it is worth considering what the origin of this function is. According to Lambrecht (2001), it is the cleft format that is responsible for marking the clefted element as non-topical: the whole cleft format indicates to the hearer that the clefted element is non-topical. This is where I would like to propose a different view: building on a large line of research on existential sentences, I argue that it is not the entire cleft format that prevents the hearer from construing a topic-comment

169 Moreover, if il y a clefts are taken to be a prototype category (see chapter 3.6), one can argue that the frequencies of the different articulations (all-focus > focus-background > multiple focus > contrastive topic-comment) correspond to a cline ranging from ‘most central’ (all-focus) to ‘most peripheral’ (contrastive topic-comment) instances of il y a clefts in terms of information structure.

198 Chapter 7 | The information structure of il y a clefts

interpretation, but the existential expression il y a. Indeed, it has been argued that existential sentences cannot introduce topical referents (Bentley 2013; Bentley et al. 2013, 2015).170 This is illustrated by the Bellunese Italian example in (264): the question in (264a) makes the towels topical. If one wishes to answer this question, it is infelicitous to make use of an existential sentence such as (264b). Instead, a canonical Subject Verb sentence must be used (264c).

(264) Bellunese Italian a. E-li andé i sugaman?

Be.3pl-3m.pl.sbj.cl where the towels

‘Where are the towels?’

b. #L’ é i sugaman te la casèla.

3m.sg.sbj.cl be.3sg the towels in the drawer

‘There are the towels in the drawer.’

c. I sugaman i é te la casèla.

the towels 3m.pl.sbj.cl be.3pl in the drawer

‘The towels are in the drawer.’

(Bentley 2013:686)

The same holds for the English translations in (264) and for French (see (265)).

(265) – A: Où sont les serviettes? ‘Where are the towels?’

– B: # Il y a les serviettes dans le tiroir.

# ‘There are the towels in the drawer.’

In other words, existential sentences and il y a clefts share the function of introducing a non-topical referent. The component that they have in common is the sentence- initial expression il y a. Therefore, I argue that it is this expression that lies at the origin

170 In a similar vein, existential sentences only introduce discourse-new constituents (the so-called ‘Novelty Condition’, e.g. Breivik 1981; Abbott 1993; Ward & Birner 1995; McNally 1997; Francez 2007; Breivik & Martínez-Insua 2008; Leonetti 2008, 2016a, see chapter 6.1.

Chapter 7 | The information structure of il y a clefts 199

of the non-topic marking interpretation function rather than the entire cleft format (i.e. including the relative clause).171

In the next section, I take the parallels between the two types of sentences (il y a clefts and il y a existential sentences) one step further by showing that existential sentences without relative clauses can express several of the same information structure articulations as il y a clefts. What this shows is that the information structure properties of il y a clefts are not unique to the cleft format: il y a clefts share these properties with il y a... qui sentences more generally.

7.5 Comparison with existential il y a sentences

In this section, I show that il y a sentences without relative clauses (henceforth referred to as ‘existential sentences’172) can also express the following articulations: (i) all-focus (section 7.5.1), (ii) focus-background (section 7.5.2) and (iii) contrastive topic- comment (section 7.5.3). A summary is presented in section 7.5.4.

7.5.1 All-focus

In the most prototypical kind of existential sentence such as (266), the whole sentence is presented as discourse-new in an ‘out-of-the-blue’ context, i.e. thetic existentials (Abbott 1993). In this sense, they are on a par with thetic il y a clefts such as (267).

(266) Il y a un grand requin blanc dans la piscine ! ‘There’s a great white shark in the swimming pool!’

171 The notion of ‘processing cue’ may also be related to the term ‘projector’, as it is used in Jullien (2014) and Pekarek Doehler (2015). However, these two authors do not look at the expression ‘il y a’ only: they consider the whole main clause ‘il y a NP’ as a segment that projects “an upcoming stretch of talk of indeterminate length and form” (Pekarek Doehler 2015:193). Furthermore, they don’t focus on information structure properties. Their analyses and mine can be seen as complementary. 172 For a diachronic corpus analysis of French existential sentences introduced by il y a, see Coy & Umbreit (2015); Coy (2016a, 2016b, 2017).

200 Chapter 7 | The information structure of il y a clefts

(267) Il y a un requin qui mange un caniche ! ‘There’s a shark that’s eating a poodle!’

7.5.2 Focus + background

There is a large body of literature about existential sentences with a so-called ‘list reading’, also referred to as ‘contextualized existentials’ (e.g. Milsark 1977; Rando & Napoli 1978; Abbott 1993; Dobrovie-Sorin & Beyssade 2004; Beaver et al. 2006; Francez 2007; Leonetti 2008; Cruschina 2012; Meulleman 2012a; Bentley 2013; Coy 2016a, 2017). In such existential sentences, the clefted element is interpreted as a value for some contextually salient variable, i.e. they are semantically specificational. For instance, the Italian sentence in (268) is a felicitous answer to the question “What does one find at the Piazza della Signoria?”.

(268) C’è la statua di Michelangelo, in Piazza della Signoria. ‘There is the statue by Michelangelo(,) in Piazza della Signoria’

(Leonetti 2008:140)

On an information structure level, the referent introduced by c’è is focal (i.e. salient). The prepositional phrase, on the other hand, is not focal but backgrounded. Similar examples can be found in French. For instance, un chat ‘a cat’ is focal in (269), whereas the prepositional phrase au grenier ‘in the attic’ that follows un chat ‘a cat’ is not focal, due to the preceding question.173

(269) Q: Y avait-il quelque chose au grenier ? R: Il n’y avait qu’un chat au grenier. Q: ‘Was there something in the attic?’ R: ‘There was only a cat in the attic.’

(Van De Velde 2005:45, my translation)

The fact that list reading existentials such as (268) and (269) are semantically specificational and have a focus-background articulation should ring a bell: il y a clefts,

173 It should be noted that this is a constructed example. In an actual communicative situation, it may be more natural to drop the repeated prepositional phrase au grenier ‘in the attic’.

Chapter 7 | The information structure of il y a clefts 201

too, can be specificational and have a focus-background articulation (section 7.3.3).174 The only difference is the linguistic encoding of the variable/background: it is expressed by a relative clause in the case of a cleft and by a prepositional phrase in (268) and (269).175

Interestingly, existential sentences display several possibilities for expressing the variable/background. Firstly, the variable/background does not necessarily follow the introduced value, it may also precede it. This is illustrated in (270). Person A on the online forum raises the variable ‘cars that cost less than €10.000’. Person B first repeats the variable by means of a fronted PP, and then uses an existential sentence that introduces several values (the Logan at Renault Dacia, the C1 at Citroen…). Person C, on the other hand, uses a specificational il y a cleft in which the variable does not precede the value, but follows it in the relative clause.

(270) – A: Je recherche des modèles de voiture à acheter neuve moins de 10 000 euros, ou aller!?

‘I’m looking for new car models that cost less than €10.000, where should I go?’

– B: A moins de 10 000 euros, il y a chez RENAULT DACIA la Logan; chez CITROEN la C1, chez PEUGEOT la 107 et chez TOYOTA la Aygo.

‘For less than €10.000, Renault Dacia has the Logan; Citroen has the C1, Peugeot has the 107 and Toyota has the Aygo.

(Lit. ‘there is at Renault Dacia the Logan, at Citroën the C1…’)

– C: bonjours. il y a la citroen c1 qui est a moins de 10 000 euros.

‘Hello. There’s the Citroën C1 that costs less than €10.000.’

(YCCQA)

174 Recall, however, that not all specificational il y a clefts have a focus-background interpretation (section 7.3.4), and not all focus-background il y a clefts are semantically specificational (section 7.3.2). I did not find examples of existential sentences that have a focus-background interpretation but which are not specificational. On the other hand, not all specificational existential sentences contain a constituent that functions as the background (see examples (271) and (272)). 175 See Davidse 2000:1121 for a similar observation about the main clause in English there and it clefts and possessive clefts (I’ve got… ) and sentences without relative clauses introduced by those same main clauses. I will return to this is chapter 12.3.1.

202 Chapter 7 | The information structure of il y a clefts

Secondly, the variable may not be expressed at all, as is the case in (271). The variable (‘which woman to marry’) has already been evoked and is not repeated in any way: only the values are presented (Louise…).

(271) [Speaking about which woman to marry] Il y a la Louise, la Sylvaine, la Claudie

‘There is Louise, Sylvaine, Claudie.’

(G. Sand, La Mare au diable, in Coy 2016a:280)

Recall, finally, that the relative clause in focus-background il y a clefts can be omitted if it contains only discourse-given information, as in (272) (chapter 3.2.2).

(272) – Il n’y a personne qui monte. ‘Nobody is going up the stairs.’

– Si, il y a Pierre (qui monte l’escalier). ‘That’s not true, there’s Pierre (who’s going up the stairs).’

(Léard 1992:45, my translation)

The result of the omission of the relative clause is thus an existential sentence similar to (271). One could argue that such list reading existential sentences are ‘shortened clefts’ with unexpressed relative clauses. For instance, Belletti (2008) speaks of ‘reduced cleft sentences’ when a speaker uses ‘it is John’ rather than ‘it is John who left’ in answer to the question ‘Who left?’, and Doetjes et al. (2004:538) refer to such instances as ‘truncated clefts’ (see also Büring 1998; Hedberg 2000). However, in light of Occam’s razor, a hypothesis that posits an unrealized structure (i.e. the ‘shortened/truncated cleft analysis’) is less desirable than a hypothesis that does not. Or, as Birner et al. (2007) put it: “the burden of proof” lies with analyses that postulate an unrealized structure. In her analysis of English that clefts, she presents several arguments against such a truncation hypothesis, concluding that the similarities between such clefts and equative that sentences without relative clauses are due to functional overlap, and one should not be derived from the other (Birner et al.

Chapter 7 | The information structure of il y a clefts 203

2007:319).176 I agree that it’s more straightforward to see focus-background il y a clefts as list reading existential sentences that have an extra relative clause, rather than stating that list reading existentials always have an unexpressed relative clause.

To conclude, if the speaker wishes to express a proposition that is semantically specificational, and in which the value that is assigned to the variable is focal (for instance in an answer to a question like “Which cars are cheaper than €10.000?” in (270)), he/she has a variety of different means at his/her disposal. Il y a clefts with a focus-background articulation are but one of those means: existential sentences without relative clauses can also be used to encode the proposition. The difference between the two is that the variable is (optionally) repeated as a sentence-final or sentence-initial prepositional phrase in the list reading existentials, while it is encoded by the relative clause in specificational/focus-background il y a clefts.

7.5.3 Topic + comment (double contrast)

Recall that several il y a clefts display a double contrast (1% of the data, section 7.3.5.2). A very similar type of existential sentence can be found, illustrated in (273)- (275). To the best of my knowledge, such existential sentences have not been analyzed before.

In all three cases, two or more referents are contrasted with each other (les oiseaux ‘the birds’, les hiboux ‘the owls’ etc. – moi ‘me’ in (273); toi ‘you’ – eux ‘them’ in (274); les autres ‘the others’ – moi ‘me’ in (275)).177 Moreover, all three existential sentences in (273)-(275) are either explicitly or implicitly followed by a second contrast, namely the property that sets the contrasted referents apart from each other: sleeping versus eating strawberries in (273), celebrating Candlemas in a normal way versus celebrating it with pancakes on your face, i.e. not in a normal way in (274) (the picture above this message shows why they are not “normal”: the people have pancakes on their faces) and saying “GREAT” versus saying “Aaargh” in (275).

176 I present her analysis in more detail in chapter 12.3.1. 177 Note that Y’a toi qui fête la chandeleur in (274) is a cleft sentence.

204 Chapter 7 | The information structure of il y a clefts

(273) A cette heure ci les oiseaux dorment, les hiboux hibernent, les pingoins dorment aussi, les humain(e)s dorment aussi <3 Et puis ya moi : Allonger dans mon lit les cheveux tel Aicha Kendicha en #train de manger des fraises ^^ ‘At this hour the birds are sleeping, the owls are hibernating, the penguins are also sleeping, human beings are also sleeping <3 And then there’s me: lying in my bed, my hair like Aichi Kendicha’s, eating strawberries ^^’ (www)

(274) [No prior context, introducing a picture of three people and a dog with pancake masks on their faces] Y'a toi qui fête la chandeleur normalement, puis y'a eux...

‘There’s you who celebrates Candlemas in a normal way, then there’s them... (www)

(275) L’incroyable positive attitude des américains: J’ai tout de suite été frappée par le fait de voir à quel point ils sont positifs et enthousiastes pour tout. Tout est « AWSOME », « AMAZING », « GREEEAAT ». Ca donne souvent des situations du genre « vous avez 5 pages à rendre pour demain » les autres: « GREAT! » et puis il y a moi « Aaargh ». ‘The incredibly positive attitude of the Americans: I was immediately struck by the fact that they are so positive and enthusiastic about everything. Everything is “AWESOME”, “AMAZING”, “GREEAAT”. That often leads to situation like “you’ve got 5 pages to hand in for tomorrow” the others: “GREAT!” and then there’s me “Aaargh”.’ (www)

The difference between the doubly contrastive existential sentences in (273) - (275) on the one hand and the doubly contrastive il y a clefts in section 7.3.5.2 is the way in which the second contrast is expressed: in the cleft instances, the second contrast is encoded by the relative clause, in (273), it is encoded by a participial clause, in (274) it is not encoded linguistically and has to be inferred by the reader from the discourse context, and in (275), it is encoded by the quote “Aaargh”. 178

178 Existential sentences of this type can also be found in English, as in (i): (i) She said there were people struggling for their lives and then there's you -- you have all these opportunities and you're throwing it all away. (www) Here, two referents (people – you) and two properties (struggling for their lives – throwing away opportunities) are contrasted. See also chapter 10.6.5 about similar c’è sentences in Italian.

Chapter 7 | The information structure of il y a clefts 205

Clearly, these existential sentences perform the same discourse function, and the properties that are expressed about the referents (e.g. me and “Aaargh” in (275)) can be seen as comments about (contrastive) topics. This can be represented as follows:

(276) Then there’s [me]CONTR. TOPIC : [lying in my bed…]COMMENT

(277) Then there’s [me]CONTR. TOPIC [“Aaargh”]COMMENT

(278) Then there’s [them]CONTR. TOPIC [picture evoking abnormal behavior]COMMENT

It should be noted that ‘aboutness’ is usually seen as a relation that holds between linguistic constituents inside the same sentence, often between a nominal phrase (Klaus in (279)) and a verbal phrase (like bananas). The comments in (276) - (278) show that information structure notions can also be used in less canonical utterances.

(279) KlausTOPIC [likes bananas]COMMENT

Note, finally, that the presence of an explicit contextually evoked contrast in all these instances supports Leonetti’s (2016b) claim that contrast is a necessary licensing factor for the appearance of pronouns in existential sentences. I already argued in section 7.3.5.2 that contrast can be seen as a survival strategy for the aboutness-topics in the few double contrast il y a clefts found in the corpus data. Thus, the fact that those il y a clefts and the existential sentences discussed here both ‘need’ contrast constitutes another parallel between the two sentence types.

7.5.4 Summary

In the previous three sections, I showed that existential sentences introduced by il y a can express almost all of the same main information structure articulations and can appear in the same types of discourse contexts as il y a clefts. In many instances, the cleft relative clause has a function that can also be performed by another (non-) linguistic form, e.g. a prepositional phrase, a quote or a picture. On a functional level, il y a clefts should thus be seen as closely related to existential sentences: they do not

206 Chapter 7 | The information structure of il y a clefts

have unique discourse functions that existential sentences cannot perform. 179 Therefore, when analyzing their discourse functions, it is useful to see il y a clefts as a subtype of the broader category of il y a sentences, as is done by authors such as Lyons (1968); Van De Velde (2005) and Dobrovie-Sorin & Beyssade (2012).

7.6 Hypothesis: the link between ‘non-topics’ and ‘suboptimal subjects’

In this section, I comment on the link between the two hypotheses that I have defended so far concerning the motivations for the use of il y a clefts, as discussed in Chapter 6 and in this chapter. These can be summarized as follows:

(280) Generalizations (this chapter) a. Corpus data: il y a clefts never introduce non-contrastive topics

b. Hypothesis: il y a clefts are used to signal that the hearer should not expect a topical constituent in order to facilitate processing

(281) Generalizations (Chapter 6) a. Corpus data: il y a clefts always introduce constituents that would make suboptimal preverbal subjects, for a variety of reasons (discourse-new, indefinite, …)

b. Hypothesis: il y a clefts are used to avoid suboptimal subjects from appearing in preverbal subject position in order to facilitate processing

These generalizations can in fact be seen as two sides of the same coin. Two observations from previous studies help elucidate this link. Firstly, there is a general cross-linguistic tendency that preverbal subjects preferably encode topics, and this also holds for French180 (e.g. Lambrecht 1988a, 1994). In other words, if a constituent

179 The only exception seems to be the introduction of spatiotemporal topics, which are, however, very infrequent in the corpus data. 180 This is especially true of spoken French, which is argued to be more discourse-configurational than written French. Hence, spoken French makes extensive use of special syntactic constructions such as left-dislocations and clefts in order to mark the information structure status (topical or focal) of a constituent (Lambrecht 1988a, 1994; De Cat 2007; Klein 2012).

Chapter 7 | The information structure of il y a clefts 207

is non-topical, it will also be a suboptimal preverbal subject. Since practically all the clefted constituents in the corpus data are non-topical (Generalization 280a), it follows that they are also suboptimal preverbal subjects (Generalization 281a).

A second important piece of the puzzle is the following. It has been shown that the same factors that influence the acceptability of a constituent as a preverbal subject also influence the acceptability of a constituent as a topic: ideally, both subjects and topics are optimally ‘identifiable’ for the hearer (e.g. Bentley 2013). Therefore, properties such as discourse-givenness, definiteness, quantification and partitivity increase the ‘subject acceptability’ and the ‘topic acceptability’ 181 of a given constituent. Inversely, constituents that are low on the ‘topic-worthiness scale’ are also low on the ‘subject acceptability’ scale (e.g. indefinites). Consequently, we can further generalize that the constituents introduced by the il y a clefts in my corpus data display properties that make them suboptimal potential preverbal subjects and suboptimal potential topics.182

I propose that the two hypotheses come together in the following way. On a very general level, when the speaker has a certain proposition in mind, he/she often disposes of several syntactic configurations to package this proposition. The preference (especially in spoken French) to start the utterance with an identifiable, topical constituent can be seen as a type of pressure that pushes the speaker towards using the il y a sentence as a ‘tool’ in order to avoid the competing canonical SV structure.183 Since the speaker would not use this extra tool unnecessarily, the hearer is able to infer that there must be some special reason why the speaker didn’t simply start the sentence with a regular topic, as one would normally expect. This special

181 The more commonly used term is ‘topic-worthiness’, defined as “the potentiality that a given argument be realized as a topic” (Bentley 2013:685) 182 Notice, however, that this is different from stating that those constituents are not topics: being a topic or not is a binary distinction, whereas there are degrees to which a constituent is a good potential topic. 183 As stated in Dufter (2006): “In the unmarked, ‘canonical’ case, focus is located on the predicate (Lambrecht 1994) or the verbal phrase (…). Other focus-background articulations are pragmatically marked and typically require signaling by means of specific coding templates that are available in that language” (Dufter 2006:43) (“Im unmarkierten, ,kanonischen‘ Fall liegt Fokus auf dem Prädikat (Lambrecht 1994) beziehungsweise auf der Verbalphrase (...). Andere Fokus-Hintergrund- Gliederungen sind pragmatisch markiert und erfordern typischerweise eine Signalisierung mit den jeweils in einer Sprache dafür vorgesehenen Kodierungsmitteln”)

208 Chapter 7 | The information structure of il y a clefts

reason is to signal that the following constituent is not fit for the preverbal subject position, because it is not identifiable enough/not topical. In other words, the hearer interprets the use of the tool (il y a) as a processing cue. If the constituent that the speaker wants to introduce is topical, the tool cannot be used: this would be both unnecessary and confusing for the hearer. The only exception is a contrastive topic, which also bears a focus feature, and is therefore marked.

Notice, however, that there can be varying degrees of pressure on the speaker when he/she picks the il y a tool or some competing structure. Therefore, there is no absolute complementary distribution between the types of propositions that are denoted by il y a clefts and those that are denoted by canonical SV sentences in French.184 For instance, the type of register (written/spoken, formal/informal) is a crucial factor: in a spoken interaction, there is presumably a higher need to facilitate processing and thus to use processing cues such as il y a. Consider for instance the authentic utterance in (282), which appears at a news website, and in which a discourse-new, indefinite NP (a poodle) is a preverbal subject. This utterance is much more likely to be encoded by means of an il y a cleft in a spoken dialogue, because the preverbal subject is indefinite, non-topical and focal, and hence a suboptimal preverbal subject.

(282) Un caniche a attaqué un croisé-Stafford samedi matin à Luxembourg-ville. ‘A poodle attacked a crossbreed Stafford on Saturday morning in the city of Luxemburg.’ (www)

There are also other external factors that either push the speaker towards using il y a or towards a different encoding in case the constituent that would appear in subject position is discourse-new and indefinite. Consider for instance (283). The character Toto enumerates several events that he observes in the parking lot. The first event is encoded by an il y a cleft (il y a une voiture… ‘there’s a car…’), the second event is encoded by a canonical SV sentence (une ambulance… ‘an ambulance…’), and the

184 Similarly, Michaelis & Francis (2007) argue that English avoids lexical, non-topical preverbal subjects, but their corpus nonetheless contains examples of utterances with such ‘bad’ subjects. In their view, those utterances violate the Principle of the Separation of Role and Reference for the sake of speaker economy (i.e. less effort for the speaker) at the expense of the hearer’s economy (i.e. ease of processing).

Chapter 7 | The information structure of il y a clefts 209

third event is again encoded by an il y a cleft (il y a des gens… ‘there are some people…’).

(283) – Toto, pourquoi n'irais-tu pas sur le balcon. Comme ça tu nous raconteras tout ce qui se passe en bas! (…)

‘Toto, why don’t you go to the balcony. Then you can tell us about everything that’s happening below!’

– Toto: Il y a une voiture qui se fait remorquer par une dépanneuse dans le parking... Tiens, une ambulance vient de passer à toute vitesse... Il y a des gens qui rendent visite aux voisins d'à côté...

Toto: ‘There’s a car that’s getting towed by a tow truck in the parking lot… Hey, an ambulance just passed by at high speed… There are some people who are visiting the neighbors next door…’ (YCCQA)

Considered by itself, the second event displays the same properties as the first and the third event (e.g. non-topical indefinite constituent, all-focus), and could easily be encoded by an il y a cleft as well (there’s an ambulance that…). However, it is likely that this sentence is encoded by means of a canonical SV sentence in order to avoid too many il y a clefts from appearing one after the other. Summing up, there are both internal factors (e.g. degree of subject acceptability of the constituent, topicality) and external factors (e.g. register, stylistics) that influence the speaker’s (unconscious) choice of using an il y a sentence or a competing structure instead. Some more additional functions will be discussed in Chapter 9.

One promising way to approach the hypothesis about il y a as a processing cue would be to use experimental methods comparing il y a clefts and their canonical equivalents. For instance, one could perform a self-paced reading experiment in order to determine whether, all other things being equal, il y a clefts are processed faster than their SV equivalents. I leave this hypothesis for future research. However, similar experiments have already been conducted on expressions related to il y a, indirectly confirming the hypothesis given here for those structures. I will briefly present the methods and results of these studies in the next section.

210 Chapter 7 | The information structure of il y a clefts

7.7 Psycho-linguistic evidence about related constructions

The link between information structure and processing is relatively under-researched (see e.g. Reichle 2014 for an overview), but several studies have confirmed that (in)felicitous focus marking has an effect on sentence processing. Cowles (2003), for instance, presents the so-called Information Structure Processing Hypothesis (ISHP): “[A]s a separate component in the grammar, information structure is used during sentence processing by both speakers and comprehenders to facilitate language processing” (Cowles 2003:36). She provides results from several experimental designs to support this hypothesis. Other studies also confirm that unexpected or illicit focus structures evoke the same processing signatures in the brain as syntactic and semantic anomalies (cf. Cowles et al. 2007; Stohlterfoht et al. 2007; Reichle 2009). Furthermore, overt (prosodic or syntactic) focus marking has processing benefits for speech perception, memory and attention (see Sauermann et al. 2013 for an overview).

Two studies are particularly relevant for the purposes of this dissertation, because they investigate the equivalents of il y a in Dutch (er ‘there’) and in English (there is). Grondelaers et al. (2002), using corpus data and psycholinguistic findings from two self-paced reading experiments, show that Dutch er functions as a so-called accessibility marker (Ariel 1990) whose insertion in a sentence depends on how predictable the following NP is (see also Grondelaers et al. 2009). The experiment features four different conditions (284): firstly, the expression er is either present or absent (it is syntactically optional in all test sentences), and secondly, the NP is either predictable or unpredictable with respect to the prior context. The context sentence preceding the target sentence states that the luggage of a famous actress contains cosmetics. This context sentence is followed by the prepositional phrase in the beauty case (and, depending on the condition, by er ‘there’), and the following target NP is either a lipstick (284a,b) or a headset (284c,d). Since a lipstick is a type of cosmetics product, this NP is predictable or ‘congruent’, whereas a headset is unpredictable or ‘incongruent’ in a beauty case.

Chapter 7 | The information structure of il y a clefts 211

(284) [Context sentence preceding target sentence: De bagage van de beroemde actrice bevatte heel wat cosmetica.

‘The famous actress’s luggage contained quite a bit of cosmetics.’]

a. Predictable, + ER

In de beautycase zat ER een lippenstift en in een koffer zat ook wat rouge.

in the beauty case sat THERE a lipstick and in a suitcase sat also some rouge

‘In the beauty case there was a lipstick and in a suitcase there was also some rouge.’

b. Predictable, no ER

In de beautycase zat een lippenstift en in een koffer zat ook wat rouge.

in the beauty case sat a lipstick and in a suitcase sat also some rouge

‘In the beauty case there was a lipstick and in a suitcase there was also some rouge.’

c. Unpredictable, + ER

In de beautycase zat ER een koptelefoon en in een koffer zat ook wat rouge.

in the beauty case sat THERE a headphone and in a suitcase sat also some rouge

‘In the beauty case there was a headphone and in a suitcase there was also some rouge.’

d. Unpredictable, no ER

In de beautycase zat een koptelefoon en in een koffer zat ook wat rouge.

in the beauty case sat a headphone and in a suitcase sat also some rouge

‘In the beauty case there was a headphone and in a suitcase there was also some rouge.’

(Grondelaers et al. 2002:17, my translation and capitalization).

The findings of the experiment show that when the NP is unpredictable (a headset, (284c,d)]), reading times are significantly faster when er is present than when it is absent. The authors therefore state that er has a facilitating effect on reading times in the unpredictable condition. In sentences with predictable NPs (a lipstick (284a,b)), there is no statistically significant effect between sentences with er or without er. The authors conclude from this that er is best seen as a ‘warning sign’ that the speaker uses “because [he] knows that the hearer constantly generates hypotheses about

212 Chapter 7 | The information structure of il y a clefts

possible sentence continuations, and because he fears that the hearer will not (or not sufficiently) be able to infer or preconceive the subject in those sentences on the basis of the available contextual cues.” (Grondelaers et al. 2002).185 The parallel with the non-topic marking hypothesis is obvious: the best topics are those that are highly predictable.

The second relevant study is the visual-world eye-tracking experiment performed by Besserman (2014) (also reported in Besserman et al. 2015), which investigates English existentials and subject verb sentences such as the ones in (285).

(285) [Context sentence: A nurse was discussing new procedures with the doctor.] a. Existential condition: there was

There was a sad patient with a broken leg in the reception area, waiting for her turn.

b. Control condition: Subject Verb word order

That day a sad patient with a broken leg was in the reception area, waiting for her turn.

(Besserman 2014:18)

The procedure goes as follows: the test subjects look at a picture with the three characters that appear in the auditory stimulus. They hear the context sentence over their headphones, followed by the target sentence of either the existential condition (285a) or the control condition (285b). The context sentence always introduces two of the three characters in the picture, in this case a nurse and a doctor. The target sentence introduces the third character (the girl with the broken leg, i.e. the patient), which is the only hearer-new character. The authors measure the proportion of looks to the discourse-new entity (the girl with the broken leg). Besserman et al. (2015) conclude that there is a significant effect of sentence type (existential or Subject Verb word order) on the proportion of looks to the new entity: “participants were significantly more likely to look at the unmentioned entity in the Existential condition

185 My translation. “Omdat de spreker weet dat de hoorder voortdurend hypotheses genereert over mogelijke zinscontinueringen, en omdat hij vreest dat de hoorder het onderwerp in de betreffende zinnen niet (of niet voldoende) zal kunnen infereren of preconcipiëren op basis van de beschikbare contexthulpen.” (Grondelaers et al. 2002:19)

Chapter 7 | The information structure of il y a clefts 213

than in the Control condition” (Besserman et al. 2015:1). Therefore, they state that “Existentials function as indicators of newness during comprehension” (Besserman et al. 2015:1). Again, this conclusion is in favor of the non-topic marking hypothesis: discourse-newness is normally not associated with topics.

To summarize, several studies confirm that (the explicit marking of) information structure has an influence on processing. More specifically, Grondelaers et al. (2002) show that sentences with unpredictable subjects have faster reading times when er (the Dutch equivalent of French y ‘there’) is inserted than when er is left out, and Besserman et al. (2015) conclude that English existential sentences function as indicators of newness, a property that is highly related to non-topicality. In other words, both studies are in line with the hypothesis according to which existential expressions are used to facilitate processing by announcing to the hearer that something unexpected or new is coming. The connection between il y a clefts and existential sentences on the one hand and online sentence processing on the other is a promising subject for future psycholinguistic research.

7.8 Conclusion

In this chapter, I presented an in-depth analysis of the information structure properties of il y a clefts. There were two main goals. Firstly, to apply notions such as topic and focus on the corpus data so as to determine the information structure properties/ articulations of il y a clefts. Secondly, to evaluate the hypothesis that il y a clefts are used in order to mark the constituent they introduce as ‘non-topical’.

Information structure articulations

It turns out that in terms of information structure, il y a clefts are richer than previously assumed, because the range of different articulations they can express is not limited to ‘all-focus’ and ‘focus-background’. I discussed the following articulations:

214 Chapter 7 | The information structure of il y a clefts

i. Thetic il y a clefts with semantically informative relative clauses (i.e. all-focus: (286)) or with semantically light relative clauses (which can be considered as in- between instances of all-focus and focus-background (287)).

(286) – Toto, pourquoi n'irais-tu pas sur le balcon. Comme ça tu nous raconteras tout ce qui se passe en bas! (…)

‘Toto, why don’t you go to the balcony. Then you can tell us about everything that’s happening below!’

– Toto: Il y a une voiture qui se fait remorquer par une dépanneuse dans le parking...

Toto: ‘There’s a car that’s getting towed by a tow truck in the parking lot.’

(YCCQA)

(287) Lorsque je veux transmettre un courrier ou répondre à un mail, il y a un message d’erreur qui apparaît. ‘When I want to send a message or answer an email, there’s an error message that appears.’ (YCCQA)

ii. Focus-background il y a clefts (288): the whole or only part of the clefted constituent is focal; the relative clause is the background.

(288) – A: Avec quels logiciels libres et gratuits puis-je lire des fichiers vidéos encodés en MP4?

‘With which freeware programs can I open video files encoded in MP4?’

– B: La meilleure sources de logiciel libre est [www]. En ce qui concerne la lecture de mp4 il y a VLC qui les ouvrent très bien.

‘The best source for freeware programs is [www]. As for opening mp4 files, there’s VLC that opens them very well. (YCCQA)

Chapter 7 | The information structure of il y a clefts 215

iii. Il y a clefts in which the clefted element is focal, but it is unclear whether the relative clause contains a focal element or not, i.e. which are ambiguous between a focus-background and a multiple focus interpretation (289).

(289) – A: Quelle est le pire film qui a été fait? ‘What is the worst movie ever made?’

– B: Bandits avec Bruce Willis. C’es tellement c-on ce film ! Et y a rien qui se passe. Y a que les dernières cinq minutes qui sont originales…

‘Bandits with Bruce Willis. That movie is soo bad! And nothing happens. Only the last five minutes are original.’

(Lit. ‘There’s only the last five minutes that are original.’) (YCCQA)

iv. Il n’y a que clefts introducing a contrastive spatiotemporal topic (290).

(290) [Context: newspaper article about Turkey] Mais l'essentiel de la marque qu'Atatürk a imposée à son pays semble indélébile. Personne d'ailleurs n'y parle sérieusement de remettre son œuvre en question. Et il n'y a qu'en Turquie que le dirigeant du parti intégriste a pu faire alliance, sans que personne parmi ses partisans s'en choque, avec une femme aussi visiblement occidentalisée que Tansu Çiller, économiste made in USA.

‘But the essence of the mark left by Atatürk on his country seems inerasable. Incidentally, nobody there speaks seriously of questioning his work. And there’s only in Turkey that the leader of the integrist party can start an alliance, without any of his partisans being shocked, with a woman who is as clearly westernized as Tansu Çiller, economist made in the USA.’ (Le Monde)

v. Il y a clefts with a contrastive topic – comment articulation (291).

(291) « Nous, on a de la chance, on a des employeurs corrects. Mais il y a plein de petites jeunes qui se font avoir », poursuit Lina. ‘“Us, we’re lucky, we have good employers. But there are plenty of youngsters who are getting ripped off”, continues Lina.’ (Le Monde)

216 Chapter 7 | The information structure of il y a clefts

I observed that in some cases, one and the same il y a cleft can be underspecified for the information structure articulation it expresses. Such cases support previous studies that report problems when confronting information structure notions with authentic corpus data (e.g. Ritz et al. 2008; Lahousse 2011; Marzo et al. 2012; Lahousse & Borremans 2014; Stede & Mamprin 2014:1722). This presumably doesn’t lead to communication problems in those instances, because the information update of the proposition in the Common Ground is practically the same.186

More generally speaking, the data show that there is no one-to-one mapping of form (il y a cleft) and information structure properties, given the variety of articulations that il y a clefts can express. Rather, il y a clefts predominantly share the characteristic of introducing a discourse-new, non-topical, focal referent (as expected), but the il y a cleft template is underspecified when it comes to whether or not the relative clause is also focal and whether or not there is an explicit (double) contrast. This is in line with the idea put forward by authors such as Matić et al. (2014:3), according to which a grammatical form may often constrain the possible information structure articulations to a few possibilities, but the articulation that the utterance ultimately has is dependent on discourse context/pragmatic reasoning. Such an instance of underspecification may even be more efficient in terms of the linguistic system: Piantadosi et al. (2012) argue that when the context suffices to disambiguate between different interpretations, linguistic underspecification is often more efficient than a one-to-one mapping of form and meaning, which would be superfluous.

186 In this respect, it is important to remember that ambiguities arise in written data. An analysis of the prosody of il y a clefts may show whether such ambiguities also characterize spoken language.

Chapter 7 | The information structure of il y a clefts 217

The non-topic marking hypothesis

Given that the corpus data contain only very few instances of topical clefted elements, which are always contrastive, i.e. marked, I argued that the data support the hypothesis according to which il y a clefts announce to the hearer that a non-topical constituent is coming. However, I also showed that il y a clefts share this function with existential sentences more generally. Hence, this function is not due to the whole cleft format, but to the existential expression il y a.187

I then compared the main information structure articulations of il y a clefts and existential sentences (e.g. Il y a un requin dans la piscine ‘There’s a shark in the swimming pool’), showing that there are extensive parallels. In most cases, il y a sentences without relative clauses can be used for the same discourse functions (and with the same information structure properties) as il y a clefts. In those instances, the function performed by the cleft relative clause is encoded by other means, such as a prepositional phrase. This shows that on a functional level, there is no strict distinction between il y a clefts and the more general category of il y a sentences without relative clauses.

Moreover, I evaluated the link between (i) the generalization that il y a clefts introduce constituents that would make suboptimal preverbal subjects (Chapter 6) and (ii) the observation that il y a clefts introduce non-topical constituents (this chapter). The hypothesis that il y a clefts are used to avoid bad preverbal subjects and the hypothesis that they are used to announce to the hearer that a non-topical constituent is coming were shown to be two sides of the same coin.

Finally, I presented some psycholinguistic studies about expressions that are similar to il y a, which support the non-topic marking hypothesis.

In what follows, I will compare the information structure properties of il y a clefts analyzed in this chapter with the information structure of c’est clefts, as reported in previous analyses.

187 The fact that the cleft format in general (i.e. not only il y a clefts but also c’est clefts etc.) does not mark the clefted element as non-topical is also supported by previous analyses of c’est clefts introducing topical constituents, see Chapter 8.

------

Chapter 8

Information structure: c’est clefts vs. il y a clefts

This chapter188 presents a comparison of the information structure properties of il y a clefts and c’est clefts (‘it is’) by contrasting the findings presented in Chapter 7 with prior studies of c’est clefts. 189 I will follow previous analyses in assuming that c’est clefts are characterized by the same general definition of clefts that was presented in Chapter 1 (e.g. Lambrecht 2001; Dufter 2009b). Put briefly, this means they express a single proposition that could also be conveyed by a canonical equivalent (i.e. the decleftability criterion), as illustrated in (292). Furthermore, the relative clause is neither restrictive (witness the possibility of the proper name Pierre as antecedent) nor appositive.

(292) C’est Pierre qui adore le poulet à la crème. ‘It’s Pierre who loves chicken with cream.’

~ Pierrre adore le poulet à la crème.

~ ‘Pierre loves chicken with cream.’ (Dufter 2006:32, my translation)

188 Parts of this chapter will be published in Karssenberg & Lahousse (forthcoming-a). 189 In contrast to il y a clefts, c’est clefts have received a considerable amount of attention. See, among others, Hupet & Tilmant 1986; Clech-Darbon et al. 1999; Katz 2000b, 2000a; Lambrecht 2001; Rialland et al. 2002; Muller 2003b; Muller 2003a; Doetjes et al. 2004; Blanche-Benveniste 2006; Dufter 2006; Hobæk Haff 2006; Scappini 2006; Rouquier 2007; Bouchard et al. 2008; De Stefani 2008; Dufter 2008; Jullien 2008; Mertens 2008; Carter-Thomas 2009; Dufter 2009a, 2009b; Destruel 2012; Mertens 2012; Destruel 2013; Lahousse & Borremans 2014; Reichle 2014; Jacob 2015; Rouquier 2015. 220 Chapter 8 | Information structure: c’est clefts vs. il y a clefts

The general goal of this chapter is to pinpoint and explain the differences and similarities between the two cleft types in order to come to a better understanding of the way they relate to each other. I will first show that il y a clefts and c’est clefts can express the same information structure articulations (section 8.1). I will then explain that the two types differ extensively in several respects (formal, distributional and discourse-functional, section 8.2). Hence, despite considerable similarities, the two cleft types are clearly different animals.

8.1 Information structure: same articulations, different distribution

In what follows, I show that c’est clefts can have the same information structure articulations as the ones discussed in Chapter 7 for il y a clefts (see Table 30).190 However, the distribution of those interpretations is very different for the two cleft types, as will become evident in section 8.1.6.

Articulation Il y a clefts C’est clefts

All-focus Section 7.3.2 Section 8.1.1

Focus-background Section 7.3.3 Section 8.1.2

Multiple focus Section 7.3.4 Section 8.1.3

(Contrastive) spatiotemporal topic + comment Section 7.3.5.1 Section 8.1.4

(Contrastive) aboutness-topic + comment Section 7.3.5.2 Section 8.1.5

Table 30: Information structure articulations of il y a clefts & c’est clefts

190 For literature on the different information structure types of English it clefts, see among others Prince 1978; Declerck 1983, 1988; Hedberg 1990; Delin & Oberlander 1995; Hedberg 2000; Lambrecht 2001; Delin & Oberlander 2005; Den Dikken 2013.

Chapter 8 | Information structure: c’est clefts vs. il y a clefts 221

8.1.1 All-focus

Two types of all-focus c’est clefts can be distinguished: those introducing a prepositional clefted element (293) and those introducing a (pro)nominal clefted element (294) (see Clech-Darbon et al. 1999; Dufter 2006; Huber 2006; Scappini 2006; Dufter 2008, 2009b). The first type is characteristic of more formal French such as political speeches (see for instance Katz 2000a:270). In both cases, the clefted element is not the topic: the relative clause does not express a comment about the clefted element. On the referential level, all the information expressed by the sentence is discourse-new.

(293) All-focus, type 1: with prepositional clefted element [Beginning of a speech]

C’est avec beaucoup de tristesse que l’on suit les événements qui se déroulent au Nigéria.

‘It is with great sorrow that we follow the events in Nigeria.’ (Dufter 2009b:105)

(294) All-focus, type 2: with nominal clefted element [Tu sembles inquiète. Qu’est-ce qui se passe?]

[‘You look worried. What happened?’]

C’est le petit qui est tombé dans l’escalier

‘The young one fell down the stairs.’ (Clech-Darbon et al. 1999:84)

8.1.2 Focus-background

The focus-background articulation is the most well known articulation of c’est clefts (see Smits 1989a; Kiss 1998; Clech-Darbon et al. 1999; Katz 2000a; Lambrecht 2001; Rialland et al. 2002; Doetjes et al. 2004; Dufter 2008; Mertens 2008; Destruel 2013; Lahousse & Borremans 2014). For instance, in (295), the relative clause expresses discourse-given information (be at the top of their class in French) and the clefted element provides the salient and discourse-new information (not her sons).

222 Chapter 8 | Information structure: c’est clefts vs. il y a clefts

(295) Le proviseur lui dit: votre fille, madame, est la première en français. Ma mère ne dit rien, rien, pas contente parce que c’est pas ses fils qui sont les premiers en français. ‘The head master said to her: your daughter, Madam, is the top of her class in French. My mother didn’t say anything, she was not happy because it is not her sons who are top of their classes in French.’ (Lahousse & Borremans 2014:807)

In a similar vein, the relative clause in (296) expressed discourse-given information, whereas the clefted element is focal. Note, also, that focus-background c’est clefts are often associated with a contrastive reading of the clefted element. In (297), for instance, toi ‘you’ is contrasted with respect to moi ‘me’.

(296) Tu as vu ma voiture? C’est mon frère qui a fait ça. ‘Did you see my car? It’s my brother who did that.’ (Katz 2000a:264)

(297) C’est toi qui l’as cassé, ce n’est pas moi. ‘It’s you who broke it, it’s not me.’ (Blanche-Benveniste 2006:273, my translation)

8.1.3 Multiple focus

It clefts and their cross-linguistic equivalents can have a ‘multiple focus’ articulation with a first focus on the clefted element and another one on the relative clause (Carlson 1983; Huber 2002, 2006; Dufter 2009b; Hedberg 2013; Büring 2016, see also Hedberg 2013 about ‘emphatic repetition clefts’). An example is presented in (298): the clefted elements (Max; Eva) as well as the instruments in the relative clauses (the piano; the violin) are focal. Note that this cleft is used to correct the preceding statement that links Eva and Max with the wrong instruments.

(298) [Speaker A: Eva plays the piano and Max the violin.] Speaker B: No, it is MAX who plays the piAno and Eva who plays the VIolin.

(Huber 2006:569, capitalization by the author)

According to Dufter (2009b), multiple focus c’est clefts can also be found in French, although he does not present examples of this type.

Chapter 8 | Information structure: c’est clefts vs. il y a clefts 223

8.1.4 Spatiotemporal topic + comment

It has been noted that c’est clefts can introduce anaphoric spatiotemporal expressions that function as topics. For instance, Dufter (2009b) analyzes the clefted element Iraqi Kurdistan in (299) as a ‘frame topic’/spatiotemporal topic.

(299) C’est également dans le Kurdistan irakien que se présente pour l’Union européenne l’occasion exceptionnelle d’apporter une contribution substantielle à la tâche urgente de reconstruire la Mésopotamie. ‘It is also in Iraqi Kurdistan that the European Union is being presented with an outstanding opportunity to make a substantial contribution to the urgent task of reconstructing Mesopotamia.’ (Dufter 2009b:102, no prior context)

In a similar vein, c’est clefts can introduce short anaphoric adverbials such as là ‘there/then’ (300) and ainsi ‘that way/thus’ (301). Lahousse & Lamiroy (2015) analyze c’est ainsi que ‘it is thus that’ clefts such as (301) as topic-comment clefts.191

(300) En quarante-deux, les bombardements! Et c’est là que beaucoup de gens sont partis de Toulon. ‘In 1942, the bombings! And it’s then that many people left Toulon.’

(Blanche-Benveniste 2006:282, my translation)

(301) Nous sommes restés trois semaines sans nous voir et c’est là qu’on s’est vu. ‘We didn’t see each other for three weeks and it’s there that we saw each other.’

(Scappini 2013:90, in Lahousse & Lamiroy 2015:5, my translation)

191 Blanche-Benveniste (2006:278) states that the clefted element in (300) is not focal, but she doesn’t explicitly argue that it is a (spatiotemporal) topic.

224 Chapter 8 | Information structure: c’est clefts vs. il y a clefts

8.1.5 Aboutness-topic + comment

The existence of c’est clefts with an aboutness-topic + comment interpretation has been acknowledged by several authors (e.g. Doetjes et al. 2004; Blanche-Benveniste 2006; Mertens 2012; Lahousse & Borremans 2014). For instance, in (302), the clefted pronoun elle ‘her’ refers to the previously introduced Sophie, and the relative clause expressed new information about her. The same holds for (303): the clefted element lui ‘he’ refers to Philippe, and the relative clause expresses information about this referent.192

(302) Sophie tu la connais depuis longtemps, enfin c’est elle que tu connaissais d’abord ‘Sophie you’ve known her for a long time, well it was her that you knew first.’ (Dufter 2008:49, translation slightly adapted)

(303) Françoise, je te présente Philippe. C’est lui qui a écrit sa thèse sur Sade. ‘Françoise, this is Philippe. It’s he who wrote his dissertation about Sade.’

(Katz 2000a:265)

8.1.6 Distribution

Section 8.1 showed that c’est clefts have the same information structure potential as il y a clefts: they express the same articulations as the il y a clefts analyzed in chapter 7.3. However, the two cleft types are very different in terms of the distribution of the different information structure articulations. Table 31 presents an overview of this distribution, based on my corpus analysis of il y a clefts (chapter 7.3.1) and the results from previous corpus studies on c’est clefts reported by Dufter (2008, 2009b).193 This overview shows that while il y a clefts mainly express all-focus, topic-comment being the least frequent, c’est clefts display almost the exact opposite pattern: they are

192 Katz (2000a) does not analyze this example as a topic-comment cleft. 193 These studies do not provide percentages of the different articulations.

Chapter 8 | Information structure: c’est clefts vs. il y a clefts 225

frequently used with a topic-comment and a focus-background articulation,194 and only rarely occur with an all-focus interpretation.

Articulation Il y a clefts C’est clefts

All-focus 83% Infrequent195

Focus-background 16% Frequent

Multiple focus ? Infrequent Only attested once in (Contrastive) spatiotemporal il n’y a que clefts Frequent topic + comment (infrequent) (Contrastive) aboutness-topic 1% Frequent + comment

Table 31: Il y a clefts vs. c’est clefts: distribution of information structure articulations

In the next section, I present several syntactic and pragmatic factors that help explain why the two cleft types display such a clear difference in distribution, and why they are usually not interchangeable in the same discourse contexts.

8.2 Differences between il y a clefts and c’est clefts

The differences between il y a clefts and c’est clefts that I will address are related to prepositional and adverbial clefted elements (section 8.2.1), text type/genre (8.2.2), exhaustiveness (8.2.3), their distribution in spoken/written French (8.2.4) and their function as markers of non-topicality (8.2.5).

194 Similarly, Mertens (2012:134) remarks that c’est clefts with an anaphoric clefted element and a relative clause containing discourse-new information are much more frequent than focus- background clefts in spoken corpora, although he doesn’t quantify this observation. (“Contrairement à ce qu’on pourrait croire, dans les corpus oraux les clivées du deuxième type sont nettement plus fréquentes que celles du premier”, Mertens 2012:134) 195 Dufter (2008:50) states that his corpora contain only few of such instances (‘Type 3’).

226 Chapter 8 | Information structure: c’est clefts vs. il y a clefts

8.2.1 Formal difference: prepositional & adverbial clefted elements

One crucial difference between il y a clefts and c’est clefts that has important implications for the distribution of the different functional uses of the two clefts is the general ban on non-(pro)nominal clefted elements in il y a clefts (304) (Pierrard 1985:46; Léard 1992:55). This ban can only be lifted by means of the restrictive particle ne… que ‘only’ (305). C’est clefts, on the other hand, are not subject to such a constraint (306).

(304) * Il y a en bateau que Jean voyage. * ‘There is by boat that Jean travels.’ (Pierrard 1985:46, my translation)

(305) Il n’y a qu’en bateau que Jean voyage. ‘Jean only travels by boat.’ (Lit. ‘There’s only by boat that Jean travels.’)

(306) C’est en bateau que Jean voyage. ‘It’s by boat that Jean travels.’

This formal difference between il y a clefts and c’est clefts is particularly relevant in light of the fact that several corpus analyses show that adverbial and prepositional clefted elements are very frequent in c’est clefts (Katz 2000a; Carter-Thomas 2009:15; Dufter 2009b; Jacob 2015). Moreover, Dufter (2008) reports an overall increase of adverbial clefted elements over time, going from 16,4% of all clefted elements (12th and 13th century) to 21,5% (14th and 15th century) to 28,1% in written and 32,8% in spoken French of the 20th century (Dufter 2008:48). Many of these clefted elements refer to the spatiotemporal settings of the sentence, i.e. those clefts have a spatiotemporal topic-comment articulation (see (299)). Given the grammatical constraint described above, il y a clefts are unable to perform this function unless ne… que is present.196 Another consequence is that il y a clefts cannot be used as paragraph openers in the same way as all-focus c’est clefts such as (293) (‘it is with great sorrow that…’), as shown in (307).

196 What’s more, recall that only 13% of the il n’y a que clefts in the corpus data contain clefted elements that are not (pro)nominal (chapter 5.3.1). Thus, it would be mistaken to state that il n’y a que clefts are used to compensate for the ban on non-(pro)nominal clefted elements in il y a clefts without ne… que ‘only’.

Chapter 8 | Information structure: c’est clefts vs. il y a clefts 227

(307) [Based on (293)] *Il y a avec beaucoup de tristesse que l’on suit les événements qui se déroulent….

*‘There’s with great sorrow that we follow the events that are taking place...’

8.2.2 All-focus il y a/c’est clefts: text type/genre

Apart from all-focus c’est clefts containing an adverbial clefted element, there’s also a second type of all-focus c’est cleft that is more similar to typical all-focus il y a clefts. In such instances, the c’est cleft expresses an event involving a human referent, as in (308) and (309).

(308) [Tu sembles inquiète. Qu’est-ce qui se passe ?] [‘You look worried. What happened?’]

C’est le petit qui est tombé dans l’escalier.

‘The young one fell down the stairs.’ (Clech-Darbon et al. 1999:84)

(309) [Beginning of a joke, no prior context] C’est un français qui va en Corse.

‘It’s a Frenchman who goes to Corsica.’ (Scappini 2006:317, my translation)

Interestingly, this type of c’est cleft can be replaced by an il y a cleft without a clear change in meaning, as in (310) and (311).197

197 Conversely, it is generally awkward to replace an eventive il y a cleft by an eventive c’est cleft in a narrative setting, as in (i) – (ii): (i) (...) subitement je vois une lumière blanche à la surface de l'eau et y'a une main qui me saisit par le cou et qui me relève. ‘ (…) Suddenly I see a white light above the water and there’s a hand that grabs my throat and that lifts me up.’ (YCCQA) (ii) ?? [same context] …et c’est une main qui me saisit par le cou et qui me relève. ?? [same context]… ‘and it’s a hand that grabs my throat and that lifts me up.’

228 Chapter 8 | Information structure: c’est clefts vs. il y a clefts

(310) [Based on (308)] Il y a le petit qui est tombé dans l’escalier

‘The young one fell down the stairs.’

(311) [Based on (309)] Il y a un français qui va en Corse.

‘A Frenchman goes to Corsica.’

It seems that the main differences between all-focus c’est clefts such as (308) and (309) on the one hand and typical eventive il y a clefts on the other are (i) their relative frequency with respect to other information structure articulations and (ii) the types of discourse genres they typically occur in. Concerning the first point, all-focus c’est clefts with eventive semantics have been noted to be very infrequent (Dufter 2008, 2009b). In contrast, the analysis of the corpus data in chapter 7.3.1 showed that all-focus il y a clefts are by far the most frequent type (84%) of all the possible information structure articulations (see Table 31). As for the genre that such c’est clefts occur in, Scappini (2006) observes that they are mainly used as a type of fixed formula to introduce a joke (see also Dufter 2006:51).198 All-focus il y a clefts, on the other hand, are not restricted to a specific genre or discourse context such as ‘jokes’. Il y a clefts do occur much more frequently in spoken than in written French. However, I am unaware of a quantitative study of c’est clefts reporting the number of eventive all-focus c’est clefts in spoken vs. written French. Hence, it is impossible to make claims about possible differences between the two clefts with respect to spoken/written French (for more on this, see section 8.2.4).

The fact that all-focus c’est clefts are so infrequent in comparison to il y a clefts clearly shows that the two clefts have specialized for different discourse functions over time.199 I leave it to future research to present an explanatory account for this difference in development between the two cleft types.

198 Note, however, that il y a clefts can also occur at the beginning of a joke (i). Hence, c’est clefts and il y a clefts are not in complementary distribution concerning the text type ‘joke’. (i) Y'a un mec qui rentre dans une pharmacie et il dit: "Bonjour, (…)." ‘There’s a guy who enters a pharmacy and says: “Hello, (…)”.’ (www) 199 See Dufter (2008) for a diachronic study of (the information structure of) c’est clefts.

Chapter 8 | Information structure: c’est clefts vs. il y a clefts 229

8.2.3 Focus-background il y a/c’est clefts: exhaustiveness

There is an extensive literature on the so-called ‘exhaustiveness reading’ in clefts introduced by c’est and their cross-linguistic equivalents.200 It has been observed that when a c’est cleft is semantically specificational, the cleft implies that the introduced value is the only possible value for the variable.201 For instance, in (312), the use of the c’est cleft indicates that Klaus is the only person who is dancing. This is supported by the fact that it is infelicitous to add ‘and his brother as well’.

(312) Specificational c’est cleft: exhaustiveness implicature [Who is dancing?]

C’est Klaus qui danse, # et son frère aussi.

‘It’s Klaus who is dancing, # and his brother as well’

In contrast, no such implication is present in specificational il y a clefts such as (313). This is why Lambrecht (2001:504) refers to such il y a clefts as “nonexhaustive specificational clefts” (see also Léard 1992).

(313) Specificational il y a cleft: no exhaustiveness implicature [Who is dancing?]

Il y a Klaus qui danse, et son frère aussi.

‘There’s Klaus who is dancing, and his brother as well.’

It should be noted that the sense of exhaustiveness that is characteristic of c’est clefts can be added to il y a clefts by means of the restrictive particle ne… que, as in (314). However, the resulting il y a cleft is not interchangeable with the c’est cleft in (312), because the restriction – the fact that only Klaus is dancing – is made lexically explicit in (314) (witness the obligatory only adverb in the English translation), whereas it is

200 See Horn 1981; Kiss 1998; Drenhaus et al. 2011; Destruel 2012, 2013; De Cesare & Garassino 2015. 201 There is no consensus about the origin of this interpretive effect in clefts (Drenhaus et al. 2011). Some argue that it’s part of the inherent semantics of clefts (e.g. Atlas & Levinson 1981; Kiss 1998). However, an increasing number of researchers convincingly argue that it should be seen as a generalized conversational implicature (e.g. Declerck 1988; Hedberg 1990; Dufter 2006, 2009b; De Cesare & Garassino 2015). An important argument for this position is that the exhaustiveness interpretation can be cancelled as well as reinforced (see cited works for examples). I adhere to the view that the exhaustiveness reading is a pragmatic implicature.

230 Chapter 8 | Information structure: c’est clefts vs. il y a clefts

implicit in the c’est cleft in (312). Therefore, the c’est cleft in (312) and the il n’y a que cleft in (314) do not express the same proposition on a semantic level.

(314) [Who is dancing?] Il n’y a que Klaus qui danse, # et puis son frère aussi.

‘Only Klaus is dancing, # and his brother as well.’

For the same reason, an il n’y a que cleft cannot be transformed into a c’est cleft without inducing a difference in interpretation, as shown in (315) - (316). In this context, the fact that only the writer’s son has the name Stewan must be lexically encoded by ne… que ‘only’: the observation that he has that name (in the c’est cleft in (316)) is not informative enough, because this information has already been evoked in the preceding context.202

(315) Connaissez-vous la signification, l'origine du prénom de mon fils STEWAN (…), Merci? j'ai l'impression qu'il n’y a que lui qui porte ce prénom. ‘Do you know the meaning, the origin of my son’s first name ‘Stewan’ (…), thank you? I have the impression that only he has that name.’

(Lit. ‘I have the impression that there’s only he who has that name’) (YCCQA)

(316) … # J’ai l’impression que c’est lui qui porte ce prénom. … # ‘I have the impression that it’s him who has that name.’

Summing up, given the presence/absence of an exhaustiveness reading in focus- background c’est/il y a clefts respectively, the two cleft types are not interchangeable and hence do not compete with one another when the speaker ‘decides’ which linguistic encoding to use in order to express a certain proposition.

202 Note that it is possible to use a c’est cleft in combination with ne… que ‘only’, as in (i). I conducted a search query in the YCCQA corpus for all instances of ce n’est que… qui/que, and only obtained 19 ce n’est que clefts, which is very infrequent with respect to the 173 il n’y a que clefts extracted from this corpus. Moreover, 11 of these ce n’est que clefts have a temporal value, as in (ii). (i) Ce n’est que Klaus qui danse. ‘It is only Klaus who is dancing.’ (ii) (…) en 1897, le Parc des Princes était un stade vélodrome où se déroulait, donc, des courses cyclistes. Ce n'est que le 28 février 1899 qu'il s'y déroula le premier match de football. ‘(…) in 1897, the Parc des Princes was a velodrome in which cycling contests took place. It wasn’t until February 28th, 1899 that the first soccer match took place there.’ (YCCQA)

Chapter 8 | Information structure: c’est clefts vs. il y a clefts 231

8.2.4 Spoken vs. written French

As noted by Dufter (2008), there is no consensus about the frequency of c’est clefts in spoken vs. written French. According to some, there’s no difference between spoken and written French, whereas others maintain that c’est clefts are characteristic of spoken French, and yet others make the opposite claim (see references in Dufter 2008:39). Dufter (2008) puts these hypotheses to the test and shows that c’est clefts are almost 2.5 times more frequent in the spoken corpus (C-ORAL-ROM) than in the written corpus (Frantext) he consulted.

This difference is much less pronounced than the one found for il y a clefts across the three corpora used in this dissertation (Chapter 4): il y a clefts are approximately 95.5 times more frequent in the spoken corpus CFPP than in the journalistic corpus Le Monde. Moreover, I showed that 83% of the il y a clefts in Le Monde are used within citations (chapter 4.1.1). In contrast, De Cesare et al. (2016) report that (only) 39% of the c’est clefts in the journalistic corpus they used appear in reported speech (De Cesare et al. 2016:169). In other words, while journalists perceive il y a clefts as characteristic of spoken French, it is not clear whether the same holds for c’est clefts.

Note, furthermore, that certain rhetorically charged uses of c’est clefts, such as all-focus c’est clefts with adverbials (‘it is with great sorrow that…’) seem typical of a more formal register of French. In contrast, the Le Monde corpus does not contain il y a clefts that have a special rhetoric purpose and which are absent from the other two corpora. In other words, there are no special ‘formal’ uses of il y a clefts that are only found in the Le Monde corpus and which do not occur in the informal YCCQA and CFPP corpora.

All in all, it seems that although il y a clefts are clearly typical of spoken French, c’est clefts are less clearly characteristic of either spoken or written French.

232 Chapter 8 | Information structure: c’est clefts vs. il y a clefts

8.2.5 Markers of non-topicality

Recall that it has been claimed that clefts in general are used as markers of non- topicality, i.e. they indicate that the clefted element should not be interpreted as a topic (e.g. Lambrecht 2001, see chapter 7.2.2). Several authors criticize this stance, because French c’est clefts can also introduce (aboutness/spatiotemporal) topics (see sections 8.1.4 and 8.1.5). C’est clefts and il y a clefts differ in this respect: regular (non- contrastive) topics turned out not to occur in the corpus data of il y a clefts (chapter 7.4), whereas the use of anaphoric adverbials (functioning as spatiotemporal topics) is reported to be frequent in c’est clefts (section 8.1.4), as in (317):

(317) Nous sommes restés trois semaines sans nous voir et c’est là qu’on s’est vu. ‘We didn’t see each other for three weeks and it’s there that we saw each other.’ (Scappini 2013:90, in Lahousse & Lamiroy 2015:5, my translation)

This is also evident from the fact that certain uses of c’est clefts, such as c’est ainsi que ‘it is thus that’ and c’est pour ça que ‘it’s for that reason that’ are “arguably on the verge of lexicalization” (Dufter 2009b:104, see also Lahousse & Lamiroy 2014, 2015). Hence, while it is difficult to maintain the hypothesis that one of the main functions of c’est clefts is to avoid a topic-comment interpretation, I argued that il y a clefts do have this core function, which they share with all il y a sentences (see chapter 7.4).

I wish to emphasize that I’m not implying that information structure in general plays no role in the motivation for c’est clefts. However, given the frequency of anaphoric adverbials functioning as topics in c’est clefts, it is hard to see the expression c’est (or c’est clefts as a whole) as a processing cue that warns the hearer about non-topical constituents.

Chapter 8 | Information structure: c’est clefts vs. il y a clefts 233

8.3 Conclusion

A comparison of il y a clefts and c’est clefts revealed that the two types have the same information structure potential. However, the two cleft types differ extensively in several respects:

i. Distribution of the different information structure articulations: All-focus is predominant in il y a clefts but rare in c’est clefts; topic-comment is almost non-existent in il y a clefts but frequent in c’est clefts.

ii. Formal difference: c’est clefts frequently introduce adverbial and prepositional clefted elements, il y a clefts can only do so when the restrictive particle ne… que ‘only’ is present.

iii. Text type/genre: all-focus c’est clefts mostly occur at the beginning of a joke, whereas the discourse context of all-focus il y a clefts is not restricted to a specific text type or genre.

iv. Exhaustiveness: focus-background c’est clefts carry a pragmatic implicature of exhaustiveness, while il y a clefts do not.

v. Spoken vs. written French: il y a clefts are clearly a spoken phenomenon, whereas c’est clefts are less clearly characteristic of either the spoken or the written modality in terms of frequency. Furthermore, c’est clefts have certain functions that are mostly found in formal/written contexts (e.g. it is with great pleasure that…), whereas il y a clefts do not have specific formal/written uses.

vi. Non-topicality marking: the corpus data of il y a clefts support the hypothesis that il y a functions as a marker of non-topicality, but it’s less straightforward to see c’est as a similar processing cue, given the frequency of anaphoric (spatiotemporal) topics in c’est clefts.

All in all, it seems that il y a clefts and c’est clefts have specialized very differently and hardly ever compete for the expression of the same proposition in a particular discourse context. What they do have in common as members of the ‘cleft family’ other than the defining properties of decleftability and a special relative clause is that they are often motivated by information structure. In the next chapter, I will show that il y a clefts also resemble c’est clefts in a different respect: it has been argued that the

234 Chapter 8 | Information structure: c’est clefts vs. il y a clefts

use of c’est clefts cannot always be explained in terms of information structure alone. Indeed, c’est clefts also have other functions that can make them more attractive than their SV counterpart in certain discourse context (e.g. Blanche-Benveniste 2006; Dufter 2009b). I will show that the same holds for il y a clefts, which have additional functions besides the ones already discussed in Chapter 6 (avoiding bad preverbal subjects) and in Chapter 7 (announcing non-topics).

------

Chapter 9

Other pragmatic functions of il y a clefts

Clefts have generally been analyzed as constructions motivated by the expression of information structure, or, more specifically, marking constituents as focal or non- topical. In Chapter 7, I showed that those notions are indeed important for explaining why a speaker uses an il y a cleft rather than a competing structure. This chapter sets out to address a question raised, among others, by Dufter (2006): it’s clear that clefts can be used for focus-marking purposes, but is that all they’re good for?203 As the title of this chapter suggests, the answer to this question is no. My analysis will thus prove to be in line with Jullien (2007), who shows that ‘il y a NP’ structures (among which il y a clefts) can be used for a variety of interactional purposes such as negotiating turn- taking and ‘try-marking’.204

Several authors have recognized that the cleft format can also be exploited for other discourse functions than marking the clefted element as focal or non-topical (e.g. Delin 1989; Delin & Oberlander 1995; Blanche-Benveniste 2006; Dufter 2006; Bicler & Davidse 2008; Dufter 2009b). For instance, in their analysis of it clefts and wh- clefts, Delin & Oberlander (1995) state:

203 “Dass nun die Satzspaltung ein wichtiges Mittel der Fokusmarkierung darstellt, ist unbestritten. Unklarheit herrscht jedoch weiterhin hinsichtlich der Frage, ob dieses syntaktische Format ausschließlich dazu dient, den abgespaltenen Ausdruck als Fokus und den subordinierten Satz als Hintergrund zu markieren” (Dufter 2006:43, my emphasis) 204 See also Pekarek Doehler (2015), who analyses instances of ‘il y a NP’ without a relative clause from an interactional perspective. She argues that these sentences are used for a variety of functions, such as floor holding, highlighting the noteworthiness of the upcoming talk and indicating turn-taking organization. This is in line with interaction-oriented research undertaken by Elwys De Stefani on c’est clefts (De Stefani 2008, 2009). 236 Chapter 9 | Other pragmatic functions of il y a clefts

“(…) we take clefts, like many other pragmatically marked syntactic constructions, to convey not a single semantic or pragmatic feature but a bundle of such features, acting jointly to determine the discourse functions a sentence type can serve. […] it seems plausible to assume that speakers choose a sentence type that best fits their conversational goals in a given context, and that a range of factors, rather than a single one, determines this choice” (Delin & Oberlander 1995:470, my italics).

In a similar vein, Dufter (2009b:108) states:

“By all odds, there must be more to clefting than signaling focus”, and shows on the basis of corpus data that clefts can be used for a variety of functions, including “sentence-related aims such as focus and operator scope marking, but also discursive goals such as enhancing cohesion, marking transitions, or differentiating between propositional content that is under discussion and information that is not” (Dufter 2009b:115, see also Blanche-Benveniste 2006; Dufter 2006; De Stefani 2008, 2009; Jacob 2015).

In what follows, it will become clear that a careful examination of il y a clefts and their canonical equivalents in my corpus data allows to draw a similar conclusion: there’s more to il y a clefts than marking constituents as non-topical. The goal of this chapter is to present several functions of il y a clefts that have so far remained unnoticed, and which help explain why the speaker uses a cleft rather than its canonical counterpart.

The structure of this chapter is as follows. In section 9.1, I show that il y a clefts may serve the function of enhancing discourse coherence, as has also been argued for certain clefts introduced by it and c’est. Section 9.2 discusses il y a clefts whose function is to signal that a singular indefinite NP does not refer to a singular, specific referent. I then argue that il y a clefts can be used to reinforce a negation (section 9.3). Finally, I discuss additional pragmatic functions of the relative clause in il y a clefts (section 9.4).

Chapter 9 | Other pragmatic functions of il y a clefts 237

9.1 Discourse coherence

Several authors observe that clefts introducing anaphoric expressions may increase textual coherence in a way that their canonical equivalents cannot (e.g. ‘cohesive c’est clefts’ in Krötsch & Sabban 1990; Dufter 2009b, see also Declerck 1984; Herslund 2005; Blanche-Benveniste 2006; Lahousse & Borremans 2014; Lahousse & Lamiroy 2014; Jacob 2015; Lahousse & Lamiroy 2015; De Cesare et al. 2016:452 inter alia about such clefts). For instance, Delin (1989) argues that it clefts may preserve referential continuity, while other potential ‘rival sentence types’ (wh-clefts and canonical SV sentences) create a discontinuous chain of reference. She illustrates thus by means of the opposition in (318) vs. (319): the cleft brings the anaphoric adverb there to the beginning of the sentence, thus presenting a more coherent discourse continuation.

(318) Part of Magritte’s school holidays was spent with grandmother and an aunt at the town of Soignies. It was there, whilst playing one day with a small girl in an abandoned cemetery, that he first became aware of painting as a special and somehow mysterious activity. (…) (Delin 1989:177)

(319) (…). He first became aware of painting as a strange and mysterious activity while playing one day with a small girl in an abandoned churchyard there. (…) (Delin 1989:178)

Delin (1989:178) therefore concludes that the serial ordering of constituents can be an important factor in the choice between it clefts, wh-clefts and SV sentences. The same line of reasoning can be applied to il n’y a que clefts introducing an anaphoric adverbial, as in (320). By virtue of clefting the manner adverbial comme ça ‘like that’, the link with the preceding discourse is made clearer than if the adverb were in sentence-final position (321).

238 Chapter 9 | Other pragmatic functions of il y a clefts

(320) – A: Faire des concessions signifie-t-il être une bonne poire? ‘Does making concessions mean being too good-natured?

– B: non!. (…) Les 2 partenaires doivent faire des concessions, il n'y a que comme cela que le couple peut durer!

‘No! (…) Both partners have to make concessions, that’s the only way the couple can last!

(Lit. ‘There’s only like that that the couple can last!’) (YCCQA)

(321) Les deux partenaires doivent faire de concessions, le couple peut seulement durer comme cela ! ‘The two partners have to make concessions, the couple can only last like that!’

In both versions, the expression seulement comme ça ‘only like that’ is focal: it is salient information that the speaker draws attention to. Hence, the difference between the two cannot be captured in terms of focality. Note furthermore that it would also be grammatically possible to front the adverbial phrase, as in (322). However, when they are intended to have a manner adverb interpretation, expressions such as comme ça ‘like that’ and ainsi ‘thus, so’ are dispreferred in sentence-initial position, because in this position, they usually have a sentence adverb reading rather than an intrapredicative/narrow scope interpretation (Molinier & Levrier 2000 inter alia205). The same holds for the adverb seulement ‘only’ (see Lamiroy & Charolles 2004). Hence, (322) is unnatural.

(322) Seulement comme ça le couple peut durer ! ‘Only like that the couple can last!’

205 See Charolles & Vigier (2005) about differences in discourse effect of adverbials in preverbal and postverbal position in general. See Karssenberg (2013) and Karssenberg & Lahousse (2014, forthcoming-b) about the manner adverb ainsi ‘thus, so’ and the influence of position in the sentence on its interpretation. See also Lahousse & Lamiroy (2014, 2015) about c’est ainsi que ‘it is thus that’ clefts.

Chapter 9 | Other pragmatic functions of il y a clefts 239

Another example is presented in (323): again, the anaphoric manner expression in the il n’y a que cleft creates a smoother link with the preceding discourse than the SV equivalent would (324). 206

(323) plus tu les laves souvent, plus ils vont devenir gras. essaie de te les laver tous les 3 jours, c dur je sais mais ya que comme ca qu'ils vont moins graisser. ‘The more you wash your hair, the more it becomes greasy. Try to wash your hair every three days, it’s hard I know, but it’s the only way it’ll become less greasy.’ (Lit. ‘There is only like that that it’ll become less greasy.’) (YCCQA)

(324) … ils vont seulement moins graisser comme ça. … ‘it’ll only become less greasy like that’

Increasing discourse coherence can be seen as a function rooted in the more general domain of ‘discourse processing’. However, discourse coherence cannot be equated with either ‘topic’ (aboutness) or ‘focus’ (salience).

Another thing to note is that il n’y a que clefts with anaphoric expressions regularly contain a relatively long relative clause, as illustrated in (325). This may also contribute to the preference of the cleft format over a canonical equivalent in such cases, due to the general tendency to place heavier constituents after lighter constituents, i.e. ‘heavy load last’ (see e.g. Behagel 1909; Quirk et al. 1985; Hawkins 1994; Wasow 2002 about the ‘Principle of End Weight’).

206 One could argue that in the canonical equivalent of (323), the clefted element comme ça ‘like that’ would be in sentence-initial position. However, as explained above, manner adverbs that have an intrapredicative interpretation preferably follow the verb rather than appearing in sentence- initial position, where they are usually associated with a sentence adverb interpretation (see Molinier & Levrier 2000).

240 Chapter 9 | Other pragmatic functions of il y a clefts

(325) [Context: newspaper article about Turkey] Mais l'essentiel de la marque qu'Atatürk a imposée à son pays semble indélébile. Personne d'ailleurs n'y parle sérieusement de remettre son œuvre en question. Et il n'y a qu'en Turquie que le dirigeant du parti intégriste a pu faire alliance, sans que personne parmi ses partisans s'en choque, avec une femme aussi visiblement occidentalisée que Tansu Çiller, économiste made in USA.

‘But the essence of the mark left by Atatürk on his country seems inerasable. Incidentally, nobody there speaks seriously of questioning his work. And there’s only in Turkey that the leader of the integrist party can start an alliance, without any of his partisans being shocked, with a woman who is as clearly westernized as Tansu Çiller, economist made in the USA.’

(Le Monde)

To conclude, the discussion of the examples above shows that the speaker’s choice of the cleft format may be influenced by (i) the desire to present a coherent discourse and by (ii) restrictions pertaining to the sentence-initial/sentence-final position of certain adverbs or (iii) to the Principle of End Weight.207 These factors make the cleft format more attractive than the canonical SV sentence in such contexts.

207 There are also cases in which the transformation of an il n’y a que cleft into its canonical equivalent leads to a possibly unwanted change in register, as in (i). The use of ‘seul + NP’ in (ii) is characteristic of a more formal register of French (see Lamiroy & De Mulder 2011:309), not the kind of French that is usual on an online discussion forum (i.e. the YCCQA corpus). (i) – A: Comment faire pour effacer de ma tête l'image du cadavre que j'ai retrouvé hier (…)? ‘What should I do to get rid of the image in my head of a dead body I found yesterday?’ – B: je crois qu'il n’y a que le temps qui te fera oublier cette image. ‘I think there’s only time that will make you forget that image.’ (YCCQA) (ii) Je crois que seul le temps te fera oublier cette image. ‘I think that only time will help you forget that image.’

Chapter 9 | Other pragmatic functions of il y a clefts 241

9.2 Signaling that a singular indefinite does not refer to a single referent

In this section, I show that il y a clefts can be used to clearly indicate that a singular indefinite NP has narrow scope with respect to a universal quantifier, whereas the canonical SV counterpart does not yield this intended interpretation as easily.

In certain il y a clefts, a singular indefinite NP is used in combination with an adverbial expression such as toujours ‘always’ or chaque jour ‘every day’ (appearing in front of il y a or the clefted element) in order to express a habitual event, as in (326) and (327):

(326) « Dans ce milieu, poursuit-il, n'importe quel cadet qui brille un peu est, un jour ou l'autre, confronté au problème du dopage. A un moment ou à un autre, il y a toujours un type qui essaie de l'approcher. » ‘“In this social environment, he continues, any youngster who stands out a little will be confronted with drugs one day or another. At some point, there’s always some guy who tries to approach him.” (Le Monde)

(327) Derrière Thomas Levet (3), Marc Pendaries (72, 69, 70, 71, 282, 2) et Olivier Edmond (74, 69, 71, 68, 282, 2) ont également fini sous le par, dans les trente premiers. « Chaque semaine, il y a désormais un Français qui finit dans les vingt premiers », rapporte Thomas Levet. ‘Behind Thomas Levet (3), Marc Pendaries (72, 69, 70, 71, 282, 2) and Olivier Edmond (74, 69, 71, 68, 282, 2) also ended up with a birdie, among the first thirty places. “Every week, there’s now a Frenchman who finishes among the top twenty”, reports Thomas Levet.’ (Le Monde)

In these examples, the singular NPs (un type ‘a guy’, un Français ‘a Frenchman’) do not refer to specific, singular entities: it is unlikely that the same guy always approaches the youngsters about drugs in (326) and it goes against our world knowledge to believe that the same Frenchman ends up high in the scores every week in (327). In other words, the existential quantifier has narrow scope with respect to the universal quantifier (always > there is a guy; every week > there is a Frenchman). Interestingly, however, this intended reading is less straightforward in the SV equivalents of these il y a clefts in (328) and (329).

242 Chapter 9 | Other pragmatic functions of il y a clefts

(328) À un moment ou à un autre, un type essaie toujours de l’approcher. ‘At one point or another, a guy always tries to approach him.’

(329) Chaque semaine, un Français finit désormais dans les vingt premiers. ‘Every week, a Frenchman now ends up among the top twenty.’

This is in line with Kurtzman & MacDonald’s (1993) “single reference principle”, according to which indefinite NPs in subject or topic position are by default taken to refer to singular entities.208 I would thus argue that the use of the il y a cleft in these examples is (at least partially) motivated by the need to clearly signal that the indefinite NP has narrow scope with respect to the temporal expression (always, every week…): by putting the NP in a position that is not a subject or a topic position, the NP can obtain the intended interpretation more easily. In other words, although the intended narrow scope reading of the indefinite NP is possible in the SV sentences in (328) - (329), it is signaled more clearly by the original il y a clefts in (326) - (327), by virtue of the syntax/word order of those clefts.

The same analysis holds for the example in (330), in which un chameau ‘a camel’ does not refer to the same camel every time.

(330) « Il tombe, le chameau. Le type le remet debout. Il tombe. (…). Un petit chameau au milieu d'une longue file de chameaux du même métal. Mais eux ne tombent pas, sur l'étagère du magasin de souvenirs de Nazareth. Seulement, il y a toujours un chameau qui tombe, dans les meilleurs alignements, les dispositifs les plus sûrs. » ‘“It falls, the camel. The guy puts him back upright. It falls. (…) A small camel in the middle of a long line of camels of the same metal. But they don’t fall, on the shelf of Nazareth’s souvenir shop. It’s just that there’s always a camel that falls, in the best alignments, the safest circumstances.” ’ (Le Monde)

208 See also Cresti (1995) and Ebert & Endriss (2005), inter alia about the link between topicality and wide/narrow scope readings. Ebert & Endriss (2005) argue that the topical status of an indefinite NP causes this NP to have wide scope with respect to another quantifier. To the extent that topical status and preverbal subject position are strongly correlated in French (and in other languages), this line of reasoning is compatible with Kurtzman & MacDonald’s (1993) analysis.

Chapter 9 | Other pragmatic functions of il y a clefts 243

Again, the intended narrow scope reading of a camel (always > there is a camel) is clearer in the original sentence introduced by il y a than in the SV equivalent in (331), because preverbal indefinite subjects are more likely to be interpreted as referring to specific, singular entities. Moreover, in this example, the use of the il y a cleft structure allows to place always in front of a camel, whereas the most natural place for toujours ‘always’ in the SV equivalent is sentence-final. The order always + a camel corresponds to the intended scope reading, which may therefore make this reading clearer than in the inverse order in the SV sentence (a camel + always). The same holds for the example in (326), where toujours ‘always’ precedes the clefted element un type ‘a guy’.

(331) Seulement, un chameau tombe toujours. ‘But a camel always falls.’

To conclude, I have shown that il y a clefts can favor an interpretation in which the singular indefinite NP introduced by the existential quantifier il y a has narrow scope with respect to the universal quantifier (always etc.). This is in contrast with their declefted equivalents, which do not indicate the intended reading as clearly.

244 Chapter 9 | Other pragmatic functions of il y a clefts

9.3 Reinforcing negation

In what follows, I argue that il y a clefts can embody a type of reinforced negation, i.e. a negation that is more emphatic than a more canonical way of encoding negation. The difference between reinforced, neutral and attenuated ways of expressing negation was already analyzed by Brunot (1922) (see also Barnicaud et al. 1967; Floricic & Mignon 2007). He mentions repetitions (non, non et non ‘no, no and no’), the addition of adverbs (sûrement pas ‘definitely not’) and swearwords (diantre non! ‘devil no!’), as examples of reinforced negation.209 In my view, a number of il y a clefts introducing negative expressions such as rien ‘nothing’ in my corpus data are also motivated by the need to reinforce a negation (see e.g. (332) and (333)).

(332) il y a une frange de l'électorat qui bascule de droite à gauche, au rythme des déceptions, c'est ça qui permet l'alternance et nous a valu des cohabitations. Mais il faut reconnaître qu'il n'y a rien ni personne, à gauche, pour l'instant, qui puisse convaincre des électeurs potentiels. ‘There’s a fringe of the electorate that swings from the right to the left, at the rhythm of disappointments, that’s what allows the alternation and has given us cohabitations. But we have to acknowledge that there is nothing and no- one, on the left, for the moment, who can convince potential voters.’ (YCCQA)

(333) Chaque malheur est important mais il n'y a rien que l'on ne peut faire en dehors d'espérer pour les pays en guerre. ‘Every tragedy is important but there is nothing we can do except hope for the best for the countries at war.’ (YCCQA)

These sentences can be declefted, as shown in (334) and (335). However, the canonical sentences express negation less emphatically and therefore have a less emotionally charged reading210 than the il y a cleft originals they are based on.

209 Note that nowadays in the linguistic literature on French, the term ‘reinforced negation’ is generally used to refer to the diachronic development of a two-particle negation system (e.g. ne à ne… mie/ne… pas, see for instance Mosegaard Hansen & Visconti 2009). 210 See also Bicler & Davidse (2008) about the ‘pragmatic function of conveying emotional involvement’ in it clefts.

Chapter 9 | Other pragmatic functions of il y a clefts 245

(334) Mais il faut reconnaître que rien ni personne, à gauche, pour l’instant, ne peut convaincre des électeurs potentiels. ‘But we have to acknowledge that nothing and no one on the left, for the moment, can convince potential voters.’

(335) Chaque malheur est important mais on ne peut rien faire en dehors d’espérer pour les pays en guerre. ‘Every tragedy is important but we can do nothing except hope for the best for the countries at war.’

The same holds for the examples presented in (336) and (338), which express negation in a more emphatic way than their canonical equivalents in (337) and (339), because the negation is put in a special syntactic position.

(336) La tristesse touche l'os, la moelle. Il n'y a rien qui soit aussi profond que la tristesse. ‘Sadness touches bone and marrow. There’s nothing that’s as profound as sadness.’ (YCCQA)

(337) Rien n’est aussi profond que la tristesse. ‘Nothing is as profound as sadness.’

(338) chacun aime à sa manière, et le montre avec des façons différentes. après il n'y a rien qui puisse mesurer l'amour. ‘Everyone loves in his/her own way and shows it in different ways. And there’s nothing that can measure love.’ (YCCQA)

(339) Rien ne peut mesurer l’amour. ‘Nothing can measure love.’

In other instances, the emphatic effect is achieved by the combination of the negative adverb pas ‘not’ in combination with a singular indefinite NP or pronoun expressing a minimal degree (pas un jour ‘not a day’ in (340), pas un ‘not one’ in (341), pas une seule personne ‘not a single person’ in (342)):

246 Chapter 9 | Other pragmatic functions of il y a clefts

(340) Le monde est de pire en pire. Il n'y a pas un jour qui passe sans qu'il ne soit pire que le jour d’avant. ‘The world is getting worse and worse. There isn’t a day that goes by without it being worse that the day before.’ (YCCQA)

(341) – A: Que peut on offrir, nous le peuple de France, à Sarko, le 28 janvier, jour de ses 54 ans ? (…)

‘What can we, the people of France, offer Sarko for his 54th birthday on January 28th?’

– B: Faire son anniversaire au resto du coeur! Y a pas un qui a les couilles de l'inviter.

‘Celebrate it at a Restaurant of Love [French charity restaurant]! There’s not a single one who has the balls to invite him.’ (YCCQA)

(342) – A: En quoi croient les tribus primitives? Dieu va t-il les détruire? ‘What do primitive tribes believe in? Is God going to destroy them?’

– B: Tout peuple a reçu un messager pour le guider, il n'y a pas une seul personne qui n'a pas entendu parler de Dieu ou ses Prophètes.

‘All peoples have received a messenger to guide them, there isn’t a single person who hasn’t heard of God or his Prophets.’ (YCCQA)

Such cases closely resemble negated existential sentences such as the ones in (343), discussed by Cappelle et al. (2016).

(343) a. Y avait pas un mot de vrai dans son histoire. ‘There wasn’t a grain of truth in what he said’

b. There wasn’t a damn thing to V.

c. Il n’y a pas l’ombre d’un doute.

‘There isn’t a shadow of a doubt’ (Cappelle et al. 2016)

Chapter 9 | Other pragmatic functions of il y a clefts 247

According to the authors, the denial of a minimal degree (a grain of truth, a damn thing, …) “reinforces the absence of anything higher” (Cappelle et al. 2016). The same analysis applies to (340) - (342): by denying ‘a single day’ and ‘a single person’, anything higher on the scale (e.g. ‘a longer stretch of time’, ‘several people’) is also excluded. Similarly, Quirk et al. (1985:1407) note that English there clefts can be used “as a means of emphasizing a negative” in instances such as (344). Denying a minimal degree is also possible without the cleft format, but the presence of il y a or there is in the examples discussed in this section allow to make this denial more emphatic.

(344) There’s nothing I can do about it. (Quirk et al. 1985:1407)

To conclude, I’ve shown that il y a clefts may be used to reinforce a negation, joining a number of other linguistic constructions and expressions that can be used to achieve this goal. One could argue that the emphatic effect is an instance of ‘focalization’, defined in terms of ‘emphasis’ or ‘that which the speaker wishes to draw the hearer’s attention to’. However, this would require a gradient conception of focus, because in both the cleft and the SV equivalent, the negative expressions can be seen as more emphatic with respect to the other words in the sentence. 211 The difference between the sentence types is that the emphasis seems to be more pronounced in the cleft, which is likely due to its biclausal (i.e. syntactic) nature. Therefore, it is safer to speak of reinforced negation as a type of emphasis without equating this phenomenon with the term ‘focus’.

211 The notion of ‘focus’ is often used as a binary, non-gradient property: a linguistic expression is either focal or non-focal (e.g. Lambrecht 1994). See Doherty (2005) for a discussion about gradience in relation to the definitions of focus and salience.

248 Chapter 9 | Other pragmatic functions of il y a clefts

9.4 Pragmatic functions of the cleft relative clause

Cleft relative clauses have received a lot of attention in the literature on clefts (see chapter 1.2). The main questions that researchers have tried to answer concern the morphosyntactic classification of these clauses (relative clause, pseudo-relative clause, small clause) and the question whether they are extraposed or not.212 Less research has been conducted about the function of the cleft relative clause. That is what I zoom in on in the present section, first presenting some previous accounts of pragmatic functions of appositive relative clauses (section 9.4.1), and then showing that the cleft relative clause in specificational il y a clefts may have the same functions (section 9.4.2).

9.4.1 Previous accounts: pragmatic functions of appositive relative clauses213

Relative clauses have often been classified into different categories based on their main function. The most predominant function in the literature is the ‘restrictive’ one, i.e. certain relative clauses restrict the reference of their antecedent, so that this referent can be identified by the interlocutor (e.g. Comrie 1981; Keenan 1985). However, in many other cases, the relative clause does not restrict the reference of the antecedent in any clear sense. For instance, the antecedent of the appositive relative clause in (345) is assumed to be uniquely identifiable already.

(345) Who is it ? – it’s John, who wants to borrow some eggs. (Lambrecht 2001:495)

On a functional level, this type of relative clause is claimed to be of the ‘continuative’ or ‘narrative-advancing’ type (e.g. Lambrecht 2001; Muller 2006), meaning that it enables ‘a movement in the narrative’ (see e.g. Lambrecht 2001; Furukawa 2005; Muller 2006; Loock 2007).214

212 See e.g. Belletti (2008); Reeve (2010, 2011); Belletti (2012); Reeve (2012); Casalicchio (2013c). 213 Parts of this section will be published in Karssenberg (accepted). 214 A French example of a continuative relative clause is presented in (i). (i) Il ouvrit la porte, qu’il referma aussitôt. ‘He opened the door, which he closed again immediately.’ (Muller 2006:331, my translation)

Chapter 9 | Other pragmatic functions of il y a clefts 249

Loock (2006, 2007, 2012, 2013) also identifies two other functions that appositive relative clauses can fulfill. The first one is what he refers to as the ‘relevance’ function: “By adding supplementary information about an antecedent without restricting its reference, the speaker makes the antecedent and/or the subject-predicate relation relevant within the discourse. The relevance of the whole utterance (…) is thus optimised” (Loock 2007:345, see also the ‘grounding function’ of relative clauses in Fox & Thompson 1990). Loock sees the use of such appositive relative clauses as a discursive strategy used by the speaker out of consideration for the addressee. For instance, the appositive relative clause in (346) adds information that legitimizes Tony Sewell’s presence in the newspaper article as a person entitled to give an opinion.

(346) Tony Sewell, who has just finished an inquiry into soaring levels of exclusions among black pupils from a London school, claimed that too much concern with money and consumer goods was almost as damaging to black pupils’ chances as racism. (Loock 2007:347)

A third function that appositive relative clauses can perform according to Loock is the expression of the speaker’s opinion, judgment or comment (i.e. the ‘subjectivity’ function), indicated by means of markers of modality such as modal auxiliaries or lexical expressions denoting an evaluation, as in (347) (Loock 2007:353). Here, the appositive relative clause is a slot that allows the writer to express his/her judgment about the men’s team.

(347) The men’s 4 x 100 m team, who might not have qualified anyway, went out in the heats when they bungled a change-over, straying out of the prescribed area. (Loock 2007:353)

To conclude, previous studies show that appositive relative clauses may perform several pragmatic functions. The functional possibilities that relative clauses have to offer may influence the speaker’s choice to use a construction with a relative clause rather than a different linguistic structure. In the following sections, I will show that the relative clause in il y a clefts can also perform several of these pragmatic functions.

250 Chapter 9 | Other pragmatic functions of il y a clefts

9.4.2 Functions of the relative clause in il y a clefts

Among the three main functions of appositive relative clauses (continuative, expression of relevance, expression of subjectivity), the first can be attributed to prototypical il y a clefts expressing an event: by virtue of their eventive semantics, such il y a clefts have a narrative-advancing function. However, the continuative function is not a factor that influences the speaker’s choice between an eventive cleft and its canonical equivalent, because it is present in both eventive il y a clefts and their canonical equivalents. This is because clefts and their SV equivalents express the same proposition, and hence, if the cleft has eventive semantics (see (348a)), the SV equivalent will also have eventive semantics (see (348b). Therefore, both the cleft and the SV sentence can be claimed to make the narrative advance.

(348) a. Il y a le chat qui arrive. ‘There’s the cat who’s coming.’

b. Le chat arrive.

‘The cat is coming.’

Things are different when it comes to specificational clefts expressing a value-variable relationship. It has been noted that in such instances (e.g. (349) and (350)), the cleft relative clause can be omitted, because it merely repeats discourse-given information. This is why Léard (1992) and Belletti (2008) present the relative clauses in brackets:

(349) – A: Qui est-ce qui a parlé ? ‘Who spoke?’

– B: C’est Jean (qui a parlé).

‘It is Jean (who spoke)’ (Belletti 2008)

(350) – A: Il n’y a personne qui monte. ‘Nobody is going up the stairs.’

– B: Si, il y a Pierre (qui monte l’escalier).

‘That’s not true, there’s Pierre (who’s going up the stairs).’

(Léard 1992:45, my translation)

Chapter 9 | Other pragmatic functions of il y a clefts 251

However, this raises a question that, to the best of my knowledge, has not yet been addressed: why would the speaker resort to a cleft with a seemingly superfluous relative clause rather than using a list-reading existential sentence without a relative clause? The analyses of appositive relative clauses presented in the previous section may provide insight here. I hypothesize that when uttering specificational il y a clefts rather than their ‘shorter versions’ without relative clauses, the speaker uses the relative clause as a slot to express relevance and/or subjectivity.

As for the first function (expressing relevance), this may be especially pertinent in cases where the relative clause repeats a discourse-given variable (quasi) verbatim, as in (351). In my view, the speaker repeats the variable in order to emphasize the relevance of his/her utterance with respect to the QUD. Put somewhat informally, by repeating the question, the speaker indicates: “See, I’m really answering your question”.

(351) – A: Je recherche des modèles de voiture à acheter neuve moins de 10 000 euros, ou aller!?

‘I’m looking for new car models that cost less than €10.000, where should I go?’

– B: bonjours. il y a la citroen c1 qui est a moins de 10 000 euros.

‘Hello. There’s the Citroën C1 that costs less than €10.000.’ (YCCQA)

However, it’s important to note that in most specificational il y a clefts, the relative clause does not repeat a discourse-given variable verbatim: it usually contains at least a small piece of new information. In other words, the type of specificational cleft with a completely discourse-given relative clause, often presented in analyses of clefts (see (349) and (350)), is quite rare in my corpora. Consequently, the relative clause usually cannot be omitted without altering the original proposition expressed by the sentence. In many cases, the extra piece of information is an evaluation on behalf of the speaker/writer, as illustrated in (352) - (354). In all three examples, the writer uses the relative clause to express his/her (positive) judgment about the clefted element (homeopathy is really good, I love Serpico, VLC opens them very well), i.e. the subjectivity function.

252 Chapter 9 | Other pragmatic functions of il y a clefts

(352) – A: Je vais bientôt passer un examen, et j'ai le trac. pouvais vous me donner des conseils?

‘I’ve got exams soon, and I’m really nervous. Can you give me some advice?’ – B: détends-toi, fais aussi du sport, ça permet de bien éliminer le stress. Il y a aussi l'homéopathie qui fonctionne vraiment super bien! ‘Relax, do sports too, it helps to get rid of stress. There’s also homeopathy that works really well!’ (YCCQA)

(353) – A: Quel film pouvez-vous me conseiller ? ‘What movie do you recommend?’

– B: Shining ! Même réalisateur que Orange Mécanique, vraiment un très bon film je trouve ! (…) et de Sydney Lumnet il y a Serpico avec Al Pacino (son premier rôle) que j’adore!

‘The Shining! Same director as A Clockwork Orange, really a good movie in my opinion! (...) and from Sydney Lumnet there’s Serpico with Al Pacino (his first role) that I love! (YCCQA)

(354) – A: Avec quels logiciels libres et gratuits puis-je lire des fichiers vidéos encodés en MP4?

‘With which freeware programs can I open video files encoded in MP4?’

– B: La meilleure sources de logiciel libre est [www]. En ce qui concerne la lecture de mp4 il y a VLC qui les ouvrent très bien.

‘The best source for freeware programs is [www]. As for opening mp4 files, there’s VLC that opens them very well. (YCCQA)

To conclude, rather than superfluous repetitions of a discourse-given variable, the relative clause in specificational il y a clefts are slots that the speaker can use to express relevance and/or subjectivity. These functions may explain why speakers resort to the cleft format instead of a syntactically less complex list-reading existential sentence.

Chapter 9 | Other pragmatic functions of il y a clefts 253

9.5 Conclusion

It is often assumed that clefts are motivated by marking focus or non-topicality. However, authors such as Delin & Oberlander (1995) and Dufter (2009b) argue that a variety of factors, rather than focus/non-topic only, may determine why a speaker uses a cleft instead of a canonical SV sentence. In this chapter, I supported the multifunctional analysis of clefts in general by showing that the same holds for il y a clefts. A close inspection of these clefts and their canonical counterparts in their original context revealed that il y a clefts may have several functions that are absent or signaled less clearly in the canonical equivalents. These functions, which to the best of my knowledge have not been acknowledged in previous studies on il y a clefts, are the following: increasing textual coherence, signaling a narrow scope reading of a singular NP and reinforcing a negation. Furthermore, I identified additional pragmatic functions that the relative clause in specificational il y a clefts can perform (expressing relevance and subjectivity), which helps explain why speakers use the cleft format rather than a list reading existential sentence in those discourse contexts.

All in all, il y a clefts prove to be multifunctional sentences whose use is not only motivated by the desire to avoid indefinite expressions from appearing in preverbal subject position (Chapter 6) or to announce non-topical constituents to the hearer (Chapter 7). In this sense, my analysis is in line with a more general tendency in the literature on information structure: it is becoming increasingly clear that certain word order patterns which used to be considered as dedicated to the expression of focus or topic are in fact multifunctional (see for instance Heidinger 2016 about postverbal constituent order in Spanish, and Bianchi et al. 2015 about focus fronting in Italian). More generally speaking, I agree with Matić & Wedgwood (2012) when they argue that researchers often oversimplify linguistic phenomena by equating them with the expression of focus, disregarding other potentially pertinent motivations that the phenomena may have.

------

Part III: Conclusion

In this Part of the dissertation, I investigated the motivations of il y a clefts, i.e. the question why people use extra structure (il y a… qui) rather than expressing the same proposition by means of an utterance with canonical SV word order. Drawing on prior analyses of il y a clefts, indefinite preverbal subjects and existential sentences, I hypothesized that il y a… qui sentences are motivated by the need to avoid ‘bad’ preverbal subjects (Chapter 6). The corpus data led me to conclude that this hypothesis is basically correct, and that the constituents introduced by il y a display varying degrees of preverbal subject acceptability. I also showed that whether or not an il y a... qui sentence can be declefted depends to a large extent on the subject acceptability of the introduced constituent, which in turn is the outcome of an interaction between different semantic, morphosyntactic and contextual factors. In other words, decleftability is not some special property of sentences that have traditionally been considered as clefts.

I then turned my attention to the link between clefts and information structure, by evaluating two widespread assumptions:

i. Clefts are used to avoid a topic-comment interpretation of the sentence.

ii. Clefts are only motivated by the function of focality and/or non-topicality

In Chapter 7, I examined il y a clefts in light of the claim that clefts avoid a topic- comment interpretation (i.e. assumption (i)). On the basis of the corpus data, I concluded that il y a clefts indeed almost always introduce non-topical constituents. Hence, it is possible to see il y a clefts as markers of non-topicality. However, I argued that it is not the cleft structure that is the source of the function of marking a constituent as non-topical, but the existential expression il y a, because il y a clefts share this function with il y a sentences in general. I also showed that existential sentences without a relative clause can express almost all of the same articulations as 256 Part III | Conclusion

il y a clefts. Moreover, the qualitative analysis of the corpus data revealed that (i) il y a clefts can express a larger variety of different articulations than previously assumed and (ii) some instances remain ambiguous with respect to the mapping op focus.

In Chapter 8, I compared the information structure properties of c’est clefts and il y a clefts. It turns out that those two cleft types have the same potential in terms of the articulation they can express, but the distribution of those articulations is very different. I presented several factors that (partially) explain those differences in distribution.

In Chapter 9, I argued that it is too simplistic to see il y a clefts and focus/non- topic marking as a single form-function pairing (assumption (ii)), because functions such as increasing textual coherence and expressing reinforced negation may also help explain the choice of the cleft format over the canonical equivalent. In this sense, my analysis is in line with Jullien’s (2014) interactional analysis of il y a clefts and related structures and with a more general tendency to acknowledge that structures we once thought were mainly or even exclusively motivated by topic or focus marking are more multifunctional than that.

More generally speaking, I showed that a detailed, discourse-functional scrutinization of corpus data allows us to arrive at a more precise understanding of the information structure and discursive properties of il y a clefts. Following Delin & Oberlander (1995); Dufter (2009b) and colleagues, I thus hope to stimulate more in- depth analyses of the versatility of other (less well known) cleft types and related word order patterns by means of authentic language data.

Part IV

Italian c’è clefts (vs. French il y a clefts)

------

Chapter 10

Italian c’è clefts (vs. French il y a clefts)

10.1 Introduction215

In this chapter, I investigate the Italian counterpart of the il y a cleft, the c’è cleft (355) (see Berruto 1986; Berretta 1995; De Cesare 2007; Sornicola 2010; Marzo & Crocco 2015).216

(355) C’è il gatto che ha fame. there.is the cat that has hunger

‘The cat is hungry.’ (Berretta 1995, my translation)

Like il y a clefts, c’è clefts are said to be typical of spoken rather than of written Italian. This is confirmed by the corpus study presented in Marzo & Crocco (2015), who compare the frequency of c’è clefts in two spoken corpora and one written corpus.217 As noted by Marzo & Crocco (2015), c’è clefts are relatively under-researched. Indeed, c’è clefts have received less attention than French il y a clefts and Italian è clefts (the Italian counterpart of English it clefts and French c’est clefts)218, and have not often been analyzed on the basis of corpus data (but see Marzo & Crocco 2015).

215 Parts of this chapter will be published in Karssenberg et al. (forthcoming). 216 Throughout this chapter, the term ‘c’è cleft’ is used to refer to clefts introduced by the singular form and the plural form of the cleft introducer (c’è ‘there is’ and ci sono ‘there are’). 217 The oral corpora used in their study are LIP and CLIPS, the written corpus is CORIS/CODIS, see Marzo & Crocco (2015) for more details. 218 About Italian è clefts, see e.g. D'Achille et al. (2005); Schöpp (2005); Roggia (2008); De Stefani (2009); De Cesare (2012); Belletti (2013); De Cesare & Garassino (2015, 2016); De Cesare et al. (2016). 260 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

The goal of this chapter is two-fold: to contribute to the understanding of c’è clefts in their own right, and to compare c’è clefts with il y a clefts. Some of the main results of the analysis of il y a clefts so far are summarized in (i)-(v). In this chapter, I will determine whether the same results hold for c’è clefts.

i. Distinguishing between clefts and cleft lookalikes. Il y a clefts display different types of delimitation ambiguities. (Chapter 3)

ii. Information structure - a. Il y a clefts do not introduce ‘regular’ topics; if they introduce a topic, it’s marked (e.g. due to contrastiveness). (Chapter 7)

iii. Information structure - b. Il y a clefts can express several information structure articulations. (Chapter 7)

iv. Link with existential sentences. Il y a clefts display important discourse- functional parallels with il y a sentences that lack a relative clause. (Chapter 7)

v. Discourse functions. Il y a clefts can have additional (pragmatic) functions, e.g. reinforcing a negation. (Chapter 9)

Comparing il y a clefts and c’è clefts may also have wider implications for research on clefts as a cross-linguistic category. È clefts and c’est clefts are sometimes (implicitly or explicitly) assumed to be language-specific instantiations of a single cross-linguistic category of clefts (e.g. research presented in Belletti 2008, 2012 in the generative cartographic framework). One could assume the same position for c’è clefts and il y a clefts, i.e. that they are instances of a cross-linguistic category of ‘existential clefts’ (that is, clefts introduced by existential expressions like il y a, c’è and there is). The hypothesis of a single cross-linguistic construction would be stronger if the two cleft types (introduced by c’è and by il y a) turn out to be similar in many respects. I will show that they indeed display extensive similarities. I will then discuss whether those parallels should be seen as evidence that il y a clefts and c’è clefts are two instantiations of a single cross-linguistic category with the same mental representation in speakers’ mental grammars.

Note that the fact that this chapter about Italian c’è clefts follows the analysis of French il y a clefts may seem to suggest that I ‘merely’ tested whether all discourse- functional properties of French also hold for Italian. In some cases, however, I first

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 261

identified a given property of the Italian data and then went back to check whether the same could be found in French. For instance, I first identified several discourse functions of c’è clefts that cannot be explained by referring to the information structure notions ‘topic’ and ‘focus’ (see section 10.7), and then identified the same functions in the French data (see Chapter 9). In other words, I did not simply try to see whether Italian behaves like French: the comparison between the French data and the Italian data was fruitful ‘in both directions’. 219

The structure of this chapter is as follows: in section 10.2, I give an overview of previous analyses of c’è clefts. In section 10.3, I present the corpus I used (the journalistic corpus La Stampa). I then show (section 10.4) that classifying c’è… che sentences as either clefts or cleft lookalikes leads to the same ambiguities as those observed in Chapter 3 for il y a… qui sentences (see (i) above). In section 10.5, I compare the frequency of c’è clefts in La Stampa with the frequency of il y a clefts in Le Monde. Section 10.6 is devoted to the information structure of c’è clefts (see (ii)-(iv)), and section 10.7 presents two additional functions of c’è clefts (see (v)). Finally, I comment on the status of French il y a clefts and Italian c’è clefts as a cross-linguistic category (section 10.8).

10.2 Previous analyses

I will briefly present previous classifications of c’è sentences (section 10.2.1) and discuss the status of ‘presentational c’è sentences’ as clefts (section 10.2.2). I then give my definition of c’è clefts (section 10.2.3). In section 10.2.4, I present previous accounts of the information structure of c’è clefts. Finally, I comment on the expected frequency difference of c’è clefts with respect to French il y a clefts (section 10.2.5).

219 The heuristic usefulness of cross-linguistic comparison has also been acknowledged by e.g. Lamiroy (1984, 2014): “[la comparaison linguistique] rend le dispositif d’analyse plus puissant : elle peut suggérer, d’une part, de nouvelles hypothèses pour les faits constatés ; elle peut, d’autre part, inciter à réexaminer des hypothèses existantes.” (Lamiroy 1984:224) and “(…) un des points d’intérêt de la comparaison linguistique consiste dans le fait qu’elle oblige à se poser de nouvelles questions, qu’elle stimule donc la recherche.” (Lamiroy 1984:229-230)

262 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

10.2.1 Semantic and functional classifications of c’è sentences

C’è sentences such as C’è il gatto che ha fame ‘The cat is hungry’ are generally referred to as ‘presentational c’è sentences’, because they present a new entity or event (e.g. c’è presentativo in De Cesare 2007; costruzione presentative in Marzo & Crocco 2015, see also Berruto 1986; Berretta 1995; Sornicola 2010; Cruschina 2012; Casalicchio 2013b, 2016; Cruschina 2016). In this section, I will follow those authors by using the term ‘presentational c’è sentence’ rather than ‘c’è cleft’. In the next section, I will motivate the term ‘c’è cleft’, which will be used in the rest of this chapter.

Different classifications of c’è sentences in general have been proposed, in which presentational c’è sentences such as (356) figure more or less prominently.

(356) ‘Presentational c’è sentence’ C’è il gatto che ha fame.

there.is the cat that has hunger

‘The cat is hungry.’ (Berretta 1995, my translation, no prior context)

A distinction is often made between presentational c’è sentences like (356) and c’è sentences with an existential (357) or locative reading (358).220

(357) Existential c’è sentence Ci sono nove pianeti nel sistema solare.

‘There are nine planets in the solar system.’ (Casalicchio 2013b)

(358) Locative c’è sentence C’è un ciliegio nel giardino.

‘There’s a cherry tree in the garden.’ (Sornicola 2010:118)

220 The distinction between existential and locative c’è sentences is not always straightforward and one could argue that it is not always relevant (see discussion about this issue in French il y a sentences in chapter 1.3.1). For instance, one could argue that the c’è sentence in (357) has a locative rather than an existential value: it locates nine planets in the solar system. However, an indication that this sentence is existential is the fact that ci sono ‘there are’ can be replaced by esistono ‘there exist’, whereas this replacement leads to an awkward result in the locative c’è sentence in (358). In my view, there is no clear-cut distinction between locative and existential sentences in Italian or in French. See also the discussion in section 10.2.3.

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 263

Although there is no general consensus about which properties distinguish presentational/locative/existential c’è sentences (an issue aptly referred to as ‘the boundary problem’ by Cruschina 2016), the following differences have been noted:

i. C’è is semantically empty in presentational instances, but has an existential/locative value in (357) & (358) (e.g. Berretta 1995; De Cesare 2007; Cruschina 2012; Casalicchio 2013b; Cruschina 2015b; Marzo & Crocco 2015; Cruschina 2016).

ii. Presentational c’è sentences contain an obligatory pseudo-relative clause (De Cesare 2007; Cruschina 2016).

iii. The constituent introduced by c’è in presentational c’è sentences tends to be specific and is preferably definite (Cruschina 2016:28).

iv. Presentational c’è sentences, in contrast to existential-locative c’è sentences, have eventive semantics (e.g. Cruschina 2015b; Marzo & Crocco 2015).221

v. Presentational c’è sentences, in contrast to existential-locative c’è sentences, can be transformed into a semantically equivalent counterpart with SV or VS word order, i.e. they are decleftable (Cruschina 2016:28).

vi. The relative clause in presentational c’è sentences must be a subject relative clause, whereas (restrictive/appositive) relative clauses combining with existential/locative c’è sentences can be object relative clauses (Cruschina 2016:35).222

These properties are also summarized in Table 32:

221 It is probably more accurate to state that presentational c’è sentences have a stage-level (rather than an eventive) interpretation in order to include instances such as (356) (there’s the cat that’s hungry), see chapter 1.3.2 about stage-level predicates as a criterion for il y a clefts. 222 Cruschina (2016) presents this property as a syntactic test. However, it is difficult to see how this test could be operationalized on authentic language data, since changing a subject relative clause into an object relative clause inevitably changes the proposition expressed by the utterance.

264 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

Presentational C’è sentences with locative or c’è sentences existential interpretation Semantics of c’è Lexically empty Locative or existential value Relative clause Pseudo-relative obligatory No pseudo-relative Constituent introduced Tends to be non-specific, Tends to be specific, definite by c’è indefinite Semantics of whole Eventive Not eventive sentence Sentence cannot be Sentence can be transformed Decleftability transformed into canonical into canonical counterpart counterpart The pseudo-relative clause is Subject/object/adjunct relative (Pseudo-)relative clause necessarily a subject relative clauses are possible clause

Table 32: Properties of presentational, locative and existential c'è sentences in previous studies

10.2.2 An instance of the cleft category?

As far as I know, there are no authors who readily accept to see utterances such as (359) as clefts: in the Italian literature, the label ‘cleft’ is mostly reserved for sentences that are introduced by è ‘(it) is’ such as (360).223

(359) C’è il gatto che miagola. ‘The cat is meowing.’ (Lit. ‘There’s the cat that is meowing.’)

(Casalicchio 2013b:1, my translation)

(360) È cleft Il ritorno di Jobs alla Mela di Cupertino avviene sul nire del 1996 [...]. È in questa seconda fase che Jobs dà il meglio di se stesso in tutti i campi.

‘The comeback to Apple in Cupertino happens at the end of 1996 [...]. It is in this second phase that Jobs gives his best in every field.’

(Repubblica.it, 6.10.2011, Garassino 2016:191)

223 See De Cesare et al. (2016: Part II, section 1) for more references about Italian è clefts.

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 265

Some authors do note the parallels between è clefts and presentational c’è sentences (e.g. Sornicola 2010; Marzo & Crocco 2015; Cruschina 2016), but they nevertheless refrain from using the label ‘cleft’. For instance, Cruschina (2016:3) uses the term ‘presentational ci-sentence’ to refer to sentences such as (359), which he sees as an “‘extension’ of the existential construction which resembles clefts” (my italics). He also uses the term ‘non-prototypical cleft’ and ‘presentational clefts’ as opposed to ‘clefts proper’ (i.e. Italian è ‘it is’ clefts, see Cruschina 2016:38). Similarly, according to Sornicola (2010:122), presentational c’è sentences are not clefts, because they have different pragmatic properties.224 However, this argument is not very convincing, because definitions of clefts generally do not include a pragmatic criterion.

10.2.3 My definition of c’è clefts

In my view, what licenses the designation of ‘cleft’ for c’è sentences such as (359) is that they fulfill the two criteria that are usually attributed to clefts: decleftability and a relative clause that is neither restrictive nor appositive.225 The examples in (361) show that c’è clefts can be transformed into a semantically equivalent counterpart with SV (361b) and/or VS (361b) word order226.

(361) a. C’è una signora che ti cerca. ‘There’s a lady who’s looking for you.’

(Serianni 2000:178-179, in De Cesare 2007:128, my translation)

b. Una signora ti cerca.

‘A lady is looking for you.’

(Serianni 2000:178-179, in De Cesare 2007:128, my translation)

c. Ti cerca una signora.

‘A lady is looking for you.’

224 Sornicola is not very explicit about what these pragmatic differences are, stating that clefts are ‘marked’, while presentational c’è sentences are not marked (see Sornicola 2010:122). 225 For an analysis of pseudo-relatives behind c’è, see Casalicchio (2013c: 124-138). 226 Italian, in contrast to French, can use VS word order to express propositions with an all-focus interpretation (e.g. Lambrecht 1994; Lahousse & Lamiroy 2012). In French, the VS word order pattern has become highly restricted (see e.g. Lahousse 2003, 2005, 2006, 2011).

266 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

The fact that the relative clause in such instances is not restrictive is evident from examples with proper nouns such as the one presented by Cruschina (2016) in (362):

(362) [Context: John and Mary are talking about Silvio’s birthday and, incidentally, Mary says:] C’è Lena che compie gli anni lo stesso giorno.

there.is Lena that finishes the years the same day

‘Lena has her birthday on the same day.’ (Cruschina 2016:23)

Therefore, the term c’è cleft will be used to refer to utterances such as (359), (361a) and (362) in the rest of this dissertation.

Moreover, a note is in order about the distinction between existential, locative and presentational instances of c’è sentences. I believe these designations should not be seen as referring to discrete categories, contrary to what often seems to be implied.227 For instance, according to Cruschina (2016:33), the example in (363) is locative, and not presentational, because the underlined c’è sentences express events that clearly take place at the locations the kitchen and on TV (see Casalicchio 2013b for a similar analysis228).

(363) Sono lì, è domenica pomeriggio, entro in cucina e c’è Cecilia che sta guardando la televisione e mi dice: guarda, c’è Paolo Bonolis che intervista Bilancia, il famoso serial killer. ‘I am there, it’s a Sunday, I go into the kitchen and there is Cecilia who is watching the television and tells me: look, Paolo Bonolis is there, interviewing Bilancia, the famous serial killer.’ (Cruschina 2016:33)

227 See also Bentley (2013:700), who notes that the distinction between Romance existential sentences and presentational sentences (i.e. c’è clefts in my terminology) is subtle. 228 Casalicchio (2013b) presents the example of a French il y a cleft with a locative interpretation in (i), and argues that such sentences differ structurally from true ‘presentational’ instances. According to him, utterances such as (i) should therefore not be seen as instances of clefts. This analysis is at odds with my definition of clefts, because I take decleftability and the presence of a cleft relative clause to be sufficient criteria for clefthood (see chapter 3.7). (i) (Dans la cuisine), il y a Marie qui fait la vaisselle. ‘(In the kitchen), there’s Marie who’s doing the dishes.’ (Casalicchio 2013b, my translation)

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 267

However, it is clear that both underlined sentences also express an ongoing event, and the relative clause is not restrictive, witness the fact that the antecedents are proper nouns. Furthermore, the sentences can be declefted without a change in interpretation (364). These properties are all characteristic of presentational c’è sentences.

(364) … Cecilia sta guardando la televisione e mi dice: guarda, Paolo Bonolis intervista Bilancia … ‘Cecilia is watching TV and tells me: look, Paolo Bonolis is interviewing Bilancia’

In other words, the c’è sentences in (363) fulfill the diagnostics that I use for the definition of clefts. It just so happens that these sentences express two events that are spatially determined.229 This is no surprise, since the semantics of ‘location’ and ‘existence’ and the pragmatics of ‘presentation’ are of course closely related.230

10.2.4 Information structure

C’è clefts are often seen as ‘focalizing structures’ (Berretta 1995; Dardano 1996; De Cesare 2007; Sornicola 2010; Marzo & Crocco 2015). In all examples of c’è clefts presented in the literature, the whole sentence expresses new information (365), and the clefted element is not topical in the sense of ‘that which the sentence is about’.

(365) ALB: posso chiederti quanti anni hai ? mi sembri/molto/giovane dalla voce // ‘Can I ask you how old you are? Your voice sounds very young.’

MIC: infatti // dodici anni //

‘In fact, twelve years.’

229 In my view, it is more useful to state that examples such as (363) combine a locative and a ‘presentational’ value, rather than classifying them as either presentational or locative. 230 See also footnote 221 about the distinction between existential and locative c’è sentences.

268 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

ALB: hai dodici anni // sei un giovane + scusa // c’è Giuseppe che si sta soffiando il naso/in sotto

‘You’re twelve years old, you’re young, sorry, there’s Giuseppe who’s blowing his nose in the back’

(Lablita, Il “DJ Time” di Albertino, De Cesare 2007:137, my translation)

As far as I can tell, it has not yet been investigated whether c’è clefts can also express other information structure articulations (e.g. focus-background, topic-comment), which have been reported in analyses of other types of clefts (see chapter 7.3 about il y a clefts and chapter 8.1 about c’est clefts).

10.2.5 Frequency: French clefts vs. Italian clefts

There is a widespread assumption, promoted most explicitly by Lambrecht (1994, 2001), according to which the frequency of clefts across languages is in a direct relationship with the existence of other focus-marking strategies in those languages (but see Dufter 2009b for criticism, see also chapter 9). For instance, according to Lambrecht (1994), French and Italian speakers have a higher need to use constructions such as il y a clefts or possessive clefts than English speakers, because the latter can use prosody much more freely to achieve the same focalizing effect (see also Lambrecht 1988a:138-139). This is illustrated in (366): Lambrecht (2001) argues that the use of the possessive cleft in French is ‘obligatory’ in this context (366a), while in English, it is possible to express this proposition with SV word order and a pitch accent on ‘foot’ (366b) (i.e. in situ focus marking). As for Italian, Lambrecht states that the use of possessive clefts is optional but not obligatory (366c).231

231 See also Bentley et al. (2015:164-167), whose analysis of different Romance varieties shows that sentence-initial foci only occur in certain dialects (such as the dialect of Buscemi, see also Bentley 2013:696), and are highly restricted in terms of prosody, syntax, semantics and discourse context. Bentley (2013) provides examples that are similar to (366) in (i) and (ii). In English, the focal NP your sister can remain in preverbal subject position if it receives a pitch accent (i). However, this is ‘clearly infelicitous’ in Italian (iia). On the other hand, the c’è sentence in (iib), in which the focal NP is not in preverbal subject position, is a better candidate to express this proposition. (i) Who is in the kitchen? – YOUR SISTER (is in the kitchen).

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 269

(366) Context: Why are you walking so slowly? a. French: Possessive cleft

J’ai mon PIED qui me fait MAL.

b. English: SV with prosodic prominence

My FOOT hurts.

c. Italian: VS word order / possessive cleft

Mi fa male un PIEDE. / Ho un PIEDE che mi fa MALE. (Lambrecht 2001:487)

This line of reasoning leads to the prediction that c’è clefts will be less frequent than il y a clefts, because Italian can express all-focus utterances not only by means of c’è clefts, but also by means of VS word order (see (366c) and (367)) (see e.g. Lahousse & Lamiroy 2012; De Cesare et al. 2016:32), whereas the VS word order pattern has become highly restricted in French (see e.g. Lahousse 2003, 2005, 2006, 2011).

(367) Arriva Maria. ‘Maria is coming.’ (De Cesare 2007:127, my translation)

As far as I know, the prediction that c’è clefts are less frequent than il y a clefts has not yet been verified by means of corpora. Studies about Italian è clefts and French c’est clefts (i.e. Italian and French it clefts) may provide insight into the frequency of clefts more generally. For instance, Roggia (2008) reports that French c’est clefts are much more frequent (334 per 300.000 words) than Italian è clefts (97 per 300.000 words) in the spoken corpus C-ORAL-ROM. De Cesare et al. (2016) also analyze the frequency of French and Italian clefts introduced by c’est and è, and state that Italian è clefts occur less frequently than French c’est clefts (De Cesare et al. 2016:168). However, the difference in frequency that they found between the two languages is much less striking than in other studies:232 French c’est clefts are 1.35 times more frequent than Italian è clefts. The authors argue that a variety of factors such as text type and sociolinguistic factors such as the degree of formality of the text influence the number

(ii) Chi c’è in cucina? ‘Who is in the kitchen?’ a. TUA SORELLA (#è in cucina). ‘Your sister (# is in the kitchen).’ b. C’è TUA SORELLA (in cucina). ‘There is your sister in the kitchen.’ (Bentley 2013:695) 232 See also Dufter (2009b) for a quantitative analysis of è clefts in contrast to other languages.

270 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

of clefts. In this respect, their analysis is in line with the one presented by Dufter (2009b), who criticizes the hypothesis that the frequency of clefts across languages can be accounted for solely as a compensation for the lack of other information structure strategies.233 The corpus data presented in the section 10.5 will further support this view.

10.3 Corpus and extraction

The La Stampa corpus contains the editions of the daily newspaper La Stampa (‘The Press’) from 1996 to 1998 and has an approximate size of 75 million tokens (see Gaeta & Ricca 2002 for more details). I limited the search to the editions of the year 1998 (approximately 25 million tokens), on a par with the French Le Monde (1998) corpus.234 The extraction was done by Daniela Guglielmo (University of Salerno), who used the software program Nooj (see also Karssenberg et al. forthcoming). This led to a total of 2616 sentences introduced by the singular form c’è ‘there is’ and 687 tokens with the plural form ci sono ‘there are’, all followed by a relative clause introduced by che ‘that’. I manually classified all plural forms as clefts or cleft lookalikes, as well as the first 1500 instances of the singular c’è forms. This led to the identification of 152 singular c’è clefts and 79 plural ci sono clefts.235 Section 10.5 will present a comparison of these results with the frequency of French il y a clefts, but I will first address the delimitation ambiguities that I encountered during the coding process.

233 One clear piece of evidence for the fact that the frequency of clefts depends on a variety of factors rather than on the presence or absence of focus marking devices alone is the fact that Portuguese employs clefts much more often than Italian and Spanish, despite the many properties Portuguese has in common with these languages (Dufter 2009b:112). 234 The two corpora are comparable in that they are both based on a national daily newspaper. 235 I used the same definition as the one presented in chapter 3.7 for il y a clefts: both the decleftability criterion and the type of relative clause criterion are necessary, but not sufficient. In other words, only c’è… che sentences that correspond to a canonical equivalent with SV or VS word order AND which do not contain a restrictive relative clause are coded as clefts (see chapter 10.4.2).

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 271

10.4 Delimitation ambiguities

Recall that not all il y a… qui sentences from the corpus data could straightforwardly be classified as clefts or as cleft lookalikes based on criteria from previous studies (chapter 3.4 & 3.5). I identified two main reasons for this:

i. Sometimes the type of relative clause (restrictive or other) is unclear/underspecified (see chapter 3.4)

ii. In many cases, the sentence contains a restrictive relative clause (indicating it is not a cleft), but it can also be declefted, i.e. transformed into a canonical equivalent (indicating it is a cleft) (see chapter 3.5)

In what follows, I will show that the same two delimitation problems occur in the Italian data, i.e. ambiguous relative clause (section 10.4.1) and contradictory results of criteria (section 10.4.2).

10.4.1 Ambiguous relative clauses

In chapter 3.4, I showed that in certain cases, it is unclear whether a relative clause in an il y a… qui sentence is restrictive or not, because the aspectual interpretation of the utterance is underspecified. Given the general assumption that clefts do not contain restrictive relative clauses (RRCs), this apparent ambiguity has consequences for the classification of such cases as clefts or cleft lookalikes. In what follows, I show that the same delimitation problems occur in the Italian data.

Notice, first, that the interpretation of c’è sentences as clefts or as cleft lookalikes, like il y a clefts, can be context-dependent (see also De Cesare et al. (2016:155-157). Thus, considered in its context, the c’è sentence in (368a) cannot be declefted (368b) and the relative clause is interpreted as restrictive: it restricts the reference of ‘cup’ to ‘broken cup’. Hence, (368a) is a cleft lookalike with an existential- locative interpretation. In contrast, the same word sequence is interpreted as a cleft in the context in (369a), witness the fact that it can be declefted without a change in meaning (369b): both sentences express the event ‘the cup broke’.236

236 The examples in (368) and (369) are translations of the French examples in Pierrard (1985:48).

272 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

(368) C’è cleft lookalike with RRC a. Che cosa c’è sul tavolo?’ – C’è la tazza che si è rotta.

‘What’s on the table?’ – ‘There’s the cup that broke.’

b. Che cosa c’è sul tavolo?’ – # Si è rotta la tazza.

‘What’s on the table?’ – # ‘The cup broke.’

(369) C’è cleft a. Che cosa è successo? – C’è la tazza che si è rotta.

‘What happened?’ – ‘There’s the cup that broke.’

b. Che cosa è successo? – Si è rotta la tazza.

‘What happened?’ – ‘The cup broke.’

The fact that the interpretation as a cleft or as a cleft lookalike can be context- dependent opens up the door to ambiguous cases in underspecified or unclear discourse contexts.

The corpus data indeed contain instances in which the relative clause can either be analyzed as a cleft relative clause expressing an event (in which case the utterance is a cleft), or as a RRC (in which case the utterance is not a cleft). This is illustrated by the c’è… che sentence in (370). The sentence can express the fact that a certain event did not take place (nobody intervened), which can also be encoded by a canonical counterpart with SV word order, as in (371). Under this interpretation, the sentence is a cleft.

(370) "Non potevo neanche chiedere aiuto - chiarisce – (…). Vedevo i passanti distogliere lo sguardo da noi, far finta di non accorgersi di quanto avveniva e tirare dritti per la loro strada. Non c'è stato nessuno che si sia sentito in dovere di intervenire per difendere Irene.” “I couldn’t even ask for help”, she clarifies, “(…). I saw passers-by who looked away from us, pretending that they didn’t realize what was happening and continuing their way. There was nobody who felt they had to intervene to defend Irene.’ (Le Stampa)

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 273

(371) Declefted sentence: cleft reading (eventive semantics) Nessuno si è sentito in dovere di intervenire per difendere Irene.

‘Nobody felt they had to intervene to defend Irene.’

However, in a different reading, one could analyze the relative clause as restrictive: in this case, the relative clause does not have an eventive interpretation, but specifies the type of person that was not present at that particular time and place, namely a person who feels like he/she should intervene, as stated in (372).

(372) RRC reading (stative semantics) At that time and place, there was not a single person with the following property: ‘person who felt they had to intervene’

As a second illustration, consider (373). The underlined c’è sentence can again be interpreted as an ongoing event (374a) (many people want to convert the Jews)237 or as the description of a subtype of people (374b) (people with a certain desire: to convert the Jews).238

(373) In Ricorderò domani si celebra anche una sorta di processo agli Stati Uniti: "Onde antisemite si muovono anche lì. Ci sono gli "uomini liberi" nel Montana, dei fascisti; c'è chi, come l'islamico fondamentalista Farrakhan, predica contro gli ebrei; c'è insomma un sacco di gente che vuole convertire gli ebrei. ‘[The book] ‘Ricorderò domani’ also celebrates a kind of process against the United States: “Anti-Semitic waves move there as well. There are the “free men” in Montana, fascists, there are those, like the fundamentalist Islamist Farrakhan, who preach against the Jews; in sum, there are a lot of people who want to convert the Jews.’ (La Stampa)

237 This interpretation would be in line with a recurrent topic on the online forum, i.e. conspiracy theories about the ongoing persecution of either Jews or Arabs. In this sense, ‘many people want to convert the Jews’ would not simply express a desire, but an ongoing action. 238 An indication in favor of a cleft analysis is the fact that this sentence presents the conclusion of a paragraph. This function has been attributed to clefts (e.g. Dufter 2009b).

274 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

(374) a. Declefted sentence: cleft interpretation (eventive semantics)239 Insomma, un sacco di gente vuole convertire gli ebrei.

‘In sum, a lot of people who want to convert the Jews.’

b. RRC interpretation (stative semantics)

C'è insomma un sacco di gente con il desiderio di convertire gli ebrei.

‘In sum, there are a lot of people with the desire to convert the Jews.’

To conclude, in some cases, the aspectual interpretation of the sentence remains underspecified. In the case of an eventive interpretation, the sentence can be analyzed as a cleft containing a cleft relative clause. In the case of a stative interpretation, on the other hand, the relative clause is restrictive, and hence, the sentence is not a cleft according to standard definitions of clefts.

10.4.2 Decleftable + RRC

In chapter 3.5, I discussed il y a sentences that could be declefted, but at the same time contain a RRC.240 The tests in (376a) and (376b) show that the relative clause is restrictive, but at the same time, this il y a… qui sentence can be transformed into a canonical equivalent, as in (376c).

(375) (…) j'pense qu'ils pouvaient pas vraiment faire autrement enfin c'est vrai que y avait+ y avait beaucoup d'élèves qui étaient motivés mais y en avait aussi une bonne partie qui n'en avait rien à faire ‘(…) I think that they couldn’t really do anything differently well it’s true that there were there were many students who were motivated but there was also a large group that didn’t care.’ (CFPP)

239 Notice that this example contains a quantified NP (un sacco di gente ‘a lot of people’). In chapter 3.5, I argued that such instances are often problematic for diagnostics for clefthood. 240 See chapter 3.5, example (131) for more discourse context.

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 275

(376) a. Il y avait beaucoup d’élèves motivés. ‘There were many motivated students.’

b. Il y en avait beaucoup, des élèves qui étaient motivés.

‘There were many of them, motivated students.’

c. Beaucoup d’élèves étaient motivés.

‘Many students were motivated.’

Cases such as (131) are problematic for the general assumption that the relative clause criterion and the decleftability criterion yield the same results when determining whether a given sentence is a cleft or not, i.e. the assumption that sentences with RRCs cannot be declefted and that decleftable sentences don’t contain RRCs.

Italian c’è sentences sometimes lead to the same situation in which the two criteria give conflicting results. For instance, the sentence in (377) can be declefted, as shown in (378).

(377) Abbiamo cominciato un ragionamento su un nuovo soggetto politico e dobbiamo continuare. Anche perché c'è una percentuale crescente di cittadini che non ha modo di identificarsi con nessuno dei partiti esistenti rappresentati in Parlamento. ‘We started a debate about a new political topic and we must continue. Also because there’s an increasing percentage of citizens who have no way of identifying themselves with any of the existing parties represented in Parliament.’ (La Stampa)

(378) SV equivalent possible Anche perché una percentuale crescente di cittadini non ha modo di identificarsi con nessuno dei partiti esistenti rappresentati in Parlamento.

‘Also because an increasing percentage of citizens have no way of identifying themselves with any of the existing parties represented in Parliament.’

However, the relative clause can be analyzed as restrictive: it expresses a property that characterizes its antecedent, and can be replaced by the adjectival phrase impossibilitata a identificarsi ‘with no way of identifying themselves’ (379):

276 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

(379) Adjectival equivalent à relative clause is a RRC Anche perché c’è una percentuale crescente di cittadini impossibilitata a identificarsi con nessuno dei partiti esistenti rappresentati in Parlamento.

‘Also because there’s an increasing percentage of citizens with no way of identifying themselves with any of the existing parties...’

A second example is presented in (380). The relative clause in this c’è… che sentence expresses a state rather than an ongoing event: it specifies a subtype of people, namely Germans who speak Italian. This indicates that the relative clause is restrictive, as confirmed by the fact that the relative clause can be replaced by the prepositional phrase con una buona conoscenza della nostra lingua ‘with a good knowledge of our language’ (381). However, the sentence can also be declefted (see (382)).

(380) Per ultimo devo smentire anche la sua ultima tesi che nei Paesi come la Francia, la Germania e l'Inghilterra ben difficilmente si troverebbe gente che parla l'italiano; almeno per la Germania, dove io vivo da ben 26 anni, questo non vale, perché qui c'è tantissima gente che sa la nostra lingua. ‘Lastly, I must also deny his final argument that in countries such as France, Germany and Great-Britain, it would be hard to find people who speak Italian; at least in Germany, where I’ve been living for 26 years, this is not true, because here there are many people who know our language.’ (La Stampa)

(381) Prepositional equivalent à relative clause is a RRC perché qui c'è tantissima gente con una buona conoscenza della nostra lingua.

‘because here there are many people with a good knowledge of our language.’

(382) SV equivalent possible … perché qui tantissima gente sa la nostra lingua.

‘… because here many people know our language.’

It is no coincidence that the NPs introduced by c’è in the examples above are both quantified (an increasing percentage of citizens, many people). This shows that the same line of reasoning as the one presented in Chapter 6 for il y a sentences holds for

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 277

c’è sentences. 241 Similarly to what I proposed for French, I argue that the c’è sentences presented above can be declefted because the (indefinite) constituents they introduce are acceptable enough as preverbal subjects thanks to the presence of a quantifier.

Examples such as (377) and (380) also show that, as is the case with il y a clefts, it is crucial to be explicit about the way in which the diagnostics are applied. As I explained in section 10.3, when classifying the c’è… che sentences as clefts or as cleft lookalikes, I used the type of relative clause diagnostic and the decleftability diagnostic in the same way as in French: both are necessary, but neither is sufficient by itself. In other words, I only classify c’è sentences as clefts if they do not contain a restrictive relative clause AND if they can be declefted. Therefore, I do not analyze instances such as (377) and (380) as c’è clefts, despite the fact that they can be declefted. However, if one sees c’è clefts as a prototype class, such decleftable c’è… che sentences with RRCs can be seen as peripheral instances of c’è clefts.

10.5 Frequency: Italian vs. French

Table 33 presents the normalized frequency of (singular) c’è clefts in La Stampa in comparison to the relative number of il y a clefts in Le Monde.

Le Monde La Stampa

Corpus size in words 25.700.000 14.325.000242

Total number of clear (il y a/c’è) clefts 81 152

Number of (il y a/c’è ) clefts per 500.000 words 1.6 5.3243

Table 33: Normalized frequency of il y a clefts (Le Monde) and c'è clefts (La Stampa)

241 In this respect, it is important to note that Italian, like French, is a language whose canonical word order is SVO. 242 As stated above, I randomly selected and coded the first 1500 c’è sentences on a total of 2616 extracted tokens. This represents (1500/2616=) 57,3% of all extracted tokens. 57,3% of the entire number of words of the used corpus (25 million) is (25.000.000*.573=) 14.325.000 words. 243 This number does not include plural ci sono clefts, which have a normalized frequency of 2 per 500.000 words (79 clefts in the entire 25 million word corpus).

278 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

What Table 33 shows is that, contrary to the prediction made in section 10.2.5, the number of c’è clefts in La Stampa is not inferior to the number of il y a clefts in the comparable French corpus Le Monde. To the contrary, c’è clefts (5.3 per 500.000 words) are more than three times more frequent than il y a clefts (1.6 per 500.000 words). This is unexpected with respect to the prediction that c’è clefts are less frequent than il y a clefts, which was based on the observation that Italian speakers can also use VS word in order to mark a proposition as all-focus. The findings in Table 33 can thus be seen as further evidence for the fact that from a typological point of view, it is too simplistic to equate the frequency of clefts with the availability of other focus marking strategies (e.g. prosody, VS word order…), as argued in detail by Dufter (2008, 2009b) and De Cesare et al. (2016) (see section 10.3).244 Note, however, that my results do not constitute solid evidence in favor of the idea that c’è clefts are more frequent than il y a clefts, because the frequencies are quite low in both corpora. Rather, the corpus data fail to confirm the prediction that c’è clefts are less frequent than il y a clefts.

A plausible explanation for this unexpected difference between French and Italian is that there is a considerable register difference between informal and journalistic French (e.g. Ayres-Bennett & Carruthers 2001:294). The Italian journalistic register, on the contrary, is evolving towards a more dynamic register that integrates several structures from spoken language and differs substantially from formal Italian (Antonelli 2011). More specifically, journalistic texts are representative of what is called “neo-standard” Italian (italiano neo-standard, see e.g. Marzo & Crocco 2015; De Cesare et al. 2016:16-17 & 35-39; De Pascale & Marzo 2016; Cerruti et al. 2017). As Cerruti et al. (2017) put it:

“At least since the 1980s, it has been claimed that the standard norm of Italian is undergoing substantial changes over the whole Italian peninsula. Italian scholars have interpreted these changes as a process of restandardization, whereby the traditional standard is converging towards spoken, informal and regional varieties. Such a convergence has led to the emergence and diffusion of so- called neo-standard Italian (…).” (Cerruti et al. 2017:4)

244 These findings also confirm the value of consulting corpora in order to verify claims about the frequency of clefts intra- and cross-linguistically more generally.

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 279

As Cerruti et al. argue, journalistic Italian is often seen as one of the clearest examples of neo-standard Italian (Cerruti et al. 2017:9 & 14-15, see also Antonelli 2011). Hence, the appearance of c’è clefts (which are generally seen as characteristic of spoken language) in journalistic Italian is not surprising. This is in contrast with journalistic French, which has not been influenced as extensively by spoken French.

10.6 Information structure

As explained in chapter 7.2.2, several authors have argued that the purpose of clefts is to signal to the hearer that a non-topical or focal constituent is coming (e.g. Lambrecht 1994, 2001). This presumably facilitates processing, because preverbal subjects are preferably topical and non-focal in many languages, among which French and Italian. This hypothesis has been criticized in studies on it/c’est clefts, but I argued that my corpus data do confirm it for il y a clefts: these clefts never introduce a regular topic and the clefted element is always focal. When they do introduce a topic, this constituent is contrastive and, hence, simultaneously topical and focal.

As predicted, the same holds for the c’è clefts in La Stampa. Most c’è clefts have an all-focus articulation (section 10.6.1) or, less frequently, a focus-background articulation (section 10.6.2). The few topical constituents introduced by c’è are marked, because they are either contrastive or partitive (section 10.6.3). Hence, just like il y a clefts, c’è clefts display a wider range of information structure articulations than previously assumed. In section 10.6.4, I present an interim conclusion about the link between c’è clefts and information structure.

Moreover, in chapter 7.5, I showed that il y a clefts display close parallels with existential sentences in terms of the information structure articulations they can express. In section 10.6.5, I show that the same link between c’è clefts, existential sentences and information structure holds in Italian.

280 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

10.6.1 All-focus

In the most typical instances of all-focus c’è clefts analyzed in previous studies, the sentence expresses an event involving an animate referent. Neither the event nor the referent involved in this event are mentioned in the preceding discourse context. The large majority of c’è clefts in La Stampa are indeed of this type. Thus, in (383) - (384), the cleft expresses an ongoing event involving an unknown man committing violent acts, and the cleft in (385) introduces four unknown men waiting for the speaker in a harbor. The events and the referents (a man, a man, four guys) are discourse-new.

(383) Nell'auto dietro a Vignola, c'era un collega, munito di telefono cellulare che ha dato l'allarme: "C'è un uomo che spara per strada.” ‘In the car behind Vignola, there was a colleague with a cellphone who raised the alarm: “There’s a man shooting in the streets.”’ (La Stampa)

(384) "Va bene, allora spostatevi in via Gioberti, c'è un uomo che dà in escandescenze e tenta di abbattere la porta di casa". ‘“Alright, then hurry to ‘via Gioberti’, there’s a man who’s throwing a tantrum and is trying to break down the door of his house”.’ (La Stampa)

(385) Al momento mi accorgo che dentro c’erano quattro tipi che mi aspettavano (…). ‘At this moment I realize that inside there were four guys who were waiting for me (…).’ (La Stampa)

The corpus data contain a considerable number of c’è clefts with an all-focus articulation in which the events are presented as values for a certain variable. What is interesting about these cases is that they combine two functions of c’è clefts: (i) introducing a referent involved in some event (associated with the all-focus articulation), and (ii) enumerating instances of a class (the enumerative or ‘listing’ function, see Rando & Napoli 1978). Consider for instance (386) and (387), which introduce an event (e.g. Ganz che ritrova l’Inter ‘Ganz who goes back to Inter’ in (386), i capigruppo della Camera che decidono… ‘the leaders of the House who decide…’ in (387)) that is an instance of a discourse-given class (cose ‘[ongoing] things’ in (386);

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 281

initiative ‘initiatives’ in (387)). These instances are then followed by other instances of the same class (Ravanelli che… ; Antonio Di Pietro che…).

(386) [Beginning of article] Nella speranza che possa dare torto agli assenti (solo 22.406 biglietti venduti), e che, soprattutto, possa aver luogo, visto l'incombente nebbione, questo derby di servo della tv e dei suoi dispotici orari ruota attorno a una vigilia piena di cose. C'è Ganz che ritrova l'Inter, Ravanelli che (probabilmente) ritorna, Ronaldo che stuzzica la Juve, e c'è, soprattutto, Nicola Berti che annuncia: vado al Tottenham. ‘Hoping that it can prove the absentees wrong (only 22.406 tickets sold) and, above all, that it can take place despite the threatening fog, this Derby Cup Italy – servant of TV and its despotic time schedule – revolves around the day before, which is full of things. There’s Ganz who finds Inter again, Ravanelli who (probably) comes back, Ronaldo who teases Juve, and there’s, above all, Nicola Bertu who announces: I’m going to Tottenham.’ (La Stampa)

(387) [Beginning of article] Con il primo aprile scatta, improvviso, un gran fervore di iniziative sotto le volte di Montecitorio. Ci sono i capigruppo della Camera che decidono di approvare le riforme della Costituzione entro luglio (…). C'è Antonio Di Pietro che si presenta nella sala stampa di Montecitorio, assieme al professore Sartori (…) per annunciare una proposta di legge di iniziativa popolare (…). Tanto movimento sembra indicare (…). ‘Now that April 1st has passed, there is suddenly a great fervor of initiatives in the House of Deputies. There are the leaders of the House who decide to approve the reforms of the constitution by July. There’s Antonio Di Pietro who appears in the pressroom of the House of Deputies, together with Professor Sartori, to announce a bill of popular initiative. So much movement seems to indicate (…).’ (La Stampa)

The enumerative function is usually associated with clefts that have a focus- background articulation (e.g. Davidse 2000; Lambrecht 2001, see also chapter 1.3.4). However, the examples above show that speakers are able to combine the

282 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

enumerative semantics of c’è with the all-focus information structure articulation and the expression of an event.

To summarize, like il y a clefts, many c’è clefts express an all-focus articulation in which the clefted element is not topical. In some cases, the whole event expressed by the c’è cleft is presented as an instance of a more general class (i.e. the enumerative function).

10.6.2 Focus + background

To the best of my knowledge, it has not yet been acknowledged that c’è clefts, like il y a clefts and there clefts, can be semantically specificational and have a focus- background articulation. The La Stampa corpus only contains less than a dozen of such cases, which may explain why this type of c’è cleft has so far remained unnoticed.245 As an illustration, consider (388), in which the relative clause (who can take on this role) relates to the role of ‘representing minors’ and is therefore discourse- given. The c’è cleft is semantically specificational: ‘take on this role (of representing minors)’ is the variable and the cleft indicates that this variable is already satisfied by the value ‘the prosecutor’.246

(388) Molti chiedono anche l'istituzione di una figura che rappresenti il minore: l'avvocato dei bambini. "Non sono d'accordo - osserva l'onorevole Lucidi - perché c'è già il pm che può rivestire questo ruolo, magari si può rafforzare il suo potere d'intervento". ‘Many people ask the institution to provide a figure who represents minors: a children’s advocate. “I disagree – observes Mr. Lucidi – because there’s already the prosecutor who can take on this role, maybe his right to intervene can be strengthened”.’ (La Stampa)

245 It is possible that more instances of focus-background c’è clefts can be found in other registers, such as the online Yahoo Questions and Answers forum (Italian version). 246 Observe that all focus-background c’è clefts presented here contain a focus particle (già ‘already’, solo ‘only’, non solo ‘not only’). I comment on this below.

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 283

In terms of information structure, il pm ‘the prosecutor’ can be seen as focal, because the new and salient information that is provided by the sentence is that the prosecutor already takes up the role of representing minors.

In a similar way, the c’è cleft in (389) evokes the variable people who are afraid of communists, and the value that fulfils this variable is ‘only Berlusconi (still is)’.

(389) [“Why do you think the right-wing politician said outrageous things about homosexuals?”] "A me pare che questa gente stia cercando qualcosa di qualificante per far presa sugli elettori. E' un modo di compattare la destra agitandogli davanti un drappo colorato. Ormai l'anticomunismo non basta più, c'è solo Berlusconi che ha ancora paura dei comunisti in questo Paese... E allora perché non puntare sugli omosessuali?”

‘It seems to me that these people are trying to find something to put pressure on the voters. It’s a way to unite right-wing politics, waving a colored drape in front of them. Anticommunism is not enough anymore now, there’s only Berlusconi who’s still afraid of communists in this country... So why not focus on homosexuals?’ (La Stampa)

‘Only Berlusconi’ is contrasted with respect to ‘everyone else’, as implied by the preceding statement that anticommunism is not enough anymore (for people in general).

As a final illustration of the focus-background type, consider (390). Since this cleft is the first sentence of an article, nothing can be assumed to be discourse-given. However, the use of non solo ‘not only’ indicates the council is not the only entity that is ‘trembling’ (i.e. ‘has internal problems’). The following discourse context then elaborates on this: the Torino shopkeepers are also internally divided.

284 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

(390) [Beginning of article] 247 Non c'è solo la giunta che vacilla. Anche tra i commercianti torinesi si è aperta una profonda spaccatura con posizioni contrapposte. ‘There’s not only the council that’s trembling. Even among the shopkeepers in Torino a profound split has emerged with opposing positions.’ (La Stampa)

Thus, ‘not only the council’ is contrasted with ‘the Torino shopkeepers as well’, which can be seen as two values for the overarching variable ‘people/entities that are trembling/internally divided’.

Something that sets the focus-background c’è clefts presented above apart from their French equivalents is that all of them contain a focus particle (già ‘already’, solo ‘only’, non solo ‘not only’).248 These particles reinforce the contrast that is evoked between the clefted element and some contextually salient alternative (e.g. not only the council vs. also the shopkeepers are trembling/divided in (390)). Most French focus- background il y a clefts, on the other hand, do not contain such an explicit contrast. For instance, the clefted element le Magellan Maestro 3250 in (104) is not contrasted with another contextually salient GPS system.

(391) – A: Est-ce qu'il y a des GPS a reconnaissance vocal disponible au Québec? ‘Are there GPS systems with speech recognition for sale in Quebec?’

– B: Oui il y a le Magellan Maestro 3250 qui a la reconnaissance vocale tel qu'expliquer sur le site de CAA Québec: www

‘Yes there’s the Magellan Maestro 3250 that has speech recognition as explained on the CAA Quebec website: www’ (YCCQA)

247 I do not have information about the title of this newspaper article. 248 Note, however, that not all instances of focus-background c’è clefts contain such a focus particle, as illustrated by (i). In this cleft, the relative clause refers to previously mentioned nominations, and the clefted element specifies who takes care of this task (my daughter). (i) Alla domanda sulle nomine in programma lunedì nel consiglio Mediaset, Berlusconi è stato evasivo: “C'è mia figlia che se ne occupa.” ‘When asked about the appointments scheduled next Monday at the Mediaset council, Berlusconi was evasive: “There’s my daughter who takes care of that.”’ (La Stampa)

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 285

An analysis of c’è clefts in other corpora may prove whether the presence of explicit indications that favor a contrastive interpretation is a peculiarity of c’è clefts in La Stampa or whether it is a property of focus-background c’è clefts in general.249

10.6.3 Contrastive/partitive topic + comment

I identified a small number of c’è clefts in the Le Stampa corpus with topical clefted elements. In all instances, the topic is atypical, because it is either contrastive or partitive (see prediction [ii] in the Introduction of this chapter).

Recall that il y a clefts can introduce a contrastive aboutness-topic followed by a comment about this constituent, encoded by the relative clause (chapter 7.3.5.2).250 It turns out that c’è clefts can have the same articulation. A first example is given in (392). The clefted element il portiere ’the goalkeeper’ in (392) is mentioned in the preceding discourse context, and is the general topic of the newspaper article. A contrast is evoked between the middlefielder and the defender on the one hand (who could make a mistake) and the goalkeeper on the other hand (who can make up for it).251

(392) [Beginning of article] Lido Vieri, preparatore dei portieri del Toro, è anche uno studioso di questo ruolo speciale, diverso da tutti gli altri. Forse, il più delicato sotto il profilo psicologico. Se un attaccante, un centrocampista o un difensore sbagliano, c'è il portiere che può metterci una pezza. Se sbaglia lui, non c'è rimedio. E il portiere è uno degli elementi-cardine, si punti al primo posto o si lotti per la salvezza.

249 If explicit focus markers are indeed more frequent across registers in Italian than in French, one could hypothesize that this is because focus-background c’è clefts are not (yet) as well established in Italian as focus-background il y a clefts are in French, and that c’è clefts therefore need a focus particle to ‘reinforce’ the focal nature of the clefted element. 250 This is illustrated by (i): (i) « Nous, on a de la chance, on a des employeurs corrects. Mais il y a plein de petites jeunes qui se font avoir », poursuit Lina. ‘”Us, we’re lucky, we have good employers. But there are plenty of youngsters who are getting ripped off”, continues Lina.’ (Le Monde) 251 This contrast is further reinforced by the postverbal subject lui ‘him’ (referring to the goalkeeper) in the sentence following the cleft.

286 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

‘Lido Vieri, trainer of the goalkeepers of Toro, also studies in depth this special role, different from all the others. Maybe even the most delicate one psychologically. If an attacker, a midfielder or a defender make a mistake, there’s the goalkeeper who can make up for it. If he makes a mistake, there’s no remedy. And the goalkeeper is one of the pivotal elements, he is in first position and struggles for salvation.’ (La Stampa)

The topic test in (393) shows that the goalkeeper can be seen as that which the sentence is about. Therefore, the goalkeeper can be analyzed as a contrastive topic.

(393) – A: Per quanto riguarda il portiere, cosa c’è di speciale a proposito di lui? ‘Regarding the goalkeeper, what’s special about him?’

– B: Se un attaccante, un centrocampista o un difensore sbagliano, c'è il portiere che può metterci una pezza.

‘If an attacker, a midfielder or a defender make a mistake, there’s the goalkeeper who make up for it.’

As I argued for il y a clefts introducing a contrastive topic (chapter 7.3.5), contrast can be seen as a survival strategy for the topical NP in a hostile environment (behind il y a/c’è).252 This analysis is confirmed by the fact that without the se ‘if’-clause which includes the referents with respect to which the goalkeeper is contrasted, the c’è cleft no longer passes the topic test:

(394) – A: Per quanto riguarda il portiere, cosa c’è di speciale a proposito di lui? ‘Regarding the goalkeeper, what’s special about him?’

– B: # C’è il portiere che può salvare il gioco.253

# ‘There’s the goalkeeper who can save the game.’

252 See Leonetti (2013) for this line of reasoning about nominal expressions in other sentence types. 253 The original relative clause che può metterci una pezza ‘who can make up for it’ was replaced by che può salvare il gioco ‘who can save the game’, because otherwise the sentence may be judged inappropriate due to the anaphoric link that is implied by ‘make up for it’.

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 287

As a second example, consider (395). Here, the two referents il ministro Bindi ‘minister Bindi’ and il Polo ‘the Pole’ (a political alliance between several parties) are contrasted: they have opposing reactions to the scandal about a doctor’s unorthodox methods. These reactions are expressed by the two relative clauses: minister Bindi is more careful, while il Polo is completely behind the new method, and uses it to make a bigger point.254

(395) [Beginning of article] “Il professor Di Bella e la sperimentazione” Il metodo adottato dal professor Luigi Di Bella contro il cancro sta ormai divenendo sempre più materia di scontro politico. Da una parte c'è il ministro Bindi che cerca di avviare un percorso di sperimentazione stabilito in un decreto appositamente approvato. Dall'altra il Polo che, pur non entrando nel merito della cura (come potrebbe?), giura sulla totale affidabilità del metodo del professore di Modena, facendo del metodo stesso un simbolo della libertà di cura. ‘”Professor Di Bella and experimentation” The method used by professor Luigi Di Bella against cancer is becoming more and more a subject of political debate. On the one hand, there’s minister Bindi who is trying to pave the way for an experimental path that would be approved in a special decree. On the other hand il Polo [political alliance] which, while not commenting on the merits of the treatment (how could they?), swears on the complete reliability of the professor’s method, turning the method itself into a symbol of freedom of treatment.’ (La Stampa)

A final illustration of ‘topic + comment’ is presented in (396). This example differs from the ones presented above, because the clefted element is not a contrastive topic, but a partitive one. In this c’è cleft, the clefted element una ‘one (of them)’ refers to i giornalisti ‘the journalists’ in the previous sentence.255

254 The contrasts expressed in (392) and (395) are not encoded in the cleft construction, but can be derived from the context: taken out of context, neither of these c’è sentences is ‘inherently’ contrastive. In other words, rather than stating that c’è clefts are special linguistic constructions dedicated to the expression of contrast in these instances, it is more accurate to claim that the NP introduced by the c’è cleft receives a contrastive reading in certain contextual environments. 255 There are no other referents in the prior context that una ‘one of them’ could plausibly refer to.

288 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

(396) L'ultima volta che un asteroide precipitò sulla Terra, nella penisola dello Yucatan, provocò l'era glaciale e la fine dei dinosauri. Naturalmente i giornalisti ne escono malconci. C'è una che chiede: "Riuscirete ad evacuare la città prima dell'arrivo dell'asteroide?". Le rispondono: "Riusciremmo, se non fossimo continuamente interrotti dalle vostre domande". ‘The last time an asteroid crashed on earth, in the Yucatan Peninsula, it caused the ice age and the extinction of dinosaurs. Of course, the journalists do not remain unharmed. There’s one [of them] who asks: “Will you be able to evacuate the city before the asteroid hits?”. They answer her: “We would be, if we weren’t continuously interrupted by your questions.”’ (La Stampa)

In this instance, the topic test indicates that the c’è sentence can answer a question about the referent, and that the referent can therefore be considered a topic.

(397) – A: Per quanto riguarda i giornalisti? ‘What about the journalists?’

– B: Beh, c'è una che chiede: …

‘Well, there’s one [of them] who asks:…’

However, una ‘one [of them]’ is a rather atypical topic, because it is a specific indefinite. According to Erteschik-Shir (2007), such topics are topical and focal at the same time. For instance, the NP a person I know in (398a) is the topic of the sentence, and has the ‘subordinate information structure’ in (398b). A person is focal, but the first person subject ‘I’ in I know is topical (see (398b)) and therefore licenses the use of the whole constituent a person I know as the topic of the sentence in (398a). Erteschik-Shir (2007:52) argues that the information structure of ‘a person I know’ is equivalent to the information structure of the sentence in (398c).

(398) a. A person I know is famous.

b. A personfoc [Itop know_]

c. Itop know [a person]foc

(Erteschik-Shir 2007:52, indication of ‘focus’ and ‘topic’ by the author)

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 289

The same analysis can be applied to una in (396), although the modifier that makes the indefinite specific by relating it to some previously established referent (‘I’ in (398a), ‘of the journalists’ in (396)) is left implicit in Italian, as indicated by the square brackets in (399).

(399) Unafoc [dei giornalistitop]

‘Onefoc [of.the journaliststop]’

In this sense, Italian differs somewhat from other languages such as English and French, in which a modifier relating the pronoun to a previously established referent (of them, en ‘of.it’) is almost obligatory in such contexts (400) - (401):256

(400) English (There’s) one of them (who) asks

(401) French Il y en a une qui demande

EXPL there of.it has one that asks

‘There’s one of them who asks: …’

To conclude, the topics introduced by c’è discussed above are contrastive (see (392) & (395)) or partitive (396). In all three cases, their interpretation implies a focal feature. Hence, there are no regular aboutness-topics in the c’è clefts in my corpus data.

256 Interestingly, several native speakers indicate that for them, the cleft in (396) would be more natural if such a modifier is added, i.e. ce n’è una che chiede ‘there is one of them who asks’. This is in line with the fact that a Google search for the word sequence “ce n’è uno che” yields ‘about 250.000 results’, whereas the search string “c’è uno che” yields ‘about 4.510 results’.

290 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

10.6.4 Interim conclusion: c’è clefts and information structure

Given the findings presented in the previous sections, c’è clefts seem to have the same information structure properties as il y a clefts. Like il y a clefts, c’è clefts display a larger array of possible information structure articulations than previously assumed (that is, not only all-focus). Importantly, there are no regular topics in the corpus data of c’è clefts, and the few topical constituents are contrastive or partitive and, hence, bear a focus feature. Moreover, it is hard to construe an example of a c’è cleft containing a non-contrastive and non-partitive topic, as in (402).257

(402) – A: Che cosa sta facendo il gatto? ‘What is the cat doing?’

– B: # C’è il gatto che miagola.

# ‘There’s the cat that’s meowing.’

Given the fact that c’è clefts do not and seemingly cannot introduce regular topics (without a focus feature), I hypothesize that c’è clefts are processing cues that announce to the hearer that a non-topical and focal constituent is coming. However, as I claimed for French (chapter 7.4), the fact that all c’è sentences (i.e. c’è clefts and c’è sentences without relative clauses) introduce focal rather than topical constituents (see Bentley 2013; Bentley et al. 2015) indicates that it’s not the cleft format that functions as a processing cue, but rather the expression c’è only. Bentley (2013) illustrates the fact that existential c’è sentences cannot introduce regular topics by means of the constructed example in (403). The existential sentence in (403b) is infelicitous, she argues, because the introduced NP i sugaman ‘the towels’ is topical due to the preceding discourse context (‘where are the towels?’). In order to answer this question, a sentence with SV word order must be used, as in (403c).

257 In my view, the difficulty to invent a given type of sentence is, however, a less convincing argument than the absence of that sentence type in corpus data.

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 291

(403) Bellunese Italian a. E-li andéi sugaman?

‘Where are the towels?’

b. #L’é i sugaman te la casèla.

‘There are the towels in the drawer.’

c. I sugaman i é te la casèla.

‘The towels are in the drawer.’ (Bentley 2013:686, glosses omitted)

In the next section, I take the parallels between il y a sentences and c’è sentences one step further by showing that the different information structure articulations expressed by c’è clefts can also be expressed by c’è sentences without relative clauses. In this sense, c’è clefts are again similar to il y a clefts, which also have the same information structure potential as existential sentence (chapter 7.5).

10.6.5 C’è clefts and existential sentences: the same articulations

Firstly, all-focus c’è clefts such as (404) are on a par with typical instances of existential-locative c’è sentences such as (405) and eventive existentials such as (406), which also have an all-focus interpretation.

(404) All-focus c’è cleft (eventive semantics) [Che succede? ‘What’s happening?]

C’è un gatto che miagola.

‘There’s a cat that’s meowing.’

(405) All-focus locative c’è sentence – Che succede?

‘What’s happening?’

– C’È IL GATTO IN CUCINA. Fallo uscire.

‘The cat is in the kitchen. Show him out.’ (Bentley 2013:700)

292 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

(406) Eventive existential c’è sentence Ci fu un terremoto.

‘There was an earthquake (then).’ (Cruschina 2016:15, no prior context)

Secondly, specificational c’è clefts such as (407), in which the clefted element (Berlin) is focal and the relative clause is the background, resemble list reading c’è sentences like (408) (see Leonetti 2008 for more examples of this type). In this list reading c’è sentence, alcuni negozi ‘a couple of shops’ is focal, whereas the PP sul lungomare ‘on the promenade’ constitutes the background due to the preceding question.

(407) Focus-background c’è cleft (specificational semantics) – A: Quali sono le tue città tedesche preferite?

‘Which are your favorite German cities?’

– B: C’è Berlino che mi piace, e anche Amburgo e Monaco.

‘There’s Berlin that I like, and also Hamburg and Munich.’

(408) List reading c’è sentence – Che cosa c’è sul lungomare?

‘What is there on the promenade?’

– Ci sono ALCUNI NEGOZI sul lungomare.

‘There are a couple of shops on the promenade.’ (Bentley 2013:695)

Finally, recall that c’è clefts can introduce a contrastive topic, as in (409). In such instances, the clefted element (the goalkeeper) is contrasted with respect to another referent (an attacker, a midfielder or a defender).

(409) C’è cleft introducing a contrastive topic Lido Vieri, preparatore dei portieri del Toro, è anche uno studioso di questo ruolo speciale, diverso da tutti gli altri. Forse, il più delicato sotto il profilo psicologico. Se un attaccante, un centrocampista o un difensore sbagliano, c'è il portiere che può metterci una pezza. Se sbaglia lui, non c'è rimedio.

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 293

‘Lido Vieri, trainer of the goalkeepers of Toro, also studies in depth this special role, different from all the others. Maybe even the most delicate one psychologically. If an attacker, a midfielder or a defender make a mistake, there’s the goalkeeper who can make up for it. If he makes a mistake, there’s no remedy.’ (La Stampa)

Similar instances of c’è sentences without relative clauses can be found, as illustrated in (410). In these cases, io ‘me’ is contrasted with some contextually salient referent (‘beautiful girls’ in (410a) and (410b), ‘girls with boyfriends’ in (410c)). In the first two cases, the property that distinguishes ‘me’ from the other referents is contextually salient, but not repeated by the c’è sentence (‘I am not beautiful’). The c’è sentence in (410c) does linguistically express what opposes ‘me’ with respect to ‘the other girls’ (apart from the fact that they have a boyfriend): the adjectival phrase da sola ‘alone’ and the PP in questa casa di merda ‘in this goddamn house’.

(410) Existential c’è sentences introducing contrastive topics a. Ci sono certe ragazze bellissime. E poi ci sono io.

‘There are certain beautiful girls. And then there’s me.’

b. Perché ci sono ragazze così belle, e poi ci sono io?

‘Why are there such beautiful girls, and then there’s me?’

c. Tutte che passano il sabato con il proprio ragazzo, e poi ci sono io, da sola, in questa casa di merda.

‘All the girls who spend Saturday with their boyfriend, and then there’s me, alone, in this goddamn house.’ (www)258

The above-described parallels between c’è clefts and c’è sentences without relative clauses indicate that the ability of expressing certain information structure articulations should not be seen as a unique property of the c’è cleft format. Rather, the existential main clause c’è can be combined with different building blocks (relative clauses, prepositional phrases…) in order to achieve similar effects.259

258 https://www.tumblr.com/search/"e+poi+ci+sono+io", last consulted on July 4th 2017. 259 I will return to this in the next chapter, where I analyze il y a clefts (and c’è clefts) as compositional structures consisting of different building blocks (Chapter 11).

294 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

10.7 Additional functions of c’è clefts

In Chapter 9, I showed that contrary to what often seems to be assumed, il y a clefts are not only used to introduce non-topics and express certain information structure articulations: they can also have other pragmatic functions.260 The same holds for c’è clefts, which can be used to reinforce a negation (section 10.7.1) and to signal that a singular NP should not be interpreted as referring to a single entity (section 10.7.2).

10.7.1 Reinforcing negation

In chapter 9.3, I argued that il y a clefts can be used to reinforce a negation.261 The same is true of c’è clefts containing negations such as non c’è niente che ‘there’s nothing that’ (411), non c’è un milligrammo che non… ‘there’s not a (single) milligram that he didn’t…’ (412)262 and non c’è uno che ‘there’s not a (single) person that’ (413):

(411) E allora perché l'"Avvenire" vi accusa di essere succubi della sinistra? "C'è chi ha l'abitudine di considerare anti-costituzionale tutto ciò che non li trova d'accordo. Mi spiace, ma nella nostra Costituzione non c'è niente che impedisca una legge sulla fecondazione eterologa.” ‘Then why does “Avvenire” accuse you of being submissive to the left? “There are people who have the habit of considering as ‘anti-constitutional’ anything that they don’t agree with. I’m sorry, but in our constitution there’s nothing to stop a law on heterologous artificial insemination.”’ (La Stampa)

(412) Ammettiamolo: non c'è un milligrammo della sua fama che non sia meritato. ‘Let’s admit it: there isn’t a milligram of his fame that he didn’t earn.’ (La Stampa)

260 As I mentioned in the introduction of 0, the question whether clefts are only motivated by focus- marking is explicitly raised by authors like Delin & Oberlander (1995, 2005) and Dufter (2006) (“Dass nun die Satzspaltung ein wichtiges Mittel der Fokusmarkierung darstellt, ist unbestritten. Unklarheit herrscht jedoch weiterhin hinsichtlich der Frage, ob dieses syntaktische Format ausschließlich dazu dient, den abgespaltenen Ausdruck als Fokus und den subordinierten Satz als Hintergrund zu markieren”, Dufter 2006:43, my boldface). 261 See chapter 9.3 for a discussion of whether this function should be seen as a case of focalization. 262 Observe that this utterance contains a double negation (there’s not a single… that he didn’t…).

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 295

(413) Rabbia per uno scippo che ai ladri frutterà pochi soldi e a lei tanti grattacapi: le hanno rubato 50 mila lire in contanti, un telefonino da rivendere a qualche ricettatore e i documenti. E rabbia per l'indifferenza della gente. "Quando mi sono rialzata tutti hanno visto com'ero conciata, ma nessuno ha mosso un dito. Non c'è stato uno che si sia avvicinato e mi abbia teso la mano”. ‘Anger about a robbery that the thieves earn little money with and that leaves her with a big headache: they stole 50 thousand lira in cash, a mobile phone to sell and some other goods and documents. And anger about the indifference of the people around her. “When I got up, everybody could see how dirty I was, but nobody lifted a finger. There wasn’t a single person who came closer and gave me a hand”.’ (La Stampa)

Such sentences have a SV equivalent263 in which the negative quantifier niente ‘nothing’ or nessuno ‘nobody’ is sentence-initial ((414) & (416)), or the double negation (non c’è… che non… ‘there isn’t… that isn’t…’) is replaced by ogni X ‘every X’ (see (415)).264

(414) Niente impedisce une legge sulla fecondazione eterologa. (= (411)) ‘Nothing stops a law about heterologous artificial insemination.’

(415) Ammettiamolo: ogni milligrammo della sua fama è meritato. (= (412)) ‘Let’s admit it: every milligram of his fame is well earned.’

(416) Nessuno si è avvicinato e mi ha teso la mano. (= (413)) ‘Nobody came closer and gave me a hand.’

263 Informants indicate that it is infelicitous to transform (411) and (413) into sentences with VS word order, as shown in (i) and (ii). The utte rance in (412) does correspond to a VS equivalent (iii). (i) ???/* Impedisce une legge sulla fecondazione eterologa niente. (ii) ??? Si è avvicinato e mi ha teso la mano nessuno. (iii) Ammettiamolo: è meritato ogni milligrammo della sua fama. 264 Recall that declefting sometimes requires a lexical adjustment (see chapter 6.4.5, see also Lambrecht 2001:499; Dufter 2006:38).

296 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

The propositional content of the SV equivalent remains the same, but the cleft version differs from the SV version in that the cleft reinforces the negation by placing it in a special syntactic position. Other forms of reinforced negation in Italian have also been reported, such as double negation (Godard & Marandin 2006) and the expression (non) mica (Cinque 1991; Mosegaard Hansen & Visconti 2009).265 Sentences that contain these forms of reinforced negation often express a speaker’s subjective attitude towards some state of affairs (see e.g. Godard & Marandin 2006: 178).266 C’è clefts with negative expressions are also found in subjective contexts.267 For instance, in (413), the speaker expresses surprise and frustration about the fact that no one helped her out.

10.7.2 Signaling that a singular indefinite does not refer to a single referent

Recall that il y a clefts can be used to indicate that a singular indefinite referent should not be interpreted as referring to a single referent (chapter 9.2). I argued that in such clefts, by virtue of their syntax, the existential quantifier has narrow scope with respect to a universal quantifier, and that this scope relation is less clear in the SV equivalent.268

265 While the use of mica is a lexical way to achieve reinforced negation, the use of c’è clefts such as (411) - (413) makes the negation more emphatic by means of the biclausal syntax of clefts. 266 The authors illustrate this with the example of double negation (nessuna… non ‘nobody… didn’t’) in (i). According to the authors, informants report that B’s answer in (i) is not a “straightforward answer” but expresses a speaker’s attitude (Godard & Marandin 2006:178). (i) A: I tuoi studenti hanno riuscito l'esame? (‘Did your students do well at their exam?’) B: Nessuna questione di SINTASSI (quei cretini) non sono riusciti a risolvere! Le due questioni di semantica, invece, tutti le hanno risolte. (‘None of the syntax question did they manage to answer, the idiots, but the two semantic questions, they all solved them’) (Godard & Marandin 2006:178) 267 Notice also the use of the subjunctive in the relative clauses in (411) - (413). 268 One of the examples of such il y a clefts is repeated in (i). In this example, un type ‘a guy’ does not necessarily refer to the same guy each time the habitual event occurs. (i) « Dans ce milieu, poursuit-il, n'importe quel cadet qui brille un peu est, un jour ou l'autre, confronté au problème du dopage. A un moment ou à un autre, il y a toujours un type qui essaie de l'approcher. »

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 297

The same function can be attributed to c’è clefts. Consider for instance (417). A singular indefinite NP is used in combination with the adverbial expression ogni giorno ‘every day’ and the sentence therefore expresses a habitual event.

(417) [Interview between author of the article and a politician] Da quando si parla di giustizia in Bicamerale, ogni giorno, c’è un procuratore che ripete: non ci lasciano lavorare, i processi non si fanno per colpa della classe politica... "Se si riferisce alle dichiarazioni di Borrelli al Corriere, non mi interessano. (…)". Mi riferisco anche a Colombo.

‘Ever since there is talk about justice in the Bicameral, every day there’s a prosecutor who repeats: they don’t let us work, the trials don’t take place because of the political class…”If you’re referring to the statements made by Borrelli to Corriere, I’m not interested in them. (…)” I also refer to Colombo.

(La Stampa)

In this example, the singular NP does not refer to a single specific referent: each time the event occurs (a prosecutor complains), it is not necessarily the same prosecutor who repeats his/her complaint. The fact that the NP does not refer to a single prosecutor is confirmed by the following discourse context: the interviewee assumes that the interviewer refers to Borrelli (“if you are referring to the statements made by Borrelli in the newspaper Corriere, …”), but the interviewer/author of the article states that he/she also refers to Colombo. In other words, in (417), un procuratore ‘a prosecutor’ refers to at least two people. Informants indicate that this reading is less straightforward in the SV equivalent of the cleft in (418).269 The preverbal subject un procuratore ‘a prosecutor’ is more likely to be interpreted as referring to a single, specific referent.

‘“In this social environment, he continues, any youngster who stands out a little will be confronted with drugs one day or another. At some point, there’s always some guy who tries to approach him.” (Le Monde) 269 In all the examples discussed in this section, transformation into a VS equivalent is deemed unnatural or grammatically incorrect by native speakers of Italian: (i) ? Ogni giorno, ripete un procuratore: non ci lasciano lavorare. = (417) (ii) ?? L’aveva sempre detto qualcuno. = (419) (iii) ??? Tirava sempre fuori “gli scheletri dall-armadio” qualcuno. = (420) (iv) ??? Va sempre ad analizzare le canzoni con la lente d’ingrandimento qualcuno. = (421)

298 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

(418) Ogni giorno, un procuratore ripete: non ci lasciano lavorare. ‘Every day, a prosecutor repeats: they don’t let us work.’

As explained in the analysis of similar il y a clefts (chapter 9.2), this observation is in line with Kurtzman & MacDonald’s (1993) “single reference principle”, according to which indefinite NPs in subject or topic position are taken to refer to a singular entity that takes wide scope over quantifying expressions such as every. Therefore, I argue that the use of the c’è cleft in this example is motivated by the need to signal that the NP should not be interpreted as referring to a single, specific entity. The cleft achieves this goal by putting the NP in a position that is not a subject or a topic position. In other words, although the intended ‘non-singular’ reading is possible in the SV sentence in (418), it is signaled more clearly by the c’è cleft in (417).

The same analysis holds for the examples in (419)-(421), in which qualcuno ‘someone’ does not necessarily refer to a single, specific person.

(419) E mentre Cimoli, alle sei del pomeriggio va in prefettura per il doveroso vertice, dalla Fisafs-Cisal rilanciano: "Quella è una tratta a rischio, li avevamo avvertiti!". C’è sempre qualcuno che l’aveva detto. [End of article] ‘And while Cimoli goes to the prefecture at six in the afternoon for the obligatory summit, those at Fisafs-Cisal repeat: “That’s a risky bill, we warned you!”. There’s always someone who (already) said so.’ (La Stampa)

(420) Diceva Sergio Escobar, che dopo neppure due anni ha lasciato l'Opera per il Piccolo Teatro di Milano, che ad ogni nuova iniziativa, ad ogni conquista, c'era sempre qualcuno che tirava fuori "gli scheletri dall'armadio". ‘Sergio Escobar, who left the Opera for the Piccolo Teatro di Milano after less than two years, said that with every new initiative, every victory, there was always someone who pulled the skeletons from the closet.’ (La Stampa)

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 299

(421) Per quanto riguarda il plagio di "Blu", Zucchero ha detto: "Non voglio fare la parte del perseguitato, ma ogni volta che esce un mio disco c'è sempre qualcuno che va ad analizzare le canzoni con la lente d'ingrandimento. ‘As for the plagiarism of “Blu”, Zucchero said: “I don’t want to play the part of the victim, but every time an album of mine is released, there’s always someone who goes and analyzes the songs with a magnifying glass.’ (La Stampa)

Again, the intended interpretation (the NP does not refer to a singular entity) is signaled more explicitly in the original sentences introduced by c’è than in their SV equivalents in (422).270

(422) a. Sempre qualcuno l’aveva detto. (=(419)) ‘Someone always (already) said so.’

b. ad ogni conquista, sempre qualcuno tirava fuori “gli scheletri dall’armadio” (=(420))

‘but with every victory, someone always pulled out the skeletons from the closet.’

c. Ogni volta che esce un mio disco sempre qualcuno va ad analizzare le canzoni con la lente d’ingrandimento. (=(421))

‘Every time one of my albums is released, someone always goes and analyzes the songs with a magnifying glass.’

To conclude, I’ve shown that c’è clefts can signal unambiguously that a singular indefinite NP should not be interpreted as referring to a single entity. This is in contrast with their SV equivalents.271

270 Note that these SV sentences seem pragmatically odd, although they do express the same proposition as their cleft counterparts. 271 It should be noted that the contrast between the intended readings (i.e. the extent to which it is clear that the singular indefinite may refer to several entities) varies according to the context.

300 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

10.8 The nature of the cross-linguistic label ‘existential clefts’272

“What is a cross-linguistic category? What is it FOR?”

(Matić & Wedgwood 2012:131, their capitalization)

One of the outcomes of this chapter is that c’è clefts are strikingly similar to il y a clefts in many respects: they display the same delimitation ambiguities (section 10.4), they have the same variety of information structure articulations (section 10.6) and both types of clefts can be used for functions such as reinforcing negation and indicating that a singular indefinite NP should not be interpreted as referring to a singular entity (section 10.7). This leads to the question whether they should be considered as two language-specific instantiations of a single cross-linguistic category, and if so, what the ontological status of this category is. In light of this question, I believe it’s useful to take into consideration the distinction between ‘cross-linguistic categories’ and ‘comparative concepts’, as proposed by Haspelmath (2010), among others. I will briefly present these notions in section 10.8.1 and then discuss how they could be used for the analysis of il y a clefts and c’è clefts in section 10.8.2.

10.8.1 ‘Comparative concepts’ vs. ‘cross-linguistic categories’

As Haspelmath (2010) explains, it is frequently assumed (especially by researchers working in generative approaches who postulate the existence of a Universal Grammar, or ‘UG’) that there’s a set of universally available linguistic categories (e.g. adjective, WH-movement, accusative case) and that each language contains a selection of them (e.g. Newmeyer 2007). In our case, one could postulate that there is a cross-linguistic ‘existential cleft category’, and that both French and Italian have it in their repository. Put differently, c’è clefts in Italian and il y a clefts in French could be seen as instantiations of the same cross-linguistic category.

272 I use the term ‘existential cleft’ as a cover term for clefts introduced by existential expressions such as there is, il y a and c’è across languages (see Collins 1992; Davidse & Lahousse 2014 for similar designations). This term is more neutral than the often used ‘presentational cleft’, which usually also refers to clefts introduced by other expressions, such as voici/voilà ‘see here/see there’, and to certain uses of c’est clefts.

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 301

However, the very notion of cross-linguistic categories has been criticized by various authors, among whom Dryer (1996, 1997); Gil (2001); Cristofaro (2009); Haspelmath (2010); Matić & Wedgwood (2012) and Haspelmath (2014). For instance, Matić & Wedgwood (2012) state about linguistically similar phenomena that “as in any domain, the existence of comparable outputs should not be assumed to imply identical inputs, nor identical mechanisms (…)” (Matić & Wedgwood 2012:129). These authors, and many others, show that assumptions about cross-linguistic validity are often empirically and theoretically flawed.273

In light of the discussion about the existence of cross-linguistic categories, Haspelmath (2010) argues in favor of a neat distinction between the categories that are used to describe particular languages (‘descriptive categories’) and more general linguistic notions that he calls ‘comparative concepts’, which are used in comparative studies. Importantly, comparative concepts are not psychologically real (speakers don’t use them to acquire a language: they use the language-specific categories), and they do not presuppose a common underlying structure:

“Comparative concepts allow linguists to identify comparable grammatical phenomena in different languages, but by identifying a phenomenon in a particular language as a match of a comparative concept, nothing is claimed about the way in which that phenomenon should be analyzed within the language (what kind of descriptive category should be used for it)”

(Haspelmath 2010:673-674).274

273 See for instance the discussion about the identification of VPs in English, Japanese, Hungarian and Welsh in Haspelmath (2014:495). The diagnostics that are used to identify VPs in English (VP preposing and VP ellipsis) often cannot be applied to other languages, yet VPs are frequently assumed to be universal. In other words, Haspelmath points out that it is problematic to identify categories across languages when it is not possible to apply the same criteria. 274 See also Haspelmath (2014:492) about the value of ‘nonaprioristic approaches’ in comparative syntax: “In the nonaprioristic approach, researchers compare languages with the goal of finding general properties shared by all or most languages, but they make no a priori assumptions about the kinds of categories and constructions that languages might have or about the kinds of explanations that might account for the generalizations. (…) in this tradition linguists who analyze a language are urged to do justice to the language by describing it ‘in its own terms’ (Boas 1911), rather than by means of categories adopted from some other language (i.e., a priori categories).” (Haspelmath 2014:492)

302 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

In other words, comparative concepts are tools for the linguist to make generalizations and come to a deeper understanding of language: they are defined

“in such a way that they allow [linguists] to capture interesting generalizations, or simply salient different language types. Thus, comparative concepts cannot be right or wrong, at least not in the same sense that innate cross-linguistic categories can be right or wrong. They can only be more or less productive, in that they allow the formulation of more or less interesting subdivisions and generalizations” (Haspelmath 2010:678, my emphasis).275

Some comparative concepts may be more interesting or cover a broader dataset than others, but the choice of a particular concept always depends on one’s goals, and, as Haspelmath (2010:680) notes, since linguists will always have a multiplicity of goals, there will never be a single, perfect way to define a concept.276 Importantly, the choice of comparative concepts is thus a methodological choice, not an assumption about the nature of language (Haspelmath 2014:495). This is very different from positing an underlying cross-linguistic category with the assumption that this category is part of the speaker’s mental grammar.277

A final note is in order about the explanatory potential of cross-linguistic analyses. One can account for cross-linguistic tendencies without assuming that the phenomena in question are instances of the same cross-linguistic category: even if a researcher does not see very similar linguistic elements in two languages as cross- linguistic instantiations of the same category, he/she can still provide an explanatory account of why the two languages dispose of those very similar linguistic elements. For instance, many researchers who do not postulate the existence of a UG explain phenomena that occur across languages by appealing to general cognitive constraints and functional pressures (see e.g. Goldberg 2003:222 for references).

275 For explicit applications of the notion ‘comparative concept’, see for instance Jary & Kissine (2016) about the imperative and Hetterle (2015: section 2.3.2) about adverbial clauses. 276 This does not mean, of course, that every researcher should make up their own definition without taking into consideration previous definitions, which would harm the comparability of the different analyses of the same phenomenon. 277 This view is also in line with the distinction between classificatory labels and categories as proper components of a speaker’s mental grammar in Cristofaro (2009:443). In Haspelmath’s terms, the components of a speaker’s mental grammar are specific to the speaker’s native language.

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 303

10.8.2 Existential clefts as a ‘comparative concept’

As explained in the introduction of section 10.8, the clear parallels between il y a clefts and c’è clefts license the use of a cross-linguistic label for the two clefts on a descriptive level: I proposed to refer to them as ‘existential clefts’. The question that arises in light of the distinction between descriptive categories and comparative concepts presented by Haspelmath (2010) is the ontological status of the label ‘existential cleft’. The first option is to argue that the parallels constitute enough evidence for assuming that il y a clefts and c’è clefts are cross-linguistic instantiations of a single linguistic category that has the same representation in the mental grammars of speakers of French and speakers of Italian. The second option, in line with Haspelmath’s (2010) proposal, is to state that we should not assume that the parallels between the two cleft types necessarily reflect exactly the same mental representation, because we never have direct access to speakers’ mental grammars and therefore cannot verify this claim. One might object to the second stance that this is overly cautious in cases where the parallels seem extensive. On the other hand, how can we determine in an objective way how similar two patterns must be before we accept to see them as instantiations of the same mental category?

In my view, it is safer to adopt Haspelmath’s position and refrain from positing a cross-linguistic category of existential clefts with the same mental representation in French and Italian. Instead, il y a clefts and c’è clefts can be seen as two instances of the comparative concept ‘existential cleft’, intended as a purely descriptive label. The main reason for this is that although the evidence for a single category with the same mental representation seems compelling, we have no way of verifying whether it really exists.

Moreover, seeing il y a clefts and c’è clefts as instances of a comparative concept does not entail that we should dismiss the parallels between the two patterns: the fact that two different languages dispose of the same pattern is in need of explanation. I would argue that the fact that both Italian and French have a cleft format introduced by an existential expression can be explained in terms of (i) similar functional needs and (ii) the availability of similar ‘building blocks’: both languages preferably avoid starting a sentence with a non-topical constituent (i.e. a functional need). Furthermore, both languages dispose of existential main clauses (introduced by

304 Chapter 10 | Italian c’è clefts (vs. French il y a clefts)

il y a and c’è) and cleft relative clauses, the two building blocks of existential clefts. In the next chapter, I will present a more elaborate proposal of il y a clefts as compositional structures consisting of two main building blocks (an existential main clause followed by a cleft relative clause) (Chapter 9).

10.9 Conclusion

In this chapter, I investigated c’è clefts by means of corpus research in order to determine to what extent c’è clefts are different from or similar to il y a clefts. C’è clefts turn out to display striking parallels with il y a clefts, as summarized below:

Like il y a clefts,

i. C’è clefts display different types of delimitation ambiguities.

ii. C’è clefts do not introduce ‘regular’ topics; if they introduce a topic, it is marked (e.g. due to contrastiveness).

iii. C’è clefts can express several information structure articulations rather than ‘all-focus’ only.

iv. C’è clefts display important discourse-functional parallels with c’è sentences that lack a relative clause.

v. C’è clefts can have additional functions, e.g. reinforcing a negation.

Moreover, and surprisingly, the number of c’è clefts is higher in La Stampa than the number of il y a clefts in the comparable Le Monde corpus. This runs against the hypothesis that Italian clefts are less frequent than French clefts, because Italian can also use VS word order to encode all-focus (Lambrecht 1994). I argued that this surprising outcome may be due to the fact that Italian newspapers are characterized by a more informal register than French ones, leading to a relatively higher number of structures judged ‘informal’, such as c’è clefts.

I also argued that the extensive similarities between French il y a clefts and Italian c’è clefts make it empirically desirable to refer to them with the same term, ‘existential cleft’. In my view, this term may best be seen as a ‘comparative concept’ in

Chapter 10 | Italian c’è clefts (vs. French il y a clefts) 305

the sense of Haspelmath (2010), rather than assuming that existential clefts are a psychologically real cross-linguistic category of which il y a clefts and c’è clefts are two instantiations.

In the next and final part of the dissertation, I will address the issue of whether il y a clefts are compositional structures, and whether they should be seen as constructions in the sense of Construction Grammar approaches. I will also comment on the status of il y a clefts and c’è clefts as a cross-linguistic construction.

Part V

Il y a clefts: compositionality and status as a construction

------

Chapter 11

The compositionality of il y a clefts

11.1 Introduction

One major goal in the field of linguistics is to come to a better understanding of the architecture of speakers’ and hearers’ mental grammars: what are the building blocks, how do those building blocks relate to each other, which processes occur when utterances are produced and processed, are they innate or not, etc. Since we have no direct access to mental grammars, we must use indirect evidence such as corpus data, psycholinguistic experiments and theoretical arguments, to construct hypothetical answers to those intriguing questions. There are heated discussions about the nature of mental grammars, especially between adherents and opponents of generative approaches. One issue that is relevant in such discussions is the notion of ‘compositionality’. According to most generative approaches, language is maximally compositional, which means that the meanings of sentences can (always or almost always) be derived from their subcomponents: speakers use combinatory rules to put together several building blocks that each have their own meaning, and the result is a sentence with a certain overall meaning (see e.g. Chomsky 1965, 1995; Carnie 2013). Hearers then interpret the sentence using the knowledge they have of the rules and the lexical-semantic meanings of the different building blocks. Put very simply, the meaning of (423) can be derived from the meanings of all the separate words (i.e. the building blocks) and the way in which those words are combined by means of combinatory rules. For instance, a sentence usually consists of at least a subject (we) and a VP (are taking…), a VP may contain an object (our Volkswagen van), etc.

(423) We’re taking our Volkswagen van to go camping. 310 Chapter 11 | The compositionality of il y a clefts

However, scholars who do not adhere to the generative framework often emphasize that many aspects of language cannot be analyzed in a compositional way, i.e. as the mere combination of building blocks according to a set of rules (e.g. Croft 2001; Goldberg 2006). Idioms are a case in point: the meaning of burying the hatchet cannot be derived from the separate meanings of burying, the and hatchet. Hence, the idiom burying the hatchet must be stored ‘as a whole’ in the speaker’s and hearer’s mental grammar, otherwise they would not be able to produce it or understand its meaning. Generative scholars also agree that idioms are stored ‘as a whole’. However, contrary to standard generative assumptions, many constructionist scholars argue that idioms are not the only types of linguistic expressions that are stored as a whole in our mental grammars: they identify many different types of constructions that vary in their degree of schematicity, such as the ‘ditransitive construction’ or the ‘what’s X doing Y? construction’ (see Goldberg 2003).

The issue of compositionality has also been addressed in the literature on clefts. According to some authors (e.g. Huddleston 1984; Katz 2000a; Lambrecht 2001), it/c’est/there clefts are ‘non-compositional’: clefts should be seen as a global pattern whose meaning is more than the sum of its parts. Other researchers argue that clefts are ‘compositional’: the main meanings and functions of clefts can be derived from their subcomponents (e.g. Clech-Darbon et al. 1999; Doetjes et al. 2004). There has been only little discussion about the compositionality of il y a clefts (Lambrecht 1986, 2001; Jullien 2014). In Lambrecht’s view, il y a clefts cannot be analyzed compositionally, while Jullien argues that they are compositional. In this chapter, I will give more evidence in support of Jullien’s view, i.e. I will argue that il y a clefts are compositional. In this way, I hope to provide some insight into the way il y a clefts may be stored in our mental grammar.

The structure of this chapter is as follows: section 11.2 presents an overview of previous studies that investigate the compositionality of il y a clefts. In section 11.3, I present further evidence in favor of a compositional analysis of il y a clefts.

Chapter 11 | The compositionality of il y a clefts 311

11.2 Compositionality: previous analyses of il y a clefts

Lambrecht explicitly investigates the compositionality of il y a clefts in his analyses of ‘presentational clefts’ (Lambrecht 1986, 2001) (section 11.2.1) and of the ‘Presentational Relative Construction’ (Lambrecht 2000a, 2002) (section 11.2.2). Jullien (2014) also addresses the issue directly and indirectly, and presents an analysis that is in some respects at odds with Lambrecht’s (section 11.2.3).

11.2.1 Il y a clefts are non-compositional (Lambrecht)

Lambrecht (1986, 2001) argues that eventive il y a clefts and possessive clefts such as (424) are instances of ‘presentational clefts’.278

(424) J’ai les yeux qui m’font mal. ‘My eyes hurt.’ (‘I have the eyes that hurt.’)

(Lambrecht 1986:115, my literal translation)

Lambrecht presents several arguments in favor of a non-compositional, constructional analysis of presentational clefts. The main idea he posits is the following:

“The [matrix] clause [i.e. il y a NP or j’ai NP] is meaningful only in combination with the following qui-clause. The conclusion to draw from all this is that in the [presentational cleft] construction neither the matrix clause nor the embedded clause can be independently generated entities. Both depend on the presence of the other, and it is the combination of the two that creates the construction.” (Lambrecht 1986:119, my emphasis)

The evidence he proposes in favor of this analysis is related to restrictions on tense and negation in the main clause (j’ai, il y a).279 Regarding tense, Lambrecht states that il y a clefts can display ‘frozen present tense’, which indicates that il y a is used as a fixed expression in those cases (see also Lagae & Rouget 1998:321; Choi-Jonin & Lagae

278 Note that Lambrecht does not discuss the (non-)compositionality of specificational il y a clefts. 279 “The tense and negation phenomena illustrated here are manifestations of the above- mentioned non-compositionality of the construction” (Lambrecht 1986:119)

312 Chapter 11 | The compositionality of il y a clefts

2005:48). In (425) for instance, the expression il y a remains in present tense, whereas the relative clause is in past tense.

(425) Chaque fois qu’il passait des aviateurs il’y’a la tante qu’elle demandait des nouvelles ‘Every time pilots came by the aunt would ask for news’

(Deulofeu 1984:116, in Lambrecht 1986:118)280

However, Lambrecht also admits that il y a isn’t ‘frozen’ in all cleft instances with past tense (see also Lagae & Rouget 1998:321). Hence, this argument is of limited power.281 Furthermore, it isn’t very clear how the ‘fixed’ nature of il y a is an argument against the compositionality of the cleft as a whole. Several generative, compositional analyses also assume that cleft introducers such as it’s, c’est, there’s are (partly) grammaticalized/desemanticized (e.g. Chomsky 1977). Put differently, the fact that il y a can be ‘frozen’ in present tense, and, more generally speaking, is desemanticized, might indicate that ‘il y a NP’ is no longer analyzed compositionally. However, this does not entail that the whole ‘il y a NP + cleft relative clause’ pattern is non- compositional.

The second indication of the non-compositional nature of presentational clefts that Lambrecht presents is the ban on negation in the main clause, as shown in (426). According to him, this is a sign of the ‘semantically non-autonomous character’ of the main clause, which depends on the relative clause.

(426) *Y’a pas le téléphone qui sonne. ‘There isn’t the phone that’s ringing.’ (Lambrecht 1986:118, my translation)

280 This is not a typical instance of an il y a cleft due to the relative clause (lit. ‘there is the aunti that shei asked’ rather than ‘there is the aunt who asked’). Lambrecht does not comment on this. 281 Note, also, that the fact that the main clause remains in present tense when the relative clause contains a passé composé may have a semantic motivation. The passé composé often has a resultative interpretation indicating that some action still holds or has consequences for the utterance time. For instance, the present tense of il y a in (i) is not at odds with the past tense in the relative clause, since the arrival of the brothers is still relevant when the il y a cleft is uttered. (i) Il y a tes frères qui sont arrivés. ‘There are your brothers who arrived’

Chapter 11 | The compositionality of il y a clefts 313

However, recall that I presented counter-examples to the supposed impossibility of negation in il y a clefts in chapter 1.3.3.3, where I explained which conditions do license negation in those instances.282 The awkwardness of (426) is only due to the difficulty to come up with an appropriate context. Furthermore, Lambrecht himself notes that negation is ruled out in instances such as (426) because the only purpose of this utterance is “to pose the NP referent in the discourse”, hence it “makes no sense to negate Y’a le téléphone” (Lambrecht 1986:118). Following this line of reasoning, the limited possibility of negation characterizes all sentences whose purpose is to introduce a new referent, including instances without relative clauses, such as (427) and (429). Indeed, it makes no sense to negate the presence of a man in (428) or of things in (430) when the purpose of these utterances is to assert the presence of a man/things respectively. These negated il y a sentences, although possibly appropriate in a different context, are no longer appropriate in the original discourse context in which they assert the presence of a man and of things.

(427) Il y a un homme dans la chambre. ‘There is a man in the room.’ (Bentley 2015:109)

(428) # Il n’y a pas un homme /d’homme dans la chambre. # ‘There is not a man in the room.’

(429) Y a des trucs ça nous a servi à rien. ‘There are things they were useless.’ (Pekarek Doehler 2015:180)

(430) # Y a pas de trucs ça nous a servi à rien. # ‘There are not things they were useless.’

Consequently, this restriction on negation (which is not absolute) characterizes the main clause ‘il y a + NP’ (as also acknowledged by Bentley 2013:697-698; Bentley et al. 2015:169), and not the il y a cleft construction as a whole. Furthermore, it isn’t due to

282 See (i) for an example of an eventive il y a cleft with a negation: (i) quand on surf sur le net le temps passe vite donc on passe plus de 4 h sans s’eb redre compte surtout quand y’a pas ma mere qui me cris dessus :-C (…) ‘When you surf online, time goes by quickly, so you spend more than 4 hours without realizing it, especially when there isn’t my mother who yells at me :-C (…) (www)

314 Chapter 11 | The compositionality of il y a clefts

the semantic non-autonomy of the main clause with respect to the relative clause, but to the main clause’s function of introducing a referent in the discourse.

On a more general level, the idea that the main clause is not generated independently of the relative clause (Lambrecht 1986:119) is, I believe, contradicted by the ‘emergent’ nature of clefts in naturally occurring speech and the ‘malleability’ of syntactic structures such as clefts, as shown in interactional analyses of clefts (Jullien 2014 and Pekarek Doehler 2015 about il y a clefts, De Stefani 2008 about c’est clefts). For instance, hesitations between a relative clause introduced by the relativizer qui ‘that’ on the one hand and a new clause with an anaphorically linked personal pronoun (e.g. ils ‘they’) show that the speaker does not necessarily ‘plan’ to utter an entire il y a cleft, as in (431). In such instances, the speaker first utters the main clause and then constructs the continuation.

(431) A: y a rien qui est aménagé spécialement pour vous\ ‘there’s nothing that is arranged specially for you’

B: .. non bon i:- s:- y a les les les autres les camarades de classe là ils qui m’aident aussi qui me qui me prêtent leur no:tes ou qui m’expliquent si j’ai pas compris/ (…)

.. ‘no, well, th:- ye:- there are the the the others the schoolmates there they who help me as well who me who lend me their notes or who give me explanations if I don’t understand (…)’

(Jullien 2014:305, my translation, his underlining and italics, my boldface)

As Jullien (2014) sees it, the main clause ‘il y a NP’ can project a continuation, but this continuation may be realized by different means, among which a relative clause (I will return to this idea in section 11.2.3). Therefore, I argue that although the relative clause requires the presence of a main clause (as is usually the case with relative clauses in general), the main clause ‘il y a NP’ is not syntactically dependent on the relative clause: the speaker often has other continuations at his/her disposal.

Chapter 11 | The compositionality of il y a clefts 315

11.2.2 The ‘Presentational Relative Construction’283

All analyses of clefts acknowledge that there is a difference between the relative clause in clefts on the one hand and restrictive or appositive relative clauses on the other. The way in which this difference is captured differs across and within frameworks,284 but what matters for the issue of compositionality is whether or not the relative clause is seen as a relative clause type that is unique to clefts. If it has unique properties, this is an argument in favor of treating clefts as a non- compositional construction whose properties cannot be derived from its subcomponents. If, however, the relative clause in clefts is also found in non-cleft contexts, the properties of the relative clause cannot be seen as evidence for a special non-compositional cleft construction, because they share those properties with other sentence types.

Lambrecht (2000a, 2002) does not postulate a special il y a cleft relative clause, but sees the relative clause in il y a clefts as an instance of a more general type called the ‘Presentational Relative Construction’, which can also occur in non-cleft sentence types such as the ones in (432).

(432) Examples of the ‘Presentational Relative Construction’ a. Il était une fois une belle princesse qui vivait dans un vieux château.

‘Once upon a time there was a beautiful princess who lived in an old castle.’

b. J’ai eu mon beau-frère qui a fait Paris-Nice.

‘I’ve had my brother-in-law who’s done Paris-Nice.’

c. Y a le téléphone qui sonne.

‘The phone is ringing.’

d. Je vois le facteur qui arrive.

‘I see the mailman coming.’ (Lit. ‘I see the mailman who’s coming.’)

(Lambrecht 2000a:49, my translation)

283 Again, this analysis does not hold for specificational il y a clefts. 284 See General Introduction, section 2 and footnote 28.

316 Chapter 11 | The compositionality of il y a clefts

Although the label ‘Presentational Relative Construction’ might be taken to refer to the relative clauses in the sentences in (432), Lambrecht uses the term to refer to the sentences as a whole. He states that the discursive function of this construction is to present a new entity in the discourse and at the same time express a new piece of information about this entity (Lambrecht 2000a:49). 285 What distinguishes this construction from other types of sentences is the function of the relative clause, which is neither restrictive nor appositive but predicative.286 In a similar vein, König & Lambrecht (1999) analyze the French ‘predicative relative clause’ as a construction in its own right that is “irreducible to any other construction” and which “does not behave as other relatives do” (König & Lambrecht 1999:6).287

Other authors also acknowledge that (subtypes of) relative clauses are constructions, i.e. ‘relative constructions’ (Diessel & Tomasello 2000; Goldberg 2003:220; 2006:9; Diessel 2009). More specifically, Diessel (2009) argues that different types of relative clauses can be seen as a ‘network of relative clause constructions’. One of those relative clause types is the ‘presentational relative’, illustrated in (433) (see Diessel & Tomasello 2000:137).

(433) Here's a tiger that's gonna scare him. (Nina 3;1, Diessel & Tomasello 2000:137)

285 One could, however, argue that these discourse-functional properties are due to the fact that the relative clause combines with a presentational main clause. 286 The fact that relative clauses in il y a clefts are predicative is also acknowledged by Furukawa (1996a) and Lagae & Rouget (1998). 287 Some remarks are in order here about the uniformity of this relative clause type. In his earlier (1986) paper, Lambrecht gives several arguments for treating the relative clauses in possessive clefts (432b) and il y a clefts (432c) on the one hand and the relative clause in perception verb utterances such as (432d) on the other as two separate types of relatives clauses. However, he seems to abandon this distinction in the (2000a, 2002) papers. What is important to realize in this respect is that the term ‘predicative/presentational relative clause’ refers to a semantic/discourse- functional type of relative clause, but not necessarily to a morphosyntactically uniform type of relative clause. Thus, although all the relative clauses in (432) share the predicative function, this doesn’t exclude the possibility that they may differ on other levels, for instance the possible relativizers that they can occur with. In other words, the ‘predicative relative clause’ may characterize a semantically uniform class, but this doesn’t entail that they all have the same morphosyntactic properties. Lambrecht (2000a) seems to assume that a uniform semantic profile suffices to identify the relative clauses in (432) as a separate construction. Whether one agrees with this depends, of course, on the definition of ‘construction’ that one adheres to (I will address the notion of construction in more detail in Chapter 12).

Chapter 11 | The compositionality of il y a clefts 317

Diessel & Tomasello (2000) follow Lambrecht (1988b) in stating that the relative clause in such utterances presents new information rather than presupposed information (as in restrictive relative clauses). They see this as “a clear indication that presentational relatives form a specific grammatical construction (in the sense of construction grammar (…))” (Diessel & Tomasello 2000:137). However, in contrast to Lambrecht (2000a:49), they use the term ‘presentational relative’ to refer to the relative clause only rather than to the whole utterance.

Interestingly, Diessel & Tomasello (2000) and Diessel (2009) show that presentational relative clauses are the earliest and most frequent type of relative clause in young children’s speech. One explanation that Diessel & Tomasello (2000:144) propose for this observation is that early presentational relative clauses appear in combination with items that are already “deeply entrenched at the time when children begin to use relative clauses” (Diessel & Tomasello 2000:144), i.e. a deictic pronoun, a copular verb and a nominal, as in (433):

“Since children use presentational [main] clauses long before the first relative clauses emerge, it can be argued that the early use of relative constructions involves a very simple procedure by which the child combines a prefabricated (main) clause (i.e. a clause of the type That’s X, There’s X, It’s X) with a second component – either a verb phrase (…) or a full relative clause (…)” (Diessel & Tomasello 2000:144)

Thus, the authors seem to propose a compositional analysis of utterances such as (433): the child uses the presentational relative clause (which is seen as a construction) as a building block to combine with a familiar main clause.

Summing up so far, it has been proposed that the relative clause in il y a clefts is an instance of a specific type of relative referred to as the ‘presentational’ or ‘predicative’ relative clause. In other words, the relative clause is not unique to il y a clefts, but can be found in combination with other main clauses as well. Therefore, the cleft relative clause in il y a clefts cannot be seen as a unique feature of a non- compositional il y a cleft construction.

318 Chapter 11 | The compositionality of il y a clefts

11.2.3 Il y a clefts are compositional (Jullien)

Jullien (2014) investigates several types of sentences with the main clause ‘il y a NP’, among which il y a clefts. One of the main goals of his dissertation is to analyze the prosody of those utterances in authentic data from an interactional perspective, in order to determine whether/how prosody plays a role in establishing the main clause as a ‘projector’. Roughly speaking, ‘projection’ means in this case that ‘il y a NP’ signals to the hearer that more is to come. Jullien indeed gives evidence that this is one of the main functions of the ‘il y a NP’ main clause, and that the whole utterance functions as a ‘compound Turn Constructional Unit’ (‘TCU’, see Lerner 1996). Important for the discussion in this chapter is that Jullien also touches upon the compositionality and constructionality of il y a clefts.

Jullien (2014) argues in favor of a compositional analysis of il y a clefts and relates his own analysis to the one proposed by Davidse (2000) for there clefts (see chapter 12.3.1). The main argument he presents is that speakers have a number of different formats at their disposal to achieve the communicative goal that il y a clefts achieve with relative clauses. What the relative clause does is express information with respect to the introduced constituent,288 but this can also be done by means of a new clause with an anaphorically linked personal pronoun.289 Hence, Jullien (2014) sees the main clause and the relative clause as building blocks that constitute the cleft as a whole, and explicitly argues that the main clause is a construction in its own right, as Davidse (2000) proposes in her analysis of there clefts. The main clause, when accompanied by the right prosody, announces an upcoming predication, and this predication can be expressed by a relative clause or by other means, depending on the interactional goals of the speaker (Jullien 2014:303 & 479). In his terminology, ‘il y a NP’ projects the realization of an activity (for instance ‘try-marking’, Jullien 2014:284), not the realization of a certain syntactic structure (e.g. a relative clause).290

288 This does not entail that such il y a clefts have a topic-comment articulation: the introduced referent is introduced as a focus (see also discussion in Lambrecht 1988a). 289 This type of il y a sentence is illustrated in (i) (see also section 12.5). (i) Il y a des gens ils ont mauvais caractère. ‘There are people they have a bad character.’ (Choi-Jonin & Lagae 2005:39, my translation) 290 “Ainsi, (…) le segment [il y a SN] projette la réalisation d’une activité plutôt que la réalisation d’une structure syntaxique (…).” (Jullien 2014:304, my emphasis)

Chapter 11 | The compositionality of il y a clefts 319

Furthermore, Avanzi (2008) shows that il y a clefts do not have a uniform prosodic profile and that they do not differ systematically with respect to related il y a sentence types. Jullien (2014:336-337) draws the same conclusion, leading him to state the following in his general conclusion:

“Although we uncovered a production format of the presentational [il y a] cleft construction in which a melodic variation occurs on the last syllable of the ‘il y a NP’ segment, we also found presentational [il y a] cleft constructions with a flat prosodic contour. The prosodic contour of the segment ‘il y a NP’, and prosody in general, thus varies depending on the activity that is engaged in, on the transitions between conversational topics and on the sequential placement of the segment. (…) We agree with the position that prosody, while organizing and segmenting speech, is not linked to syntax in a one-to-one way.”

(Jullien 2014:483, my translation)291

I take this to indicate that prosody cannot be used as an argument for the view that il y a clefts have properties that are unique with respect to other patterns.

To conclude, Jullien (2014) argues in favor of seeing il y a clefts as both compositional:292 (i) they consist of two main building blocks (a main clause and a relative clause) from which the meaning and functions of il y a clefts can be derived and (ii) the relative clause is but one of several continuations of the main clause. Moreover, (iii) il y a clefts do not have a uniform prosodic profile. In the next section, I will present further support for Jullien’s (2014) analysis of il y a clefts as compositional.

291 “Si nous avons relevé un format de production de la construction présentative clivée dans lequel une variation mélodique survient sur la dernière syllabe du segment [il y a SN], nous avons également relevé des constructions présentatives clivées avec un contour prosodique plat. Le contour prosodique du segment [il y a SN], et la prosodie en général, varie donc en fonction de l’activité engagée, des transitions des topiques conversationnels et de l’emplacement séquentiel de ce segment. (…) Nous rejoignons les positions de ces travaux concernant le fait que la prosodie, si elle organise la parole et la segmente, n’est pas reliée de manière univoque à la syntaxe.” (Jullien 2014:483, my emphasis) 292 He also argues that il y a clefts are constructions in the sense of Construction Grammar (despite their compositional nature). I discuss this in Chapter 12.

320 Chapter 11 | The compositionality of il y a clefts

11.3 Il y a clefts: further evidence for a compositional analysis

In section 11.2.1, I showed that the main arguments presented by Lambrecht against a compositional analysis are not solid and that a compositional analysis such as the one presented by Jullien (2014) is so far more convincing. In what follows, I will provide further evidence in favor of a compositional analysis by showing that two properties of the il y a cleft format as a whole can be derived from its subcomponents. These properties are: the discourse functions of il y a clefts (section 11.3.1) and the fact that they express a single proposition and can be declefted (section 11.3.2). Furthermore, I argue that il y a clefts do not have an inherent (eventive or specificational) semantics (section 11.3.3).

11.3.1 Discourse functions

As explained in detail in Chapter 7, the il y a cleft is associated with certain information structure functions: (i) marking the constituent it introduces as focal and non-topical (chapter 7.4), and (ii) expressing different information structure articulations on the sentence-level (chapter 7.3). Lambrecht attributes the first function of il y a clefts (introducing a new, non-topical constituent) to the il y a cleft format. However, existential sentences are also argued to introduce only non-topical and focal constituents (see Bentley 2013; Bentley et al. 2015).293 Therefore, I argued that this function is due to the initial expression il y a alone (see chapter 7.4). Hence, this function cannot be seen as unique to the il y a cleft format as a whole.

Secondly, I showed that existential sentences without relative clauses can also express the different information structure articulations that il y a clefts can express (all-focus, focus-background, contrastive topic-comment, see chapter 7.5). What distinguishes the sentences without relative clauses from the cleft instances expressing the same articulation is that some other linguistic expression (e.g. a PP)

293 Moreover, the fact that c’est clefts can also introduce topics shows that on a more general level, the ‘cleft format’ (introduced by either il y a or c’est) does not have the uniform function of introducing non-topics (see chapter 8.1.4 & 8.1.5).

Chapter 11 | The compositionality of il y a clefts 321

fulfills the role that the relative clause takes on in the cleft instances.294 For instance, both the existential sentence in (434) 295 and the cleft in (435) convey a focus- background articulation.

(434) Focus-background existential sentence Q : Y avait-il quelque chose au grenier ? R : Il n’y avait qu’un chat au grenier.

Q: ‘Was there something in the attic?’ R: ‘There was only a cat in the attic.’

(Van De Velde 2005:45, my translation and underlining and boldface)

(435) Focus-background il y a clefts – A: Je recherche des modèles de voiture à acheter neuve moins de 10 000 euros, ou aller!?

‘I’m looking for new car models that cost less than €10.000, where should I go?’

– B: bonjours. il y a la citroen c1 qui est a moins de 10 000 euros.

‘Hello. There’s the Citroën C1 that costs less than €10.000.’ (YCCQA)

The fact that the PP au grenier ‘in the attic’ and the relative clause qui est à… ‘that is …’ are interpreted as backgrounded rather than focal is due to the preceding discourse context. In other words, it is not necessary to stipulate that this is an effect of the ‘il y a NP PP’ construction in (434) and of the il y a cleft construction in (435).

Furthermore, the fact that il y a clefts can express a variety of different information structure articulations also shows that they are not a uniform form- function pairing. This is another argument against the idea that il y a clefts have a unique information structure-based function that other il y a sentences lack.

294 As I show in chapter 7.5, there are also instances in which context (e.g. a picture) fulfills the role that the relative clause fulfills in a cleft (see example (274)). 295 Note that (434) is a made-up example. It seems more natural to drop the prepositional phrase in the il y a sentence rather than to repeat it. However, it is nonetheless not completely awkward either. Similarly, the relative clause in (435) can also be omitted (see chapter 7.5).

322 Chapter 11 | The compositionality of il y a clefts

11.3.2 Expressing a single proposition/decleftability

One characteristic that is generally seen as a special property of clefts is that they can be declefted without a change in their semantics. In other words, they express a single proposition. In Chapter 6, I showed that decleftability can be derived from several properties of il y a clefts (most notably: a semantically light main clause that can be omitted and a constituent that is acceptable enough as a preverbal subject). Furthermore, I argued that il y a sentences that are traditionally not regarded as instances of il y a clefts can also be declefted. This is a first indication that expressing a single proposition and the possibility of declefting are not unique to (il y a) clefts. This observation is further reinforced by the fact that other sentence types (such as the there sentence in (436a)) can also be transformed into sentences with canonical word order without a change in the proposition they express (see (436b)).

(436) a. There were some non-members present. b. Some non-members were present. (Huddleston 1984:468)

The similarity with il y a/there clefts is clear: instead of a relative clause, it is the adjective present that takes on the role of the predicate once the sentence is transformed into the SV sentence in (436b). In the literature on existential and locative sentences, a similar correspondence is noted (see also Francez 2007:21; Creissels 2014:1), as shown in (437) - (438).

(437) Non c’è Gianni in giardino. not ci-is John in garden

‘There isn’t Gianni in the garden’296

(Moro 1997:138, in Cruschina 2012:83, his glosses, my translation)

(438) Gianni non è in giardino. John not is in garden

‘John is not in the garden’ (Cruschina 2012:83)

296 Note that contrary to the English translation, this utterance is felicitous in Italian.

Chapter 11 | The compositionality of il y a clefts 323

To conclude, expressing a single proposition and the possibility of being transformed into a canonical sentence type can be derived from the subcomponents of il y a clefts, and are not unique to clefts.

11.3.3 No inherent semantics

In the literature on it clefts and its cross-linguistic equivalents, it has been argued that the cleft construction has inherent specificational semantics (e.g. Davidse 2000; Lambrecht 2001). However, it has also been acknowledged that there clefts can be semantically eventive rather than specificational (e.g. Huddleston 1984; Davidse & Kimps 2016; Davidse et al. 2016) and that not all it/c’est clefts are clearly specificational (e.g. Clech-Darbon et al. 1999; Bicler & Davidse 2008). Clearly, the same holds for il y a clefts, which can be eventive as well as specificational (see Chapter 4). There doesn’t seem to be any link between eventive and specificational semantics, so it is hard to derive one from the other and propose a single (more abstract) semantic profile for all il y a clefts.297

Another option would be to abandon the idea of one uniform ‘il y a cleft construction’ and to propose two separate constructions, one semantically specificational and one semantically eventive. However, this account also runs into trouble. Firstly, the idea that the specificational semantics of il y a clefts is due to the whole il y a cleft format is problematic because il y a sentences without relative clauses can also be specificational. The specificational interpretation is due to an interaction between the discourse context, the expression il y a and the type of constituent that is introduced, not to the il y a cleft format as a whole.

Secondly, the eventive interpretation of eventive il y a clefts can also be derived compositionally from the type of relative clause: if the relative clause contains a stage-level predicate, the sentence will be interpreted as eventive. Furthermore, il y a sentences without a relative clause can also have eventive/stage-level semantics.

297 One could, however, argue that on a very abstract level, what eventive and specificational il y a clefts have in common is the ‘secondary’ semantic relation expressed by the relative clause: in eventive clefts, the relative clause encodes secondary predication (see Lambrecht 2002), and in specificational clefts, the relative clause encodes secondary specification (see Davidse & Kimps 2016).

324 Chapter 11 | The compositionality of il y a clefts

For instance, in (439), an event is first described by means of the existential sentence y a eu un écroulement ‘there was a collapse’, and is then again encoded by means of a first il y a cleft with an unfinished relative clause and a second, full il y a cleft. This shows that eventive semantics is not unique to il y a clefts: the utterance can be interpreted as eventive as long as it contains an eventive predicate (e.g. a relative clause or an eventive NP such as un écroulement ‘a collapse’.

(439) ben inattendu c'est qu'un jour + y a eu un écroulement + dans la le passage de la villa d' l'Avenir + y a des maisons qui se sont y avait vous savez en-dessous c'est des des carrières et y a une maison qui s'est effondrée ‘well unexpected it’s that one day + there was a collapse + in the passage de la villa de l’Avenir + there are houses that there were you know below it’s a quarry and there’s a house that collapsed (CFPP)

Summing up, there is no clear evidence that il y a clefts have inherent semantics that cannot be derived compositionally.

11.4 Conclusion

In this chapter, I investigated whether il y a clefts can be analyzed in a compositional way in order to come to a better understanding of how il y a clefts might be represented in speakers’ mental grammars. I argued that contrary to claims made by Lambrecht (1986, 2001), il y a clefts are compositional: the main semantic and discourse-functional properties of il y a clefts can be explained by means of the two main subcomponents, the main clause and the relative clause. I first pinpointed the weaknesses of Lambrecht’s arguments in favor of a non-compositional analysis. I then showed that two properties one might be tempted to attribute to the il y a cleft format as a whole are not unique to il y a clefts: (i) they share their main discourse functions with other sentence types, and (ii) they are not the only sentence type that can be transformed into a canonical SV equivalent. Moreover, il y a clefts do not have a single inherent semantics: they can be both eventive and specificational. Summing up, there is no clear indication of a non-predictable aspect of function or meaning.

------

Chapter 12

No evidence (yet) for il y a clefts as a construction

12.1 Introduction

In the previous chapter, I argued that il y a clefts are compositional: there are no meanings or functions of the il y a cleft format that cannot be derived from its subcomponents. In the introduction of that chapter, I explained that the issue of compositionality is often related to the notion of ‘constructions’ in the sense of Construction Grammar (e.g. Croft 2001; Goldberg 2006). Put simply, constructions can be defined as conventionalized form-meaning pairings, where ‘meaning’ is a broad term that refers to both the semantics of a given form and its possible pragmatic values (a more detailed characterization of ‘construction’ is presented below). Many scholars define constructions as necessarily non-compositional. Hence, if the meaning(s) of a given form cannot be derived from its subparts, this form is a construction (Goldberg 1995).

Most authors who investigate clefts equate the notion of construction with non-compositionality in the way described above: if clefts are non-compositional, they must be constructions (e.g. Lambrecht 2001) and, conversely, if the meaning and functions of clefts can be derived compositionally, there is no need to postulate a special construction for them (e.g. Doetjes et al. 2004). If one assumes a definition of constructions as necessarily non-compositional, it follows that il y a clefts are not constructions, since they can be analyzed compositionally (see previous chapter).

However, an increasing number of scholars in constructionist approaches use a broader definition of constructions that is not opposed to compositionality (e.g. 326 Chapter 12 | No evidence (yet) for il y a clefts as a construction

Nikiforidou 2009; Kay & Michaelis 2012). For instance, the meaning of the expression Could you pass the salt? can be derived from the meanings of the different words that make up the sentence, but Nikiforidou (2009:22) argues that it could still be considered a construction because this expression is conventionalized. Returning to clefts, if one assumes this more inclusive definition of ‘constructions’, one can argue that clefts are compositional but should nonetheless be considered constructions, as is proposed by Davidse (2000); Birner et al. (2007) and Jullien (2014).

In this chapter, I first present the narrow and the broad definition of constructions (section 12.2) and then summarize previous studies that propose to see different types of clefts as compositional constructions (section 12.3). In section 12.3.3, I discuss whether il y a clefts can also be considered constructions in the broad sense. Finally, I address the question whether il y a clefts should be seen as an instance of a higher-level (cross-linguistic) cleft construction that also includes other instances such as c’est clefts and Italian c’è clefts (section 12.5).

12.2 What are constructions?

‘Constructions’ are often defined as conventional symbolic units (Langacker 1987; Croft 2005). However, apart from this general characterization, the notion of construction has been approached in different ways. In what follows, I present a narrow (section 12.2.1) and a more inclusive (section 12.2.2) characterization of ‘construction’. 298

12.2.1 Narrow definition: constructions as necessarily non-compositional

According to Croft (2005:274), constructions are “(…) an entrenched routine (‘unit’), that is generally used in the speech community (‘conventional’), and involves a pairing of form and meaning (‘symbolic’ (…))” (Croft 2005:274). ‘Meaning’ is not to be

298 Note that yet other definitions and/or approaches to the notion of ‘construction’ are possible. In this dissertation, I limit myself to the investigation of il y a clefts in light of the two general definitions of constructions presented in sections 12.2.1 (‘constructions are non-compositional’) and 12.2.2 (‘constructions are conventionalized’).

Chapter 12 | No evidence (yet) for il y a clefts as a construction 327

understood in the strict sense as the semantics of a given form, but refers to “all of the conventionalized aspects of a construction’s function” (Croft & Cruse 2004:258, emphasis by the authors), which may include pragmatic functions of the form in question. As explained in the introduction, some authors add to this that such pairings of form and meaning are not or can no longer be analyzed compositionally (Goldberg 1995, 2003). In other words, they are more than the mere sum of their parts:

“Any linguistic pattern is recognized as a construction as long as some aspect of its form or function is not strictly predictable from its component parts or from other constructions recognized to exist” (Goldberg 2003:219; 2006:5).

Importantly, in this view, constructions characterize not only whole utterances such as clefts, but also smaller linguistic units. As Goldberg (2003) puts it, “All levels of description are understood to involve pairings of form with semantic or discourse function, including morphemes or words, idioms, partially lexically filled and fully abstract phrasal patterns” (Goldberg 2003:219). Furthermore, utterances typically involve “the combination of at least half a dozen different constructions” (Goldberg 2003:221). For instance, the sentence in (440) consists of all the constructions listed below it.

(440) What did Liza buy the child? - Liza, buy, the, child, what, did constructions (i.e. words)

- Ditransitive construction

- Question construction

- Subject-Auxiliary inversion construction

- VP construction

- NP construction (Goldberg 2003:221)

Most of the analyses of clefts that explicitly address the issues of compositionality and the status of clefts as constructions in the sense of Construction Grammar define constructions as necessarily non-compositional, as described above. Hence, arguments that clefts contain some unpredictable meaning are taken as evidence that the cleft in

328 Chapter 12 | No evidence (yet) for il y a clefts as a construction

question is a construction (e.g. Huddleston 1984; Lambrecht 1986; Katz 2000a; Lambrecht 2001; Pavey 2004:2). This is clear from statements such as the following:

“I will propose an analysis in which [clefts] are treated as grammatical constructions in the sense of construction grammar, that is, as form–function pairings whose structural and semantic properties cannot, or not entirely, be accounted for in terms of other properties of the grammar of a language (…) and which therefore require independent explanation (…)”

(Lambrecht 2001:466, his emphasis)

“[presentational clefts] are a global grammatical construction [that is] semantically and syntactically non-compositional” (Lambrecht 1986:117)

Conversely, several authors who show that the semantic and pragmatic values of clefts can be derived compositionally conclude that clefts are not constructions because their global meaning is not more than the sum of their parts (Clech-Darbon et al. 1999; Doetjes et al. 2004; Dufter 2006). For instance, Doetjes et al. (2004) state the following:

“[Clech-Darbon et al. (1999)] argue in favor of a maximally simple analysis of [c’est] clefts, which is not based on construction specific rules. Given the existence of relative clauses, the semantics of ce and basic combinatorial rules of syntax, the properties of [c’est] clefts presented in this paper can be obtained: this approach is maximally consonant with the Principles and Parameters Program, according to which constructions are descriptive artefacts” (Doetjes et al. 2004:530)

Similarly, Clech-Darbon et al. (1999) argue that their analysis of clefts shows that constructions are “a descriptive notion, with little, if any, theoretical content” (Clech- Darbon et al. 1999:108). 299 These statements reveal the tension between

299 See also Chomsky (1995): “The notion of grammatical construction is eliminated, and, with it, construction-particular rules. Constructions such as verb phrase, relative clause, and passive remain only as taxonomic artefacts, collections of phenomena explained through the interaction of the principles of UG [Universal Grammar], with the values of parameters fixed.” (Chomsky 1995:170)

Chapter 12 | No evidence (yet) for il y a clefts as a construction 329

constructionist approaches on the one hand and generative approaches on the other. The latter see the nature of language as maximally compositional: meaning is generated by combinatory rules that govern syntax, semantics and the other levels of linguistic analysis (see chapter 11.1). 300 In their view, this makes the notion of construction redundant.

In sum, the majority of analyses of clefts that debate about the question whether clefts are constructions only do so by looking at (non-)compositionality. However, there are a few exceptions to this tendency (see most notably Davidse 2000; Birner et al. 2007; Jullien 2014; Davidse & Kimps 2016). They join a number of scholars in constructionist approaches that do not define constructions as necessarily non- compositional. I present this more inclusive definition in the next section.

12.2.2 Broad definition: constructions that allow compositionality

Nikiforidou (2009) observes the following about the link between compositionality and the definition of constructions:

“The requirement of (at least partial) unpredictability of the syntactic, semantic or pragmatic properties associated with a particular pattern has dominated early constructional analyses both in the choice of the phenomena analyzed and in the resulting kind of analysis. (…) As the constructional approaches multiplied, however, the scope of constructional analysis has extended to include all conventional patterns, regardless of their degree of internal predictability”

(Nikiforidou 2009:17, my emphasis)

Langacker (2009) also agrees that Goldberg’s (1995) definition of constructions as non-compositional is too strict, and already argued in earlier works (e.g. Langacker 1987) that constructions can be derived compositionally. He states that it is arbitrary to exclude fixed but regular expressions from the category of ‘constructions’

300 See Jackendoff (1997:48) about “the doctrine of syntactically transparent semantic composition”: “all elements of content in the meaning of a sentence are found in the lexical conceptual structures (…) of the lexical items composing the sentence” (Jackendoff 1997:48, my emphasis)

330 Chapter 12 | No evidence (yet) for il y a clefts as a construction

(Langacker 2009:168, see also Langacker 2008:section 8.4.1 and Kay & Michaelis 2012:2272).

Thus, according to a number of researchers, a construction is not necessarily more than the sum of its parts: some form-meaning pairings are considered constructions despite the fact that they can be derived compositionally. What, then, licenses the label of construction in such instances? Goldberg (2003) states that patterns are stored in the speaker’s/hearer’s mental grammar as constructions “even if they are fully predictable as long as they occur with sufficient frequency” (Goldberg 2003:220). They must be conventional or familiar enough for the speaker. It may be helpful to consider a few examples to illustrate this point. Nikiforidou (2009) presents the ones in (441), among others.

(441) a. Are you crazy? (Kay 2005:695, in Nikiforidou 2009:21) b. Have you lost your mind? (Kay 2005:695, in Nikiforidou 2009:21)

c. Can you pass the salt? (Nikiforidou 2009:22)

The utterances in (441a-b) are treated as ‘interpretational constructions’ by Kay (2005) despite the fact that they “are not characterized by any particular morphosyntax (other than the interrogative pattern) [,] and the semantics of questioning the suggestor’s mental health is of course fully derivable; so is most likely the pragmatic force of challenging the wisdom of a suggested course of action (…)” (Nikiforidou 2009:21). In a similar vein, Nikiforidou states:

“[the “Can you VP?” pattern in (441c)] is by now conventionally associated with the expression of a request and is even characterized by some degree of non-detachability (cf. *Are you able to pass the salt?) and lexico-semantic features normally associated with direct requests (cf. Can you please pass the salt?). The conventionality of these cases is not undermined by the fact that such aspects of meaning may be derived, as noted, by general pragmatic principles.” (Nikiforidou 2009:22, my emphasis)

It should be noted, however, that it is not always made explicit what ‘conventionalized’ means, and, perhaps more importantly, how this can be measured

Chapter 12 | No evidence (yet) for il y a clefts as a construction 331

empirically. This notion often seems to refer to the fact that a certain pattern is routinized because of its frequency (e.g. Nikiforidou 2009). In this sense, ‘conventionalization’ is closely related to the notion of ‘entrenchment’, which can be defined as “the degree to which a linguistic structure (…) forms an established unit of the mental grammar of a speaker” (Stefanowitsch & Flach 2017:101). In other words, conventionalization refers to the degree to which linguistic structures are established at the level of the linguistic system in the speech community, whereas entrenchment refers to this degree in the individual speakers’ minds (see e.g. Schmid & Mantlik 2015:584; Stefanowitsch & Flach 2017:122).301

Both conventionalization and entrenchment can be captured empirically in a number of ways, among which figure different methods of measuring frequency302 in corpus data 303 (see Evert 2005; Schmid 2010; Stefanowitsch & Flach 2017 for details).304 When performing those measurements, one must take into consideration the complexity of the linguistic unit under investigation (i.e. monosyllabic words are less complex than multi-word constructions) as well as its degree of schematicity (which can be measured in terms of the number of different possible instantiations of

301 “Conventionalization is a theoretical construct that differs from entrenchment mainly in that it describes established linguistic structures at the level of the linguistic system itself (…). Entrenchment, in contrast, describes established linguistic structures at the level of an average speaker’s mental representation (…)” (Stefanowitsch & Flach 2017:122). “Collective processes on the macro-level of speech communities (conventionalization) have to be separated from cognitive processes on the micro-level of individual authors (entrenchment)” (Schmid & Mantlik 2015:584). 302 The link between degree of entrenchment and frequency of occurrence was already mentioned by Langacker (1987:59). The importance of frequency of occurrence to language acquisition, language use and language change in general is acknowledged by a large number of scholars in cognitive and functional approaches to the study of language (see Diessel 2007 for an overview). 303 As explained by Stefanowitsch & Flach (2017), there are (at least) two ways of conceptualizing the link between corpus data on the one hand, and the representation of language in speakers’ minds on the other. Firstly, from the corpus-as-output perspective, corpora are seen as samples of language use of a given group of speakers that are taken to be representative of a speech community: it is “a snapshot (…) of the linguistic performance [those speakers] collectively generate on the basis of the individual linguistic representations in their minds” (Stefanowitsch & Flach 2017:103). Hence, we can draw inferences about the mental representations that underlie the observed output. One can also assume the corpus-as-input perspective, according to which corpora are seen as input upon which speakers of a community base their mental representations. This model is at odds with generative approaches and is more accepted among usage-based approaches. The two perspectives on corpus data can be seen as complementary. 304 Entrenchment can also be measured by means of methods such as reaction time studies.

332 Chapter 12 | No evidence (yet) for il y a clefts as a construction

the pattern).305 In the case of complex, multi-word units (which are relevant for the discussion of clefts), Stefanowitsch & Flach discuss three types of measurements: token frequency (i.e. simply counting the number of instantiations of a pattern in a corpus), probability-based measurements (e.g. calculating the probability that word A occurs given the presence of word/pattern B) and statistical measures (e.g. log- likelihood tests and chi-square tests). They conclude that the statistical tests combine the strengths of the other two tests and are therefore “the best corpus-based approximation of entrenchment” (Stefanowitsch & Flach 2017:115). Although it lies beyond the scope of this dissertation to apply such tests on the il y a cleft data, this is certainly a promising road to pursue in order to determine whether il y a clefts are indeed frequent enough to be seen as entrenched and/or conventionalized. I will return to this in section 12.3.3.

Another way in which the role of compositionality has been captured in constructionist approaches is the concept of ‘constructional compositionality’, which is “understood in terms of form-meaning correlations which can recur in different constructions and contribute systematically to constructional meaning” (Dancygier 2005:456, see also Dancygier & Sweetser 2005; Ettlinger 2005). In their view, constructions can be used as building blocks to create new constructions in a compositional way. For instance, Ettlinger (2005) analyzes the English resultative construction as an example of constructional compositionality and argues that it “consists of a number of different, related, sub-constructions, each with its own semantic and syntactic properties” (Ettlinger 2005). He defines constructional compositionality as ‘multiple constructions combining to create a new construction’, without implying that the result is more than the sum of its parts.

In sum, according to a number of scholars, the notion of ‘construction’ may also apply to conventionalized (or ‘entrenched’) patterns that are fully predictable from their subcomponents. What makes them constructions is their conventionality.

305 For instance, the pattern ‘drive NP ADJ’ is mostly used with the first person pronoun (me) and an adjective referring to insanity (drive me mad), and is therefore less schematic than the pattern ‘color NP ADJ’ (as in color me stupid, meaning ‘consider me stupid’), which co-occurs with a larger variety of different adjectives (see Stefanowitsch & Flach 2017:118-119).

Chapter 12 | No evidence (yet) for il y a clefts as a construction 333

12.3 Clefts as compositional constructions: previous analyses

In what follows, I present three analyses that acknowledge the compositional nature of clefts, but at the same time propose to see them as constructions in the broad sense discussed in section 12.2.2 (i.e. a pattern is a construction if it is conventionalized in the linguistic community): Davidse’s analysis of there clefts and it clefts (Davidse 2000; Davidse & Kimps 2016, section 12.3.1), Birner et al.’s (2007) analysis of that clefts (section 12.3.2) and Jullien’s (2014) analysis of il y a clefts (section 12.3.3).

12.3.1 Compositional constructions: there clefts and it clefts

Davidse (2000) and Davidse & Kimps (2016) propose that there are two main building blocks that combine in a compositional way in it clefts, there cleft and English possessive clefts: the main clause and the relative clause.

As for the main clause, according to Davidse (2000), the ‘identifying’, ‘existential’ and ‘possessive’ main clauses found in clefts correspond exactly to the clause types in contexts such as (442). Therefore, they are not unique to clefts, but they do constitute constructions in their own right that encode specific semantics.

(442) What’s for supper? a. It’s spaghetti.

b. Well, there’s spaghetti.

c. Well, you’ve got/we have spaghetti. (Davidse 2000:1121)

Davidse (2000:1121) argues that all three of the main clauses in (442) are associated with a different semantic profile, but what they have in common is that they convey a quantifying value: “they express exhaustive specification of a set, enumeration, or the cardinal quantification of instances” (Davidse 2000:1101).

The second building block of English clefts is the relative clause. The relation it encodes is that of specification: it takes the introduced constituent as antecedent and indicates that this constituent is the value of a given variable. As Davidse puts it, “What the relative clause does is add a value–variable dimension as part of the

334 Chapter 12 | No evidence (yet) for il y a clefts as a construction

construction” (Davidse 2000:1127). In other words, the main clause in combination with the introduced constituent already presents a value for a discourse-given variable, and the relative clause ‘reconfirms’ this value-variable relation.

Davidse & Kimps (2016) do not propose an explicit separate analysis of eventive, non-specificational there clefts. However, I believe their compositional analysis of specificational clefts can be extended naturally to this second type of there clefts by stating that the relative clause encodes predication rather than specification in those instances. The combination of the presentational (non-quantifying) main clause with a predicative, stage-level relative clause leads to a cleft with an eventive reading.

12.3.2 A compositional construction: that clefts

Birner et al. (2007) investigate three sentence patterns: clefts introduced by that, which they refer to as th-clefts (443a), equative sentences with epistemic would and demonstrative subjects (443b), and simple equatives with demonstrative subjects (443c).

(443) a. That’s John who wrote the book. b. That would be John.

c. That’s John. (Birner et al. 2007:317)

According to the authors, the properties that these different constructions share with one another “can be straightforwardly accounted for as the sum of the constraints on their shared elements—that is, the equative construction, the demonstrative subject, and the presence of a contextually salient open proposition” (Birner et al. 2007:317) Hence, they argue that the three sentence types in (443) are “functionally complex constructions whose elements compositionally impose discourse-functional constraints on the use of the whole” and that “the discourse functions of an utterance are built up compositionally from those of its parts.” (Birner et al. 2007:317).

All three sentence types can be used to fulfill three communicative purposes: “to convey commitment to the truth of the proposition expressed (…), to mark the postcopular constituent as instantiating the focus in a salient [open proposition] (…),

Chapter 12 | No evidence (yet) for il y a clefts as a construction 335

and to equate this postcopular with either some salient discourse entity or simply with the variable itself (…)” (Birner et al. 2007:326-327). This explains why the three sentence types can often be found in the same discourse context, as shown in (444):

(444) [Context: a knock at the front door] a. That would be Christine.

b. That’s Christine who’s at the door.

c. That’s Christine. (Birner et al. 2007:325)

Furthermore, the three sentence types have three properties in common (referential ambiguity, apparent lack of number agreement 306 , and apparent lack of tense agreement), which can be derived “from the combination of the equative and the demonstrative in the presence of a sufficiently salient [open proposition]” (Birner et al. 2007:338).307 Therefore, there is no need to stipulate that th-clefts as a whole give rise to those properties, because they are due to the cleft’s subcomponents. Evidence to support this analysis comes from the fact that th-clefts share those properties with sentence types containing the same components, i.e. equative sentences with epistemic would (444b) and simple equatives with demonstrative subjects such as (444c).

Summing up, Birner et al. (2007) show that the discourse functions and constraints holding for th-clefts are not due to the cleft format as a whole, but arise compositionally from the sum of its subcomponents. They nonetheless adhere to an analysis of such clefts as complex constructions in the sense of Construction Grammar, although they do not explicitly give arguments for doing so. In Kay & Michaelis’s (2012) view, the analysis presented by Birner et al. (2007) “supports the claim that constructional and compositional modes of analysis are compatible” (Kay & Michaelis 2012:2294).

306 This is illustrated by utterances such as (i), in which singular that rather than plural those is used. (i) (…) The show centered on three young characters just starting out in life—that would be the two guys and a girl. (Birner et al. 2007:330) 307 For the sake of brevity, I will not present those properties and the way in which Birner et al. (2007) derive them from the subcomponents. The details of this derivation are not relevant for the analysis of il y a clefts. I refer the reader to Birner et al. (2007) for a detailed analysis.

336 Chapter 12 | No evidence (yet) for il y a clefts as a construction

12.3.3 A compositional construction: il y a clefts

Jullien (2014) proposes to analyze il y a clefts as compositional constructions, on a par with the analysis of English there clefts proposed by Davidse (2000). He argues that il y a clefts can be seen as a ‘highly ritualized activity’ consisting of two actions: an ‘orientation of the interlocutor’s attention to an element’ and a ‘conversational sequence about the element’.308 In this sense, Jullien sees the function of the il y a cleft as a whole as a single routinized activity despite the fact that the two sub-actions can clearly be mapped onto the two main building blocks of the pattern, i.e. the main clause (orientation to an element) and the relative clause (predication about this element). Jullien then relates this two-piece activity to Tomasello et al.’s (2007) study of pointing in child-adult interaction and suggests that presentational clefts emerge as the linguistic equivalent of one of the most crucial interactions performed by children with their parents: pointing to an object in order to achieve joint attention with the parent, and then continuing the interaction about this object, for instance by making a request (Jullien 2014:482).309 These two actions (bringing something to the other person’s attention and then expressing something about this element or doing something with it) are so closely connected that they can be seen as one routinized, two-step interaction according to Jullien (2014). Hence, it can be argued that the two building blocks taken together (the il y a main clause and the relative clause) are indeed part of a conventionalized construction in the broad sense described in section 12.2.2.310

Moreover, I believe Jullien provides another piece of evidence in favor of a constructional analysis, although he doesn’t present it as such. He gives examples of collaborative production, i.e. cases in which interlocutors finish the first speaker’s utterance. This is illustrated in (445): the child (speaker A) produces a relative clause in order to ‘finish’ the ‘il y a NP’ main clause uttered by the adult (speaker S).

308 “Lorsque les activités accomplies sont des activités ritualisées du type [orientation de l’attention des interlocuteurs vers un élément]-[séquence conversationnelle à propos de cet élément], le format syntaxique peut être hautement grammaticalisé et sédimenté, comme la construction présentative clivée.” (Jullien 2014:482) 309 Tomasello et al. (2007:719) argue that the action of pointing and using linguistic structures to introduce new topics can be seen as the same ‘building blocks’ for children. 310 Jullien (2014) himself does not explicitly equate ‘routinized’ with ‘entrenched’.

Chapter 12 | No evidence (yet) for il y a clefts as a construction 337

(445) Collaborative production of an il y a cleft S: il y a deux enfants alors/ .

so there are two children

(short pause)

S: [c’est ça/

[that’s it

A: [qui se rencontrent

[who meet each other

S: qui se rencontrent\ d’accord .

that meet each other right

(Child: 6;04, Jullien 2014:292, my translation)

What this shows is that the interlocutor (the child) is able to produce a continuation of the il y a main clause, i.e. a relative clause (Jullien 2014:294).311 In my view, this can be seen as evidence that the il y a cleft is highly predictable and may therefore be stored as an entrenched pattern in the speaker’s mental grammar.312

However, apart from this indication, Jullien does not present clear empirical evidence that il y a clefts are indeed entrenched in the speaker’s mental grammar or conventionalized at the level of the speech community.

311 Jullien also argues that the fact that this expansion is realized by a six-year-old confirms that the use of the presentational cleft starts at an early age, as also reported by Jisa & Kern (1998). See also Jourdain & Lahousse (in preparation) about the acquisition of French il y a clefts and c’est clefts. 312 See also Lamiroy & Klein (2016), who see the fact that speakers can ‘finish’ the beginning of fixed expressions as evidence that those expressions are part of the speaker’s memory (e.g. native speakers know that un froid de… ‘a cold of’ is followed by canard ‘duck’, meaning ‘a bitter cold’).

338 Chapter 12 | No evidence (yet) for il y a clefts as a construction

12.4 Il y a clefts: a construction or not?

In Chapter 11, I argued that il y a clefts are compositional: there are no indications that they are more than the sum of their subcomponents. Consequently, there is no strong evidence in favor of a construction in the narrow, non-compositional sense (see definition of ‘construction’ in section 12.2.1). However, as illustrated in section 12.3, a linguistic pattern can also be seen as a construction if it is conventionalized enough (definition of ‘construction’ in section 12.2.2). This implies that in order to answer the question whether il y a clefts are constructions in the broad sense, we must somehow determine whether they are conventionalized enough.

It often seems to be assumed that the degree of conventionalization can be captured in terms of a pattern’s frequency, which can in its turn be measured empirically by looking at corpus data (see section 12.2.2). As already announced in that section, it lies beyond the scope of this dissertation to measure the degree of conventionalization of il y a clefts in terms of frequency, for instance by means of a collostructural analysis (see e.g. Gries 2003). I did present normalized frequencies of il y a clefts per 500.000 words in the different corpora in Chapter 5. These frequencies were useful to determine differences between the corpora, but they are less reliable to determine overall conventionalization of il y a clefts in the French language in general.313 Of course, one could state that on an intuitive level, il y a clefts are frequent enough to be seen as entrenched in the speakers’ minds and conventionalized in the language system. However, intuition must ideally be backed up by empirical evidence.314

313 See for instance Stefanowitsch & Flach (2017:110) about the problems of ‘raw’ frequency counts for measuring entrenchment, which is directly related to conventionalization (according to Stefanowitsch & Flach (2017:122), entrenchment and conventionalization differ theoretically but are measured in the same way, see section 12.2.2). One would need to take into consideration, for instance, the individual frequencies of all the different words occurring as clefted element and inside the relative clause. For example, if the utterance il y a les oiseaux qui chantent ‘there are the birds that are singing’ is one instantiation of il y a clefts in the corpus data, we would also need to know how often les oiseaux ‘the birds’ and chanter ‘to sing’ appear in utterances that are not clefts in order to compute reliable test statistics. 314 Suppose that future statistically oriented corpus-based research provides evidence that il y a clefts are indeed conventionalized and can thus be seen as a construction in the broad sense described in section 12.2.2. In this case, il y a clefts can be seen as an example of ‘constructional compositionality’, i.e. a construction consisting of several sub-constructions, “each with its own semantic and syntactic properties” (Ettlinger 2005, see chapter 12.2.2): the existential main clause

Chapter 12 | No evidence (yet) for il y a clefts as a construction 339

Note that specific instantiations of il y a clefts may be more conventionalized (because more frequent) than others. Two good candidates for a routinized use of il y a clefts, due to their possible frequency in spoken French, are the expressions y’a le téléphone qui sonne ‘the phone is ringing’ and y’a quelque chose qui cloche ‘there’s something wrong’. It is indeed possible (even plausible) that these utterances are stored as semi-fixed expressions315 in the lexicon, alongside expressions such as Could you pass the salt?, which is analyzed as a construction for the same reason by Nikiforidou (2009). However, the fact that those specific instantiations (i.e. tokens) are relatively conventionalized does not necessarily say much about the degree of conventionalization of il y a clefts in general (i.e. the type).

To conclude, whether or not il y a clefts are indeed conventionalized (and, hence, can be seen as a construction or not) depends on how ‘conventionalization’ is defined and measured. Certain instantiations of il y a clefts with particular lexical items (e.g. y’a quelque chose qui cloche ‘there’s something wrong’) may be more conventionalized than others. However, we cannot use such instantiations to make generalizations about il y a clefts in general. More (statistically-oriented corpus-based and/or experimental) research is necessary to come to a better understanding of whether il y a clefts in general are really conventionalized enough to call them full- fledged constructions. In other words, I cannot answer the question whether il y a clefts are constructions with a straightforward ‘yes’ or ‘no’.

would be a sub-construction (as also argued for English there clefts by Davidse 2000, see section 12.3.1), and the cleft relative clause would also be a construction in its own right (as posited by König & Lambrecht 1999; Diessel & Tomasello 2000; Lambrecht 2000a, 2002, see section 11.2.2). In this case, il y a clefts would thus be a relatively ‘transparent’ construction, meaning that its sub- components can easily be recognized as such by the hearer (see Hilpert & Diessel 2017:61 about the effect of entrenchment on the transparency of complex constructions). However, one may wonder what the usefulness of such a construction would be, if it can be computed in a compositional way. 315 See Lamiroy (2003); Balibar-Mrabti & Vaguer (2005); Lamiroy (2008); Klein & Lamiroy (2011) and Lamiroy & Klein (2016) inter alia about fixed and semi-fixed expressions in French.

340 Chapter 12 | No evidence (yet) for il y a clefts as a construction

12.5 Il y a clefts as an instance of a higher-level (cross-linguistic) ‘cleft construction’?

As I explained in the General Introduction of this dissertation, clefts are often regarded as a family that not only includes different members within the same language (depending on the main clause expression, e.g. c’est & il y a), but which also characterizes a uniform pattern that can be found across languages (e.g. il y a clefts & c’è clefts). In light of the analysis presented in this chapter, the view of clefts as a family gives rise to (at least) two questions:

I. Should il y a clefts and c’è clefts be seen as cross-linguistic instances of the same construction, given their discourse-functional similarities?

II. Should il y a clefts and c’est clefts be seen as members of a higher-level (French) cleft construction?

I. A cross-linguistic ‘existential cleft construction’?

As for the link between il y a clefts and c’è clefts, I argued in chapter 10.8.2 that the similarities between il y a clefts and c’è clefts can likely be explained in terms of similar functional needs (e.g. avoiding sentence-initial non-topics) and the availability of similar building blocks (existential main clause + relative clause). Thus, I see no need to stipulate an innate ‘existential cleft construction’ in a Universal Grammar in order to explain why both French and Italian dispose of this pattern.316

316 Rather, one could capture the similarities between the two constructions in terms of a shared ‘conceptual space’ (Croft 2005) or ‘semantic map’ (Haspelmath 2003): “Radical Construction Grammar recognizes that virtually all formal grammatical structure is language-specific and construction-specific. (…) There are universals underlying the grammatical diversity of the world’s languages. But the universals are functional, that is, semantic/pragmatic/discourse-functional. As a number of typologists have now proposed, the distributional patterns of categories defined by constructions can be mapped onto a conceptual space that is hypothesized to hold for all languages (see Croft 2001:91-92; 2003: ch. 5-6; Haspelmath 2003 and references therein).” (Croft 2005:309, my emphasis). This is in line with the general view of constructionist approaches on cross-linguistic similarities (see e.g. Goldberg 2003:222). In this view, monolingual speakers only have mental representations of one language in their mind, which are not innate but are acquired by means of generalizations about recurrent patterns in the input the speaker is confronted with. From this perspective, it makes no sense to speak of a psychologically real cross-linguistic construction except, perhaps, in the case of plurilingual speakers who generalize over several languages.

Chapter 12 | No evidence (yet) for il y a clefts as a construction 341

Note, also, that the fact that we find the same sentence type with the same functions in different languages (il y a clefts, c’è clefts, there clefts) can also been seen as an indication of the compositional nature of this sentence type. A given sentence type or expression is less likely to be widespread across languages if it is an idiosyncratic construction (e.g. the idiom kick the bucket) than if its meaning and functions can be computed compositionally.

II. An overarching French cleft construction?

As for the second question (‘Should il y a clefts and c’est clefts be seen as members of a higher-level French cleft construction?’), I showed in chapter 8.2 that there are some parallels and many differences between il y a clefts and c’est clefts. I believe it is not necessary to postulate a special overarching cleft construction in order to explain the similarities between il y a clefts and c’est clefts. Rather, they use the same mechanism (the combination of a main clause and a relative clause), and differences naturally arise due to semantic differences between the main clauses, as already argued for by others (e.g. Davidse 2000). As far as I know, there is no way of determining with certainty whether speakers really have an abstract representation of ‘clefts in general’ with the format ‘Main clause expression + clefted constituent + relative clause’ in their mental grammar.317 In the absence of in-depth studies on this issue, I would not postulate a higher-level cleft construction in French that c’est clefts and il y a clefts would be related to in terms of an inheritance hierarchy.

However, it is reasonable to assume that there is a network of different uses of il y a sentences (among which il y a clefts) that display discourse-functional overlap and therefore compete with each other when the speaker constructs his/her utterance. These different sentence types include il y a sentences such as (446) and (447), as well as sentences introduced by semantically and/or pragmatically similar main clauses such as voici/voilà ‘see here/there’ and j’ai ‘I have’ (448).

317 In other words, linguists may well generalize over different clefts introduced by different main clauses, but there does not seem to be compelling evidence that naïve speakers make the same generalizations.

342 Chapter 12 | No evidence (yet) for il y a clefts as a construction

(446) il y a des gens ils avaient acheté un tapis, ben, ils ont dû se déshabiller mais vraiment jusqu’à la petite culotte, quoi ‘There were people they bought a carpet, well, they had to get undressed but really right down to their underwear’

(Choi-Jonin & Lagae 2005, my translation)

(447) y a y a quelque chose d’important à à voir est-ce que la langue est faite pour s’exprimer avec ou sans fautes ‘There’s something important: is language made for expressing yourself with or without mistakes?’

(Pekarek Doehler 2015:180-181, additional transcription omitted)

(448) J’ai ma copine qui habite à Paris. My friend lives in Paris. (Lit. ‘I have my friend who lives in Paris.’)

(Conti 2010:63, my translation)

As Pekarek Doehler (2015) puts it, the il y a main clause is “part of a network of interrelated constructional schemata that (…) are connected by family resemblances” (Pekarek Doehler 2015:195). In contrast, c’est clefts and il y a clefts hardly ever compete for the expression of a given proposition in a given context: they display little discourse-functional overlap (see chapter 8.2). Therefore, it is unlikely that the speaker, when ‘choosing’ which encoding to use in order to express a certain proposition, first ‘decides’ to use a general cleft format and then chooses which specific type (c’est or il y a). Instead, il y a clefts enter into competition with functionally similar patterns such as ‘il y a NPi ilsi ’ ‘there are NPi theyi’, as stated by Jullien (2014) and Pekarek Doehler (2015). Examples such as (449), in which the ‘il y a NP’ main clause is followed by a personal pronoun that is coreferential with the introduced NP (ils ‘they’) and by the relativizer qui ‘that’ (introducing a relative clause) can be seen as an indication that the two continuations (a new clause introduced by ils ‘they’ and a relative clause introduced by qui ‘that’) are indeed in competition with each other (see also chapter 11.2.3 about this example).

Chapter 12 | No evidence (yet) for il y a clefts as a construction 343

(449) A: y a rien qui est aménagé spécialement pour vous\ ‘there’s nothing that is arranged specially for you’

B: .. non bon i:- s:- y a les les les autres les camarades de classe là ils qui m’aident aussi qui me qui me prêtent leur no:tes ou qui m’expliquent si j’ai pas compris/ (…)

.. ‘no, well, th:- ye:- there are the the the others the schoolmates there they who help me as well who me who lend me their notes or who give me explanations if I don’t understand (…)’

(Jullien 2014:305, my translation, his underlining and italics, my boldface)

Summing up, there is no conclusive evidence in favor of a psychologically real overall ‘cleft construction’ that is underspecified for the type of main clause in our mental grammar. However, the use of the term ‘cleft’ as a cover term does have its merits: it is descriptively useful to refer to sufficiently similar patterns within and across languages by means of the same term as a way to stimulate new discoveries (see discussion about comparative concepts in chapter 10.8).

344 Chapter 12 | No evidence (yet) for il y a clefts as a construction

12.6 Conclusion

In this chapter, I argued that whether or not il y a clefts are seen as constructions in the sense of Construction Grammar remains speculative. Given that il y a clefts are compositional (see Chapter 11), they are not constructions in the narrow, non- compositional sense (i.e. a form-meaning pairing whose meaning is more than the sum of its parts). Apart from this narrow definition, a number of scholars define ‘constructions’ in a broader sense as patterns that are conventionalized. I illustrated several accounts of clefts that use this broader definition of constructions.

However, with respect to il y a clefts, I argued that since there is no convincing empirical evidence that il y a clefts are conventionalized, I cannot conclude that il y a clefts are constructions in the broad sense. It is true that some instantiations (i.e. tokens) of il y a clefts, such as il y a quelque chose qui cloche ‘there’s something wrong’ may be relatively routinized due to their frequency. However, this does not allow us to make claims about the conventional nature of il y a clefts in general (i.e. as a type). Future research using methodologies such as statistically- oriented corpus analyses or self-paced reading may provide more evidence in favor or against the existence of an ‘il y a cleft construction’ in our mental grammar.

From a cross-linguistic perspective, I argued that the similarities between il y a clefts and Italian c’è clefts can be explained by the fact that both languages dispose of the two necessary building blocks (existential main clause + relative clause) and are subject to the same functional pressures (most notably the need to avoid sentence- initial non-topics). Hence, there is no need to postulate a special cross-linguistic ‘existential cleft construction’, although this term can be descriptively useful for comparative linguistic analyses.

In a similar vein, I argued that there is no evidence for an overarching ‘French cleft construction’ in speakers’ mental grammars that would include different French cleft types such as c’est clefts, il y a clefts and voici/voilà clefts. The similarities and differences between those cleft types can be explained by referring to their subcomponents. Again, however, I did acknowledge the descriptive usefulness of such a term for linguistic analysis: analyzing different cleft types together and drawing on studies of one cleft type (e.g. c’est clefts) to investigate another cleft type (e.g. il y a clefts) can lead to new insights.

General conclusion

The general goal of this dissertation was to arrive at a better understanding of French il y a clefts (450). It has been claimed that il y a clefts are characteristic of spoken French and that their main function is to introduce a discourse-new entity or event. In this sense, they represent an interesting object of study for investigating the interaction between word order patterns and discourse-functional motivations.

(450) Il y a cleft – Toto, pourquoi n'irais-tu pas sur le balcon. Comme ça tu nous raconteras

tout ce qui se passe en bas! (…)

‘Toto, why don’t you go to the balcony. Then you can tell us about everything that’s happening below!’

– Toto: Il y a une voiture qui se fait remorquer par une dépanneuse dans le parking...

Toto: ‘There’s a car that’s getting towed by a tow truck in the parking lot.’

(YCCQA)

Although il y a clefts have received some attention, also on the basis of corpus data (see most notably Lagae & Rouget 1998; Choi-Jonin & Lagae 2005; Verwimp & Lahousse 2016), they have not been subject to a systematic quantitative and qualitative corpus analysis of the scale of this dissertation before. My corpus data allowed to empirically confirm several hypotheses and observations from previous analyses, for instance the fact that il y a clefts predominantly introduce discourse- new, non-topical constituents. Moreover, I presented a more nuanced analysis of il y a clefts that diverges from prior accounts in several respects. In what follows, I present an overview of the main contributions of this dissertation.

346 General Conclusion

Part I – Delimitation of il y a clefts

After summarizing previous accounts of il y a clefts (Chapter 1), I presented the corpus data that were used for the analysis of French il y a clefts presented in this dissertation (Chapter 2). The data were extracted from different corpora: spoken French (Corpus de Français Parlé Parisien des années 2000), informal written French (Yahoo Contrastive Corpus of Questions and Answers) and written journalistic French (Le Monde). A total number of 8570 il y a sentences were extracted and manually coded. Of these, 5439 were put aside as noise because they lack a relative clause, and 3131 il y a… qui sentences were analyzed in more detail.

In previous analyses of il y a clefts, two main criteria are presented that characterize il y a clefts: (i) the sentence must correspond to a canonical equivalent with SV word order expressing the same proposition (i.e. the ‘decleftability’ criterion), and (ii) the relative clause must not be a restrictive relative clause (‘RRC’) (i.e. the ‘type of relative clause’ criterion). These two criteria indeed apply to the example in (450): the relative clause is not restrictive, and the whole sentence can be declefted, i.e. transformed into a canonical SV counterpart, as shown in (451).

(451) Canonical SV equivalent of (450) (= decleftability criterion) … une voiture se fait remorquer par une dépanneuse dans le parking.

‘….a car is getting towed by a tow truck in the parking lot.’

I applied the criteria from previous analyses on the corpus data in order to distinguish between il y a… qui sentences that are clefts and those that are not clefts (i.e. ‘cleft lookalikes’) (Chapter 3). I tried to disentangle the different diagnostics that have been proposed for il y a clefts in order to determine whether those diagnostics allow to make clear-cut distinctions in all 3131 il y a… qui sentences in my dataset. Apart from clear il y a clefts such as (450), a considerable number of il y a… qui sentences in my corpus data (40%) remain unclear with respect to the two main diagnostics described above: (i) in certain cases, it is unclear whether the relative clause is restrictive or not (see (452)). (ii) In other cases, the sentence can be declefted, but it also contains a RRC (see (453)). In these cases, the ‘decleftability criterion’ indicates that the sentence is a cleft, but the ‘type of relative clause criterion’ indicates that the sentence is not a cleft.

General Conclusion 347

(452) Unclear il y a… qui sentence (Type 1: Type of relative clause is unclear) Pourquoi y'a plein de gens qui posent des questions débiles au lieu de chercher sur Google ?? (YCCQA)

– Interpretation 1. ‘Why is it that so many people are asking stupid questions instead of googling it?’ = cleft reading

– Interpretation 2. ‘Why are there so many people who ask stupid questions instead of googling it?’ = RRC reading

(453) Unclear il y a… qui sentence (Type 2: decleftable + RRC) y avait beaucoup d'élèves qui étaient motivés mais y en avait aussi une bonne partie qui n'en avait rien à faire

‘There were many students who were motivated but there was also a large group that didn’t care.’

~ Beaucoup d'élèves étaient motivés. ‘Many students were motivated.’ (CFPP)

Given that previous analyses of il y a clefts do not explicitly state whether the two criteria are sufficient and/or necessary, it is unclear whether sentences such as (452) and (453) should be considered as clefts or not. Consequently, I argued that there are several possible definitions of il y a clefts, depending on whether one sees the two main diagnostics as sufficient and/or necessary. Since most previous studies of il y a clefts seem to assume that il y a clefts can never contain RRCs (e.g. Lambrecht 1986, 2001), I decided to adopt a definition of il y a clefts in which both criteria must be met, i.e. they are both necessary and neither of them is sufficient by itself. According to this definition, il y a… qui sentences that can be declefted but which at the same time contain a RRC (such as (453)) are not instances of clefts. I refer to this class as ‘decleftable il y a… qui sentences with RRC’ and explain in Part III which properties characterize these instances. The way I classify the data is presented in Figure 7. Using this definition of il y a clefts, I identified 576 il y a… qui sentences in my data as instances of il y a clefts.

348 General Conclusion

All il y a… qui sentences

Decleftable Not decleftable

Restrictive relative clause Cleft relative clause (RRC) (not RRC)

“Decleftable il y a… qui “Il y a cleft” “Il y a cleft lookalike” sentence with RRC”

Figure 7: Classification of different il y a... qui sentence types

Part II – Distribution, semantic and morphosyntactic properties

I analyzed the il y a clefts in terms of their distribution throughout the different corpora, and their main semantic and morphosyntactic properties. Besides ‘regular’ il y a clefts (Chapter 4), I used a separate query to extract and subsequently analyze il y a clefts containing the ne… que ‘only’ particle (Chapter 5).

As for the distribution of il y a clefts, I was able to empirically confirm the statement made in previous analyses that il y a clefts are characteristic of spoken French rather than of written French (e.g. Blanche-Benveniste 1997:93; Choi-Jonin & Lagae 2005; Willems & Meulleman 2010). As shown in Table 34, il y a clefts are by far most frequent in the spoken corpus CFPP (191 il y a clefts per 500.000 words). The journalistic corpus Le Monde contains only 2 il y a clefts per 500.000 words, whereas the informal written corpus YCCQA contains approximately 23 il y a clefts per 500.00 words and is thus in between the other two corpora. The fact that over 80% of the il y a clefts in Le Monde occur within a citation in the text is another indication that il y a clefts are characteristic of spoken French rather than of (formal) written French.

General Conclusion 349

Le Monde YCCQA CFPP Total

Corpus size in words 25700000 6100000 550.000 32.350.000

Total number of il y a clefts 81 285 210 576 Number of il y a clefts 2 23 191 9 per 500.000 words Table 34: Normalized frequency of il y a clefts throughout the corpora

I then turned my attention to the semantic and morphosyntactic properties of il y a clefts. Previous studies of il y a clefts are almost exclusively devoted to il y a clefts that express an event (such as (450)), but the existence of a second semantic subtype of il y a clefts has also been acknowledged a few times (e.g. Lambrecht 2001; Verwimp & Lahousse 2016, in line with what e.g. Davidse 2000; Davidse & Kimps 2016 propose for English there clefts). This subtype of il y a clefts has specificational rather than eventive semantics. For instance, the il y a cleft in (454) specifies a certain value (homeopathy) for a discourse-given variable (good ways to get rid of stress).

(454) Specificational il y a cleft – A: Je vais bientôt passer un examen, et j'ai le trac. pouvais vous me donner

des conseils ?

‘I have an exam soon and I’m stressed out. Can you give me some advice?’

– B: détends-toi, fais aussi du sport, ça permet de bien éliminer le stress. Il y a aussi l'homéopathie qui fonctionne vraiment super bien !

‘Relax, do some sports, that enables you to get rid of the stress. There’s also homeopathy that works really well! (YCCQA)

Specificational il y a clefts such as (454) are overall much less frequent (n= 94, i.e. 16%) than eventive il y a clefts (n= 482, i.e. 84%), which may explain why this subtype has received little attention. There is, however, a clear difference in distribution across the corpora: specificational il y a clefts are much more frequent in the YCCQA and CFPP corpora than in Le Monde (21%; 16% and 2% of all il y a clefts respectively), which I explained in terms of the interactive set-up of those corpora: specificational il y a

350 General Conclusion

clefts mostly occur in answers to questions, which are frequent in YCCQA and CFPP. The Le Monde corpus, on the other hand, does not have a ‘question-answer’ setup.

Moreover, I identified several morphosyntactic and semantic differences between eventive and specificational il y a clefts. In the specificational clefts, 59% of the clefted constituents are definite, while only 11% of eventive il y a clefts contain definite constituents (despite the fact that many of the examples of ‘prototypical il y a clefts’ in previous analyses contain proper names such as Jean, see e.g. Lambrecht 1986, 1988a, 2002). I argued that the discourse function of specificational il y a clefts (specifying a value for a variable) explains the relatively high number of definite clefted constituents. Finally, I identified a subclass of specificational il y a clefts that have also been identified in analyses of English there clefts (see Davidse 1999, 2000 and Davidse & Kimps 2016). Such instances of specificational il y a clefts quantify a variable, as in (455): the variable is ‘girls who are looking for a guy who is not a pain in the ass’, and the clefted element expresses the quantity ‘999.999’.

(455) Specificational il y a cleft of the “quantifying” type – A: Y a t'il 1 fille sympa qui cherche 1 mec pas chiant ?

‘Is there a nice girl who’s looking for a guy who’s not a pain in the ass?’

– B: Il y a environ 999'999 de filles qui cherche un mec pas chiant

‘There are about 999.999 girls who are looking for a guy who’s not a pain in the ass.’ (YCCQA)

In Chapter 5, I used a separate targeted query to extract il y a clefts with the restrictive particle ne… que ‘only’, i.e. il n’y a que clefts, such as (456).

(456) Il n’y a que cleft y'a que toi qui comprend ou c'est moi qui pige rien à rien?

‘Are you the only one who understands or is it me who doesn’t understand a thing?’ (Lit. ‘Is there only you who understands…’) (YCCQA)

I showed that il n’y a que clefts are the least frequent in the Le Monde corpus, which is in line with the frequency differences found for il y a clefts without ne… que ‘only’. Moreover, although il n’y a que clefts, contrary to il y a clefts without ne… que ‘only’,

General Conclusion 351

are able to introduce prepositional phrases, adverbials and conditional/temporal clauses in the clefted position, as illustrated in (457), they do not do so very often: 87% of the clefted elements in il n’y a que clefts are (pro)nominal (i.e. pronouns, NPs or proper nouns).

(457) Il n’y a que cleft + clausal clefted element Pourquoi ya que quand je dis des conneries et des trucs qu'ils ne veulent rien dire qu'on me repond?

‘Why do people only answer when I say stupid things and things that don’t mean anything?’

(Lit. ‘Why is there only when I say stupid things and things that don’t mean anything that people answer?’) (YCCQA)

The large majority of clefted constituents in il n’y a que clefts are definite (89%), which can be explained by the fact that most il n’y a que clefts are of the specificational subtype, providing a specific value (encoded by a definite NP) for a given variable.

On a more general level, Part II gives a more complete and accurate description of the distribution and the morphosyntactic and semantic properties of il (n’)y a (que) clefts than previous analyses, which are predominantly based on only few (and sometimes made-up) examples (e.g. Lambrecht 1986). Most important in this respect are: (i) the distinction between eventive and specificational il y a clefts, which is not always acknowledged; (ii) the fact that the large majority of clefted constituents in eventive il y a clefts are indefinite rather than definite, and (iii) the fact that il n’y a que clefts mostly introduce (pro)nominal constituents rather than prepositional phrases and adverbials.

352 General Conclusion

Part III – Functions of il y a clefts

The main goal of this part of the dissertation was to examine the functions of il y a clefts by means of a corpus analysis, in order to explain what motivates speakers to use such clefts rather than their SV counterpart. In order to do so, I verified two hypotheses from previous analyses: (i) il y a clefts (and il y a… qui sentences more generally) are used in order to avoid suboptimal subjects from appearing in preverbal subject position (e.g. Lambrecht 1986; Cappeau & Deulofeu 2001), and (ii) il y a clefts are used in order to introduce focal and/or non-topical constituents (e.g. Lambrecht 1994, 2000a, 2001). Moreover, I identified a number of additional functions of il y a clefts that had not been reported before.

In Chapter 6, I investigated the hypothesis according to which il y a… qui sentences are used in order to avoid suboptimal subjects from appearing in preverbal subject position. I did so by systematically comparing il y a… qui sentences with their SV counterpart, which also allowed me to gain more insight in the phenomenon of ‘decleftability’. In this chapter, I considered not only instances of il y a clefts, but the broader category of decleftable il y a… qui sentences (see Part I). I argued that my corpus data support the hypothesis under investigation, because all il y a… qui sentences in the corpus data introduce constituents that would make suboptimal preverbal subjects in a SV sentence. For instance, the NP certains hommes ‘some men’ in (458) is a suboptimal preverbal subject in the canonical equivalent in (459), because it is indefinite. Indefinite NPs are unnatural in this position, especially in spoken French (Van De Velde 1995). However, this NP is nonetheless not excluded from the preverbal subject position thanks to the partitive determiner certains ‘some’.

(458) – A: Pourquoi nous les hommes, on as peur que les femmes nous aiment pour de vrai ??

‘Why are we, men, scared that women really love us?’

– B: Il ne faut pas généraliser. Il y a certains hommes qui ont vraiment besoin de sentir aimés (…).

‘You shouldn’t generalize. There are some/certain men who really feel the need to be loved (…).’ (YCCQA)

General Conclusion 353

(459) Certains hommes ont vraiment besoin de se sentir aimés. ‘Some men really feel the need to be loved.’

However, I also showed that ‘subject acceptability’ is a gradient phenomenon: there is a variety of interacting factors that influence to what extent a constituent is acceptable as a preverbal subject (e.g. definiteness, quantification, discourse- newness, specificity). Moreover, I argued that the function of avoiding suboptimal preverbal subjects can shed light on the frequency difference of il y a clefts in spoken and written French (see Chapter 4): I hypothesized that the conditions that a preverbal subject must fulfill are stricter in spoken French than in written French, because in spoken French, there is a higher need to facilitate online language processing for the interlocutor than in writing. Therefore, spoken French resorts to il y a… qui sentences more frequently than written French.

By systematically determining for each il y a… qui sentence in the corpus data whether it could be transformed into a SV counterpart and what licenses or blocks this transformation, I also clarified the phenomenon of decleftability: I explained that whether or not this transformation is possible depends to a large extent on the subject acceptability of the introduced constituent. Hence, decleftability is not due to a special (syntactic) cleft structure, but can be derived from other properties. Moreover, given that subject acceptability is a gradient phenomenon, it follows that some sentences can be declefted more easily (relatively high acceptability) than others (low acceptability), as illustrated in Figure 8:

Declefting is easy Constituent is highly acceptable as a preverbal subject

Declefting is relatively easy Constituent is relatively acceptable as a preverbal subject

Declefting is difficult Constituent is hardly acceptable as a preverbal subject

Figure 8: Link between decleftability and subject acceptability

354 General Conclusion

In Chapter 7, I presented an in-depth analysis of the information structure properties of all il y a clefts. Previous accounts only presented two articulations for il y a clefts: all- focus (see (450)) and, in a few analyses, focus-background (see (454)). I showed (i) that the focus-background articulation is not as infrequent as one might think given that it has not been acknowledged often in previous studies, and (ii) that the corpus data display a wider variety of possible articulations than all-focus and focus-background: il y a clefts also sporadically express contrastive topic-comment; they can have a double focus articulation; and in certain cases in the written corpora, the exact articulation seems to be underspecified.

I also confirmed the hypothesis that il y a clefts are used to introduce non- topical constituents and argued that (i) announcing to the hearer that a non-topical constituent is coming and (ii) avoiding suboptimal subjects from appearing in preverbal subject position are two sides of the same coin. From both perspectives, il y a clefts can be seen as a strategy for optimizing language processing. However, the il y a cleft is not the only type of il y a sentence that has the function of announcing a non-topical constituent: I argued that this function is due to the existential expression il y a, since il y a sentences without relative clauses generally do not introduce topical constituents either (see (460)).

(460) [– A: ‘Where are the towels?’] – B: # Il y a les serviettes dans le tiroir.

# ‘There are the towels in the drawer.’

I also showed that existential sentences without relative clauses can express the same information structure articulations as il y a clefts. It follows from this that in terms of information structure, il y a clefts do not have unique functions that other il y a sentences lack.

As for the few cases (1%) in which il y a clefts do introduce (spatio-temporal and aboutness) topics, these are not regular but contrastive topics (i.e. simultaneously topical and focal, see e.g. Erteschik-Shir 2007). This is illustrated in (461), where the clefted constituent plein de jeunes ‘plenty of youngsters’ is contrasted with nous ‘us’. Hence, instances such as (461) should not be seen as compelling evidence against the hypothesis that il y a functions as a processing cue that signals

General Conclusion 355

the absence of a topic. I argued that contrast can be seen as a survival strategy for topics in a hostile environment (i.e. the position behind il y a) (see Leonetti 2013 for a similar analysis of contrastive NPs in other positions). In other words, topics are normally excluded from the position behind il y a, and require some additional licensing factor: contrast.

(461) Contrastive topic-comment il y a cleft « Nous, on a de la chance, on a des employeurs corrects. Mais il y a plein de petites jeunes qui se font avoir », poursuit Lina.

‘“Us, we’re lucky, we have good employers. But there are plenty of youngsters who are getting ripped off”, continues Lina.’ (Le Monde)

In Chapter 8, I contrasted the information structure properties of il y a clefts and c’est clefts. It turns out that the two cleft types have the same information structure potential. However, there are substantial differences between the two French clefts in terms of the frequency of the different information structure articulations. I argued that several morphosyntactic and pragmatic properties of the two cleft types (partially) explain those differences. For instance, previous corpus studies (e.g. Dufter 2006, 2009b) show that c’est clefts very frequently introduce spatiotemporal topics that are encoded by anaphoric adverbials or prepositional phrases, as in (462).

(462) C’est également dans le Kurdistan irakien que se présente pour l’Union européenne l’occasion exceptionnelle d’apporter une contribution substantielle à la tâche urgente de reconstruire la Mésopotamie. ‘It is also in Iraqi Kurdistan that the European Union is being presented with an outstanding opportunity to make a substantial contribution to the urgent task of reconstructing Mesopotamia.’ (Dufter 2009b:102, no prior context)

However, il y a clefts are grammatically unable to introduce adverbials unless the ne… que ‘only’ particle is present. As a consequence of this restriction, only very few instances of il y a clefts (which contain ne… que ‘only’) introduce spatiotemporal topics.

356 General Conclusion

In Chapter 9, I examined additional functions of il y a clefts. Clefts are often presented as a textbook example of a structure that is motivated by the function of focalization. I presented several discourse functions of il y a clefts that are not immediately linked to information structure and which their canonical counterparts do not have: reinforcing a negation (see (463)); enhancing discourse coherence and signaling a narrow scope reading of a singular indefinite NP.

(463) La tristesse touche l'os, la moelle. Il n'y a rien qui soit aussi profond que la tristesse. ‘Sadness touches bone and marrow. There’s nothing that’s as profound as sadness.’ (YCCQA)

I also argued that the relative clause in specificational clefts have certain pragmatic functions (expressing subjectivity and/or relevance) that explain why the speaker uses a cleft rather than an existential sentence without a relative clause in those contexts. On a more general level, my analysis is in line with an increasing tendency to see clefts as multifunctional utterances whose motivation cannot be reduced to the expression of information structure (e.g. Delin 1989; Delin & Oberlander 2005; Dufter 2006).

In general, it has become clear that there are a number of (inter-related) reasons that make the speaker opt for an il y a cleft rather than for its canonical SV equivalent. Moreover, although my analysis confirms the importance of information structure in order to explain the use of il y a clefts, it would be too simplistic to see il y a clefts as mere focus markers.

General Conclusion 357

Part IV – Italian c’è clefts (vs. French il y a clefts)

In this part of the dissertation, I adopted a comparative perspective, contrasting il y a clefts with their under-researched Italian counterpart, the c’è cleft (De Cesare 2007; Marzo & Crocco 2015), in order to pinpoint the differences and similarities between the two languages. For the analysis of Italian c’è clefts, I analyzed a total number of 2187 c’è… che ‘there is… that’ sentences extracted from the journalistic corpus La Stampa. 231 of these sentences were identified as clefts according to the same definition used for the analysis of il y a clefts (i.e. the sentence must correspond to a canonical equivalent expressing the same proposition, and the relative clause must not be restrictive). I showed that c’è clefts are strikingly similar to French il y a clefts: (i) c’è clefts display the same delimitation ambiguities (cleft vs. cleft lookalikes) as il y a clefts, (ii) they can express the same information structure articulations (see (464) - (466)), and (iii) c’è clefts also have additional functions that are not directly related to information structure, such as reinforcing negation.

(464) All-focus c’è cleft (with eventive semantics) Nell'auto dietro a Vignola, c'era un collega, munito di telefono cellulare che ha dato l'allarme: "C'è un uomo che spara per strada.”

‘In the car behind Vignola, there was a colleague with a cellphone who raised the alarm: “There’s a man shooting in the streets.”’ (La Stampa)

(465) Focus-background c’è cleft (with specificational semantics) Molti chiedono anche l'istituzione di una figura che rappresenti il minore: l'avvocato dei bambini. "Non sono d'accordo - osserva l'onorevole Lucidi - perché c'è già il pm che può rivestire questo ruolo, magari si può rafforzare il suo potere d'intervento".

‘Many people ask the institution to provide a figure who represents minors: a children’s advocate. “I disagree – observes Mr. Lucidi – because there’s already the prosecutor who can take on this role, maybe his right to intervene can be strengthened”.’ (La Stampa)

358 General Conclusion

(466) Contrastive topic-comment c’è cleft [Beginning of article] Lido Vieri, preparatore dei portieri del Toro, è anche uno studioso di questo ruolo speciale, diverso da tutti gli altri. Forse, il più delicato sotto il profilo psicologico. Se un attaccante, un centrocampista o un difensore sbagliano, c'è il portiere che può metterci una pezza. Se sbaglia lui, non c'è rimedio. E il portiere è uno degli elementi-cardine, si punti al primo posto o si lotti per la salvezza.

‘Lido Vieri, trainer of the goalkeepers of Toro, also studies in depth this special role, different from all the others. Maybe even the most delicate one psychologically. If an attacker, a midfielder or a defender make a mistake, there’s the goalkeeper who can make up for it. If he makes a mistake, there’s no remedy. And the goalkeeper is one of the pivotal elements, he is in first position and struggles for salvation.’ (La Stampa)

I also examined the prediction that c’è clefts are less frequent than il y a clefts, which is based on the observation that Italian, contrary to French, disposes of a second word order pattern (VS) that achieves a similar effect as c’è clefts, namely avoiding a non- topical constituent from appearing in preverbal subject position (see Lambrecht 1994, 2001). Interestingly, this prediction is not borne out by the corpus data: c’è clefts are slightly more frequent than il y a clefts (5.3 c’è clefts per 500.000 words in La Stampa vs. 1.6 il y a clefts per 500.000 words in Le Monde). I stated that this surprising result is probably due to the fact that journalistic Italian is characteristic of ‘neo-standard Italian’, which incorporates patterns from informal spoken Italian (such as c’è clefts). In French, on the other hand, there is a clearer difference between the journalistic register (which is relatively formal) and (informal) spoken French. Lastly, I argued that the extensive similarities between French il y a clefts and Italian c’è clefts make it empirically desirable to refer to them with the same term, ‘existential cleft’. However, we should not necessarily take these similarities as evidence that existential clefts are a psychologically real cross-linguistic category of which il y a clefts and c’è clefts are two instantiations.

All in all, the comparison between French and Italian was fruitful in the sense that insights from French allowed to identify characteristics that had not been reported for Italian and vice versa.

General Conclusion 359

Part V – Compositionality and constructionality of il y a clefts

In this more theoretically oriented part of the dissertation, I examined to what extent il y a clefts can be analyzed in a compositional way and whether they can be seen as ‘constructions’ in the sense of Construction Grammar. In this way, I tried to shed some light on the way in which il y a clefts might be represented in speakers’ mental grammar: as patterns that are stored ‘as a whole’ (i.e. a non-compositional analysis) or as as separate building blocks that are combined in a compositional way.

In Chapter 11, I addressed the question whether il y a clefts are compositional, as argued for by Jullien (2014), or non-compositional (i.e. more than the sum of their parts), as argued for by Lambrecht (1986, 2001). I first showed that the arguments in favor of a non-compositional analysis presented by Lambrecht are not convincing, because the properties he ascribes to the whole il y a cleft format (e.g. the ban on negation) can be derived from the subcomponents of il y a clefts. I then presented arguments in favor of a compositional analysis: (i) the discourse functions of il y a clefts and the fact that il y a clefts can be transformed into a canonical counterpart are not unique to il y a clefts (they share these with il y a sentences that lack a relative clause, see Part III) and can be derived from the subcomponents of il y a clefts (existential main clause + relative clause). (ii) Il y a clefts do not have a unique inherent semantics: they can be eventive or specificational. Therefore, there cannot be one single ‘il y a cleft construction’ that is associated with one semantic value. In sum, there is no clear indication of a non-predictable aspect of function or meaning.

In Chapter 12, I tried to determine whether il y a clefts can be analyzed as ‘constructions’ in the sense of Construction Grammar. According to some scholars (e.g. Goldberg 1995), constructions are necessarily non-compositional (a ‘narrow’ definition). Given the compositional nature of il y a clefts described in Chapter 11, I argued that il y a clefts cannot be seen as linguistic constructions in this strict, non- compositional sense. According to a second definition of ‘construction’, a given form- meaning pairing is a construction as long as it is conventionalized, regardless of whether it is compositional or not (e.g. Nikiforidou 2009). I presented several analyses of clefts that use this broader definition and argue that clefts are compositional constructions (e.g. Davidse 2000). However, with respect to il y a clefts, I argued that there is no empirical evidence (yet) that they are indeed conventionalized in the

360 General Conclusion

speech community. Hence, there is no clear evidence that they should be seen as constructions. More research is needed to address this issue. In this respect, methods that can measure the degree of conventionalization of il y a clefts (e.g. statistically oriented corpus analyses and psycholinguistic experiments) are particularly promising.

On a general level, the picture that emerges from this dissertation is the following. In previous studies, il y a clefts were presented as a well-delimited class of sentences with a relatively uniform discourse-functional profile (the ‘presentational function’: introducing a new referent or event). However, my data contain a considerable number of il y a… qui sentences whose status as il y a clefts or cleft lookalikes is unclear when adopting the diagnostics from previous analyses. Moreover, a close scrutinization of the discourse contexts and information structure properties of il y a clefts and their canonical equivalents has revealed that there is an intricate interplay of (morphosyntactic, semantic, information structure and contextual) factors that motivate the use of il y a clefts. What’s more, many of these properties characterize il y a sentences more generally, e.g. the function of introducing non-topics is not unique to il y a clefts but holds for existential sentences with relative clauses as well. As a consequence, il y a clefts should not be seen as a phenomenon that stands on its own. Hence, it is important not to study il y a clefts in isolation, that is, without taking into consideration other sentence types that are closely related to il y a clefts in form and/or function.

General Conclusion 361

Future perspectives

While this dissertation led to new insights and provided answers to (old and new) questions, it also pointed out a number of issues that could not yet be resolved. For instance, I hypothesized on the basis of the corpus data that il y a functions as a processing cue that announces a non-topic. In order to verify this hypothesis, one could perform psycholinguistic experiments such as self-paced reading and eye tracking, in line with previous studies of cross-linguistic counterparts of il y a (Dutch, self-paced reading: Grondelaers et al. 2002, English, eye-tracking: Besserman et al. 2015). Another issue that remains unsolved is whether or not il y a clefts are entrenched at the level of the individual speaker’s mental grammar and conventionalized at the level of the speech community. The possible entrenchment of il y a clefts could for instance be investigated by priming experiments, and the (degree of) conventionalization of il y a clefts could be measured by means of a quantitative corpus analysis (e.g. a collostructural analysis). Moreover, it would be worth looking at the link between information structure and il y a/c’è clefts from a different perspective: in this dissertation, I departed from a form (il y a/c’è cleft) and determined that this form corresponds to an all-focus information structure articulation in most instances. In contrast, one could investigate the all-focus articulation by considering a given extract of authentic language and analyzing all the different forms that express the all-focus articulation (e.g. other cleft types, VS word order in Italian). This would undoubtedly contribute to our understanding of the link between information structure and word order patterns.

In short, a lot remains to be done before we can fully comprehend il y a clefts and c’è clefts. This dissertation will hopefully inspire future research about these intriguing phenomena.

Appendix

Table of contents

Section 1 Relativizers other than qui/que/qu’ ...... 364

Section 2 Comparability across the corpora ...... 365

Section 3 Formal variability of il y a in il y a clefts ...... 367

Section 4 All il y a… qui sentences ...... 371

4.1 Subject vs. object relative clauses ...... 371

4.2 Constituent introduced by il y a: syntactic category ...... 371

4.3 Constituent introduced by il y a: (in)definiteness ...... 372

4.4 Constituent introduced by il y a: semantic category ...... 372

Section 5 Decleftable il y a… qui sentences ...... 375

Section 6 All il n’y a que… qui sentences ...... 377

6.1 Properties of the relative clause/infinitival clause ...... 377

6.2 Properties of the constituent introduced by il n’y a que ...... 380 6.2.1 Grammatical category ...... 380 6.2.2 (In)definiteness ...... 381

Section 7 Eventive & specificational il y a clefts ...... 383

7.1 Clefted constituent: syntactic category (eventive il y a clefts) ...... 383

7.2 Clefted element: (in)definiteness (eventive il y a clefts) ...... 383

7.3 Subject vs. object relative clauses ...... 383

7.4 Specificational il y a clefts: additional examples ...... 384

Section 8 Il n’y a que clefts ...... 387

364 Appendix

Section 1 Relativizers other than qui/que/qu’

(467) Pour ‘for’ + infinitive Il y a encore des gens pour croire aveuglement à leurs mensonges.

‘There are still people who blindly believe their lies.’ (YCCQA)

(468) À qui ‘to whom’ Je sais qu’il y a une à qui s’est arrivé.

‘I know there’s one woman that it happened to.’ (YCCQA)

(469) Où ‘where’ Mais il y a des moments où ça devient une vraie corvée.

‘But there are moments when it becomes a real chore.’ (YCCQA)

(470) Dont ‘of which’ Il y a vraiment des choses dont il faut que nous parlions...

‘There are really things that we have to talk about.’ (YCCQA)

(471) Avec qui ‘with whom’ Parce (…) qu'il n'y a personne avec qui la partagé.

‘Because there’s nobody with whom I can share her.’ (YCCQA)

(472) Sur lequel ‘on which’ Même si il ya bien un sujet sur lequel j'aimerai avancer.

‘Even if there is indeed a topic on which I would like to progress.’ (YCCQA)

(473) Par qui ‘by whom/which’ Il n’y a d’autre nom sous le Ciel par qui nous puissions être sauvé.

‘There is no other name under the Heavens by which we can be saved.’ (YCCQA)

(474) Pour qui ‘for whom’ Y a des gens pour qui faire 50m à pied est du sport.

‘There are people for whom walking 50 meters is doing sports.’ (YCCQA)

Appendix 365

Section 2 Comparability across the corpora

There are four groups of il y a… qui sentences that were extracted from YCCQA and CFPP but which were not extracted from Le Monde, presented in Table 35 and illustrated below.

YCCQA CFPP Total

n % n % n % Constituent longer than 4 words 207 63% 97 80% 304 67% Ne between il and y 79 24% 24 20% 103 23% Adverb between a and eu 2 1% 1 1% 3 1% Y a-t-il 42 13% 0 0% 42 9% Total 330 100% 122 100% 452 100%

Table 35: Types of il y a… qui sentences that were not extracted from Le Monde

– Constituent longer than 4 words –

In the Le Monde query, the number of words between il y a and qui/que/qu’ is limited to four. No such limit was imposed on the YCCQA and CFPP queries. Consequently, examples such as (475), which do appear in the YCCQA and CFPP data, are not extracted from the Le Monde corpus.

(475) Introduced element longer than 4 words je suis hélas quelqu'un qui cache ses sentiments, et il y a très peu de gens autours de moi qui arrive à détecter ce que je cache.

‘Unfortunately, I am someone who hides his feelings, and there are very few people around me who are able to detect what I’m hiding.’ (YCCQA)

– Ne between il and y –

In the Le Monde query, nothing can appear in between il, y and the verb avoir. Therefore, sentences such as (476), in which the negative particle ne is located in between il and y, were not extracted from the Le Monde corpus. However, they do appear in the YCCQA and CFPP data, because their queries do not contain the

366 Appendix

expletive il (since it is often dropped in informal written/spoken French). Consequently, all instances in which a negation was present (ne… pas ‘not’, ne… personne ‘nobody’, ne… aucun ‘no’ etc.) in the YCCQA and CFPP data are not considered in the quantitative comparison between the corpora. I also exclude cases in which the ne particle is dropped, as in (477).

(476) Ne between il and y Il n’y a aucun aliment par lui-même qui fait maigrir.

EXPL NEG.there has no food by itself that makes lose.weight

‘There is no food product that makes you lose weight by itself.’ (YCCQA)

(477) Negation in which ne is not expressed Et y a rien qui se passe.

‘And there is nothing that happens.’

(Normative spelling: Il n’y a rien qui se passe.) (YCCQA)

– Adverb between a and eu –

In a few instances, there is an adverb between the copula a ‘has’ and the past participle eu ‘had’, as in in (478).

(478) Adverb between a and eu Y a toujours eu ceux qui savaient c' qu'ils voulaient.

‘There have always been those who knew what they wanted.’ (CFPP)

–Y a-t-il –

Since the YCCQA query does not contain the expletive il, it also yielded tokens in which y a-t-il ‘is there’ introduces a question, as in (479). Such instances were not extracted from Le Monde, due to the absence of il in front of y a.

(479) Y a-t-il Ya til kel k'1 de bien qui pourai m'invitez pour ouvrir un compte gmail (…) ?

‘Is there someone nice who could invite me to open a gmail account (…)?’ (YCCQA)

Appendix 367

Section 3 Formal variability of il y a in il y a clefts

(480) Past tense (passé composé) Mais encore une fois ils disent ça mais à la fin y a eu quand même plusieurs fois des voisins qui sont venus manger là au moins.

‘But again, they say that, but in the end some neighbors at least did come over for dinner there several times.’

(Lit. ‘There have at least been neighbors who came over for dinner several times at least.’) (CFPP)

(481) Past tense (imparfait) au café toutes les petites cuillères étaient percées pour éviter que les parce que il y avait des jeunes qui venaient et qui faisaient chauffer de la cocaïne.

‘at the café, all the small spoons were pierced to avoid that the because there were young people who came and heated cocaine.’ (CFPP)

(482) Past tense (imparfait) + negation (…) j'suis allé voir si y avait pas quelque chose qui brûlait.

‘ (…) I went to see if there wasn’t something that was burning.’ (CFPP)

(483) Future tense Mardi soir, sur France 2, il a averti que « s'il doit y avoir des différences qui continuent à apparaître, il y aura des recompositions qui se feront, c'est certain ».

‘Tuesday night, on France 2, he warned that “if there must be differences that continue to appear, there will be recompositions that will take place, that’s for sure.”’ (Le Monde)

(484) Subjunctive mood Moi j'sais qu'j'aime bien c'quartier oui oui j'aime bien j'ai peur moi que qu'il y ait trop d'Asiatiques qui viennent s'installer.

‘Me, I know that I like this neighborhood, yes yes I like it, me I’m afraid that there are too many Asian people who come and settle down.’ (CFPP)

368 Appendix

(485) Subjunctive mood + negation Le métro c'est c'est pas très joy- c'est pas très marrant euh parce que on peut pas prendre une rame sans qu'y ait euh sans qu'y ait quelqu'un euh qui vous demande de l'argent et ça devient limite euh

‘The subway it’s it’s not very ni- it’s not very funny uh because you can’t take the subway without there being uh without there being someone uh who asks you for money and that becomes unpleasant uh’

(Lit. ‘Without that there is someone who asks you for money’) (CFPP)

(486) Il y a aurait ‘there would be’ ah la la, s il n y avait pas la religion qui est une bonne daube inventé par les hommes , il y aurait vraiment moins d abrutis sur terre !!! (…) encore heureux que le pov type qu on appel jesus christ a été crucifié sur une croix car s il avait été noyé, y aurait pas mal de conn.ards qui dormiraient avec un aquarium au dessus du lit !!!

‘oh la la, if there were no religion, which is bullshit invented by men, there would really be fewer suckers on Earth!!! (…) At least the poor guy they call Jesus was crucified on a cross because if he has been drowned, there would be quite a few idiots who slept with an aquarium above their beds!!!’ (YCCQA)

(487) Il y a aurait ‘there would be’ C’est simplement parce qu'on est un quartier moins cher c'est tout et puis la disposition j'pense que sinon y aurait déjà quelqu'un qui aurait racheté tout l'ensemble par exemple notre voisine a bien l'idée qu'elle va se racheter à un moment l'ensemble des chambres de bonnes et puis faire un duplex

‘It’s simply because this is a less expensive neighborhood that’s all, and then the disposition, I think that otherwise someone would already have bought everything, for example our neighbor had the idea that she’s going to buy all the maid’s rooms at some point and make a duplex.’

(Lit. ‘There would already be someone who would have bought everything’)

(CFPP)

Appendix 369

(488) Il doit y avoir ‘there must be’ – A: Jeu IL 2 STURMOVIK 1946? Quand je veux utiliser un avion à réaction, (…), au bout d'un temps extrêmement court les réacteurs prennent feu, et je ne peux donc pas aller plus loin. Comment faire pour remédier à ce phénomène... (…)

‘Video game IL 2 STURMOVIK 1946? When I try to use a jet aircraft (…), after an extremely short time the reactors are on fire, and so I can’t go any further. What should I do to overcome this problem… (…)’

– B: (…) Recherchez la façon sur votre jeu de réduire la puissance et réduisez celle ci au minimum pour la croisière. Il doit y avoir une temporisation préprogrammée qui met le feu aux moteurs au dessus d'un certain régime utilisé pendant un certain temps.

‘Look for a way to reduce the power of your game and reduce it to the minimum for the crossing. There must be a pre-programmed delay setting fire to the engines when a certain regime is passed for a certain time. (…)’ (YCCQA)

(489) Eventive il y a cleft with negation Papa, soupire le gamin, quand j'avais 6 ans tu m'as expliqué que le père-noël n'existait pas. A 7 ans tu m'as dit que ce n'était pas les cloches qui apportaient les oeufs de Pâques. Quand j'ai eu 8 ans tu m'as appris qu'il n'y avait pas de petite souris qui donnait une pièce quand on perdait une dent...

‘Dad, sighs the kid, when I was 6 you explained to me that Santa Claus didn’t exist. When I was 7, you told me that it wasn’t the bells that brought the Easter eggs. When I was 8 you taught me that there wasn’t a small mouse that left a coin when you lost a tooth…’ (YCCQA)

(490) Eventive il y a cleft with negation Commee ton débit descendant est correct tu vois bien la personne, mais comme ton débit montant est beaucoup plus faible la personne avec qui tu communiques reçoit une image saccadée Essaye d'abord de voir si il n'ya pas de spywares ou d'autres logiciels qui utilisent de la bande passante montante sur ton pc, et si ça

370 Appendix

ne résout toujours pas le problème, faudra que tu augmente ton débit (montant et descendant)

‘Since your download rate is correct, you see the person well, but since your upload rate is much weaker, the person you communicate with receives a jerky picture. Try first to see if there are no spywares or other software that are using the upload bandwidth on your pc, and if that doesn’t solve the problem, you’ll have to increase your upload and download speed.’ (YCCQA)

(491) Interrogative inversion (y a-t-il ‘is there’) – A: Salut , j'ai un pc portable hp/compaq nc6000 depuis 2.5 ans (finie donc la garantie) , aujourd'hui je me lève , hop , je démarre le pc , je lis des news par-ci , par-là , (wikinews, slashdot ...) , je vais déjeuner , je reviens , et que vois-je? un bel écran noir (…)

‘Hi, I’ve had a HP/Compaq nc6000 pc for 2,5 years (so no more warranty), today I get up, hop, I start the laptop, I read the news here and there (wikinews, slashdot), I have some breakfast, I come back, and what do I see? A nice black screen. (…)

– B: Ya-t-il le logo windows qui charge au démarrage puis s'éteind après le chargement? (en gros : tu allume le pc, t'entend des bip (ou pas), ya le logo hp qui apparait, ensuite le logo windows qui apparait puis l'ordi s'éteind ?)

‘Does the Windows logo load when you start your computer and then disappear after it’s charged? (basically: you turn on your laptop, you hear some beeps (or not), there’s the HP logo that appears, then the Windows logo that appears then the laptop shuts down?)

(Lit. ‘Is there the Windows logo that loads when you start your laptop?’)

(www)

Appendix 371

Section 4 All il y a… qui sentences318

4.1 Subject vs. object relative clauses

Le Monde YCCQA CFPP Total n % n % n % n % Subject relative 342 81% 995 91% 490 90% 1827 89% Object relative 80 19% 98 9% 54 10% 232 11% (see (492) - (493)) Total 422 100% 1093 100% 544 100% 2059 100% Table 36: All il y a… qui sentences: subject vs. object relative clauses

(492) moi y a certaines familles que j'connais depuis longtemps. ‘Me, there are some families that I’ve known for a long time.’ (CFPP)

(493) c'est à dire qu'y a des choses moi que j'ai jamais fait. ‘That is to say that there are things that I never did.’ (CFPP)

4.2 Constituent introduced by il y a: syntactic category

Le Monde YCCQA CFPP Total n % n % n % n % Noun phrase 361 86% 962 88% 478 88% 1801 87% Pronoun 51 12% 80 7% 40 7% 142 7% Proper noun 10 2% 51 5% 26 5% 116 6% Total 422 100% 1093 100% 544 100% 2059 100%

Table 37: All il y a… qui sentences: syntactic category of the constituent introduced by il y a

318 The term ‘il y a… qui sentence’ refers to all il y a… qui/que/ qu’ sentences extracted from the corpora, regardless of the cleft/cleft lookalike distinction.

372 Appendix

Le Monde YCCQA CFPP Total n % n % n % n % Ceux/celles ‘those’, celui/celle ‘the one’ 35 69% 21 26% 8 20% 64 37% (see (494)) Other pronouns 16 31% 59 74% 32 80% 107 63%

Total 51 100% 80 100% 40 100% 171 100%

Table 38: All il y a… qui sentences: types of pronouns

(494) Désormais, il y a ceux qui « s’en sortent » et les autres, la majorité, qui n’arrivent plus à « joindre les deux bouts ». ‘From now on, there are those who “manage” and the others, the majority, who are unable to “make ends meet”.’ (Le Monde)

4.3 Constituent introduced by il y a: (in)definiteness

Le Monde YCCQA CFPP Total n % n % n % n % Indefinite 330 78% 946 87% 448 82% 1724 84% Definite 92 22% 147 13% 96 18% 335 16% Total 422 100% 1093 100% 544 100% 2059 100% Table 39: All il y a… qui sentences: (in)definiteness of the constituent introduced by il y a

4.4 Constituent introduced by il y a: semantic category

Le Monde YCCQA CFPP Total

n % n % n % n % Human 168 40% 507 46% 254 47% 929 45% Non-human 254 60% 586 54% 290 53% 1130 55% Total 422 100% 1093 100% 544 100% 2059 100% Table 40: All il y a… qui sentences: human or non-human constituent introduced by il y a

Appendix 373

Le Monde YCCQA CFPP Total n % n % n % n % Chose, quelque 28 7% 87 8% 66 12% 181 9% chose, truc Gens, personnes, 45 11% 151 14% 101 19% 297 14% quelqu'un Other 349 83% 855 78% 377 69% 1581 77% Total 422 100% 1093 100% 544 100% 2059 100%

Table 41: All il y a… qui sentences: semantically vague constituents introduced by il y a

Most constituents introduced by il y a in CFPP refer to people in the speaker’s vicinity (family, friends, neighbors, acquaintances), and many of the non-human constituents refer to places (l’Père-Lachaise ‘the Père-Lachaise cemetery’, des quartiers ‘neighborhoods’, une rue ‘a street’) and buildings (le Pure Café ‘the Pure café’, un bar ‘a bar’, l’autre école ‘the other school’). The predominance of such expressions is due to the interviewer’s recurrent questions in the different interviews (e.g. ‘Where do you go shopping, where do you like to hang out in your neighborhood, what’s changed in the past couple of years in your neighborhood?’ etc.). In contrast, the other two corpora show a wider range of different referents. Unsurprisingly, many of the constituents introduced by il y a in Le Monde refer to political and economical concepts, as illustrated in (495).

(495) Le Monde: examples of introduced constituents Des sanctions ‘sanctions’, des nations ‘nations’, une dictature ‘a dictatorship’, des affaires ‘business transactions’, un argument ‘an argument’, des problèmes ‘problems’, des lois irréversibles ‘irreversible laws’, des vérités vraies ‘true truths’

The types of referents appearing in YCCQA belong to even more diverse semantic classes than the ones in Le Monde and CFPP. This is a consequence of the diversity of the online forum, which discusses topics ranging from cars, restaurants, sex, politics and software to homework assignments, methods of losing weight, friendship and employment. Examples of human referents are presented in (496). Unfortunately, the

374 Appendix

forum is full of racist and offensive language, and heated discussions about the Israel- Palestine issue reappear throughout the different forum subtopics. Some of the terms used in such discussion are presented in (496) under ‘Political discussions’:

(496) YCCQA: examples of human referents General: des pères ‘fathers’, des filles ‘girls’, des mecs ‘guys’

About love and sex: un Bogosse ‘a handsome guy’, des jaloux ‘jealous people’, des femmes rondes ‘round women’

Political discussions: des tarés ‘crazy people’, des crétins ‘idiots’, un ras-le-bol du bordel ‘dirty trash’ (talking about immigrants), des étrangers sympas ‘nice foreigners’, plein de juifs ‘a lot of Jews’, des ARABES ‘ARABS’, des innocents ‘innocent people’, des salopards terroristes ‘terrorist bastards’, des gros connards ignorants ‘big ignorant bastards’, les limaces rampantes de gauche ‘creeping left-wing slugs’, des grandes gueules socialistes ‘socialist loud- mouths’

Other: 1% des lecteurs du Da Vinci Code ‘1% of the readers of The Da Vinci Code’, des mathématiciens de très haut niveau ‘mathematicians of a very high level’, de véritables entrepreneurs ‘real entrepreneurs’

The constituents in (497) illustrate the wide range of different non-human referents introduced by il y a in the YCCQA corpus.

(497) YCCQA: examples of non-human referents une tentative ‘an attempt’, des logiciels ‘software programs’, une légère surtension ‘a slight overvoltage’, 5 format MPEG ‘5 MPEG formats’, des consortiums ‘consortia’, les livres d’Amélie Nothomb ‘Amélie Nothomb’s books’, certains meubles ‘some pieces of furniture’, une boutique Levis ‘a Levi’s store’, un clip de Dead Can Dance ‘a Dead Can Dance music video’, un liquide ‘a liquid’, une maladie ‘a disease’, des mensonges ‘lies’, une seule idéologie ‘a single ideology’, beaucoup de questions ‘many questions’, des matières radioactives ‘radioactive materials’, les sex toys ‘sex toys’, l’épilation à la cire ‘waxing’, des inégalités criantes ‘glaring inequalities’, les vrais téléphones ‘real phones’

Appendix 375

Section 5 Decleftable il y a… qui sentences319

Le Monde YCCQA CFPP Total

n % n % n % n % Declefting is possible in combination with 81 19% 285 26% 210 39% 576 28% eventive/specificational semantics (il y a clefts) Declefting is possible in combination with other 60 14% 247 23% 80 15% 387 19% properties Declefting is possible, 91 22% 253 23% 102 19% 446 22% but requires a lexical adjustment Not decleftable 190 45% 308 28% 152 28% 650 32%

Total 422 100% 1093 100% 544 100% 2059 100%

Table 42: All il y a… qui sentences: classes of decleftability per corpus

Le Monde YCCQA CFPP Total

Corpus size in words 25.700.000 6.100.000 550.000 32.350.000 Total number of decleftable 232 785 290 1307 il y a... qui sentences Number of decleftable il y a... qui sentences 5 64 264 20 per 500.000 words Table 43: Normalized frequency of decleftable il y a... qui sentences throughout the corpora

Within citation Not within citation Total (see (498)) (see (499))

n % n % n % Decleftable 178 77% 54 23% 232 100% Not decleftable 82 43% 108 57% 190 100% Table 44: (Non-)decleftable il y a... qui sentences appearing within a citation or not (Le Monde)

319 I define ‘decleftable il y a… qui sentences’ in Part III, Chapter 4 of the dissertation.

376 Appendix

(498) Decleftable il y a… qui sentence appearing within a citation « Des arguments plaident en ce sens mais ailleurs qu'ici il y a quelque chose qui cloche », écrivait le rédacteur en chef de ce quotidien de gauche.

‘“Some arguments support that direction, but in other places there is something that’s not right”, wrote the editor in chief of that left-wing daily.’

(Le Monde)

(499) Decleftable il y a… qui sentence not appearing within a citation MARSEILLE de notre envoyé spécial. Le Sentier, sous les toits de Paris. Forcément, il y a l'amour qui luit.

‘MARSEILLE from our special reporter. Le Sentier, under the rooftops of Paris. There must be love that’s shining.’

(Lit. ‘Surely, there is love that is shining.’) (Le Monde)

Appendix 377

Section 6 All il n’y a que… qui sentences

6.1 Properties of the relative clause/infinitival clause

Le Monde YCCQA Total n % n % n % Qui 26 60% 153 83% 179 79% Que 15 35% 26 14% 41 18% Où 'where'(see (500) 1 2% 4 2% 5 2% Pour que 'so that' (see (501)) 0 0% 2 1% 2 1% À laquelle 'to which' (see (502)) 1 2% 0 0% 1 0% Total 43 100% 185 100% 228 100% Table 45: All il n’y a que sentences: type of relativizer

(500) Où ‘where’ Je ne regarde jamais 2 fois la même chose il n'y a que DBZ ou j'ai fais 2 fois le tour des épisodes entièrement.

‘I never watch the same show twice, I only watched all the episodes of DBZ twice.’ (YCCQA)

(501) Pour que ‘so that’ Il n’y a que si quelqu'un(e) de mes proches qui est en danger pour que je devienne Infect

‘I only become vile when someone who is close to me is in danger.’ (YCCQA)

(502) À laquelle ‘to which’ À entendre les Américains, nous sommes les affreux du monde atlantique, (…) toujours incapables d'admettre qu'il n’y ait qu’une seule superpuissance à laquelle nous n'avons pas les moyens de faire contrepoids.

‘When you hear the Americans, it seems as though we are the awful people of the Atlantic world, (…) still incapable of admitting that there is only one super power that we do not have the means to counterbalance.’ (Le Monde)

378 Appendix

Le Monde YCCQA Total n % n % n % Pour 'for' + infinitive 3 16% 21 70% 24 49% (see (503) - (94)) À 'to' + infinitive 7 37% 5 17% 12 24% (see (505)) À 'to' + XP + infinitive 9 47% 4 13% 13 27% (see (506)) Total 19 100% 30 100% 49 100%

Table 46: All il n’y a que sentences: infinitival clauses

(503) Pour ‘for’ infinitive Robert Hué a rappelé aux militants communistes qu'il n’y a que 10 % des électeurs pour soutenir leurs propositions.

‘Robert Hué reminded the communist militants that only 10% of the electorate support their propositions.’ (Le Monde)

(504) Pour ‘for’ + pronoun + infinitive Il y a que toi pour savoir si tu doit l'acheter ou pas.

‘Only you can know if you should buy it or not.’ (YCCQA)

(505) À ‘to’ + infinitive L’etre humain ne peu s'empêcher de tuer quand l'autre n'est pas du même avis, ou de même religion, etc... Il n’y a qu'a regarder l'histoire, depuis la nuit des temps, les gens se font la guerre.

‘Human beings can’t stop killing when the other doesn’t have the same opinion or adheres to a different religion, etc… Just look at history, since the dawn of time, people wage war on each other.’ (YCCQA)

(506) À ‘to’ + pronoun + infinitive Rêver, y'a que sa à faire si tu veut voir une autre vision du monde.

‘Dreaming, that’s the only thing to do if you want to see a different vision of the world.’ (YCCQA)

Appendix 379

Le Monde YCCQA Total n % n % n % Subject relative (507) 26 63% 152 85% 179 81% Object relative (508) 8 20% 4 2% 12 5% Adjunct relative (509) 7 17% 23 13% 29 13% Total 41 100% 179 100% 220 100% Table 47: All il n’y a que sentences: subject vs. object relative clauses

(507) Subject relative clause Il n’ya que votre dentiste qui puisse répondre à cette question!!

‘Only your dentist can answer that question!!’ (YCCQA)

(508) Object relative clause [Après l’accouchement] j'ai repris mon poids normal,mais il faut que muscle tout ca! et apaprament il ne faut courir ,ni pratiquer de sport extreme. il ya que ca que j'aime malheuresement,mais ya t'il donc d'autres moyens?

‘[After giving birth] I went back to my normal weight, but I want to become more muscular! And apparently you shouldn’t run or do extreme sports. That’s the only thing I like, unfortunately, but are there other ways?’ (YCCQA)

(509) Adjunct relative clause Pourquoi ya que quand je dis des conneries et des trucs qu'ils ne veulent rien dire qu'on me repond?

‘Why do people only answer when I say stupid things and things that don’t mean anything?’ (YCCQA)

380 Appendix

6.2 Properties of the constituent introduced by il n’y a que

6.2.1 Grammatical category

Le Monde YCCQA Total n % n % n % Noun phrase 32 78% 114 64% 146 66% Pronoun 3 7% 32 18% 35 16% Prepositional phrase 6 15% 17 9% 23 10% (see (162) & (513)) Proper noun 0 0% 12 7% 12 5% Temporal/conditional clause 0 0% 3 2% 3 1% (see (163)) Adverb (see (512)) 0 0% 1 1% 1 0% Total 41 100% 179 100% 220 100% Table 48: All il n’y a que sentences: grammatical category of introduced constituent

(510) Prepositional phrase Les 2 partenaires doivent faire des concessions, il n'y a que comme cela que le couple peut durer!

‘Both partners have to make concessions, only like that can the couple last!’ (YCCQA)

(511) Conditional clause y'a que si tu es handicapé que tu ne peux pas franchir les portes

‘Only if you are handicapped are you unable to go through doors.’ (YCCQA)

(512) The only adverb in the corpus data (là-bas ‘over there’) y a que làba que tu auras une réponse fiable!

‘Only there will you find a reliable answer!’ (YCCQA)

(513) All prepositional phrases in both corpora Le Monde: en France ‘in France’, aux États-Unis ‘in the United States’, dans le cas d’une récession mondiale en 1999 ‘in the case of a global recession in 1999’,

Appendix 381

dans les facultés de lettres ‘in Faculties of Arts’, en Turquie ‘in Turkey’, au golf, à la pétanque et au handball ‘during golf, Petanque or handball matches’

YCCQA: comme cela/ça ‘like that’ (n=7), de cette manière ‘in that way’, dans la haute bourgeoisie ‘in high society’, pour elle ‘for her’, dans le Chien et le Loup ‘in the Dog and the Wolf’, dans les grandes mosquées ‘in big mosks’, pour les sakolatres invétérés habitués a suivre aveuglément leur chef stalinien ‘for veteran Sarkozy fanatics who are used to following their Stalin-like chef blindly’, en France ‘in France’, chez moi ‘at home’, dans ma chambre ‘in my room’, pour la religion ‘for religion’

Le Monde YCCQA Total n % n % n % Nominal (NP, pronoun, proper noun) 35 85% 158 88% 193 88% Non-nominal 6 15% 21 12% 27 12% Total 41 100% 179 100% 220 100% Table 49: All il n’y a que sentences: nominal vs. non-nominal constituents

6.2.2 (In)definiteness

Le Monde YCCQA Total n % n % n % Indefinite (see (514)) 18 51% 5 3% 23 12% Definite (see (165)) 17 49% 147 97% 164 88% Total 35 100% 152 100% 187 100% Table 50: All il n’y a que sentences: (in)definiteness of (pro)nominal constituents

(514) Il n’y a que + des N + restrictive relative clause (≠ cleft) Que vont penser ceux qui ont voté pour le FN ? Et ceux qui ont voté pour la droite? S'imagine-t-on qu'ils ne seront pas persuadés que, du côté de ceux qui parlent de « démocratie », de « pacte républicain », de « nos valeurs », il n'y a que des comédiens qui veulent préserver leur fromage ?

382 Appendix

‘What are those who voted for the FN [Front National] going to think? And those who voted for the right? Will people imagine that they are not persuaded that, on the sides of those who talk about ‘democracy’, ‘a republican pact’, ‘our values’, there are only comedians who want to hold on to what they’ve got?’ (Le Monde)

(515) Specificational il n’y a que sentence + definite NP (cleft) – A: Comment faire pour que mon copain s'interresse moins au foot?

‘How can I make my boyfriend become less interested in soccer?’

– B: Mon chéri doit etre pareil que ton copain mais je te dis juste attends il n'y a que le temsp qui pourra changer les choses... Il faut etre patiente....

‘My honey must be like your boyfriend but let me just tell you to wait, only time can change things… You have to be patient.’ (YCCQA)

Le Monde YCCQA Total n % n % n % Numeral (see (516)) 15 83% 2 40% 17 74% Des (see (157)) 3 17% 2 40% 5 22% Du 0 0% 1 20% 1 4% Total 18 100% 5 100% 23 100% Table 51: All il n’y a que sentences: determiners of indefinite NPs

(516) Il n’y a que + numeral (cleft) Tout le monde sait désormais à quelle région il appartient alors qu'il y a encore dix ans il n'y avait qu'un Français sur quatre qui en avait conscience.

‘Nowadays, everybody knows to which region he or she belongs, whereas only ten years ago, only one Frenchman in four was aware of this.’ (Le Monde)

Appendix 383

Section 7 Eventive & specificational il y a clefts

7.1 Clefted constituent: syntactic category (eventive il y a clefts)

Le Monde YCCQA CFPP Total n % n % n % n % Noun phrase 71 90% 193 85% 155 88% 419 87% Proper noun 2 3% 16 7% 4 2% 22 5% Pronoun 6 8% 17 8% 18 10% 41 9% Total 79 100% 226 100% 177 100% 482 100% Table 52: Eventive il y a clefts: grammatical category of clefted constituent

7.2 Clefted element: (in)definiteness (eventive il y a clefts)

Le Monde YCCQA CFPP Total

n % n % n % n % Indefinite 74 94% 195 86% 158 89% 427 89% Definite 5 6% 31 14% 19 11% 55 11% Total 79 100% 226 100% 177 100% 482 100% Table 53: Eventive il y a clefts: (in)definiteness of introduced element

7.3 Subject vs. object relative clauses

Eventive Specificational Total il y a clefts il y a clefts n % n % n % Subject relative 477 99% 90 96% 567 98% Object relative 5 1% 4 4% 9 2% Total 482 100% 94 100% 576 100% Table 54: Specificational and eventive il y a clefts: subject vs. object relative clauses

384 Appendix

(517) Eventive il y a cleft + object relative clause – A: Peut on annuler une paternité? Quelle est la procédure juridique et combien de temps ça prend?

‘Can one undo a paternity? What is the judicial procedure and how much time does it take?’

– B: (…) Pour toutes ces questions là de toute façon il faudra voir avec le Tribunal d'instance de ta ville. Il te faudra un dossier béton, ça sera long et onéreux. Et ça sera bien fait. Au milieu de ces histoires sordides, il y a un enfant que tu vas rejeter.

(…) ‘In any case, for all of those questions you have to go to your city’s district court. You’ll need a really solid case, it will be long and expensive. And you’ll have earned it. In the middle of this sordid business, there’s a child that you’re rejecting.’ (YCCQA)

7.4 Specificational il y a clefts: additional examples

– Examples of specificational il y a clefts with proper nouns –

(518) – A: Est-ce qu'il y a des GPS a reconnaissance vocal disponible au Québec? ‘Are there GPS systems with speech recognition in Quebec?’ – B: Oui il y a le Magellan Maestro 3250 qui a la reconnaissance vocale tel qu'expliquer sur le site de CAA Québec: [link]

‘Yes there’s the Magellan Maestro 3250 that has speech recognition, as explained on the CAA Québec website: [link]’ (YCCQA)

(519) – A: Pouvez-vous me conseiller un shampoing, un savon et gel douche bio? ‘Which organic shampoo, soap and shower gel would you recommend?’

– B: il ya douce nature qui est pas mal au niveau rapport qualité prix

‘There’s Douce Nature that’s not bad in terms of price – quality.’

(YCCQA)

Appendix 385

(520) – A: Je cherche une bonne crème de jour hydratante pour le visage. ‘I’m looking for a good hydrating facial cream.’

– B: J'aime bien la supplétive visage de Dermagor. Aussi, pour les bébés, il y a la crème hydratante Uriage qui est très bien, pas seulement pour les bébés, justement ;-)

‘I like Dermagor’s face suppletive. Also, for babies, there’s the hydrating cream Uriage that is very good, not only for babies ;-)’

(YCCQA)

(521) – A: Cherche logiciel gratuit de création de musique. ‘Looking for software to create music.’

– B: il ya Audacity qui est totalement gratuit.

‘There’s Audacity that is completely for free.’ (YCCQA)

(522) – A: C'est quoi le mieux, firefox ou internet explorer? ‘Which one is better, Firefox or Internet Explorer?’

– B: Internet explorer est plus répandu certe mais moins sécurisé, il est lourd, Firefox est plus léger et plus rapide. Mais il y a aussi Google Chrome qui est léger et encore plus rapide que Firefox

‘Internet Explorer is more widespread, yes, but less secure, it’s heavy, Firefox is lighter and faster. But there’s also Google Chrome that is light and even faster than Firefox’ (YCCQA)

(523) – A: Quelles est votre meilleure série du moment? ‘What’s your favorite TV show at the moment?’

– B: "How I Met Your Mother" c'est génial, y'a aussi "Lost" qui est bien.

‘”How I Met Your Mother” is great, there’s also “Lost” that is good.’

(YCCQA)

386 Appendix

– Examples of specificational il y a clefts with definite NPs –

(524) – A: Je vais bientôt passer un examen, et j'ai le trac. pouvais vous me donner des conseils?

‘I soon have an exam and I’m stressed out. Can you give me some advice?’

– B: détends-toi, fais aussi du sport, ça permet de bien éliminer le stress. Il y a aussi l'homéopathie qui fonctionne vraiment super bien!

‘Relax, do some sports, that enables you to get rid of the stress. There’s also homeopathy that works really well! (YCCQA)

(525) – A: Aimez-vous les chats ? ? Si oui, pourquoi ? ‘Do you like cats?? If yes, why?’

– B: Certainement que j'aime les chats. J'en ai 2 et je suis très attachée à mes deux Minets. Ils sont inséparables et très gentils. Ils sont délicats et gracieux en tout ce qu'ils font. Les chats sont toujours beaux. Et il y a aussi leur apparente indépendance qui me séduit. Bref, je me sens très proche de mes chats.

‘Of course I love cats. I’ve got two of them and I’m very fond of them. They’re inseparable and very kind. They’re delicate and graceful in everything they do. Cats are always beautiful. And there’s also their apparent independence that seduces me.’ (YCCQA)

(526) – A: Comment on fait pour trouver l'amour? je suis tout seul depuis longtmps je comprends pas, je m'intéresse aux gens mais ça donne rien

‘How do people find love? I’ve been all alone for a long time, I don’t understand, I’m interested in people but it doesn’t work’

– B: ça tombe parfois par surprise mais maintenant il y a aussi la nouvelle technologie qui aide, Internet peut t'aider à trouver l'amour.

‘Sometimes it happens without warning but now there’s also new technology that helps, the Internet can help you find love.’ (YCCQA)

Appendix 387

(527) – A: Pourqoui kurt cobain s'est t'il suicidé? ‘Why did Kurt Cobain kill himself?’

– B: (…) Il y a la drogue qui rentre en compte.

(…) ‘There’s drugs that comes into play.’ (YCCQA)

(528) – A: J'ai des cheveux qui régressent trop vite, je suis obligée de les laver tous les jours avez vous des conseils?

‘My hair becomes greasy very quickly, I have to wash it every day, do you have any advice?’

– B: (…) Sinon il y a les shampoings secs qui peuvent dépanner.

(…) ‘Otherwise there are dry shampoos that can help out.’ (YCCQA)

Section 8 Il n’y a que clefts

Le Monde YCCQA Total n % n % n % Numeral (see (516)) 15 94% 2 40% 17 81% Des 1 6% 2 40% 3 14% Du 0 0% 1 20% 1 5% Total 16 100% 5 100% 21 100%

Table 55: Il n’y a que clefts: determiners of indefinite NPs

n % n % n %

Subject relative 24 62% 147 85% 171 81% Adjunct relative 7 18% 23 13% 30 14% Object relative 8 21% 3 2% 11 5% Total 39 100% 173 100% 212 100%

Table 56: Il n’y a que clefts: subject vs. object vs. adjunct relative clauses

References

Abbott, Barbara. 1993. A pragmatic account of the definiteness effect in existencial sentences. Journal of Pragmatics 19(1). 39-55. Achard, Michel. 2015. Impersonals and other agent defocusing constructions in French. Amsterdam: John Benjamins. Akmajian, Adrian. 1970. On deriving cleft sentences from pseudo-cleft sentences. Linguistic Inquiry 1. 149-168. Alexiadou, Artemis, Paul Law, André Meinunger & Chris Wilder. 2000. The syntax of relative clauses. Amsterdam/Philadelphia: John Benjamins. AnderBois, Scott, Adrian Brasoveanu & Robert Henderson. 2010. Crossing the appositive/at-issue meaning boundary. In Proceedings of SALT 20, 328-346. AnderBois, Scott, Adrian Brasoveanu & Robert Henderson. 2013. At-issue proposals and appositive impositions in discourse. Journal of Semantics 32(1). 93-138. Anscombre, Jean-Claude. 1996. Partitif et localisation temporelle. Langue française 109(1). 80-103. Antonelli, Giuseppe. 2011. Lingua. In Andrea Afribo & Emanuele Zinato (eds.), Modernità italiana. Cultura, lingua e letteratura dagli anni settanta a oggi. Rome: Carocci. Aparici, Melina, Elisa Rosado & Joan Perera. 2016. Later development of relative clauses across discourse genres and modalities of production. In Joan Perera, Melina Aparici, Elisa Rosado & Naymé Salas (eds.), Written and spoken language development across the lifespan: Essays in honour of Liliana Tolchinsky, 201-225. Cham: Springer International Publishing. Apothéloz, Denis & Małgorzata Nowakowska. 2010. La résultativité et la valeur de parfait en français et en polonais. Cahiers Chronos 21. 1-23. Ariel, Mira. 1990. Accessing noun-phrase antecedents. London/New York: Routledge. Ashby, William J. 1995. French presentational structures. In Jon Amastae, Grant Goodall, Mario Montalbetti & Marianne Phinney (eds.), Contemporary research in Romance linguistics, 91- 104. Amsterdam/Philadelphia: John Benjamins. Ashby, William J. 1999. Au sujet de quoi? La fonction du sujet grammatical, du complément d'objet direct, et de la construction présentative en français parlé. The French Review 72(3). 481- 492. Atlas, Jay David & Stephen C. Levinson. 1981. It-clefts, informativeness, and logical form: Radical pragmatics (revised standard version). In Peter Cole (ed.), Radical Pragmatics, 1-62. New York: Academic Press. Avanzi, Mathieu. 2008. La différence entre micro- et macro-syntaxe est-elle marquée prosodiquement? L'exemple des dispositifs clivés en "il y a SN qui/Ø V". L'information grammaticale, Heck . 8-13. Ayres-Bennett, Wendy & Janice Carruthers. 2001. Problems and perspectives: Studies in the modern French language. London: Longman. Balibar-Mrabti, Antoinette & Céline Vaguer (Eds.). 2005. Le semi-figement: Linx. Revue des linguistes de l'Université Paris Ouest Nanterre La Défense. Barnicaud, G., M. A. Compare, Oswald Ducrot & A. Duval. 1967. Le problème de la négation dans diverses grammaires françaises. Langages 2(7). 58-73. Beaver, David & Brady Z. Clark. 2008. Sense and Sensitivity: How focus determines meaning. Oxford: Wiley-Blackwell. Beaver, David, Itamar Francez & Dmitry Levinson. 2006. Bad subject: (Non-)canonicality and NP distribution in existentials. In Effi Georgala & Jonathan Howell (eds.), Proceedings of SALT 15, 19-43. Ithaca, NY: Cornell University Press. Béguelin, Marie-José. 1998. Des clauses impersonnelles aux constituants phrastiques: Quelques axes de grammaticalisation. In Patrick Seriot & Alain Berrendonner (eds.), Le paradoxe du sujet. Les propositions impersonnelles dans les langues slaves et romanes, 25-41. Lausanne: Cahiers de l'ILSL. Behagel, Otto. 1909. Beziehungen zwischen Umfang und Reihenfolge von Satzgliedern. Indogermanische Forschungen 25. 110-142.

390 References

Belletti, Adriana. 2008. The CP of clefts. Rivista di Grammatica Generativa 33. 191-204. Belletti, Adriana. 2012. Revisiting the CP of clefts. In Ede Zimmermann & Günther Grewendorf (eds.), Discourse and grammar. From sentence types to lexical categories, 91-114. Berlin: De Gruyter Mouton. Belletti, Adriana. 2013. The focus map of clefts: Extraposition and Predication. In Ur Shlonsky (ed.), Where do we go from here? Chapters in syntactic cartography, 42-59. Oxford: Oxford University Press. Bentley, Delia. 2013. Subject canonicality and definiteness effects in Romance there-sentences. Language 89(4). 675-712. Bentley, Delia. 2015. Predication and argument realization. In Delia Bentley, Francesco Maria Ciconte & Silvio Cruschina (eds.), Existentials and locatives in Romance dialects of Italy, 99- 160. Oxford: Oxford University Press. Bentley, Delia, Francesco Maria Ciconte & Silvio Cruschina. 2013. Micro-variation in subject agreement: The case of existential pivots with split focus in Romance. Rivista di Linguistica 25(1). 15-43. Bentley, Delia, Francesco Maria Ciconte & Silvio Cruschina. 2015. Existentials and locatives in Romance dialects of Italy. Oxford: Oxford University Press. Berrendonner, Alain. 2003. Eléments pour une macro-syntaxe: Actions communicatives, types de clauses, structures périodiques. In Antonietta Scarano (ed.), Macro-syntaxe et pragmatique. L'analyse de l'oral, 93-110. Rome: Bulzoni. Berretta, Monica. 1995. Come inseriamo elementi nuovi nel discorso /1: 'C'è il gatto che ha fame'. Italiano e Oltre 53. 79-105. Berruto, Gaetano. 1986. Un tratto sintattico dell'italiano parlato: Il c'è presentativo. In Klaus Lichem, Edith Mara & Susanne Knaller (eds.), Parallela 2. Aspetti della sintassi dell'italiano contemporaneo, 61-73. Tübingen: Narr. Besserman, Ana. 2014. There was... something new! Do information status constraints guide hearers' expectations during online language comprehension? San Diego, CA: San Diego State University MA thesis. Besserman, Ana, Tracy Love & Lew Shapiro. 2015. Anticipatory processes in language comprehension: The English existential as an indicator of newness. Paper presented at Experimental Pragmatics [Xprag], Chicago. https://lucian.uchicago.edu/blogs/xprag2015/files/2015/06/bessermanetal.pdf. Bhatt, Rajesh & Roumyana Pancheva. 2006. The Syntax and Semantics of Aspect - Handout 1. Paper presented at LSA 130. http://web.mit.edu/rbhatt/www/lsa130/l1.pdf. Bianchi, Valentina, Giuliano Bocci & Silvio Cruschina. 2015. Focus fronting and its implicatures. In Enoch Aboh, Jeannette Schaeffer & Petra Sleeman (eds.), Romance Languages and Linguistic Theory 2013: Selected Papers from Going Romance, Amsterdam 2013, 1-20. Amsterdam: John Benjamins. Bicler, Chris & Kristin Davidse. 2008. It-clefts in casual conversational English: The weakening of their specificational meaning. In Bert Cappelle & Naoaki Wada (eds.), Distinctions in English grammar. Offered to Renaat Declerck, 260-277. Tokyo: Kaitakusha. Birner, Betty J., Jeffrey P. Kaplan & Gregory Ward. 2007. Functional compositionality and the interaction of discourse constraints. Language 83(2). 317-343. Blanche-Benveniste, Claire. 1983. Examen de la notion de subordination. In Recherches sur le Français Parlé, Vol. 4, 71-115. Aix-en-Provence: Publications Université de Provence. Blanche-Benveniste, Claire. 1990. La français parlé: Etudes grammaticales. Paris: CNRS. Blanche-Benveniste, Claire. 1997. Approches de la langue parlée en français. Paris: Ophrys. Blanche-Benveniste, Claire. 2006. Les clivées françaises de type: C'est comme ça que, C'est pour ça que, C'est là que tout a commencé. Moderna Språk 100(2). 273-287. Blasco-Dulbecco, Mylène. 1999. Les dislocations en français contemporain: Étude syntaxique. Paris: Champion. Boas, Franz. 1911. Introduction. In Franz Boas (ed.), Handbook of American Indian languages, 1-83. Washington, DC: Bureau of American Ethnology. Bock, J. Kathryn & Richard K. Warren. 1985. Conceptual accessibility and syntactic structure in sentence formulation. Cognition 21(1). 47-67.

References 391

Bouchard, Jacynthe, Fernande Dupuis & Monique Dufresne. 2008. Un processus de focalisation en ancien français: Le développement des clivées. Paper presented at the Congrès annuel de l'Association canadienne de linguistique 2007. http://homes.chass.utoronto.ca/~cla- acl/actes2007/Bouchard_Dupuis_Dufresne.pdf. Branca-Rosoff, Sonia, M. Serge Fleury, Florence Lefeuvre & Mat Pires. 2012. Discours sur la ville. Présentation du Corpus de Français Parlé Parisien des années 2000 (CFPP2000). http://cfpp2000.univ-paris3.fr. Branigan, Holly P., Martin J. Pickering & Mikihiro Tanaka. 2008. Contributions of animacy to grammatical function assignment and word order during production. Lingua 118(2). 172- 189. Breivik, Leiv Egil. 1981. On the interpretation of existential there. Language 57(1). 1-25. Breivik, Leiv Egil & Ana E. Martínez-Insua. 2008. Grammaticalization, subjectification and non- concord in English existential sentences. English Studies 89(3). 351-362. Brunot, Ferdinand. 1922. La pensée et la langue: Méthode, principes et plan d'une théorie nouvelle du langage appliquée au français. Paris: Masson. Büring, Daniel. 1998. Identity, modality, and the candidate behind the wall. In Devon Strolovitch & Aaron Lawson (eds.), Proceedings of SALT 8, 36-54. Ithaca, NY: Cornell University. Büring, Daniel. 2016. (Contrastive) topic. In Caroline Féry & Shin Ishihara (eds.), Handbook of Information Structure, 64-85. Oxford: Oxford University Press. Burr, Elisabeth. 1998. Lingua media e lingua dei giornali. In Navarro Salazar & María Teresa (eds.), Italica Matritensia. Atti del IV Convegno SILFI, Società Internazionale di Linguistica e Filologia Italiana, 115-133. Firenze: Franco Cesati. Cappeau, Paul. 2005. Les indéfinis à l'aune de l'oral. Scolia 20. 67-82. Cappeau, Paul. 2008. Il manque des indéfinis ! Ou comment l'oral nous oblige à revoir la description des indéfinis. Le français aujourd'hui 162(3). 73-83. Cappeau, Paul & José Deulofeu. 2001. Partition et topicalisation: Il y en a "stabilisateur" de sujets et de topiques indéfinis. Cahiers de praxématique 37. 45-82. Cappeau, Paul & José Deulofeu. 2006a. Le fonctionnement de divers quantifieurs indéfinis à l’oral. In Georges Kleiber, Catherine Schnedecker & Anne Theissen (eds.), La relation partie-tout, 427-448. Leuven: Peeters. Cappeau, Paul & José Deulofeu. 2006b. Les "indéfinis" en relation avec la position sujet dans trois types de constructions prédicatives. In Actes du colloque de Paris Prédication et Indéfinis, 125-138. Paris: Presses Universitaires de Paris Sorbonne. Cappelle, Bert, Anne Carlier, Benjamin Fagard & Machteld Meulleman. 2016. Negative existentials - Distinct syntactic, semantic and pragmatic properties. Paper presented at Societas Linguistica Europaea [SLE] 49, Naples. Carlson, Gregory N. 1977. Reference to kinds in English. Amherst: University of Massachusetts PhD thesis. Carlson, Lauri. 1983. Dialogue games: An approach to discourse analysis. Massachusetts: MIT. Carnie, Andrew. 2013. Syntax: A generative introduction (3rd ed.). Oxford: Blackwell. Carter-Thomas, Shirley. 2009. The French c’est-cleft: Function and frequency. In David Banks, Simon Eason & Janet Ormrod (eds.), La linguistique systémique fonctionnelle et la langue française, 127-157. Paris: L'Harmattan. . Casalicchio, Jan. 2013a. The pseudo-relatives and other correspondent constructions in the Romance languages. In Irina Windhaber & Peter Anreiter (eds.), Proceedings of the 4th Austrian students' Conference of Linguistics, 64-84. Cambridge: Cambridge Scholars Publishing. Casalicchio, Jan. 2013b. The pseudo-relatives in presentational and locative constructions with the verb esserci. Cambridge Italian Dialect Syntax-Morphology Meeting 7, Cambridge. https://www.academia.edu/4065040/The_Pseudo- relatives_in_presentational_and_locative_constructions_with_the_verb_esserci. Casalicchio, Jan. 2013c. Pseudorelative, gerundi e infiniti nelle varietà romanze: Affinità (solo) superficiali e corrispondenze strutturali. München: LINCOM Europa. Casalicchio, Jan. 2016. Secondary predicates in Romance c'è-clauses. Paper presented at the International Workshop on Non-Prototypical Clefts (16 December 2016). https://www.arts.kuleuven.be/ling/presentationalclefts/casalicchio-workshop, Leuven.

392 References

Cerruti, Massimo, Claudia Crocco & Stefania Marzo. 2017. On the development of a new standard norm in Italian. In Massimo Cerruti, Claudia Crocco & Stefania Marzo (eds.), Language and social life. Berlin/ Boston: De Gruyter Mouton. Chafe, Wallace L. 1976. Givenness, contrastiveness, definiteness, subjects and topics. In Charles N. Li (ed.), Subject and topic, 25-55. New York: Academic Press. Chafe, Wallace L. 1987. Cognitive constraints on information flow. In Russell S. Tomlin (ed.), Coherence and grounding in discourse: Outcome of a Symposium, Eugene, Oregon, June 1984, 21-51. Amsterdam: John Benjamins. Charolles, Michel. 2002. La référence et les expressions référentielles en français. Paris: Ophrys. Charolles, Michel & Denis Vigier. 2005. Les adverbiaux en position préverbale: Portée cadrative et organisation du discours. Langue française 148(4). 9-30. Choi-Jonin, Injoo & Véronique Lagae. 2005. Il y a des gens ils ont mauvais caractère. A propos du rôle de il y a. In Adolfo Murguía (ed.), Sens et références. Mélanges Georges Kleiber, 39-66. Tübingen: Gunter Narr. Chomsky, Noam. 1965. Aspects of the theory of syntax. Cambridge, MA: MIT Press. Chomsky, Noam. 1977. On wh-movement. In Peter Culicover, Thomas Wasow & Adrian Akmajian (eds.), Formal syntax, 71-132. New York: Academic Press. Chomsky, Noam. 1995. The Minimalist Program. Cambridge, MA: MIT Press. Cinque, Guglielmo. 1991. Mica: Note di sintassi e pragmatica. In Guglielmo Cinque (ed.), Teoria linguistica e sintassi italiana, 311-323. Bologna: Il Mulino. Cinque, Guglielmo. 2008. Two types of non-restrictive relatives. In Olivier Bonami & Patricia Cabredo Hofherr (eds.), Empirical issues in formal syntax and semantics 7. Papers from CFFP 2007, 99-137. http://www.cssp.cnrs.fr/eiss7/. Cinque, Guglielmo. 2013. Typological studies. Word order and relative clauses. New York: Routledge. Cinque, Guglielmo & Paola Benincà. 2010. La frase relativa. In Giampaolo Salvi & Lorenzo Renzi (eds.), Grammatica dell’italiano antico, Vol. 1, 469-507. Bologna: Il Mulino. Clark, Eve. 1978. Locationals: Existential, locative and possessive constructions. In Joseph H. Greenberg (ed.), Universals of human language, 85-126. Stanford, CA: Stanford University Press. Clark, Herbert H. & Eve V. Clark. 1977. Psychology and language. New York: Harcourt Brace Jovanovich. Clech-Darbon, Anne, Georges Rebuschi & Annie Rialland. 1999. Are there cleft sentences in French? In Laurice Tuller & Georges Rebuschi (eds.), The grammar of focus, 83-118. Amsterdam: John Benjamins. Clifton, Charles & Lyn Frazier. 2012. Discourse integration guided by the ‘Question Under Discussion’. Cognitive Psychology 65(2). 352-379. Collins, Peter C. 1991. Cleft and pseudo-cleft constructions in English. London: Routledge. Collins, Peter C. 1992. Cleft existentials in English. Language Sciences 14(4). 419-433. Comrie, Bernard. 1981. Language universals and linguistic typology: Syntax and morphology. Chicago: University of Chicago Press. Constant, Noah. 2012. Topic abstraction as the source for nested alternatives: A conservative semantics for contrastive topic. In Nathan Arnett & Ryan Bennett (eds.), Proceedings of the 30th West Coast Conference on Formal Linguistics, 120-130. Somerville, MA: Cascadilla Proceedings Project. Constant, Noah. 2014. Contrastive topic: Meanings and realizations. Amherst: University of Massachussets Amherst PhD dissertation. Conti, Virginie. 2010. La construction en avoir SN qui SV ("j’ai ma copine qui habite à Paris"): Une forme de dispositif clivé ? Linx(62-63). 63-87. Copley, Bridget & Isabelle Roy. 2015. Deriving the readings of French être en train de. In Enoch O. Aboh, Jeannette C. Schaeffer & Petra Sleeman (eds.), Romance Languages and Linguistic Theory 2013: Selected papers from 'Going Romance' Amsterdam 2013, 103–118. Amsterdam: John Benjamins. Cowles, Heidi Wind. 2003. Processing information structure: Evidence from comprehension and production. San Diego: University of California PhD dissertation.

References 393

Cowles, Heidi Wind, Robert Kluender, Marta Kutas & Maria Polinsky. 2007. Violations of information structure: An electrophysiological study of answers to wh-questions. Brain and Language 102(3). 228-242. Coy, Charlotte. 2016a. The evolution of definiteness effects with French il y a from 1300 to today. In Susann Fischer, Tanja Kupisch & Esther Rinke (eds.), Definiteness effects: Bilingual, typological and diachronic variation, 278-300. Cambridge: Cambridge Publishing Scholars. Coy, Charlotte. 2016b. Existentials in relative clauses: A contrastive corpus study of Spanish haber and French y avoir. In Carlota De Benito Moreno & Álvaro S. Octavio De Toledo y Huerta (eds.), En torno a haber: Construcciones, usos y variación desde el latín hasta la actualidad, 191-207. Frankfurt am Main: Peter Lang. Coy, Charlotte. 2017. Existenz und Lokalisierung bei französisch 'il y a': Konsequenzen für den Definitheitseffekt. Tübingen: Universität Tübingen PhD dissertation. Coy, Charlotte & Birgit Umbreit. 2015. Diskurstraditionelle Faktoren der Verwendung und Verbreitung von Existenzkonstruktionen am Beispiel von frz. il y a + strong NP. In Esme Winter-Froemel, Araceli Lopéz Serena, Álvaro Octavio de Toledo y Huerta & Barbara Frank-Job (eds.), Diskurstraditionelles und Einzelsprachliches im Sprachwandel / Tradicionalidad discursiva e idiomaticidad en los procesos de cambio lingüístico. Tübingen: Narr. Creissels, Denis. 2014. Existential predication in typological perspective. Unpublished manuscript, www.deniscreissels.fr/public/Creissels-Exist.Pred.pdf Cresti, Diana. 1995. Indefinite topics. MIT PhD dissertation. Cristofaro, Sonia. 2009. Grammatical categories and relations: Universality vs. language-specificity and construction-specificity. Language and Linguistics Compass 3(1). 441-479. Croft, William. 1990. Possible verbs and the structure of events. In Savas Tshohatzidis (ed.), Meanings and prototypes: Studies in linguistic categorization, 48-73. London: Routledge. Croft, William. 2001. Radical Construction Grammar. Oxford: Oxford University Press. Croft, William. 2003. Typology and universals (2 ed.). Cambridge: Cambridge University Press. Croft, William. 2005. Logical and typological arguments for Radical Construction Grammar. In Mirjam Fried & Jan-Ola Östman (eds.), Construction Grammar(s): Cognitive and cross- language dimensions, 273-314. Amsterdam: John Benjamins. Croft, William & Alan D. Cruse. 2004. Cognitive linguistics. Cambridge: Cambridge University Press. Cruschina, Silvio. 2012. Focus in Existential Sentences. In Valentina Bianchi & Christiano Chesi (eds.), Enjoy Linguistics! Papers offered to Luigi Rizzi on the occasion of his 60th birthday, 77- 107. Sienna: CISCL Press. Cruschina, Silvio. 2015a. Focus structure. In Delia Bentley, Francesco Maria Ciconte & Silvio Cruschina (eds.), Existentials and locatives in Romance dialects of Italy, 43-98. Oxford: Oxford University Press. Cruschina, Silvio. 2015b. Patterns of variation in existential constructions. Isogloss 1(1). 1-32. Cruschina, Silvio. 2016. Setting the boundaries: Existential sentences and presentational clefts in Italian. Paper presented at the International Workshop on Non-Prototypical Clefts (15 December 2016) https://www.arts.kuleuven.be/ling/presentationalclefts/cruschina-2016, Leuven. Culioli, Antoine. 1990. Pour une linguistique de l'énonciation. Paris: Ophrys. D'Achille, Paolo, Domenico Proietti & Andrea Viviani. 2005. La frase scissa in italiano: Aspetti e problemi. In Iørn Korzen & Paolo D'Achille (eds.), Tipologia linguistica e società: Due giornate italo-danesi di studi linguistici (Roma, 27–28 November 2003), 249-279. Firenze: Cesati. Dancygier, Barbara. 2005. Constructional compositionality and blending: The case of Polish SLVF constructions. In Marc Ettlinger, Nicholas Fleisher & Mischa Park-Doob (eds.), Proceedings of the thirtieth annual meeting of the Berkeley Linguistic Society, 456-467. Berkeley: Berkeley Linguistics Society. Dancygier, Barbara & Eve Sweetser. 2005. Mental spaces in grammar: Conditional constructions. Cambridge: Cambridge University Press. Daneš, František. 1974. Functional sentence perspective and the organization of the text. In František Daneš (ed.), Papers on Functional Sentence Perspective, 106-128. Prague: Academia.

394 References

Danon-Boileau, Laurent. 1989. La détermination du sujet. Langages 24(94). 39-72. Dardano, Maurizio. 1987. Il linguaggio dei giornali italiani. In Jader Jacobelli (ed.), Dove va la lingua italiana? , 58-65. Roma/Bari: Laterza. Dardano, Maurizio. 1996. Manualetto di linguistica italiana. Bologna: Zanichelli. Dasinger, Lise & Cecile Toupin. 1994. The development of relative clause functions in narrative. In Ruth A. Berman & Dan Isaac Slobin (eds.), Relating events in narrative: A crosslinguistic developmental study, 457-514. Hillsdale, NJ / Hove, UK: Lawrence Erlbaum Associates. Dassi, M. 2003. De la morphosyntaxe à la sémantique des présentatifs en français contemporain: Une aperception fondée sur la prose romanesque de Mongo Beti. München: LINCOM. Davidse, Kristin. 1999. The semantics of cardinal versus enumerative existential constructions. Cognitive Linguistics 10(3). 203-250. Davidse, Kristin. 2000. A constructional approach to clefts. Linguistics 38(6). 1101-1131. Davidse, Kristin. 2014. On specificational there-clefts. Leuven working papers in Linguistics. 1-34. Davidse, Kristin & Ditte Kimps. 2016. Specificational there-clefts: Functional structure and information structure. English Text Construction 9(1). 115-142. Davidse, Kristin, Ditte Kimps & Ngum Meyuhnsi Njende. 2016. Making the case for defining clefts broadly as secondary specification structures: Prosody and information structure. The second Round Table on Communicative Dynamism, Namur. Davidse, Kristin & Karen Lahousse. 2014. Specificational existential clefts in French and English. Round Table on Communicative Dynamism, Cardiff, UK, 5 September 2014. De Cat, Cécile. 2007. French dislocation: Interpretation, syntax, acquisition. Oxford: Oxford University Press. De Cesare, Anna-Maria. 2007. Sul cosidetto 'c'è presentativo'. Forme e funzioni. In Anna-Maria De Cesare & Angela Ferrari (eds.), Lessico, grammatica e testualità, tra italiano scritto e parlato, 127-153. Basel: University of Basel. De Cesare, Anna-Maria. 2012. Riflessioni sulla diffusione delle costruzioni scisse nell’italiano giornalistico odierno a partire dalla loro manifestazione nei lanci di agenzia in italiano e in inglese. Cuadernos de Filología Italiana 19. 11-39. De Cesare, Anna-Maria & Davide Garassino. 2015. On the status of exhaustiveness in cleft sentences: An empirical and cross-linguistic study of English also-/only-clefts and Italian anche-/solo-clefts. Folia Linguistica 49(1). 1-56. De Cesare, Anna-Maria & Davide Garassino (Eds.). 2016. Current issues in Romance non-canonical word orders. Syntax - Information Structure - Discourse organization. Bern: Peter Lang. De Cesare, Anna-Maria, Davide Garassino, Rocío Agar Marco, Ana Albom & Doriana Cimmino. 2016. Sintassi marcata dell'italiano dell'uso medio in prospettiva contrastiva con il francese, lo spagnolo, il tedesco e l'inglese. Uno studio basato sulla scrittura dei quotidiani online. Frankfurt am Main: Peter Lang. De Cesare, Anna-Maria, Davide Garassino, Rocío Agar Marco & Laura Baranzini. 2014. Form and frequency of Italian cleft constructions in a corpus of electronic news: A comparative perspective with French, Spanish, German and English. In Anna-Maria De Cesare (ed.), Frequency, forms and functions of cleft constructions in Romance and Germanic: Contrastive, corpus-based studies, 49-99. Berlin: De Gruyter Mouton. De Pascale, Stefano & Stefania Marzo. 2016. Gli italiani regionali. Atteggiamenti linguistici verso le varietà geografiche dell'italiano. Incontri: Rivista Europea di Studi Italiani 31(1). 61-76. De Saussure, Louis. 2014. Verbes modaux et sous-détermination. In J. Blochoviak (ed.), Mélanges offerts à Jacques Moeschler: Online publication. De Stefani, Elwys. 2008. De la malléabilité des structures syntaxiques dans l'interaction orale: Le cas des constructions clivées. In J. Durand, B. Habert & B. Laks (eds.), 1er Congrès Mondial de Linguistique Française, 703-720. Les Ulis, France: EDP Sciences. http://icar.univ- lyon2.fr/ecole_thematique/tranal_i/documents/DeStefani2008bcleft.pdf. De Stefani, Elwys. 2009. Le strutture grammaticali come epifenomeni dell'interazione sociale? Riflessioni sull'uso delle costruzioni scisse nel parlato conversazionale italiano e francese. In Sintassi storica e sincronica dell'italiano: Subordinazione, coordinazione, giustapposizione: Atti del X Congresso della Società internazionale di linguistica e filologia italiana, Basilea, 30 giugno-3 luglio 2008, 1615-1631. Firenze: Franco Cesati Editore.

References 395

De Stefani, Elwys, Anne-Sylvie Horlacher & Simona Pekarek Doehler. 2015. Time and emergence in grammar: Dislocation, topicalization and hanging topic in French talk-in-interaction. Amsterdam: John Benjamins. Declerck, Renaat. 1983. Predicational clefts. Lingua 61(1). 9-45. Declerck, Renaat. 1984. The pragmatics of it-clefts and WH-clefts. Lingua 64(4). 251-289. Declerck, Renaat. 1988. Studies on copular sentences, cleſts and pseudo-clefts. Dordrecht: Foris Publications. Delin, Judy. 1989. Cleft constructions in discourse. Edinburgh: University of Edinburgh PhD dissertation. Delin, Judy. 1992. Properties of it-cleft presuppositions. Journal of Semantics 9(4). 289-306. Delin, Judy & Jon Oberlander. 1995. Syntactic constraints on discourse structure: The case of it- clefts. Linguistics 33(3). 465-500. Delin, Judy & Jon Oberlander. 2005. Cleft constructions in context: Some Suggestions for research methodology. Unpublished manuscript.http://studylib.net/doc/7373211/cleft- constructions-in-context. Den Dikken, Marcel. 2013. Predication and specification in the syntax of cleft sentences. In Katharina Hartmann & Tonjes Veenstra (eds.), Cleft structures, 35-70. Amsterdam: John Benjamins. Desclés, Jean-Pierre & Zlatka Guentchéva. 1995. Is the notion of process necessary? In Pier Marco Bertinetto, Valentina Bianchi, Östen Dahl & Mario Squartini (eds.), Temporal reference, aspect and actionality, Vol. 1, 55-70. Turin: Rosenberg and Sellier. Destruel, Emilie. 2012. The French c’est-cleft: An empirical study on its meaning and use. In Christopher Piñón (ed.), Empirical Issues in Syntax and Semantics, Vol. 9, 95-112. Destruel, Emilie. 2013. The French c’est-cleft: Empirical studies of its meaning and use. Austin: University of Texas at Austin PhD dissertation. Destruel, Emilie & Leah Velleman. 2014. Refining contrastiveness. Empirical evidence from the English it-cleft. In Cristopher Piñon (ed.), Selected Papers from CFFP 2013, 197-214. Deulofeu, José. 1984. L'étude des langues parlées er la typologie des langues. Recherches sur le Français Parlé 5. 103-124. Deulofeu, José. 1989. Les couplages de constructions verbales en français parlé: Effet de cohésion discursive ou syntaxe de l'énoncé. Recherches sur le Français Parlé 9. 111-141. Deulofeu, José. 2000. Les commentaires sportifs télévisés sont-ils un genre au sens de la "grammaire des genres"? In Mireille Bilger (ed.), Corpus: Méthodologie et applications linguistiques, 271-295. Paris: Honoré Champion. Diesing, Molly. 1992. Bare plural subjects and the derivation of logical representations. Linguistic Inquiry 23(3). 353-380. Diessel, Holger. 2007. Frequency effects in language acquisition, language use, and diachronic change. New Ideas in Psychology 25(2). 108-127. Diessel, Holger. 2009. The emergence of relative clauses in early child language. Unpublished manuscript, http://www.ruf.rice.edu/~eivs/sympo/papers/Diessel.pdf Diessel, Holger & Michael Tomasello. 2000. The development of relative clauses in spontaneous child speech. Cognitive linguistics 11(1). 131-151. Dobrovie-Sorin, Carmen & Claire Beyssade. 2004. Définir les indéfinis. Paris: CNRS. Dobrovie-Sorin, Carmen & Claire Beyssade. 2012. Redefining indefinites. Dordrecht: Springer. Doetjes, Jenny, Georges Rebuschi & Annie Rialland. 2004. Cleft Sentences. In Francis Corblin & Henriëtte De Swart (eds.), Handbook of French semantics, 529-552. Stanford: CSLI Publications. Doherty, Monika. 2005. Topic-worthiness in German and English. Linguistics 43(1). 181-206. Drenhaus, Heiner, Malte Zimmermann & Shravan Vasishth. 2011. Exhaustiveness effects in clefts are not truth-functional. Journal of Neurolinguistics 24(3). 320-337. Dryer, Matthew S. 1996. Grammatical relations in Ktunaxa (Kutenai): Belcourt lecture delivered before the University of Manitoba on 24 February 1995. Winnipeg: Voices of Rupert's land. Dryer, Matthew S. 1997. Are grammatical relations universal? In Joan Bybee, John Haiman & Sandra A. Thompson (eds.), Essays on language function and language type, 115-143. Amsterdam: John Benjamins.

396 References

Dufter, Andreas. 2006. Zwischen Kompositionalität und Konventionalisierung: Satzspaltung mit c'est im Französischen der Gegenwart. Romanistisches Jahrbuch 57. 31-59. Dufter, Andreas. 2008. On explaining the rise of c'est-clefts in French. In Ulrich Detges & Richard Waltereit (eds.), The paradox of grammatical change: Perspectives from Romance, 31-56. Amsterdam: John Benjamins. Dufter, Andreas. 2009a. Beyond focus marking: Fine-tuning the evolution of cleft types from Latin to Modern French. Paper presented at the DGfS-Jahrestagung 31. AG 9: Focus marking strategies and focus interpretation, Osnabrück. Dufter, Andreas. 2009b. Clefting and Discourse organization: Comparing Germanic and Romance. In Andreas Dufter & Daniel Jacob (eds.), Focus and Background in Romance languages, 83- 121. Amsterdam/Philadelphia: John Benjamins. Dupont, Didier, Thierry Fauvaux & Michel Ghenet. 1994. La dynamique de l’information. Eléments de grammaire textuelle. Brussels: De Boeck-Duculot. Ebert, Christian & Cornelia Endriss. 2005. Wide scope indefinites as aboutness topics. In Cécile Meier & Matthias Weisgerber (eds.), Proceedings of the conference "sub8 - Sinn und Bedeutung". 8th Annual meeting of the Gesellschaft für Semantik, 95-110. Konstanz: Fachbereich Sprachwissenschaft der Universität Konstanz. Erteschik-Shir, Nomi. 2007. Information structure. The syntax-discourse interface. Oxford: Oxford University Press. Ettlinger, Marc. 2005. Constructional compositionality and the English resultative. In Proceedings of 41st Annual Meeting of the Chicago Linguistics Society. Chicago: Chicago Linguistics Society. Evert, Stefan. 2005. The statistics of cooccurrences: Word pairs and collocations. Stuttgart: Institut für maschinelle Sprachverarbeitung. University of Stuttgart PhD dissertation. Floricic, Franck & Françoise Mignon. 2007. Négation et réduplication intensive en français et en italien. In Franck Floricic (ed.), La négation dans les langues romanes, 117-136. Amsterdam: John Benjamins. Fox, Barbara A. & Sandra A. Thompson. 1990. A discourse explanation of the grammar of relative clauses in English conversation. Language 66(2). 297-316. Fox, Barbara A. & Sandra A. Thompson. 2007. Relative clauses in English conversation: Relativizers, frequency, and the notion of construction. Studies in Language 31(2). 293-326. Francez, Itamar. 2007. Existential propositions. Stanford: Stanford University PhD dissertation. Frascarelli, Mara & Roland Hinterhölzl. 2007. Types of topics in German and Italian. In Kerstin Schwabe & Susanne Winkler (eds.), On information structure, meaning and form: Generalizations across languages, 87-116. Amsterdam: John Benjamins. Frascarelli, Mara & Francesca Ramaglia. 2009. (Pseudo)cleft constructions at the interfaces. Unpublished manuscript, Lingbuzz. http://ling.auf.net/lingbuzz/000841 Frazier, Lyn & Charles Clifton. 1996. Construal. Cambridge, MA: MIT Press. Fuchs, Catherine. 1987a. Avant-propos. Les types de relatives: De la logique à la grammaire, de la syntaxe à la sémantique. Langages 22(88). 5-7. Fuchs, Catherine. 1987b. Les relatives et la construction de l'interprétation. Langages(88). 95-127. Furukawa, Naoyo. 1996a. Grammaire de la prédication seconde. Louvain-la-Neuve: Duculot. Furukawa, Naoyo. 1996b. Il n'y a que toi qui puisses le faire! - A propos de l'emploi thématique d'un type de proposition subordonnée. Revue Romane 31(2). 271-282. Furukawa, Naoyo. 2005. Pour une sémantique des constructions grammaticales. Brussels: Duculot. Gaeta, Livio & Davide Ricca. 2002. Corpora testuali e produttività morfologica; i nomi d'azione italiani in due annate della Stampa (1996-1997). In Roland Bauer & Hans Goebl (eds.), Paralella IX. Testo - variazione - informatica. Atti del IX Incontro italo-austriaco dei linguisti, 223-249. Wilhelmsfeld: Egert. Gapany, Joël. 2004. Formes et fonctions des relatives en français. Étude syntaxique et sémantique. Bern: Peter Lang. Gapany, Joël & Françoise Zay. 1995. Les relatives parenthétiques: Problèmes de reconnaissance et de délimitation. Scolia 125. 31-64. Garassino, Davide. 2014a. Cleft sentences: Italian-English in contrast. In Anna-Maria De Cesare (ed.), Frequency, forms and functions of cleft constructions in Romance and Germanic: Contrastive, corpus-based studies, 101-138. Berlin: De Gruyter Mouton.

References 397

Garassino, Davide. 2014b. Le frasi scisse nei testi giornalistici online: Italiano e inglese a confronto. In Enrico Garavelli & Elina Suomela-Härmä (eds.), Atti del XII Convegno Silfi (Helsinki, 18–20 June 2012), 631-640. Firenze: Cesati. Garassino, Davide. 2016. Using cleft sentences in Italian and English. A multifactorial analysis. In Anna-Maria De Cesare & Davide Garassino (eds.), Current issues in Romance non-canonical word orders. Syntax - Information structure - Discourse organization, 181-204. Bern: Peter Lang. Gazdar, Gerald. 1979. Pragmatics: Implicature, presupposition, and logical form. New York: Academic Press. Geeraerts, Dirk. 2006. Prototype theory: Prospects and problems of prototype theory. In Dirk Geeraerts (ed.), Cognitive Linguistics: Basic readings, 141-166. Berlin/New York: Mouton de Gruyter. Gil, David. 2001. Escaping eurocentrism: Fieldwork as a process of unlearning. In Paul Newman & Martha Ratliff (eds.), Linguistic Fieldwork, 102-132. Cambridge: Cambridge University Press. Giry-Schneider, Jacqueline. 1988. L'interprétation événementielle des phrases en il y a. Lingvisticae Investigationes 12(1). 85-100. Givón, Talmy. 1976. Topic, pronoun, and grammatical agreement. In Charles N. Li (ed.), Subject and topic, 151-188. New York: Academic Press. Givón, Talmy. 1978. Definiteness and referentiality. In Joseph H. Greenberg, Charles A. Ferguson & Edith A. Moravcsik (eds.), Universals of human language, Vol. 4, 291-330. Stanford, CA: Stanford University Press. Givón, Talmy. 1982. Logic vs. pragmatics with human language as the referee. Toward an empirically viable epistemology. Journal of Pragmatics 6. 81-133. Givón, Talmy. 1983. Topic continuity in discourse: An introduction. In Talmy Givón (ed.), Topic contituity in discourse, 1-42. Amsterdam/Philadelphia: John Benjamins. Givón, Talmy. 1992. The grammar of referential coherence as mental processing instructions. Linguistics 30(1). 5-56. Givón, Talmy. 2001. Syntax: An introduction (Vol. I). Amsterdam: John Benjamins. Godard, Danièle. 1988. La syntaxe des relatives en français. Paris: CNRS. Godard, Danièle & Jean-Marie Marandin. 2006. Reinforcing negation: The case of Italian. In Stefan Müller (ed.), Proceedings of the 13th International Conference on Head-Driven Phrase Structure Grammar, 174-194. Stanford: CSLI Publications. Goldberg, Adele E. 1995. Constructions: A Construction Grammar approach to argument structure. Chicago/London: University of Chicago Press. Goldberg, Adele E. 2003. Constructions: A new theoretical approach to language. Trends in Cognitive Sciences 7(5). 219-224. Goldberg, Adele E. 2006. Constructions at work: The nature of generalization in language. Oxford: Oxford University Press. Grevisse, Maurice & André Goosse. 2008. Le bon usage: Grammaire française (16 ed.). Brussels: De Boeck. Gries, Stefan Thomas. 2003. Collostructions: Investigating the interaction of words and constructions. International Journal of Corpus Linguistics 8(2). 209-243. Grondelaers, Stefan, Marc Brysbaert, Dirk Speelman & Dirk Geeraerts. 2002. Er als accessibility marker: On- en offline evidentie voor een procedurele interpretatie van presentatieve zinnen. Gramma/TTT 9. 1-22. Grondelaers, Stefan, Dirk Speelman, Denis Drieghe, Marc Brysbaert & Dirk Geeraerts. 2009. Introducing a new entity into discourse: Comprehension and production evidence for the status of Dutch er "there" as a higher-level expectancy monitor. Acta Psychologica 130(2). 153-160. Gundel, Jeanette K. 1974. The role of topic and comment in linguistic theory. Austin: University of Texas at Austin PhD dissertation. Gundel, Jeanette K. 1988. Universals of topic-comment structure. In Michael Hammond, Edith Moravcsik & Jessica R. Wirth (eds.), Studies in syntactic typology, 209-239. Amsterdam: John Benjamins.

398 References

Gundel, Jeanette K. 2006. Clefts in English and Norwegian: Implications for the grammar- pragmatics interface. In Valéria Molnar & Susanne Winkler (eds.), The Architecture of focus, 517-548. Berlin/New York: De Gruyter Mouton. Gundel, Jeanette K. & Thorstein Fretheim. 2003. Topic and Focus. In Gregory L. Ward & Laurence R. Horn (eds.), The handbook of pragmatics, 175-196. Malden, MA: Blackwell. Gundel, Jeanette K., Nancy Hedberg & Ron Zacharski. 1993. Cognitive status and the form of referring expressions in discourse. Language 69(2). 274-307. Haegeman, Liliane, André Meinunger & Aleksandra Vercauteren. 2014. The architecture of it-clefts. Journal of Linguistics 50(2). 269-296. Hamann, Cornelia & Laurice Tuller. 2015. Intervention effects in the spontaneous production of relative clauses in (a)typical language development of French children and adolescents. In Elisa Di Domenica, Cornelia Hamann & Simona Matteini (eds.), Structures, strategies and beyond: Papers in honor of Adriana Belletti, 321-342. Amsterdam: John Benjamins. Han, Chung-Hye & Nancy Hedberg. 2008. Syntax and semantics of it-clefts: A tree adjoining grammar. Journal of Semantics 25(4). 345-380. Hartmann, Jutta M. 2006. Well, there's the list reading. UiL OTS Working Papers 2006. 1-15. Haspelmath, Martin. 2003. The geometry of grammatical meaning: Semantic maps and cross- linguistic comparison. In Michael Tomasello (ed.), The new psychology of language, Vol. 2, 211-242. Mahwah, NJ: Lawrence Erlbaum Associates. Haspelmath, Martin. 2010. Comparative concepts and descriptive categories in crosslinguistic studies. Language 86(3). 663-687. Haspelmath, Martin. 2014. Comparative syntax. In Andrew Carnie, Yosuke Sato & Dan Siddiqi (eds.), The Routledge handbook of syntax, 490-508. London/New York: Routledge. Hawkins, John A. 1994. A performance theory of order and constituency. Cambridge: Cambridge University Press. Hedberg, Nancy. 2006. Topic-focus controversies. In Valeria Molnar & Susanne Winkler (eds.), The architecture of focus, 373-397. Berlin/New York: Mouton de Gruyter. Hedberg, Nancy Ann. 1990. Discourse pragmatics and cleft sentences in English. Minneapolis, MN: University of Minnesota PhD dissertation. Hedberg, Nancy Ann. 1999. The discourse function of English clefts and Mandarin shi... de constructions. Paper presented at the Workshop on the Discourse Function of Clefts, Humboldt University Berlin, 1-3 October 1999. http://www.sfu.ca/~hedberg/Berlin%20Paper%20presented+.pdf. Hedberg, Nancy Ann. 2000. The referential status of clefts. Language 76(4). 891-920. Hedberg, Nancy Ann. 2013. Multiple focus and cleft sentences. In Katharina Hartmann & Tonjes Veenstra (eds.), Cleft structures, 227-250. Amsterdam: John Benjamins. Hedberg, Nancy Ann & Lorna Fadden. 2007. The information structure of it-clefts, wh-clefts and reverse wh-clefts in English. In Nancy Hedberg & Ron Zacharski (eds.), The grammar– pragmatics interface: Essays in honor of Jeanette K. Gundel, 49-76. Amsterdam: John Benjamins. Heidinger, Steffen. 2016. El orden de los constituyentes posverbales y la expresión del foco informativo en español. In Adam Ledgeway, Michela Cennamo & Guido Mensching (eds.), Actes du XXVIIe Congrès international de linguistique et de philologie romanes (Nancy, 15-20 juillet 2013) Section 4: Syntaxe, 217-232. Nancy: ATILF. Herburger, Elena. 1994. Focus and NP quantification. In Eric Duncan, Donka Farkas & Philip Spaelti (eds.), The Proceedings of the twelfth West Coast Conference on Formal Linguistics, 517-533. Stanford: Center for the Study of Language and Information. Herslund, Michael. 2005. Clivage, structure thématique et anaphores. In Frédéric Lambert & Henning Nølke (eds.), La syntaxe au cœur de la grammaire: Receuil offert en hommage pour le 60e anniversaire de Claude Muller, 127-135. Rennes: Presses Universitaires de Rennes. Hetterle, Katja. 2015. Adverbial clauses in cross-linguistic perspective. Berlin: De Gruyter Mouton. Heycock, Caroline. 2012. Specification, equation, and agreement in copular sentences. Canadian Journal of Linguistics/Revue canadienne de Linguistique 57(2). 209-240. Hilpert, Martin. 2014. Construction grammar and its application to English. Edinburgh: Edinburg University Press.

References 399

Hilpert, Martin & Holger Diessel. 2017. Entrenchment in Construction Grammar. In Hans-Jörg Schmid (ed.), Entrenchment and the psychology of language learning: How we reorganize and adapt linguistic knowledge, 57-74. Berlin: De Gruyter Mouton. Hobæk Haff, Marianne. 2006. La construction clivée en c’est … qui/que – étude contrastive français- norvégien. In Michel Olsen & Erik H. Swiatek (eds.), XVIe Congrès des Romanistes Scandinaves. Roskilde: Roskilde University. Horn, Laurence R. 1981. Exhaustiveness and the semantics of clefts. In Victoria Burke & James Pustejovsky (eds.), Proceedings of NELS, 125-142. Amherst: GLSA. Huber, Stefan. 2002. Es-clefts und det-clefts. Zur Syntax, Semantik und Informationsstruktur von Spaltsätzen im Deutschen und Schwedischen. Stockholm: Almqvist & Wiksell. Huber, Stefan. 2006. The complex functions of it-clefts. In Valéria Molnár & Susanne Winkler (eds.), The architecture of focus, 549-578. Berlin/New York: Mouton de Gruyter. Huddleston, Rodney. 1971. The sentence in written English: A syntactic study based on an analysis of scientific texts. Cambridge: Cambridge University Press. Huddleston, Rodney. 1984. Introduction to the grammar of English. Cambridge: Cambridge University Press. Huddleston, Rodney & Geoffrey Pullum. 2002. The Cambridge grammar of the English language. Cambridge: Cambridge University Press. Hung, Yu-Chen & Petra B. Schumacher. 2014. Animacy matters: ERP evidence for the multi- dimensionality of topic-worthiness in Chinese. Brain Research 1555. 36-47. Hupet, Michel & Brigitte Tilmant. 1986. What are clefts good for? Some consequences for comprehension. Journal of Memory and Language 25(4). 419-430. Husband, E. Matthew. 2010. On the compositional nature of stativity. Michigan: Michigan State University PhD dissertation. Jackendoff, Ray. 1997. The architecture of the language faculty. Cambridge, MA: MIT Press. Jacob, Daniel. 2015. Anaphorische Spaltsätze im Französischen: Grammatik - Text - Rhetorik. In Séverine Adam, Daniel Jacob & Michael Schecker (eds.), Informationsstrukturen in Kontrast: Strukturen, Kompositionen und Strategien, 101-122. Frankfurt am Main: Peter Lang. Jacobs, Joachim. 2001. The dimensions of topic – comment. Linguistics 39(4). 641-681. Jäger, Gerhard. 1996. The stage/individual contrast revisited. In Brian Agbayani & Sze-Wing Tang (eds.), Proceedings of the West Coast Conference on Formal Linguistics 15, 225-240. University of California, Irvine: http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.56.6722. Jäger, Gerhard. 2001. Topic-comment structure and the contrast between stage level and individual level predicates. Journal of Semantics 18(2). 83-126. Japrisot, Sébastien. 2012 [1977]. L'été meurtrier. Paris: Denoël. Jary, Mark & Mikhail Kissine. 2016. When terminology matters: The imperative as a comparative concept. Linguistics 54(1). 119-148. Jeanjean, Colette. 1979. Soit y avait le poisson soit y avait ce rôti farci: Étude de la construction il y a dans la syntaxe du français. Recherches sur le Français Parlé 2. 121-162. Jisa, Harriet & Sophie Kern. 1998. Relative clauses in French children's narrative texts. Journal of Child Language 25(3). 623-652. Jourdain, Morgane & Karen Lahousse. in preparation. Les clivées dans l'acquisition du français. Jullien, Stéphane. 2007. Prosodic, syntactic and semantico-pragmatic parameters as clues for projection: The case of "il y a". Nouveaux cahiers de linguistique française 28. 279-297. Jullien, Stéphane. 2008. La construction présentative clivée dans la gestion des tours de parole: Le cas des interactions adulte – enfant. Revue Tranel 49. 101-118. Jullien, Stéphane. 2014. Syntaxe et dialogue. Les configurations syntaxiques impliquant "il y a". Neuchâtel: Université de Neuchatêl, Université Paris III - Sorbonne Nouvelle PhD thesis. Karssenberg, Lena. 2013. L'adverbe 'ainsi' en tête de phrase: Une analyse de corpus. Leuven: KU Leuven MA thesis. Karssenberg, Lena. 2016a. French il y a clefts, existential sentences and the focus-marking hypothesis. Journal of French Language Studies. 1-26. Karssenberg, Lena. 2016b. Il n'y a que Superman qui porte le slip par-dessus le pantalon: Les clivées en il n'y a que x qui. In 5e Congrès Mondial de Linguistique Française, Vol. 27. Les Ulis: EDP

400 References

Sciences. https://www.linguistiquefrancaise.org/articles/shsconf/abs/2016/05/shsconf_cmlf2016_02 009/shsconf_cmlf2016_02009.html. Karssenberg, Lena. accepted. Specificerende il y a-clefts (er is-clefts). Handelingen der Koninklijke Zuid-Nederlandse Maatschappij voor Taal- en Letterkunde en Geschiedenis (KZM). Karssenberg, Lena & Karen Lahousse. 2014. Ainsi en tête de phrase + inversion: une analyse de corpus. In Frank Neveu, Peter Blumenthal, Linda Hriba, Annette Gerstenberg, Judith Meinschaefer & Sophie Prevost (eds.), 4e Congrès Mondial De Linguistique Francaise, Vol. 8, 2413-2427. Les Ulis: EDP Sciences. www.shs-conferences.org/articles/shsconf/pdf/2014/05/shsconf_cmlf14_01266.pdf. Karssenberg, Lena & Karen Lahousse. 2017. Les SN définis et indéfinis dans les clivées en il y a. In Caroline Lachet, Luis Meneses-Lerín & Audrey Roig (eds.), Contraintes linguistiques. À propos de la complémentation nominale, 197-210. Brussels: PIE Peter Lang. Karssenberg, Lena & Karen Lahousse. forthcoming-a. The information structure of French il y a clefts & c’est clefts: A corpus-based analysis. Linguistics. Karssenberg, Lena & Karen Lahousse. forthcoming-b. On the different interpretations of sentence- initial ainsi 'so' and the competition between three types of verb-subject order. Folia Linguistica. Karssenberg, Lena, Karen Lahousse & Stefania Marzo. under review. Les clivées en voici/voilà: Une analyse de corpus. Karssenberg, Lena, Stefania Marzo, Karen Lahousse & Daniela Guglielmo. forthcoming. There's more to Italian c'è clefts than marking all-focus. Italian Journal of Linguistics 29(2). Karttunen, Lauri. 1974. Presuppositions and linguistic context. Theoretical Linguistics 1. 181-194. Katz, Stacy. 2000a. Categories of c’est-cleft constructions. Canadian Journal of Linguistics/Revue canadienne de linguistique 45(2). 253-273. Katz, Stacy. 2000b. A functional approach to the teaching of the French c’est-cleft. The French Review 74(2). 248-262. Kay, Paul. 2005. Pragmatic aspects of grammatical constructions. In Laurence R. Horn & Gregory Ward (eds.), The handbook of pragmatics, 675-700. Oxford, UK: Blackwell. Kay, Paul & Laura A. Michaelis. 2012. Constructional meaning and compositionality. In Claudia Maienborn, Klaus Von Heusinger & Paul Portner (eds.), Semantics. An international handbook of natural language meaning, Vol. 3, 2271–2296. Berlin: De Gruyter. Kayne, Richard S. 1994. The antisymmetry of syntax. Cambridge, MA: MIT press. Keenan, Edward L. 1971. Two kinds of presupposition in natural language. In Charles J. Fillmore & Terence Langendoen (eds.), Studies in Linguistic Semantics, 45-54. New York: Holt. Keenan, Edward L. 1985. Relative clauses. In Timothy Shopen (ed.), Language typology and syntactic description, Vol. 2, 141-170. Cambridge/New York: Cambridge University Press. Kehler, Andrew & Hannah Rohde. 2016. Evaluating an expectation-driven QUD model of discourse representation. Discourse Processes. Kiss, Kátalin É. 1998. Identificational focus versus information focus. Language 74(2). 245-273. Kleiber, Georges. 2001. Indéfinis: Lecture existentielle et lecture partitive. In Georges Kleiber, Brenda Laca & Liliane Tasmowski (eds.), Typologie des groupes nominaux, 47-97. Rennes: Presses Universitaires de Rennes. Klein, Jean & Béatrice Lamiroy. 2011. Routines conversationnelles et figement. In Jean-Claude Anscombre & Salah Mejri (eds.), Le figement linguistique: La parole entravée, 195-217. Paris: Honoré Champion. Klein, Wolfgang. 2012. The information structure of French. In Manfred Krifka & Renate Musan (eds.), The expression of information structure, 95-126. Berlin/Boston: De Gruyter Mouton. Koch, Peter. 2012. Location, existence, and possession: A constructional-typological exploration. Linguistics 50. 533-603. König, Jean-Pierre & Knud Lambrecht. 1999. French relative clauses as secondary predicates: A case study in Construction Theory. In Francis Corblin, Carmen Dobrovie-Sorin & Jean- Marie Marandin (eds.), Empirical issues in Formal Syntax and Semantics 2, 191-214. The Hague: Thesus. Koopman, Hilda & Dominique Sportiche. 2007. The que/qui alternation: New analytical directions. Unpublished manuscript, Lingbuzz. http://ling.auf.net/lingbuzz/000638

References 401

Krifka, Manfred. 2008. Basic notions of information structure. Acta Linguistica Hungarica 55(3). 243- 276. Krötsch, Monique & Annette Sabban. 1990. "Bleu, je veux" – Remarques sur la focalisation en français. Zeitschrift für romanische Philologie 106(1). 80-98. Kuno, Susumu. 1972. Functional sentence perspective: A case study from Japanese and English. Linguistic Inquiry 3(2). 269-320. Kuroda, Sige-Yuki. 1972. The categorical and the thetic judgment: Evidence from Japanese syntax. Foundations of Language 9(2). 153-185. Kurtzman, Howard S. & Maryellen C. MacDonald. 1993. Resolution of quantifier scope ambiguities. Cognition 48(3). 243-279. Lagae, Véronique. 1997. En quantitatif: Pronom lié à la fonction object ou à une position? Travaux de Linguistique 35. 103-114. Lagae, Véronique. 2001. J'en ai lu deux, de livres: Les structures à détachement de forme de N. In Dany Amiot, Walter De Mulder & Nelly Flaux (eds.), Le syntagme nominal: Syntaxe et sémantique, 215-231. Arras: Artois Presses Université. Lagae, Véronique & Christine Rouget. 1998. Quelques réflexions sur les relatives prédicatives. In Mireille Bilger, Karel Van den Eynde & Françoise Gadet (eds.), Analyse linguistique et approches de l'oral. Recueil d'études offert en hommage à Claire Blanche-Benveniste, 313- 325. Leuven/Paris: Peeters. Lahousse, Karen. 2003. Le sujet nominal postverbal en français moderne. Leuven: KU Leuven PhD dissertation. Lahousse, Karen. 2005. 'Focus VS': A special type of NP subject inversion. In Twan Geerts, Ivo Van Ginneken & Haike Jacobs (eds.), Romance Languages and Linguistic Theory 2003. Amsterdam: John Benjamins. Lahousse, Karen. 2006. NP subject inversion in French: Two types, two configurations. Lingua: international review of general linguistics 116(4). 424-461. Lahousse, Karen. 2007. Specificational sentences and word order in Romance: A functional analysis. Folia Linguistica 41(3-4). 357-404. Lahousse, Karen. 2008. Implicit stage topics: A case study in French. Discours. Revue de linguistique, psycholinguistique et informatique 1. 1-18. Lahousse, Karen. 2009. Specificational sentences and the influence of information structure on (anti-)connectivity effects. Journal of Linguistics 45(1). 139-166. Lahousse, Karen. 2011. Quand passent les cigognes: Le sujet nominal postverbal en français moderne. Paris: Presses Universitaires de Vincennes. Lahousse, Karen & Marijke Borremans. 2014. The distribution of functional-pragmatic types of clefts in adverbial clauses. Linguistics 52(3). 793-836. Lahousse, Karen, Christopher Laenzlinger & Gabriela Soare. 2014. Contrast and intervention at the periphery. Lingua 143. 56-85. Lahousse, Karen & Béatrice Lamiroy. 2012. Word order in French, Spanish and Italian: A grammaticalization account. Folia Linguistica 46(2). 387-415. Lahousse, Karen & Béatrice Lamiroy. 2014. Grammaticalisatie en taalvergelijking: Nederlands het is zo dat vs Frans c'est ainsi que. In Freek Van de velde, Hans Smessaert, Frank Van Eynde & Sara Verbrugge (eds.), Patroon en argument: Een dubbelfeestbundel bij het emeritaat van William Van Belle en Joop van der Horst. Leuven: Universitaire Pers Leuven. Lahousse, Karen & Béatrice Lamiroy. 2015. C'est ainsi que: Grammaticalisation ou lexicalisation ou les deux à la fois ? Journal of French Language Studies. 1-25. Lambrecht, Knud. 1986. Pragmatically motivated syntax. Presentational cleft constructions in spoken French. In 22nd Conference of the Chicago Linguistic Society. Papers from the parasession on Pragmatics and Grammatical Theory, 115-126. Chicago: Chicago Linguistic Society. Lambrecht, Knud. 1988a. Presentational cleft constructions in spoken French. In John Haiman & Sandra A. Thompson (eds.), Clause combining in grammar and discourse, 135-179. Amsterdam/Philadelphia: John Benjamins. Lambrecht, Knud. 1988b. There was a farmer had a dog: Syntactic amalgams revisited. In Proceedings of the 14th Annual Meeting of the Berkeley Linguistic Society, 319-339. Berkeley: Berkeley Linguistic Society.

402 References

Lambrecht, Knud. 1994. Information structure and sentence form: Topic, focus and the mental representations of discourse referents. Cambridge: Cambridge University Press. Lambrecht, Knud. 2000a. Prédication seconde et structure informationnelle: La relative de perception comme construction présentative. Langue française 127(1). 49-66. Lambrecht, Knud. 2000b. When subjects behave like objects: A markedness analysis of sentence- focus constructions across languages. Studies in Language 24(3). 611-682. Lambrecht, Knud. 2001. A framework for the analysis of cleft constructions. Linguistics 39(3). 463- 516. Lambrecht, Knud. 2002. Topic, focus and secondary predication. The French presentational relative construction. In Claire Beyssade, Reineke Bok-Bennema, Frank Drijkoningen & Paola Monachesi (eds.), Romance Languages and Linguistic Theory 2000, 171-212. Amsterdam/Philadelphia: John Benjamins. Lambrecht, Knud & Laura A. Michaelis. 1998. Sentence accent in information questions: Default and projection. Linguistics and philosophy 21(5). 477-544. Lamiroy, Béatrice. 1984. La valeur heuristique de la comparaison linguistique: Un exemple concernant le français, l'espagnol et l'italien. In Alain Guillet & Nunzio La Fauci (eds.), Lexique-Grammaire des langues romanes, 223-233. Amsterdam/Philadelphia: John Benjamins. Lamiroy, Béatrice. 1991. Coréférence et référence disjointe: Les deux pronoms 'en'. Travaux de linguistique 22. 41-65. Lamiroy, Béatrice. 2003. Les notions linguistiques de figement et de contrainte. Lingvisticae Investigationes 26(1). 1-14. Lamiroy, Béatrice. 2008. Le figement: À la recherche d'une définition. Zeitschrift für französische Sprache und Literatur 36. 85-99. Lamiroy, Béatrice. 2014. Contrastive linguistics as a discovery procedure. International Conference on Using Corpora in Contrastive and Translation Studies 4, University of Lancaster. Lamiroy, Béatrice & Michel Charolles. 2004. Simplement, seulement, malheureusement, heureusement. Travaux de linguistique 49(2). 57. Lamiroy, Béatrice & Walter De Mulder. 2011. Degrees of grammaticalization across languages. In Bernd Heine & Heiko Narrog (eds.), The Oxford Handbook of Grammaticalization, 302-317. Oxford: Oxford University Press. Lamiroy, Béatrice & Jean Klein. 2016. Le figement. Unité et diversité. Collocations, expressions figées, phrases situationnelles, proverbes. L'Information Grammaticale 148. 15-20. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar (Vol. 1. Theoretical prerequisites). Stanford: Stanford University Press. Langacker, Ronald W. 1991. Foundations of cognitive grammar. Vol. 2: Descriptive Application. Stanford: Stanford University Press. Langacker, Ronald W. 2008. Cognitive Grammar: A basic introduction. Oxford: Oxford University Press. Langacker, Ronald W. 2009. Cognitive (Construction) Grammar. Cognitive Linguistics 20(1). 167-176. Le Goffic, Pierre. 1979. Propositions relatives, identification et ambiguïté, ou: Pour en finir avec les deux types de relatives. DRLAV 21. 135-145. Léard, Jean-Marcel. 1992. Les gallicismes. Étude syntaxique et sémantique. Paris/Leuven: Duculot. Leonetti, Manuel. 1998. A Relevance-theoretic account of the property predication restriction. In Villy Rouchota & Andreas H. Jucker (eds.), Current issues in Relevance Theory, 141-167. Amsterdam: John Benjamins. Leonetti, Manuel. 2008. Definiteness effects and the role of the coda in existential constructions. In Henrik Høeg Müller & Alex Klinge (eds.), Essays on Nominal Determination: From morphology to discourse management, 131-162. Amsterdam: John Benjamins. Leonetti, Manuel. 2013. On contrastive readings in the interpretation of NPs/DPs. In Sofiana Chiriacescu (ed.), Proceedings of the VI Nereus International Workshop “Theoretical implications at the syntax-semantics interface in Romance”, 99-116. Konstanz: Fachbereich Sprachwissenschaft der Universität Konstanz. Leonetti, Manuel. 2016a. Definiteness effects: The interplay of Information Structure and pragmatics. In Susann Fischer, Tanja Kupisch & Esther Rinke (eds.), Definiteness effects: Bilingual, typological and diachronic variation. Cambridge: Cambridge University Press.

References 403

Leonetti, Manuel. 2016b. Pronominal pivots in existential sentences. Paper presented at the workshop Between EXISTENCE and LOCATION: Empirical, Formal and Typological Approaches to Existential Constructions, Tübingen, 2 December 2016. Leonetti, Manuel & M. Victoria Escandell Vidal. 2002. Coercion and the stage/individual distinction. In Javier Gutiérrez-Rexach (ed.), From words to discourse. Trends in Spanish semantics and pragmatics, 159-180. Amsterdam: Elsefier. Lerner, Gene H. 1996. On the 'semi-permeable' character of grammatical units in conversation: Conditional entry into the turn space of another speaker. In Elinor Ochs, Emanuel A. Schegloff & Sandra A Thompson (eds.), Interaction and grammar, 238-276. Cambridge: Cambridge University Press. Lewis, David. 1979. Scorekeeping in a language game. Journal of Philosophical Logic 8(3). 339-359. Loock, Rudy. 2006. Propositions relatives appositives et discours rapporté. In C. Delesse (ed.), Journée d'études CERTA. Discours rapporté(s): Approche(s) linguistique et/ou traductologique, 77-93. Arras: Presses de l'Université d'Artois. Loock, Rudy. 2007. Appositive relative clauses and their functions in discourse. Journal of Pragmatics 39(2). 336-362. Loock, Rudy. 2012. Appositive relative clauses and their competing allostructures in English: An information-packaging approach. In Anton Benz, Manfred Stede & Peter Kühnlein (eds.), Constraints in Discourse 3: Representing and inferring discourse structure, 207-226. Amsterdam/Philadelphia: John Benjamins. Loock, Rudy. 2013. Pour (enfin?) en finir avec les deux types de relative: La linguistique aux limites de la catégorisation. Cercles 29. 21-45. Lyons, John. 1967. A Note on possessive, existential and locative sentences. Foundations of Language 3(4). 390-396. Lyons, John. 1968. Introduction to theoretical linguistics. Cambridge: Cambridge University Press. Mak, Willem M., Wietske Vonk & Herbert Schriefers. 2008. Discourse structure and relative clause processing. Memory & Cognition 36(1). 170-181. Malchukov, Andrej & Anna Siewierska. 2011. Impersonal constructions: A cross-linguistic perspective. Amsterdam: John Benjamins. Manninen, Satu. 1997. Interpretation of Adverbials with Stage/Individual Level Predicates. In Proceedings of the Annual Postgraduate Conference, University of Edinburgh. http://www.lel.ed.ac.uk/~pgc/archive/1997/paper1915/paper1915.html. Martin, Robert. 2002. Sur le statut sémantique de il y a. Linx 47. 79-87. Marzo, Stefania & Claudia Crocco. 2015. Tipicità delle costruzioni presentative per l'italiano neostandard. Revue Romane 50(1). 30-50. Marzo, Stefania, Kris Heylen & gert De Stutter. 2012. Developments in corpus-based contrastive linguistics. In Stefania Marzo, Kris Heylen & Gert De Stutter (eds.), Corpus studies in contrastive linguistics, 1-6. Amsterdam/Philadelphia: John Benjamins. Matić, Dejan, Rik Van Gijn & Robert D. Van Valin Jr. 2014. Information structure and reference tracking in complex sentences: An overview. In Rik van Gijn, Jeremy Hammond, Dejan Matić, Saskia van Putten & Ana V. Galucio (eds.), Information structure and reference tracking in complex sentences, 1-42. Amsterdam: John Benjamins. Matić, Dejan & Daniel Wedgwood. 2012. The meanings of focus: The significance of an interpretation-based category in cross-linguistic analysis. Journal of Linguistics 49(1). 127- 163. McNally, Louise. 1997. A semantics for the English existential construction. New York: Garland Press. Meinunger, André. 1997. The structure of cleft and pseudo-cleft sentences. Texas Linguistic Forum 38. 235-246. Meinunger, André. 1998. A monoclausal approach to cleft and pseudo-cleft sentences. ZAS Papers in Linguistics 10. 89-105. Mertens, Piet. 2008. Syntaxe, prosodie et structure informationnelle: Une approche prédictive pour l'analyse de l'intonation dans le discours. Travaux de linguistique 56(1). 87-124. Mertens, Piet. 2012. La prosodie des clivées. In Sandrine Caddéo, Marie-Noëlle Roubaud, Magali Rouquier & Frédéric Sabio (eds.), Penser les langues avec Claire Blanche-Benveniste, 127- 139. Aix-en-Provence: Presses Universitaires de Provence.

404 References

Meulleman, Machteld. 2012a. Degrees of grammaticalization in three Romance languages: A comparative analysis of existential constructions. Folia Linguistica 46(2). 417-451. Meulleman, Machteld. 2012b. Les localisateurs dans les constructions existentielles: Approche comparée en espagnol, en français et en italien. Berlin: De Gruyter. Michaelis, Laura & Hartwell Francis. 2007. Lexical subjects and the conflation strategy. In Nancy Hedberg & Ron Zacharski (eds.), Topics in the grammar-pragmatics interface: Papers in honor of Jeanette K. Gundel, 19-48. Amsterdam: John Benjamins. Milsark, Gary L. 1977. Toward an explanation of certain peculiarities of the existential construction in English. Linguistic Analysis 3. 1-29. Molinier, Christian & Françoise Levrier. 2000. Grammaire des adverbes: Description des formes en -ment. Geneva: Librairie Droz. Moro, Andrea. 1997. The raising of predicates: Predicative noun phrases and the theory of clause structure. Cambridge: Cambridge University Press. Mosegaard Hansen, Maj-Britt & Jacqueline Visconti. 2009. On the diachrony of "reinforced" negation in French and Italian. In Corinne Rossari, Claudia Ricci & Adriana Spiridon (eds.), Grammaticalization and pragmatics: Facts, approaches, theoretical Issues, 137-171. Bingley: Emerald. Muller, Charles. 2003a. Naissance et évolution des constructions clivées en c'est... que...: De la focalisation sur l'objet concret à la focalisation fonctionnelle. In Peter Blumenthal & Jean- Emmanuel Tyvaert (eds.), La cognition dans le temps. Études cognitives dans le champ historique des langues et des textes, 101-120. Tübingen: Niemeyer. Muller, Claude. 2003b. Traduire les clivées du français en allemand. In Michael Herslund (ed.), Aspects linguistiques de la traduction, 149-167. Bordeaux: Presses Universitaires de Bordeaux. Muller, Claude. 2006. Sur les propriétés des relatives. Cahiers de grammaire 30(2006). 319-337. Muskens, Reinhard. 2001. Talking about trees and truth-conditions. Journal of Logic, Language, and Information 10(4). 417-455. Nahajec, Lisa Margaret. 2012. Evoking the possibility of presence: Textual and ideological effects of linguistic negation in written discourse. University of Huddersfield PhD thesis. Newmeyer, Frederick J. 2007. Linguistic typology requires crosslinguistic formal categories. Linguistic Typology 11. 133-157. Nikiforidou, Kiki. 2009. Constructional analysis. In Frank Brisard, Jan-Ola Östman & Jef Verschueren (eds.), Grammar, meaning and pragmatics, 16-32. Amsterdam/Philadelphia: John Benjamins. Njende, Ngum Meyuhnsi. 2016. Specificational it- and there-clefts with quantified NPs as value: Semantics and Pragmatics. International Workshop on Non-Prototypical Clefts, Leuven. Olson, Mari Broman. 1994. The semantics and pragmatics of lexical aspect features. Studies in the Linguistic Sciences 24(2). Patten, Amanda L. 2010. Grammaticalisation and the it-cleft construction. In Elizabeth Closs Traugott & Graeme Trousdale (eds.), Gradience, gradualness and grammaticalization, 221- 243. Amsterdam: John Benjamins. Pavey, Emma Louise. 2004. The English it-cleft construction: A Role and Reference Grammar analysis. University of Sussex PhD dissertation. Pekarek Doehler, Simona. 2015. Grammar, projection and turn-organization: (il) y a NP 'there is NP' as a projector construction in French talk-in-interaction. In Arnulf Deppermann & Susanne Günthner (eds.), Temporality in interaction, 173-199. Amsterdam: John Benjamins. Percus, Orin. 1997. Prying open the cleft. In Kiyomi Kusumoto (ed.), Proceedings of NELS 27, 337- 351. Amherst, MA: GLSA. Piantadosi, Steven T., Harry Tily & Edward Gibson. 2012. The communicative function of ambiguity in language. Cognition 122(3). 280-291. Pierrard, Michel. 1985. Il n'y a que X qui: Remarques sur la syntaxe de 'il y a' marquant l'exclusivité. Revue Romane 20. 46-55. Pinto, Manuela. 1997. Licensing and interpretation of inverted subjects in Italian. Utrecht: Universiteit Utrecht PhD dissertation. Piotrowski, Jennifer A. 2009. Information structure of clefts in spoken English. Eugene: University of Oregon PhD dissertation.

References 405

Pitavy, Christophe. 2009. "Y a-t-il un chien qui aboie … dans les didascalies ?" Theatrical discourse and enunciation. Review of Interdisciplinary Centre for Studies of Contemporary Discursive Forms(4). 123-146. Potts, Christopher. 2005. The logic of conventional implicatures. Oxford: Oxford University Press. Prat-Sala, Mercè & Holly P. Branigan. 2000. Discourse constraints on syntactic processing in language production: A cross-linguistic study in English and Spanish. Journal of Memory and Language 42(2). 168-182. Prince, Ellen F. 1978. A comparison of wh-clefts and it-clefts in discourse. Language 54(4). 883-906. Prince, Ellen F. 1981. Toward a taxonomy of given-new information. In Peter Cole (ed.), Radical Pragmatics, 223-255. New York: Academic Press. Proust, Marcel. 2008 [1913]. Du côté de chez Swann. Malesherbes: Gallimard. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech & Jan Svartvik. 1985. A comprehensive grammar of the English language. London: Longman. Rando, Emily & Donna Jo Napoli. 1978. Definites in there-sentences. Language 54(2). 300-313. Reeve, Matthew. 2010. Clefts. University College London PhD dissertation. Reeve, Matthew. 2011. The syntactic structure of English clefts. Lingua 121(2). 142-171. Reeve, Matthew. 2012. Clefts and their relatives. Amsterdam: John Benjamins. Reichle, Robert Vincent. 2009. ERP correlates of syntactic focus structure processing: Evidence from L1 and L2 French. In Jane Chandlee, Michelle Franchini, Sandy Lord & Gudrun- Marion Rheiner (eds.), BUCLD 33: Proceedings of the 33rd annual Boston University conference on language development, Vol. 2, 420-431. Somerville, MA: Cascadilla Press. Reichle, Robert Vincent. 2014. Cleft type and focus structure processing in French. Language, Cognition and Neuroscience 29(1). 107-124. Reinhart, Tanya. 1982. Pragmatics and Linguistics: An analysis of sentence topics. Philosophica 27. 53-94. Rialland, Annie, Georges Rebuschi & Jenny Doetjes. 2002. What is focused in C'est XP qui/que cleft sentences in French? In Bernard Bel & Isabelle Marlien (eds.), Speech Prosody, 595-598. Aix-en-Provence: ISCA Archive. Riegel, Martin, Jean-Christophe Pellat & René Rioul. 2011. Grammaire méthodique du français. Paris: Presses Universitaires de France. Ritz, Julia, Stefanie Dipper & Michael Götze. 2008. Annotation of information structure: An evaluation across different types of texts. In Nicoletta Calzolari, Khalid Choukri, Bente Maegaard, Joseph Mariani, Jan Odijk, Stelios Piperidis & Daniel Tapias (eds.), Proceedings of the International Conference on Language Resources and Evaluation (LREC-2008), 2137- 2142. http://www.lrec-conf.org/proceedings/lrec2008/. Roberts, Craige. 1996. Information structure in discourse: Towards an integrated formal theory of pragmatics. In Jae Hak Yoon & Andreas Kathol (eds.), OSU Working Papers in Linguistics 49: Papers in Semantics, 91-136. Columbus: Ohio State University. Roggia, Carlo Enrico. 2008. Frasi scisse in italiano e in francese orale: Evidenze dal C-ORAL-ROM. Cuadernos de Filología Italiana 15. 9-29. Rosch, Eleanor H. 1978. Principles of categorization. In Eleanor H. Rosch & Barbara Lloyd (eds.), Cognition and Categorization, 27-48. Hillsdale NJ: Erlbaum Associates. Rosengren, Inger. 1997. The thetic/categorical distinction revisited once more. Linguistics 35(3). 439-479. Rothenberg, Mira. 1979. Les propositions relatives prédicatives et attributives: Problème de linguistique française. Bulletin de la Société de linguistique de Paris 74. 351-395. Rouquier, Magali. 2007. Les constructions clivées en ancien français et en moyen français. Romania 125. 167-212. Rouquier, Magali. 2015. L'émergence des constructions clivées, pseudo-clivées et liées en français. Paris: Classiques Garnier. Ruwet, Nicolas. 1978. Une construction absolue en français. Lingvisticae Investigationes 2(1). 165- 210. Sabio, Frédéric & Christophe Benzitoun. 2013. Sur les relations entre syntaxe et discours: Dispositifs de la rection et dispositifs macrosyntaxiques. Studia Universitas Babeş-Bolyai Philologia 58(4). 97-110 . Sasse, Hans-Jürgen. 1996. Theticity. Unpublished manuscript

406 References

Sauermann, Antje, Ruth Filik & Kevin B. Paterson. 2013. Processing contextual and lexical cues to focus: Evidence from eye movements in reading. Language and Cognitive Processes 28(6). 875-903. Scappini, Sophie. 2013. Un sous-type de la construction clivée en "c’est…qu": La structure d’enchaînement: "et c’est pour ça que"…. et d'autres exemples. Studia Universitatis Babes Bolyai - Studia Philologia 58(4). 81-95. Scappini, Sophie-Anne. 2006. Etude du dispositif d'extraction en ‘c'est...qu’, différenciation entre une relative en ‘c'est...qu’ et une proposition clivée. Université d'Aix-en-Provence PhD dissertation. Schlesinger, Izchak M. 1995. Cognitive space and linguistic case. Semantic and syntactic categories in English. Cambridge: Cambridge University Press. Schmid, Hans-Jörg. 2010. Does frequency in text instantiate entrenchment in the cognitive system? In Dylan Glynn & Kerstin Fischer (eds.), Quantitative methods in cognitive semantics: Corpus-driven approaches, 101-134. Berlin: De Gruyter Mouton. Schmid, Hans-Jörg & Annette Mantlik. 2015. Entrenchment in historical corpora? Reconstructing dead authors' minds from their usage profiles. Anglia 133(4). 583-623. Schöpp, Frank. 2005. Fokus-Konstruktionen im Italienischen – mit Vergleichen zum Französischen. Beiträge zur Fremdsprachenvermittlung 43. 85-105. Schubö, Fabian, Anna Roth, Viviana Haase & Caroline Féry. 2015. Experimental investigations on the prosodic realization of restrictive and appositive relative clauses in German. Lingua 154. 65-86. Secova, Maria. 2010. Discourse-pragmatic features of spoken French: Analysis and pedagogical implications. London: Queen Mary, University of London PhD dissertation. Serianni, Luca. 2000. Italiano. Torino: Garzanti. Smits, Reinier Johannes Charles. 1989a. Eurogrammar. The relative and cleft constructions in the Germanic and Romance languages. Dordrecht: Foris Publications. Smits, Reinier Johannes Charles. 1989b. The relative and cleft constructions of the germanic and Romance languages. Dordrecht: Katholieke Universiteit Brabant PhD dissertation. Sornicola, Rosanna. 1988. It-clefts and wh-clefts: Two awkward sentence types. Journal of Linguistics 24(2). 343-343. Sornicola, Rosanna. 2010. La rappresentazione delle strutture locativo-esistenziali in un corpus di italiano. Uno studio sull’analizzabilità strutturale del discorso parlato. Vox Romanica 69. 111-140. Stalnaker, Robert. 1974. Pragmatic presuppositions. In Peter Unger (ed.), Semantics and philosophy, 197-213. New York: New York University Press. Stede, Manfred & Sara Mamprin. 2014. Information structure in the Potsdam Commentary Corpus: Topics. In Nicoletta Calzolari, Khalid Choukri, Thierry Declerck, Hrafn Loftsson, Bente Maegaard, Joseph Mariani, Asuncion Moreno, Jan Odijk & Stelios Piperidis (eds.), Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC 2016), 1718-1723. Reykjavik: European Language Resources Association (ELRA). Stefanowitsch, Anatol & Susanne Flach. 2017. The corpus-based perspective on entrenchment. In Hans-Jörg Schmid (ed.), Entrenchment and the psychology of language learning: How we reorganize and adapt linguistic knowledge, 101-127. Berlin: De Gruyter Mouton. Stohlterfoht, Britta, Angela D. Friederici, Kai Alter & Anita Steube. 2007. Processing focus structure and implicit prosody during reading: Differential ERP effects. Cognition 104(3). 565-590. Strawson, Peter F. 1964. Identifying reference and truth-values. Theoria 30. 86-99. Swets, Benjamin, Timothy Desmet, Charles Clifton & Fernanda Ferreira. 2008. Underspecification of syntactic ambiguities: Evidence from self-paced reading. Memory & Cognition 36(1). 201-216. Taylor, John R. 1995. Linguistic Categorization. Prototypes in linguistic theory (2nd ed.). Oxford: Oxford University Press. Tellier, Christine. 2003. Eléments de syntaxe du français: Méthodes d'analyse en grammaire générative (2 ed.). Montreal: Morin. Tomasello, Michael, Malinda Carpenter & Ulf Liszkowski. 2007. A new look at infant pointing. Child Development 78(3). 705-722.

References 407

Tortora, Christina. 1997. The syntax and semantics of the weak locative. Delaware: University of Delaware PhD dissertation. Traxler, Matthew J. 2014. Trends in syntactic parsing: Anticipation, Bayesian estimation, and good- enough parsing. Trends in Cognitive Sciences 18(11). 605-611. Valentini, Ada. 2016. Signor Presidente, sono due anni che aspettiamo. In Anna-Maria De Cesare & Davide Garassino (eds.), Current issues in Romance non-canonical word orders. Syntax - information structure - discourse organization, 205-226. Bern: Peter Lang. Vallduví, Enric & Elisabet Engdahl. 1996. The linguistic realization of information packaging. Linguistics 34(3). 459-520. Van De Velde, Danièle. 1995. Le spectre nominal: Des noms de matières aux noms d'abstractions. Leuven: Peeters. Van De Velde, Danièle. 2005. Les interprétations partitive et existentielle des indéfinis dans les phrases existentielles locatives. Travaux de linguistique 50(1). 37-52. Van Der Auwera, Johan. 1985. Relative 'that': A centennial dispute. Journal of Linguistics 21(1). 149- 179. Van Der Sandt, Rob. 1988. Context and presupposition. London: Croom Helm. Van Der Wal, Jenneke. 2016. Diagnosing focus. Studies in Language 40(2). 259-301. Van Kuppevelt, Jan. 1995. Discourse structure, topicality, and questioning. Journal of Linguistics 31. 109-147. Verlinde, Serge & Thierry Selva. 2001a. Corpus-based versus intuition-based lexicography: Defining a word list for a French learners' dictionary. In Corpus Linguistics 2001, 594-598. Lancaster: http://ucrel.lancs.ac.uk/publications/cl2003/CL2001%20conference/. Verlinde, Serge & Thierry Selva. 2001b. Nomenclature de dictionnaire et analyse de corpus. Cahiers de Lexicologie 79(2). 113-139. Verwimp, Lyan. 2013. Les clivées en il y a. Leuven: KU Leuven MA thesis. Verwimp, Lyan & Karen Lahousse. 2016. Definite il y a-clefts in spoken French. Journal of French Language Studies. 1-28. Ward, Gregory & Betty Birner. 1995. Definiteness and the English Existential. Language 71(4). 722- 742. Wasow, Thomas. 2002. Postverbal behavior. Stanford, CA: CSLI Publications. Wehr, Barbara. 2016. Some remarks on different classifications of cleft constructions. In Anna- Maria De Cesare & Davide Garassino (eds.), Current issues in Romance non-canonical word orders. Syntax - information structure - discourse organization, 147-180. Bern: Peter Lang. Willems, Dominique & Machteld Meulleman. 2010. "Il y des gens ils viennent acheter des aspirines pour faire de l'eau gazeuse". Sur les raisons d'être des structures parataxiques en il y a. In Marie-José Béguelin, Mathieu Avanzi & Gilles Corminboeuf (eds.), La parataxe. Tome 2: structures, marquages et exploitations discursives, 167-184. Bern: Peter Lang. Zimmerman, Malte & Caroline Féry (Eds.). 2009. Information structure. Theoretical, typological, and experimental Perspectives. Oxford: Oxford University Press.

View publication stats