Language Patterns in Spanish and Beyond: Structure, Context and Development 2020017841, 2020017842, 9780367550639, 9781003091790

Citation preview

Language Patterns in Spanish and Beyond

The scholarly articles included in this volume represent significant contributions to the fields of formal and descriptive syntax, conversational analysis and speech act theory, as well as language development and bilingualism. Taken together, these studies adopt a variety of methodological techniques—​ranging from grammaticality judgments to corpus-​based analysis to experimental approaches—​to offer rich insights into different aspects of Ibero-​Romance grammar. The volume consists of three parts, organized in accordance with the topics treated in the chapters they comprise. Part I focuses on structural patterns, Part II analyzes pragmatic ones, and Part III investigates the acquisition of linguistic aspects found in the speech of L1, L2 and heritage speakers. The authors address these issues by relying on empirically rooted linguistic approaches to data collection, which are coupled with current theoretical assumptions on the nature of sentence structure, discourse dynamics and language acquisition. The volume will be of interest to anyone researching or studying Hispanic and Ibero-​Romance linguistics. Juan J.  Colomina-​Almiñana is Instructor of Linguistics and Spanish in the Department of Foreign Languages and Literatures at Louisiana State University, USA. Sandro Sessarego is Associate Professor in the Department of Spanish and Portuguese at the University of Texas at Austin, USA.

Routledge Studies in Hispanic and Lusophone Linguistics Series Editor: Dale A. Koike University of Texas at Austin

The Routledge Studies in Hispanic and Lusophone Linguistics series provides a showcase for the latest research on Spanish and Portuguese Linguistics. It publishes select research monographs on various topics in the field, reflecting strands of current interest. Titles in the series: Interface-​Driven Phenomena in Spanish Essays in Honor of Javier Gutiérrez-​Rexach Edited by Melvin González-​Rivera and Sandro Sessarego Spanish in the United States Attitudes and Variation Edited by Scott M. Alvord and Gregory L. Thompson Spanish in Health Care Policy, Practice and Pedagogy in Latino Health Glenn A. Martínez Los castellanos del Perú historia, variación y contacto lingüístico Edited by Luis Andrade Ciudad and Sandro Sessarego Language Patterns in Spanish and Beyond Structure, Context and Development Edited by Juan J. Colomina-​Almiñana and Sandro Sessarego For more information about this series please visit:  https://​www.routledge. com/ ​ Routledge- ​ S tudies- ​ i n- ​ H ispanic- ​ a nd- ​ L usophone-​ L inguistics/​ b ook-​ series/​RSHLL

Language Patterns in Spanish and Beyond Structure, Context and Development Edited by Juan J. Colomina-​Almiñana and Sandro Sessarego Series Editor: Dale A. Koike Spanish List Advisor: Javier Muñoz-​Basols

First published 2021 by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN and by Routledge 52 Vanderbilt Avenue, New York, NY 10017 Routledge is an imprint of the Taylor & Francis Group, an informa business © 2021 selection and editorial matter, Juan J. Colomina-​Almiñana and Sandro Sessarego; individual chapters, the contributors The right of Juan J. Colomina-​Almiñana and Sandro Sessarego to be identified as the authors of the editorial material, and of the authors for their individual chapters, has been asserted in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. British Library Cataloguing-​in-​Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging-​in-​Publication Data Names: Colomina Almiñana, Juan José, editor. | Sessarego, Sandro, editor. Title: Language patterns in Spanish and beyond: structure, context and development / edited by Juan J. Colomina-Almiñana and Sandro Sessarego. Description: London; New York : Routledge, 2020. | Series: Routledge studies in Hispanic and Lusophone linguistics | Includes bibliographical references and index. Identifiers: LCCN 2020017841 (print) | LCCN 2020017842 (ebook) | ISBN 9780367550639 (hardback) | ISBN 9781003091790 (ebook) Subjects: LCSH: Spanish language–Syntax. | Spanish language–Variation. | Spanish language–Acquisition. Classification: LCC PC4361 .L36 2020 (print) | LCC PC4361 (ebook) | DDC 460–dc23 LC record available at https://lccn.loc.gov/2020017841 LC ebook record available at https://lccn.loc.gov/2020017842 ISBN: 978-​0-​367-​55063-​9  (hbk) ISBN: 978-​1-​003-​09179-​0  (ebk) Typeset in Times New Roman by Newgen Publishing UK

Per a Nicole i Ausiàs (Juan J. Colomina-​Almiñana) Per Antonella (Sandro Sessarego)

Contents

List of contributors 

x

Introduction: patterns in syntax, pragmatics and acquisition in Spanish and beyond 

1

S AN D RO SE SSA R EG O A N D JUA N J.  C O LO MI NA-​A L MIÑANA

PART I

Syntactic patterns 

11

1 Argument and adjunct coordination in Spanish 

13

MI C H AE L WI LSO N A N D RO N G  Y I N

2 The interaction of parasitic gaps and adjunct control in Spanish 

37

K AT I E VAN DYN E

3 “¿Qué traes güey?”: an analysis of Spanish clausal possession  52 C H RI ST I AN RU VA LC A BA

4 Invariable qué-​questions (IQQs) in Spanish 

75

JAV I E R F E RNÁ N D EZ SÁ N C H EZ A N D A LFR EDO GARCÍA PARDO

5 Verbless DP interrogative constructions and enclisis in Galician 

97

B RI AN M.  G RAV ELY, JR . A N D TI MOTH Y M.  GUPT ON

6 Leísmo: one le or two?  AD OL F O AU S Í N A N D FR A N C I SC O J.  FER NÁ N DE Z -​RUBIE RA

122

viii Contents PART II

Pragmatic patterns 

145

7 What’s courteous about leísmo? 

147

AD OL F O  AUSÍ N

8 Verbless exclamatives in Spanish beyond the syntax-​semantics interface: a pragmatic account 

166

ME LV I N G ON ZÁ LEZ-​R I V ER A

9 The meaning of y tó: conventional implicatures in Puerto Rican Spanish 

188

N I E V E S  RI VER A

10 Exploring extended focus and meaning in Chilean Spanish intonational plateau contours 

199

B RAN D ON M . A . RO G ER S, R A JI V R AO A N D MATT HE W BURNER

11 The intonation of yes-​no questions in Basque Spanish 

223

C AROL I NA G O N ZÁ LEZ A N D LA R A R EG LERO

PART III

Acquisition patterns 

243

12 How preschoolers acquire the null-​overt contrast in Mexican Spanish: evidence from production 

245

H AN NAH F O R SY TH E, DA N G R EESO N A N D CRIS T INA S CHMIT T

13 Acquisition of null subjects by heritage children and child L2 learners 

266

MI C H E L E   GO LD I N

14 Demonstratives in Spanish: children’s developing conceptualization of interactive space 

285

NAOMI L .  SH I N A N D JI LL P.  MO R FO R D

15 The production of ser and estar in Catalan/​Spanish bilingual children  AL E JAN D RO C U ZA A N D PED RO G U I JA R RO-​F UE NT E S

302

Contents  ix

16 Frequency and semantic prototypicality in L2 Spanish learners’ dative constructions 

329

DAV I D ABU G A BER

Index 

349

Contributors

David Abugaber is a Ph.D.  candidate in the Department of Hispanic and Italian Studies at the University of Illinois, Chicago. Adolfo Ausín is Fixed Term Associate Professor of Spanish Linguistics in the Department of Romance and Classical Studies at Michigan State University. Matthew Burner is a Ph.D.  candidate in the Department of Spanish and Portuguese at the University of Wisconsin, Madison. Juan J.  Colomina-​Almiñana is Instructor of Linguistics and Spanish in the Department of Foreign Languages and Literatures at Louisiana State University. Alejandro Cuza is Professor of Spanish and Linguistics and Chair of Linguistics at Purdue University. Francisco J. Fernández-​Rubiera is Associate Professor of Spanish Linguistics in the Department of Modern Languages and Literatures at the University of Central Florida. Javier Fernández Sánchez is a Post-​doctoral Fellow in the Instytut Anglistyki i Amerykanistyki at Uniwersytet Gdański. Hannah Forsythe is NSF Post-​ doctoral Fellow in the Computation of Language Lab at the University of California, Irvine. Alfredo García Pardo is Assistant Professor of Spanish and Linguistics at SUNY, Purchase. Michele Goldin is a Ph.D.  candidate in the Department of Spanish & Portuguese at Rutgers, the State University of New Jersey. Carolina González is Associate Professor of Spanish Linguistics in the Department of Modern Languages and Linguistics at Florida State University.

newgenprepdf

Contributors  xi Melvin González-​Rivera is Professor of Spanish Linguistics in the Department of Hispanic Studies at the University of Puerto Rico, Mayagüez. Brian M.  Gravely, Jr. is a Ph.D.  candidate in the Department of Romance Languages at the University of Georgia. Dan Greeson is undergraduate researcher in Linguistics at Michigan State University. Pedro Guijarro-​Fuentes is Full Professor of Spanish Philology at the University of the Balearic Islands. Timothy M.  Gupton is Associate Professor of Spanish Linguistics in the Department of Romance Languages at the University of Georgia. Jill P. Morford is Professor in the Department of Linguistics at the University of New Mexico. Rajiv Rao is Associate Professor of Spanish Linguistics at the University of Wisconsin, Madison. Lara Reglero is Associate Professor of Spanish Linguistics in the Department of Modern Languages and Linguistics at Florida State University. Nieves Rivera is a Ph.D.  candidate in the Department of Spanish and Portuguese at the Ohio State University. Brandon M.A. Rogers is Assistant Professor of Spanish Linguistics in the Department of Classical and Modern Languages at Texas Tech University. Christian Ruvalcaba holds a Ph.D.  in Second Language Acquisition and serves as a research coordinator in the Confluentcenter at the University of Arizona. Cristina Schmitt is Associate Professor in the Department of Linguistics and Germanic, Slavic, Asian and African Languages at Michigan State University. Sandro Sessarego is Associate Professor of Spanish Linguistics in the Department of Spanish & Portuguese at the University of Texas at Austin. Naomi L. Shin is Associate Professor in the Department of Linguistics and the Department of Spanish & Portuguese at the University of New Mexico. Katie VanDyne is a Ph.D.  student in the Department of Spanish and Portuguese at the University of Illinois, Urbana-​Champaign. Michael Wilson is a Ph.D. candidate in the Department of Linguistics at the University of Massachusetts, Amherst. Rong Yin is a Ph.D.  candidate in the Department of Linguistics at the University of Massachusetts, Amherst.

Introduction Patterns in syntax, pragmatics and acquisition in Spanish and beyond Sandro Sessarego and Juan J. Colomina-​Almiñana

Language Patterns in Spanish and Beyond: Structure, Context and Development is a collection of selected articles focusing on a number of linguistic aspects of Ibero-​Romance syntax, pragmatics and acquisition. The studies collected here were originally presented at the Hispanic Linguistics Symposium, which took place at the University of Texas at Austin on October 25–​27, 2018, a conference that was attended by more than 400 scholars, who gave a total of 305 individual and co-​authored presentations. In line with the format of the Routledge series on Studies in Hispanic and Lusophone Linguistics and with the focus of the Hispanic Linguistics Symposium, the papers we selected for this volume expand beyond monolingual varieties of Spanish to include other Ibero-​Romance languages, as well as Spanish contact dialects. The language varieties here analyzed are spoken across Spain, North America, South America and the Caribbean: Galician, Catalan, Catalonian Spanish, Castilian Spanish, Basque Spanish, Canary Islands Spanish, Puerto Rican Spanish, Chilean Spanish, Mexican Spanish and US Spanish. The main objective is to identify and examine a number of structural, contextual and developmental configurations found across the aforementioned varieties. The studies collected here take a closer look at grammatical issues that have long been investigated in the field, and offer new and challenging analyses to cast light on those research topics. In particular, this book provides new perspectives on argument and adjunct coordination, parasitic gaps, patterns of clausal possession, the nature of qué-​questions, verbless DP (Determiner Phrase) interrogative constructions and verbless exclamative structures, morphosyntactic and pragmatic aspects of leísmo, politeness, grammatical resources to convey conventional implicatures, pragmatic strategies to express extended focus intonationally, a variety of prosodic patterns concerning yes/​no questions that are associated with different prosodic contexts, aspects of the acquisition of null-​overt pronominal realizations (across native, heritage and L2 children), the development of demonstrative usage in heritage and native speakers, the acquisition of ser and estar in bilinguals, and the effects of frequency and other linguistic factors on the development of dative constructions in L2 speech.

2  Sandro Sessarego and Juan J. Colomina-Almiñana The volume consists of three parts, organized in accordance with the topics treated in the chapters they comprise. Part I focuses on structural patterns, Part II analyzes pragmatic ones, and Part III investigates the acquisition of linguistic aspects found in the speech of L1, L2 and heritage speakers. The authors address these issues by relying on empirically rooted linguistic approaches to data collection, which are coupled with current theoretical assumptions on the nature of sentence structure, discourse dynamics and language acquisition. The scholarly articles included in this volume represent significant contributions to the fields of formal and descriptive syntax, conversational analysis and speech act theory, as well as language development and bilingualism. Taken together, these studies adopt a variety of methodological techniques—​ranging from grammaticality judgments to corpus-​based analysis to experimental approaches—​to offer rich insights into different aspects of Ibero-​Romance grammar, and thus present a set of linguistic findings resulting in highly significant and innovative theoretical implications. These chapters, therefore, span across a number of theoretical and methodological perspectives, which reflect the richness of approaches characterizing the field of Ibero-​Romance linguistics. Part I focuses on syntactic patterns. It begins with Chapter 1, by Michael Wilson and Rong Yin. The authors show that in Spanish there are two different structural configurations conditioning the realization of coordinated con-​PPs “with-​PPs”: argument con-​PPs and instrumental con-​PPs. By assuming that coordinated structures are of the type (Partee & Rooth, 1983), and by proposing two different internal structures for argument and instrumental con-​PPs, the authors cast light on how in the latter type of coordinated PPs—​which contain multiple functional projections and act as adjuncts—​con can be omitted, while in argument con-​PPs, which only contain one projection, the preposition must be obligatorily spelled out. Chapter  2 further builds on the notion of adjunct. In this study, Katie VanDyne inquires about the distribution of clitic-​and passive-​licensed parasitic gaps in Spanish (Campos, 1991; García Mayo, 1995; Sheehan, 2015). She highlights that these structures are characterized by two asymmetric patterns: (1) in both clitic and passive constructions the parasitic gap antecedent can only be inanimate, since animate antecedents yield ungrammatical sentences; (2)  only passive subjects can license a parasitic gap, active ones cannot. The author accounts for these asymmetries by analyzing the nature of the controller of the adjunct subject. Thus, she points out that the category controlling PRO must always be different from the parasitic gap antecedent. In Chapter  3, Christian Ruvalcaba provides additional insights into the nature of the preposition con “with” in Spanish. The author shows that possessive verbs of the types tener, traer and estar con all share a similar base structure, which contains a stative copula and the comitative preposition con. Tener constructions (e.g. tener dinero “to have money”), which tend to express ownership or permanent possession, can be analyzed as the

Introduction  3 result of con incorporating into the simplex copula ser (Avelar, 2004, 2009; Levinson, 2011). On the other hand, traer “to bring” can be understood as the byproduct of con incorporating into the complex copula estar (Gallego & Uriagereka, 2016), which explains its temporal possession reading. Finally, the last possessive pattern, estar con, is the result of a light preposition p merging with a PP headed by con. In line with this configuration, P incorporates into p, thus spelling out con, which does not incorporate into the verb. Chapter  4, by Javier Fernández Sánchez and Alfredo García Pardo, explores the syntactic patterns of what they label invariable qué-​questions (IQQs) in Castilian Spanish. These constructions consist of two parts, divided by a prosodic break: the left periphery of the construction is characterized by a wh-​element qué, while the rightmost part is a tag, which requires confirmation, ¿qué estás, en la biblioteca? “are you in the library?” The authors present an analysis that accounts for a number of aspects of these constructions, such as the lack of thematic link between qué and the tag, monocausality and the requirement of adjacency between the verb and qué. This being said, IQQs seem to be subject to significant cross-​dialectal variation. Indeed, the authors also point out how adjacency may not be required in other varieties, such as Catalonian Spanish (Contreras & Rocas, 2007). Another study that concerns syntactic patterns and interrogative structures within the Iberian Peninsula is Chapter  5, in which Brian Gravely, Jr. and Timothy Gupton explore the nature of verbless DP interrogative constructions (VDICs) in Galician, u-​lo can? “where is the dog?” This syntactically fixed interrogative construction is minimally comprised of a locative interrogative u “where” and a third person determiner clitic, which may vary according to gender and number features:  lo, la, los, las. In line with Chapter  3, this study also builds on Gallego and Uriagereka’s (2016) analysis of the split overt copula in Romance, which characterizes estar as ser plus a null stative matrix. Thus, the authors propose that a VDIC structure may be conceived of as a DP plus a null v, which is selected by the u head base-​generated in Force. With Chapter  6, Adolfo Ausín and Francisco Fernández-​Rubiera close the part on structural patterns. Their chapter addresses another phenomenon concerning third-​person clitics in Ibero-​Romance: Spanish leísmo, or the use of le(s)—​which in standard varieties corresponds to indirect objects—​to indicate masculine human direct objects (DOs) lo(s), and in some varieties masculine inanimate DOs, or even feminine ones, la(s). A well-​known debate characterizing leísta dialects has to do with whether in these varieties there exists one single le, which is used for both IOs and DOs, or if two different le pronouns are actually available (Bleam, 1999 & Ormazabal & Romero, 2007). The authors point out the existence of what they label the “le-​for-​les” phenomenon, which consists of the common use of le instead of les in dative clitic-​doubling constructions, thus resulting in the frequent lack of agreement between the clitic and its postverbal doubled DP (Cuervo, 1955). By relying on a series of syntactic tests, which are supported by data illustrating so-​ called “fake leísmo” (Fernández-​Ordóñez, 1999), they show how the presence

4  Sandro Sessarego and Juan J. Colomina-Almiñana of le-​for-​les in these dialects can be used as an argument to show that two different le pronouns exist in leísta varieties. Part II concentrates on pragmatic patterns. The first study presented in this part, Chapter  7, deals again with leísmo, but this time from a politeness perspective. In this chapter, Adolfo Ausín elaborates on the so-​called leísmo de cortesia “courteous leísmo,” or the tendencies in some dialects, such as Canary-​Islands Spanish, to prefer the use of le rather than lo, with usted in DO position:  a usted, no le/​lo esperamos “sir, we will not wait for you” (Lorenzo Ramos, 1981). Ausín casts light on this pragmatic pattern by building on Pineda’s (2012) account of accusative/​dative case alternation. Thus, he proposes that the aforementioned le/​lo variation can be conceived as a byproduct of the speaker’s ability to choose either one or the other structure. In particular, Ausín argues that the dative option yields a less affected and thus more courteous interpretation of the argument, which is exactly why speakers select le instead of lo to express a more respectful treatment towards the addressee. Chapter 8, by Melvin González-​Rivera, provides a novel analysis of predicative noun phrases (PredNPs) in Spanish (i.e., un gilipollas este chaval “a jerk this guy”). In contrast with how these structures have been treated by previous authors in relation to the notion of scalar implicature (Zanuttini and Portner, 2003), he shows that they should be understood as verbless exclamatives (Paul and Stainton, 2006). He does so by offering a series of grammatical tests that goes well beyond syntactic and semantic analysis. In particular, by integrating a pragmatic approach to his study, he builds on the distinction between conventional and conversational implicature (Gutiérrez-​ Rexach, 2001) to demonstrate that PredNPs and exclamatives share a variety of structural and discourse properties. Nieves Rivera, in Chapter 9, brings us to Puerto Rico, where she analyzes the construction y tó, the reduced version of y todo “and all,” which is commonly used in Puerto Rican Spanish (and in some other Spanish dialects) to express unexpectedness. Even though this construction is quite common in the local dialect, no linguistic analysis of it had previously been undertaken. The author offers a series of pragmatic tests based on Potts’ (2005) pragmatic framework to show that y tó can be characterized as a conventional implicature. In Chapter  10, Brandon Rogers, Rajiv Rao and Matthew Burner provide a speaker-​ driven and pragmatic account of Chilean Spanish intonational plateau contours. The authors build on previous studies by Rogers (2013, 2016) showing how certain naturally occurring prosodic phenomena may clash with current phonological theories on the nature of intonation (Face, 2014), since such theories have usually been built on tightly controlled mechanisms of data collection and analysis. Conversely, this study brings to the table new and challenging data, which proceed from a large corpus of naturalistic interviews. After analyzing a set of intonational plateau contours in line with Gussenhoven’s (2002) Effort Code, Rogers, Rao and Burner

Introduction  5 propose that these prosodic configurations represent instances of extended focus, which can be conceived as an economic pragmatic strategy to emphasize strings of words that form thoughts. Chapter 11 closes the part on pragmatic patterns with a study by Lara Reglero and Carolina González, who provide an account of prosodic configurations in yes-​no questions in Basque Spanish. They analyze four different pragmatic contexts:  (1) information-​seeking; (2)  confirmation; (3)  surprise; and (4)  requests. In line with previous findings (Elordieta & Hualde, 2014), this work shows that the information-​seeking yes-​no questions in Basque Spanish tend to display a final circumflex contour. Nevertheless, this study also offers novel results for contexts (2–​4), which had never been analyzed for this dialect before. In particular, the authors detect high rates of elevated high tones in nuclear configurations for surprise and confirmation questions. The chapter also brings to light significant differences between Spanish-​dominant and Basque-​dominant participants: the first group presents more variation between rising and circumflex contours and displays more elevated high tones overall. The part on developmental patterns begins with Chapter 12, by Hannah Forsythe, Dan Greeson and Cristina Schmitt. The author focuses on the acquisition of the null-​overt pronominal subject contrast by Spanish-​speaking preschoolers in Mexico City. This study, in line with recent research in the field of Spanish L1 acquisition (Shin, 2012, 2016), casts light on how children acquire variable subject expression. After analyzing a corpus of naturally occurring children speech. Forsythe, Greeson and Schmitt propose that children acquire the overt-​covert contrast by initially focusing their attention on 1st-​and 2nd-​person pronouns, whose referents are easier to infer in naturalistic discourse. They appear to be able to master such contrast proficiently by the age of four and a half and generalize it to the 3rd-​person pronouns only later, by the age of six. The acquisition patterns of this overt-​covert alternation are also analyzed in Chapter 13, in which Michele Goldin explores its development in the speech of two different child populations (between four and seven years of age) in a Spanish immersion school in the US: heritage speakers of Spanish and L2 Spanish learners. These groups completed an acceptability judgment task, which was carried out to investigate their understanding of null subjects in both English and Spanish. Results show that both bilingual groups performed lower in English than a control group of monolingual English speakers, while no significant difference could be detected in their Spanish performance. These lower scores in English are explained as a prolonging of the null-​subject stage, which is attributed to the presence of Spanish input in the children’s linguistic environment, and thus to cross-​linguistic effects. The lack of a significant difference between the two groups, as far as Spanish subject pronouns are concerned, is ascribed to a delay in the development of the syntax-​pragmatic interface, which affects both heritage speakers and L2 learners transversally (Hulk & Müller, 2000; Sorace & Filiaci, 2006). Such a delay is also considered responsible for the lack of a parallel evolution between verbal morphology

6  Sandro Sessarego and Juan J. Colomina-Almiñana and pronominal usage in Spanish for both of the bilingual groups analyzed. Conversely, this is a correlation that has been found to characterize the linguistic development of children’s English skills, since the usage of subject pronouns in English is not as conditioned by discourse factors as in Spanish. In Chapter 14, Naomi Shin and Jill Morford further explore patterns of language development in heritage Spanish child speech in the US. In this case, they analyze the acquisition of demonstratives. By relying on a puzzlecompletion task to elicit the data, the authors showed how six-​to eight-​year old children have developed a clear distinction between the determiners esta “this” and esa “that” when referring to pieces located at different distances. On the other hand, this is a skill that younger children, between three and five, have not yet mastered. The experiment also reveals that adults, in contrast with both groups of children, use demonstratives to convey intersubjective space, which is expressed by the increasing use of esta to overcome a lack of shared attentional focus. In line with Küntay and Ozyürek’s (2006) study on Turkish, this chapter shows that in Spanish there is also a developmental trajectory in the acquisition of determiners. Children first learn how to encode the proximal-​distal dimension and only in a second phase do they manage to integrate the subjectively constructed space into their use of demonstratives. Chapter  15, by Alejandro Cuza and Pedro Guijarro-​Fuentes, analyzes the acquisition of verbs ser and estar in Spanish and Catalan among simultaneous bilingual children. The study builds on Arnaus Gil et al.’s results (2018) and shows that, due to input variability and Spanish influence, estar constructions tend to be overgeneralized to Catalan locatives and adjectives. Conversely, as far as Spanish is concerned, findings indicate that the estar use with adjectives and locatives is not significantly affected by Catalan grammatical effects. Moreover, in both languages, bilingual children tend to present high rates of estar with event locatives. The authors analyze these results in terms of contact-​driven effects as well as language-​internal developments. In particular, they suggest that, due to the semantic underspecification of ser and its relatively reduced frequency in the input, it takes longer for children to approximate an adult-​like competence for the copula constructions requiring its use. Chapter  16 closes the part on developmental patterns and with it the entire volume. In this study, David Abugaber relies on the written corpus of Spanish as a second language (Corpus Escrito del Español como Segunda Lengua; CEDEL2) to cast light on the acquisition of verbs that are used with the indirect-​object clitic le. A comparison with texts produced by native speakers (also contained in CEDEL2) shows that the most common verbs in L2 writing are dramatically more frequent than the corresponding forms in the native corpus. Moreover, L2 learners rely more heavily on verbs whose meanings are semantically prototypical for the different uses of le: dar, gustar and decir, for benefactive, psychological and epistemic uses respectively. These findings are explained in light of usage-​based models of L2 acquisition

Introduction  7 (Ellis, 2006a, 2006b), which stress the importance of frequency and semantic prototypicality in language development, especially in the early stages of L2 acquisition. Besides providing empirical support for this theoretical account, these findings are also relevant to the applied side of L2 acquisition, since they can be used to inform pedagogical strategies for the teaching of Spanish as a second language. In conclusion, this book provides a series of structural, discourse and developmental analyses that highlight a number of linguistic patterns across monolingual and contact varieties of Spanish, Galician and Catalan. In so doing, the present volume offers a collection of articles by well-​known scholars as well as promising researchers, whose work points to a cohesiveness in the field of Ibero-​Romance linguistics that is built on the incorporation of up-​to-​date empirical research into current theoretical proposals on the nature of syntax, pragmatics and language acquisition. We feel honored to have worked with all of them and hope this book will serve as a source of inspiration for future research and collaboration in the field.

References Arnaus Gil, L., Jiménez Gaspar, A., & Müller, N. (2018). The acquisition of Spanish SER and ESTAR in bilingual and trilingual children:  Delay and enhancement. In A. Cuza & P. Guijarro-​Fuentes (eds), Language acquisition and contact in the Iberian peninsula, 91–​124. Berlin: Mouton de Gruyter. Avelar, J. (2004). Dinâmicas morfossintáticas com ter, ser e estar em português brasileiro (Master’s thesis). Universidade Estadual de Campinas. Avelar, J. (2009). On the emergence of TER as an existential verb in Brazilian Portuguese. In P. Crisma & G. Longobardi (eds.), Historical syntax and linguistic theory, 158–​175. Oxford: Oxford University Press. Bleam, T. (1999). Leísta Spanish and the syntax of clitic doubling (Doctoral dissertation). University of Delaware. Campos, H. (1991). Silent objects and subjects in Spanish. In H. Campos & F. Martínez-​Gil (eds.), Current studies in Spanish linguistics, 117–​141. Washington, DC: Georgetown University Press. Contreras, J.M., & Roca, F. (2007). Un tipus especial d’oracions interrogatives:  les interrogatives escindides. Caplletra, 42, 145–​184. Cuervo, R.J. (1955). Apuntaciones críticas sobre el lenguaje bogotano, con frecuente referencia al de los países de Hispano-​América (9th ed.). Bogotá: Instituto Caro y Cuervo. Ellis, N.C. (2016a). Frequency in language learning and language change. In H. Behrens & S. Pfänder (eds.), Experience counts: Frequency effects in language, 239–​256. Berlin: de Gruyter. Ellis, N.C. (2016b). Online processing of verb–​argument constructions: Lexical decision and meaningfulness. Language and Cognition, 8 (3), 391–​420. Elordieta, G., & Hualde, J.I. (2014). Intonation in Basque. In S-​A. Jun (ed.), Prosodic typology II:  The phonology of intonation and phrasing, 405–​463. Oxford:  Oxford University Press.

8  Sandro Sessarego and Juan J. Colomina-Almiñana Face, T. (2014). Sp_​ToBI and the phonological analysis of Spanish intonation: A critical perspective. Studies in Hispanic and Lusophone Linguistics, 7 (1), 185–​210. Fernández-​ Ordóñez, I. (1999). Leísmo, laísmo y loísmo. In I. Bosque & V. Demonte (eds.), Gramática descriptiva de la lengua española, 1317–​1398. Madrid: Espasa Calpe. Gallego, Á., & Uriagereka, J. (2016). Estar = Ser + X. Borealis, 5 (1), 123–​156. García Mayo, M.P. (1995). The licensing of parasitic gaps. Miscelánea: A Journal of English and American Studies, 16, 125–​140. Gussenhoven, C. (2002). Intonation and interpretation: Phonetics and phonology. In B. Bel & I. Marlien (eds.), Proceedings of Speech Prosody 2002, 47–​57. Aix-​en-​ Provence: Laboratoire Parole et Langage. Gutiérrez-​Rexach, J. (2001). Spanish exclamatives and the semantics of the left periphery. In Johan Rooryck, Y.d. Hulst & J. Schroten (eds.), Selected papers from ‘Going Romance’, 99, 167–​194. Amsterdam/​Philadelphia: John Benjamins. Hulk, A., & Müller, N. (2000). Bilingual first language acquisition at the interface between syntax and pragmatics. Bilingualism:  Language and Cognition, 3 (3), 227–​244. Küntay, A.C., & Ozyürek, A. (2006). Learning to use demonstratives in conversation:  What do language specific strategies in Turkish reveal? Journal of Child Language, 33, 303–​320. Levinson, L. (2011). Possessive WITH in Germanic: HAVE and the role of P. Syntax, 14, 355–​393. Lorenzo Ramos, A. (1981). Algunos datos sobre el leísmo en el español de Canarias. In M. Alvar (ed.), I Simposio Internacional de Lengua Española [1978], 175–​180. Las Palmas: Ed. del Excmo. Cabildo Insular de Gran Canaria. Ormazabal, J., & Romero, J. (2007). The object agreement constraint. Natural Language & Linguistic Theory, 25 (2), 315–​347. Partee, B., & Rooth, M. (1983). Generalized conjunction and type ambiguity. In R. Bäuerle, C. Schwarze & A. von Stechow (eds.), Meaning, use and interpretation of language, 361–​383. Berlin: de Gruyter. Paul, I., & Stainton, R. (2006). Really intriguing, that PredNP! Philosophy publications. Available online: http://​westernlinguistics.ca/​Publications/​CLA2006/​ Pineda, A. (2012). Double object constructions and dative /​accusative alternations in Spanish and Catalan: A unified account. Borealis – An International Journal of Hispanic Linguistics, 2 (1), 57–​115. Potts, C. (2005). The logic of conventional implicatures. New  York:  Oxford University Press. Rogers, B.M.A. (2013). The extent of tonal events: Intonational hat patterns in Chilean Spanish. Estudios de fonética experimental, 22, 171–​192. Rogers, B.M.A. (2016). When theory and reality collide:  Exploring Chilean Spanish intonational plateaus (Unpublished doctoral dissertation). University of Minnesota. Sheehan, M. (2015). Subjects, null-​subjects and expletives in Romance. In S. Fischer & S. Gabriel (eds.), Manuals of Romance linguistics (MRL): Grammatical interfaces. Berlin: de Gruyter. Shin, N.L. (2012). Variable use of Spanish subject pronouns by monolingual children in Mexico. In K.L. Geeslin & M. Díaz-​Campos (eds.), Selected proceedings of the 14th Hispanic Linguistics Symposium, 130–​141. Somerville, MA: Cascadilla Proceedings Project.

Introduction  9 Shin, N.L. (2016). Acquiring constraints on morphosyntactic variation:  Children’s Spanish subject pronoun expression. Journal of Child Language, 43 (4), 914–​947. Sorace, A., & Filiaci, F. (2006). Anaphora resolution in near-​native speakers of Italian. Second Language Research, 22, 339–​368. Zanuttini, R., & Portner, P. 2003. Exclamative clauses at the syntax-​semantics interface. Language, 79 (1), 39–​81.

Part I

Syntactic patterns

1 Argument and adjunct coordination in Spanish* Michael Wilson and Rong Yin

1.  A puzzle: Spanish coordination with(out) P repetition Spanish shows an interesting asymmetry in coordination possibilities in the context of PPs. Consider the following sentences: (1) a. Me  caso   con  Juan o  *(con) Pedro.    1s.C L   1s.marry  with   Juan or  *(with Pedro “I will marry Juan or Pedro.”   b. Se  puede pagar con  tarjeta de crédito o ?(con) la  app    3.C L 3s.can   pay  with credit card    or ?(with the  app “One can pay by credit card or with the app.” In some cases, repeating the preposition is obligatory, while not repeating is ungrammatical, as in (1a). In other cases, such as (1b), both repeating and not repeating the preposition are possible (though repeating it is preferred, perhaps for prescriptive reasons). A Google search revealed naturally occurring examples of the latter type without P-​repetition: (2) a. Si van a pagar con  tarjeta de crédito  o  cheque de viajero,      if  3p.go to  pay   with credit card    or  traveler’s check    cuenten=le        al  gerente        3p.tell.S B JV =3sDAT.C L     to.the manager “If you will be paying with a credit card or a traveler’s check, please inform the manager.” (www.lanacion.com.co/​2016/​03/​18/​policia-​lanzo-​plan-​semana-​santa-​segura-​y-​en-​paz/​)    b. Las  perlas  se  pueden  pegar  con  pegamento  o  cinta  adhesiva      the  pearls 3.C L 3p.can  stick  with glue       or tape   adhesive      en las caras de distintas cajas.  on   the   sides  of different boxes “The pearls can be attached to the sides of different boxes with glue or tape.” (https://​books.google.com/​books?id=A4U2KRbLVioC&pg=PA161)

This differs from English, where it is always possible to not repeat the preposition:

14  Michael Wilson and Rong Yin (3)  a. I dealt with the problem or (with) the solution. b. You can pay with a credit card or (with) the app. An additional fact is that in (1a), the con-​PP is an argument of the verb. When casarse takes an internal argument, it must be a con-​PP: (4) a. * Me   caso     Juan. 1s.C L  1s.marry Juan    b.  * Me   caso    {a   /​por medio de /​junto   con}  Juan.  1s.C L 1s.marry {to by means of  together   with   Juan The DP inside the con-​PP also combines in a semantically idiosyncratic way with the verb; that is, con does not carry the meaning of instrumenthood or comitativity that it does when it introduces an adjunct, as in (1b). Note in particular that con here cannot be paraphrased with alternative phrases that introduce instruments (por medio de) or comitative accompaniers (junto con). Here, we focus on deriving the contrast in the case of instrumental con, though we offer some speculation about how our analysis might extend to comitative con in our conclusion. We propose the following in order to explain the facts in (1)–​(3): argument PPs are syntactically and semantically simple, while instrumental adjunct PPs are syntactically and semantically complex. Assuming a theory of coordination à la Partee and Rooth (1983), this explains the contrast in (1): (1a) allows for only one kind of coordinate structure in this theory, which involves Gapping in the second disjunct; while (1b) allows for two kinds of coordinate structures, one of which is like that available for (1a), and the other of which is possible because of the more complex structure of instrumental PPs. The first option results in P-​repetition for both sentences, since the contrasted term must move to escape being Gapped; pied-​piping is obligatory in Spanish, so the preposition must move as well. The second option for (1b), due to the more complex structure of instrumental con-​PPs, involves coordination within the functional projection of the PP; this “low” coordination results in the variant of (1b) without P-​repetition. To explain the contrast between Spanish and English in (1a) and (3a), we appeal to the fact that pied-​piping is obligatory in Spanish and optional in English, so it is possible for the contrasted term to move out of the ellipsis site in English without carrying along the preposition. We also briefly offer an explanation about why the usual prohibition of P-​stranding under Gapping in English is circumvented in these kinds of structures. The rest of the chapter is organized as follows:  in §2, we present our assumptions about coordination. §3 presents our analysis of the syntax and semantics of adjunct con-​PPs, and §4 presents how to derive (1b) taking into account the results of §§2–​3. In §5, we present how to derive (1a) and (3) taking into account §2 along with the difference between pied-​piping in English and Spanish; we also present in this section our explanation of why P-​stranding is possible in these cases of Gapping in English. §6 concludes.

Argument and adjunct coordination in Spanish  15

2.  A quick note on coordination As a baseline, we assume a theory of coordination with the following characteristics. To start, coordinators have one basic semantic type:  ⟨⟨s, t⟩, ⟨⟨s, t⟩, ⟨s, t⟩⟩⟩, where s is the type of events, and t the type of truth-​values. Coordinators may conjoin other types by means of type-​shifters; however, type-​shifters are costly, and can only be used as a last resort (Partee & Rooth, 1983).1 These two facts will create a strong preference for coordinators to operate on constituents of type ⟨s, t⟩ if this is possible (cf. Hirsch, 2017). In addition, we assume that there are two ands in English, and two y-​es in Spanish: one is the logical conjunction operator “∧,” which has the properties described above. The other is a sum formation operator “⊕,” which forms non-​atomic entities (Link, 1983). To avoid the potential confounds of the existence of this homophonous sum formation operator, we use disjunction in our examples, for which no such confound exists.

3.  Complex PPs It has been widely noted that spatial PPs are internally complex, consisting of a number of distinguishable functional projections (Cinque & Rizzi, 2010; Radkevich, 2010; Roy & Svenonius, 2009; Svenonius, 2007, 2010, among others). One motivation for this can be seen in the distinction between Place PPs and Path PPs. The former specify the location of an object in a place, while the latter specify trajectories relative to a place. (5) a. b. c. d.

The elephants remained in the boat. They cast a wistful glance to the shore. The boat drifted further from the beach. Their ears sank down several notches.

(Place) (Pathgoal) (Pathsource) (Path↓) (Svenonius 2010, (1))

Path and Place heads are not mutually exclusive; they can co-​occur, and when they do, Path occurs outside of Place. (6) a. ná gmá tábèl to on table “onto the table”

(Zina Kotoko) (Holmberg, 2002, cited in Svenonius, 2010)

b.

16  Michael Wilson and Rong Yin This sort of structure makes sense semantically, as paths are always specified relative to a place. In some cases, a preposition may lexicalize Path and Place simultaneously (Fábregas, 2007): (7) a. Silenciosamente flotaba hacia la puerta. Silently 3s.floated.imp towards the door “She floated silently towards the door.” (Mario Vargas Llosa, La tía Julia y el escribidor, p. 261; cited in Fábregas, 2007, p. 170, (7)) b.

c.

In (7c), the complex head consisting of [haciapath haciaplace] gets spelled out as the single surface word hacia.2 In some cases, this kind of incorporation of a Place head into a Path head is more phonologically obvious. For example, consider English into, which might have the following structure and derivation: (8) a. The cat walked intopath+place the room. b.

c.

Argument and adjunct coordination in Spanish  17 3.1  Decomposing con/​with We propose that something like (8b–​c) is the right structure for non-​argument con/​with. While it makes little sense semantically to decompose this instrumental con/​with into a Path and a Place, it has a similarly complex meaning, though of a different sort. We begin by considering a simplified example of instrumental con in (9): (9) Juan paga la cuenta con la tarjeta de crédito. Juan 3s.pay the bill with the credit card “Juan pays the bill with the credit card.” It has been widely noted that vPs modified by instrumental con/​with have a complex event structure, which involves relationships between at least three distinct events (Jerro, 2017; Koenig et al., 2007; Mari, 2006; Rissman 2013): • • • •

e1: The subject acts on the instrument. e2: The instrument acts on the theme. e3: The theme is affected. e1 causes e2, which causes and/​or facilitates3 e3.

In other words, the meaning of a sentence like (9) could be (preliminarily and schematically) semantically represented as follows: (10)  ∃e,e1,e2,e3: (e1 ∈ e) ∧ (e2 ∈ e) ∧ (e3 ∈ e) ∧ (e) AG (e, j) ∧ pay(e) ∧ TH (e, the.bill′) ∧ ( e1) AG (e1, j) ∧ TH (e1, the.CC′) ∧ CAUS E( e1, e2) ∧ ( e2) AG( e2, the.CC′) ∧ TH (e2, the.bill′) ∧ CAU SE (e2, e3) ∧ (e3) TH (e3, the.bill′) ∧ pay(e3) Existing approaches to the semantics of instrumentals tend to encode the entire complex relationship between e1, e2, and e3 in one functional head. For example, Jerro (2017) gives the following semantics for the Kinyarwanda morpheme –​ish, which has instrumental and causative uses (slightly adapted for presentational purposes): (11) ⟦ -​ish ⟧ = λx.λP⟨e,st⟩.λy.λe.P(e, y) ∧ ∃e′: e′ ⊂ e ∧ AG (e′, x) Jerro assumes a theory of event structure wherein verbs describe events that have causally related subparts. In other words, what it means to say that e′ ⊂ e is to say that e′ is part of the causal chain of e. If it is the initial event in e, it causes the second event in e, which causes the third, and so on. If it is not the initial event, it is caused by event e′-​1, and causes e′+1, unless it is the final event in e.

18  Michael Wilson and Rong Yin Thus, while the lexical entry here only appears to make reference to two events, rather than the three posited above, the third event is implicit in the meaning of “⊂” as used here. What this lexical entry does, then, is take an entity argument, the instrument, and an ⟨e,st⟩ function P. The result describes events of P which contain an event that has x as its agent. In other words, this says that x is an agent of a subevent in the causal chain of an event of P. Assuming con/​with have the same meaning as  –​ish, we get (12) as the semantics for (9): (12)  λe.pay(e) ∧ TH (e, the.bill′) ∧ AG (e, j) ∧ ∃e′: e′ ⊂ e ∧ AG (e′, the.CC′) This describes events of paying the bill with Juan as their agent, which contain subevents in their causal chains that have the credit card acting as an agent. Given the definition of a causal chain, this means that the credit card will ultimately cause the bill to be paid. We propose decomposing Jerro’s denotation of with into simpler parts, much like the decomposition of spatial Ps presented earlier. In particular, we note that there is already a head that introduces an agent argument:  v° (Kratzer, 1996). We therefore propose that con/​with has the following structure and semantics: (13) 

All we have done here is taken Jerro’s denotation for –​ish, and parceled out its meaning into a part that introduces an event with an agent (v°)4,5 and a part that says that event is part of the causal chain of another event (C AU SE -​ PA RT °). C AU S E -​PA RT P will then combine with a projection of the main clause’s v°, as in (14). (14)

Argument and adjunct coordination in Spanish  19 We assume that con/​with are the spellout of a complex head [CAUSE-​PART v°], following head movement of v°:6,7,8 (15)

4.  Adjunct coordination with(out) P-​repetition: vP coordination Now let us take stock of what we have done in the context of our theory of coordination. Assuming coordination can apply to any node on a tree, we have introduced a new locus of coordination inside instrumental con/​with-​ phrases: the vP below CAUS E-​PA RT P. This phrase is of type ⟨s, t⟩, and so can be coordinated without using type shifters—​in other words, without inducing a cost (recall §2). We re-​present the structure and semantic types in (16): (16) 

Coordinating the vP in (16) would result in the following structure: (17) 

20  Michael Wilson and Rong Yin How is such a structure pronounced? Given that con is the spellout of the complex head [C AUS E-​PA RT v°], we propose that v° across-the-board moves to incorporate into C AUS E-​PA RT : (18)

The resulting head is spelled out as con, but since there is only one instance of [C AUS E-​PA RT v°], con is only spelled out once. This leads to the case in (1b) when the P is not repeated. How can we derive the structure where con is repeated? Given the types in (16), we cannot coordinate CAUS E-​PA RT Ps except as a last resort, since they are not of type ⟨s, t⟩. Instead, we can coordinate the main clause’s vP, since this forms a constituent of type ⟨s, t⟩: (19)

Argument and adjunct coordination in Spanish  21 Coordinating this vP would give us the following: (20) 

How do we avoid pronouncing the subject and VP twice? We follow Schwarz (1999) and propose that Gapping applies to the second conjunct.9 Gapping is a process that can apply in coordinate structures to elide material from a repeated vP: (21) Álex toca el violín y Marta △ el piano. Alex 3s.play the violin and Marta the piano “Alex plays the violin and Marta the piano.” (Brucart & MacDonald, 2012, p. 582, (10)) One kind of analysis of Gapping treats it as ellipsis of a vP constituent from which contrasted terms have moved (e.g., Johnson, 2014; Johnson, 2018); we will adopt this treatment. We also assume a “small conjuncts” account of Gapping; see Johnson (2018), §4.1 for arguments in favor of this view, which we avoid repeating here for reasons of space.10 Note that this analysis will require us to assume that Gapping can elide a vP, but that it is not vP-​ellipsis, as Spanish lacks that process (we thank a reviewer for pointing this out to us). It seems clear though, that despite its being able to elide vPs, Gapping cannot be the same process as vP-​ellipsis. We leave aside the question of what exactly it is that is responsible for this mysterious difference between the two processes, though see Johnson (2009) for an argument about what this difference is, which we believe the present account could easily incorporate. We do not present our analysis in terms of Johnson’s (2009) because its predictions do not differ from the simpler analysis we will assume with regards to the facts we examine here. We follow Schwarz (1999) and assume that Gapping can produce structures with only one contrasted term.11 In the case of (20), that will mean that in order to pronounce the contrasted DP la app, it must move out of the gapped constituent. Because pied-​piping of prepositions is obligatory in Spanish,

22  Michael Wilson and Rong Yin when la app moves to escape the gapped constituent, it must pied-​pipe the C AUS E-​PA RT P with it. (22)

Within each disjunct, movement of v° to C AU SE -​PART ° occurs, leading to con being spelled out twice. Thus, both the structures without repeated Ps following (18), and the structures with repeated Ps following (22) are generated. (We leave aside the question of whether C AU SE -​PART P undergoes a similar movement in the first conjunct in cases like (22), compatible with a syntactic identity condition on ellipsis; or stays in situ, compatible with a semantic identity condition on ellipsis.)

5.  Argument coordination: also vP coordination We now turn to the question of why argument coordination requires P-​ repetition. Given that argument Ps are idiosyncratically selected and lack their usual semantic meaning, we assume they are structurally simple: (23)

We remain agnostic on the semantic type of these PPs, except to note that they are clearly not of type ⟨s, t⟩.12 What this means is that there is no counterpart to (18) for argument PPs. Instead, argument coordination is similar to (20): it is (matrix) vP coordination, which avoids the need for a type-​shifter. This is shown in (24):

Argument and adjunct coordination in Spanish  23 (24)

As before, material in the second conjunct can be gapped. In order for the contrasted DP, Pedro, to escape the ellipsis site, it moves out. Since pied-​ piping is obligatory in Spanish, con comes with it, as in (25): (25)

With this in mind, it is interesting to compare Spanish to English. In English, coordination of prepositional arguments allows for not repeating the P, as shown in (3a), repeated as (26): (26) I dealt with the problem or (with) the solution. (=(3a)) We assume that the same constraints on coordination in Spanish hold for English. Instead, the difference between the two reduces to the obligatoriness of pied-​piping in Spanish. In Spanish structures like (25), con must move out of the ellipsis site because pied-​piping is obligatory in Spanish. In an equivalent English sentence, the preposition may remain in the ellipsis site, because pied-​piping is optional. This is shown in (27):

24  Michael Wilson and Rong Yin (27)

5.1  The P-​stranding restriction on English Gapping13 Our analysis of (26) in terms of Gapping initially seems to cause a problem, given the generalization that English usually prohibits P-​stranding in Gapping structures:14 (28) a. Charley writes with a pencil and John writes with a pen. b. * Charley writes with a pencil and John writes with a pen. (Hankamer, 1979, p. 18, cited in Johnson, 2014, p. 12, (41a–​b)) If this is so, why can with be stranded in (27), leading to its not being pronounced twice, when it cannot be so stranded in (28)? We propose that two facts explain the contrast between (27) and (28): first, Abe and Hoshi’s (1997) Crossing Constraint on Gapping (adapted from Pesetsky, 1982); and second, the fact that the availability of P-​stranding depends on whether movement is short or long:  short movement disallows P-​stranding, while sufficiently long movement allows it. Let us first discuss the Crossing Constraint on Gapping: (29) Crossing Constraint on Gapping (CCG): A contrasted element cannot cross another contrasted element. (Abe & Hoshi, 1997; Pesetsky, 1982)

(30) Crossing: A crosses B if B asymmetrically c-​commands one of A’s traces, and A asymmetrically c-​commands B.15

Argument and adjunct coordination in Spanish  25 Consider how (29) rules out the ungrammatical gap in (31): (31) a. * John dealt with Mary, and with Bill, Susan. b. * John dealt with Mary, and [[with Bill]i [Susan [dealt ti]]] Here, both with Bill and Susan are contrasted elements in the coordinate with Gapping. Susan asymmetrically c-​commands the trace of with Bill, but with Bill asymmetrically c-​commands Susan (shown in (31b)). As such, with Bill crosses Susan. Since these are both contrasted elements, (29) rules out the structure in (31b).16 The consequence of this is that in structures involving two contrasted terms, one of which is a subject, and the other a PP, the PP will not be able to move to a higher position than the subject. In other words, a PP that moves out of a gapped constituent will have to undergo a short movement inside the vP that cannot take it above the subject. This fact matters because there is independent evidence that P-​stranding is disallowed for short movements of this sort, even in languages which allow it in long movements (here, movements above Spec,vP). For instance, Object Shift in English and Icelandic disallows P-​stranding (Holmberg, 1986; Johnson, 1991; Thráinsson, 2001), while this is allowed in longer movements such as wh-​ movement, tough-​movement, and topicalization, as well as in pseudo-​passives (Jónsson, 2008; Maling & Zaenen, 1985, 1990; Thráinsson, 2001, 2007). (32) a. I will put up the book. b. I will put the booki up ti. c. I will put up with the book. d. * I will put the booki up with ti. (33) a. b. c. d.

Whati will you put up with ti? That booki is hard to put up with ti. Thati, I won’t put up with ti. Old textbooksi are only barely put up with ti by students.17

(34) Icelandic: a. Þá máluðu allir strákarnir stundum bílana rauða. then painted all boys-​the sometimes cars-​the red b. Þá máluðu allir strákarnir bílanai stundum ti rauða. then painted all boys-​the cars-​the sometimes red “Then all the boys sometimes painted the cars red.” (Thráinsson, 2001, p. 153, (15a–​b)) c. Jón talaði ekki [PP við  Maríu]. Jon spoke not  [PP with  Mary “John didn’t speak to Mary.” d. * Jón talaði Maríui ekki [PP við ti]. (Thráinsson, 2001. p. 151, (6a–​b))

26  Michael Wilson and Rong Yin (35)  Icelandic: a. Hvaði eruð þið að tala um What are you to talk about “What are you (pl.) talking about?”

ti? (Jónsson, 2008, p. 404, (1a))

b. Sigrúnui hef ég aldrei talað við ti. Sigrun have I never spoken to “S igrun, I have never spoken to.” (Thráinsson, 2007, p. 345, (7.6b)) For example, (32a–​b) show that Object Shift is possible across a particle for a DP in English, while (32c–​d) show that this cannot occur if such movement would strand a preposition.18 (33) shows various examples of longer movements which do allow P-​stranding in English. For Icelandic, (34a–​b) shows Object Shift of a DP bílana “the cars” across an adverb stundum “sometimes,” while (34c–​d) show that Object Shift is blocked if the moved DP would strand a preposition. (35) shows examples of longer movements that allow P-​stranding in Icelandic. These facts show that short movements disallow P-​stranding, even if long movements allow it.19 We can form even closer pairs by comparing verbal to nominalizing uses of –​ing in English. One way in which these differ is that the verbal use of –​ing introduces objects normally, while the nominal –​ing requires using of to do so. Object Shift can only occur with verbal –​ing: (36) a. John hastily looking up the reference bothered me. b. John hastily looking the referencei up ti bothered me. c. John’s hasty looking up of the reference bothered me. d. * John’s hasty looking the referencei up of ti bothered me. The meanings of the verbal and nominal structures here seem to be very similar. Holmberg (1986, p. 199) gives a similar near minimal pair for Swedish tro, which can occur with or without a preposition (he does not offer an English free translation for these examples): (37) a.

Jag tror inte på det. I believe not in it b. * Jag tror deti på inte ti. c. Jag tror deti inte ti. (Holmberg, 1986, p. 199, (118a–​b); (121))

Argument and adjunct coordination in Spanish  27 We again see that this sort of short movement cannot strand a preposition. But when there is no preposition, Object Shift is possible. We have now shown that when there are two contrasted terms in a constituent with a gap, and one of those terms is the subject, any contrasted PP can only undergo short movement, due to the CCG. Short movement does not allow P-​stranding. This accounts for why P-​stranding is disallowed when a subject is contrasted in a gapped constituent, even though English generally allows P-​stranding. Furthermore, this predicts that when the subject is not contrasted, a contrasted PP will move past it; thus, P-​stranding should be possible in cases when the subject is not contrasted, shown in (38): (38) a. ? On Monday, John dealt with the problem, and on Tuesday, the solution. b.  John dealt with the problem on Monday, and the solution on Tuesday. In (38), with can be stranded, regardless of whether it occurs before or after the other contrasted term (provided the order of the contrasted terms corresponds to the order of their correlates). We find a slight oddness in (38a), but it seems more acceptable than (28b), where a subject is the other contrasted term; (38b) is, however, completely acceptable. This offers further support for the idea that, in a gap, movement across the subject is key to the availability of P-​stranding in languages like English. Now, we have an explanation for why P-​stranding is allowed in a structure like (39), despite our analysis in terms of Gapping: (39) I dealt with the problem or (with) the solution. (=(26)) As discussed above, we treat this as a case of Gapping with one contrasted term. This means that the non-​contrasted subject will remain in the second conjunct, and the contrasted term must move past it to escape being gapped. This is thus a case of long movement by our definition, and so P-​stranding will be allowed exactly as shown in (27). We thus have an account for why P-​ stranding is disallowed in canonical Gapping structures involving a contrasted subject and a contrasted PP like those in (28), but allowed in cases like (38)–(39) in English. Of course, we are not claiming here that the only thing influencing the availability of P-​stranding is the distance that a targeted DP moves. Clearly, Spanish does not allow P-​stranding even for long distance movement. We are merely claiming that in languages that do allow P-​stranding, it seems to be only allowed in long movements. In other languages, like Spanish, P-​stranding will be ruled out altogether for some independent reason, and this will be what prevents it in (25).

28  Michael Wilson and Rong Yin

6.  Conclusion and future directions Summarizing, we have shown that in Spanish, there is a contrast between coordination in argument PPs and instrumental adjunct PPs: argument PPs require P-​repetition, while instrumental adjunct PPs allow for both repetition and non-​repetition of P. We assume that coordination has a basic type of ⟨st, ⟨st, st⟩⟩ (Partee & Rooth, 1983), and propose that non-​argument con/​with can be decomposed into at least two functional projections, the lower of which is of type ⟨s, t⟩. This allowed us to derive non-​repetition of P in instrumental adjunct coordination via coordination of this lower, vP projection. Repetition of P in instrumental adjuncts is derived by coordination of the main clause vP. Argument PPs, in contrast, lack the complex functional structure of instrumental adjunct PPs, so only the main clause vP can be coordinated to avoid a type-​shifting penalty. Coordination of the main clause vP is followed by Gapping:  in Spanish, this results in P-​repetition due to obligatory pied-​ piping; in English, this can result in either repeating P or not repeating P, as pied-​piping is optional in Gapping structures that involve long movement of the PP, a claim for which we provided independent evidence. Future research should address whether we might want to take a stricter approach to coordination à la Hirsch (2017), whose theory is like ours, but doesn’t allow type-​ shifters. However, disallowing type-​ shifters would not affect our present analysis, since we always coordinate constituents of type ⟨s, t⟩. In addition, we offer some speculation regarding how our analysis might be extended to account for comitative and DP-​internal uses of adjunct con: (40)  Quiero helado con fresas o ?(con) frambruesas. 1s.want ice.cream with strawberries or ?(with raspberries “I want ice cream with strawberries or raspberries.” We observe again here that comitative con is optional (though preferred as before). A starting point for a decompositional analysis of comitative con/​ with should note that its meaning is more complex than that of the and of logical conjunction or sum formation, despite claims to the contrary (in, e.g., Koenig et al., 2007). Consider the following contrast (Al Khalaf, 2018): (41) a. A neighbor was cleaning with her pet. b. # A neighbor and her pet were cleaning. (Al Khalaf, 2018, (13)) (40a) is acceptable under a comitative reading, but it is not possible to paraphrase it as (40b), since this latter sentence would require the pet to be cleaning—​which is decidedly odd. The same contrast exists in Spanish:

Argument and adjunct coordination in Spanish  29 (42)  Context: A woman’s neighbor’s newborn son suddenly began having trouble breathing. The woman’s and the neighbor’s cars were broken down, but the child clearly needed medical attention quickly. The neighbor was still recovering from the birth, so… a. La mujer corrió    una milla al   hospital con the  woman  3s.run.PR ET   a   mile  to.the  hospital  with el  recién nacido para conseguir tratamiento the  newborn     for   get     treatment “The woman ran a mile to the hospital with the newborn to get treatment.” b. #La mujer y  el recién nacido corrieron  una milla the  woman and the newborn   3p.run.P RE T  a  mile al  hospital para conseguir  tratamiento. to.the hospital for  get     treatment “The woman and the newborn ran a mile to the hospital to get treatment.” We leave a detailed syntactic and semantic analysis of the decomposition of comitative and DP-​internal con, and its possible integration with the analysis presented here, for future work.

Notes * Michael Wilson and Rong Yin, University of Massachusetts Amherst. For questions, contact the first author at [email protected]. Before all else, we would like to thank our Spanish consultants, Helena López Palma and Jorge Guzman, who provided judgements on Spanish data. This chapter owes an especially great debt to many conversations with Kyle Johnson; whatever herein has merit certainly originated in his advice. We also thank those who gave us feedback on this material at the UMass Syntax Workshop and our audience at the Hispanic Linguistics Symposium 2018 at the University of Texas at Austin. We finally extend our gratitude to two anonymous reviewers who provided very helpful comments, which led to significant revisions in the analysis in §§4–​5 in particular, and helped sharpen our thinking throughout. Having given credit where it’s due, we take responsibility for all errors. 1 One way of thinking about how to derive such a constraint would be to treat it as the result of a learning process that favors positing a maximally transparent syntax/​ semantics mapping. In other words, a language learner would avoid positing new semantic rules (i.e., type-​shifters) if independently motivated syntactic processes could derive the correct meaning and desired surface word order. Previewing the analysis to come, we note that Gapping and movement out of an ellipsis site are independently motivated syntactic operations, so positing a type-​shifter when the combination of these could derive the correct result would be avoided, despite the apparent structural complexity that results. We thank Kyle Johnson (personal communication) for this suggestion. 2 Fábregas (2007) couches his analysis in terms of Ramchand’s (2008) system, in which lexicalization is defined in terms of contiguous spans of heads (see also Ramchand,

30  Michael Wilson and Rong Yin 2018). For our purposes, there is no crucial difference between this system, and a more traditional one involving head movement. We illustrate Fábregas’s analysis using head movement for ease of exposition. 3 We assume for present purposes that facilitating events can be understood as events causing a particular event token of e3. See Paul Portner’s personal communication cited in Koenig et al. (2007, p. 214) for more details. 4 Putting a v° inside con/​with may give us a handle on the instrument/​subject alternation: if an instrument is introduced by v° inside its adjunct, it is unsurprising that it could be introduced by a “normal” v°. However, there may need to be a restriction that the agent in the main clause be the initiator of e1, since facilitating instruments cannot be subjects. (i) a.

The knife cut the meat. /​El cuchillo cortó la carne. (Spanish data from Fernández-​Soriano & Mendikoetxea, 2013, (15a)) b. # The brush scrubbed the table. /​# El cepillo frotó la mesa. (Spanish data from García-​Pardo, 2015, (5b))

We suggest verbs like cortar/​cut can encode both “reduced” causal chains starting with (what would normally be) e2 and “expanded” causal chains starting with e1. In contrast, verbs like frotar/​scrub may only encode the expanded causal chains, leaving them infelicitous with instrument subjects that cannot be agents of e1. An anonymous reviewer further suggests a relevant difference between (i-​a) and (i-​b) may be that the verbs in (i-​a) encode a change of state, while those in (i-​b) do not; we think this may correlate in some way with the verb’s ability to encode different sizes of causal chains (the change of state contrast cannot be the only relevant factor, since instrumental adjuncts can occur with non-​change of state verbs). We must also ensure this observation about which verbs allow instrumental subjects holds in the general case. We leave these tasks for future research. 5 An anonymous reviewer notes a potential problem for using v° to introduce both instruments and agents, as agents tend to be animate and instruments, inanimate. However, what counts as an agent may simply be vague (see Williams, 2015, ch. 2, p. 32, and ch. 7, §7.2 for a critical review of this idea); what we call instrument subjects may simply be the interpretation we get for an inanimate “agent.” This reviewer further asks why we do not treat instruments as receiving a “cause” interpretation. We avoid this because in semantic theories of causation, CAUSE is a relation that applies to events rather than entities, and instruments are clearly entities. We are also unaware of any theory that satisfactorily links up a thematic role “causer” with semantic theories of causation; a causer is clearly not intended to be the agent of a causing event, but defining it in other ways makes it difficult to integrate it with a semantic theory of causation. See also fn. 4, where we suggest that it may be the verb that is ambiguous, rather than the thematic relation of the subject. 6 An anonymous reviewer asks if there is independent evidence for the decomposition we present in this section; namely, whether vP+DP can ever be realized as a PP without C AU SE-​PART, similarly to how PlacePs can occur without PathPs. We note that while this would indeed strengthen the explanatory reach of our analysis, it is not strictly speaking necessary. For comparison, analyses of the English double object construction often invoke a covert preposition, which may incorporate into the verb (Beck & Johnson, 2004; Harley, 2002, 2011; Pesetsky, 1995). However, this preposition is not thought to be overtly realized

Argument and adjunct coordination in Spanish  31 alone in English. Pesetsky (1995) argues for multiple kinds of null prepositions in analyses of the double object construction, object experiencer verbs, causative constructions, and some uses of adjectives in English; these prepositions are also not realized in English without incorporation into some other head. Finally, Landau (2010) presents an argument that experiencer arguments are universally introduced by prepositions which may be overt in some languages, but are often not in others. As such, the fact that a preposition may only be pronounced by incorporation (or not at all) is independently motivated in a variety of domains (also, cf. Fábregas’s 2007 analysis of hacia as consisting of two parts, above). However, we also note it might be possible to analyze passive and nominal uses of by as spelling out v° to the exclusion of C AU SE-​PART (cf. Collins, 2005), in line with our reviewer’s suggestion (but see Bruening, 2013 for a different treatment of passive by-​phrases). 7 An anonymous reviewer notes that typologically, agents seem to be introduced by a functional head higher than one that introduces causes, while our analysis places v° below C AUS E -​PART in contravention of this sequence. However, we note that this tendency clearly does not apply to cases of embedded causation, in which the agent of a caused event appears below the functional head introducing the relation between the causing event and the caused event (Johnson 2019 reports data from Muskogee Creek exemplifying this pattern, which holds more generally as well). We are dealing with exactly such a case of embedded causation here: the instrument is introduced as the agent of a subevent in the causal chain of the predicate, not as the agent of the full event. 8 An anonymous reviewer asks why instrumental con is spelled out the same as comitative con, despite comitative con presumably consisting of different components. We claim this is simply a case of accidental homophony, almost certainly for historical reasons (Heine & Kuteva, 2006, ch. 5). In fact, instrumental markers and comitative markers are not syncretic in most (non-​ European) languages (Heine & Kuteva, 2006, ch. 5); Mandarin Chinese, for instance, uses yong “use” for instruments and dai “carry” for comitative accompaniers. This broader picture gives us reason to doubt that there is a deep reason that these two distinct uses of con/​with happen to be spelled out in the same way. 9 We are indebted to Kyle Johnson (personal communication) for this suggestion. 10 Our approach should also work under a large conjuncts account as well, so the choice does not appear to be crucial. The only difference regards our analysis of the supposed impossibility of P-​stranding in English Gapping in §5.1, which is controversial (see fn. 14). 11 Under some approaches, this might be equivalent to Stripping. See Johnson (2018) for discussion. 12 Two anonymous reviewers note examples like the following that show that there may be more nuance to the argument coordination cases: (i) a. Me  aburro con las películas y  (con) los conciertos 1s. C L   1s.bore  with  the  movies    and  (with)   the   concerts “I get bored with movies and concerts.”

32  Michael Wilson and Rong Yin b. Rompí       con  la  rutina  y  (?con) la monotonía  diarias 1s.break.PRET  with  the routine and (?with  the monotony  daily.FEM.PL “I broke with the daily routine and monotony.” c. Anoche, soñé con un elefante y (con) un tigre. Last.night 1s.dream.PRET with an elephantand (with a tiger “Last night, I dreamt about an elephant and a tiger.” These sentences feature argument coordination, but con-​ repetition is only optional, and even dispreferred in some cases, such as (i-​b). We would like to mention three possibilities: first, these are cases of conjunction rather than disjunction, which are plausibly receiving a sum formation reading rather than a predicate conjunction reading, which could be obscuring the contrast we focus on (see §2 for details)—​particularly in the second case, with two singular conjuncts but plural agreement on the adjective. Second, these sentences feature objects which receive non-​specific interpretations, whereas the objects in our sentences receive specific interpretations. Non-​specific nominal expressions have been proposed to be predicates rather than arguments (Chierchia, 1998; but cf. Borer, 2005). In this case, these expressions would thus have to integrate syntactically and semantically in a different way with the rest of the sentence than expressions of type e, which could lead to a different coordination strategy. We thank a reviewer for this second suggestion. Third, it may the case that not all argumental cons are structurally simple; it seems plausible, for instance, that the con in soñar con may carry more meaning—​and, by extension, more structure—​than the con of casarse con does. Obviously, this last sort of explanation is not ideal, but it may be justified if independent evidence could be found to support it. We leave aside the details for future work, due to reasons of space. 13 This section in particular owes much to suggestions from and conversations with Kyle Johnson (personal communication). 14 But see Kuno (1976), Steedman (1990), Vanden Wyngaerd (2007), and Johnson (2018) for counterexamples and counterarguments; if they are ultimately correct, that would simplify our explanation of why English does not require P-​repetition for arguments. 15 Abe & Hoshi (1997) define crossing in linear terms. We define it structurally, more in line with Pesetsky (1982). 16 Note that we must say more to allow the grammatical (i): (i)  With Mary, John dealt, and with Bill, Susan. Presumably, we have to allow no more crossings of contrasted elements than those that their correlates in the antecedent display; there is also a constraint that the linear order of contrasted elements in a gapped coordinate must match the linear order of their correlates in the antecedent (Johnson, 2018), which may or may not be reducible to a properly formulated version of the CCG. We leave the exact formulation of this refinement of the CCG for future work. 17 We use put up with in all examples in (33) for consistency, despite some oddness in the pseudo-​passive. However, pseudo-​passives are not all consistently degraded; cf. The bed was slept on last night, which is completely acceptable. Additional factors influence the availability of pseudo-​passives that likely account for (33d)’s slight

Argument and adjunct coordination in Spanish  33 oddness (such as the affectedness of the deep object). But it is clear regardless that pseudo-​passives in English allow P-​stranding; indeed, it is one of their defining characteristics. 18 Note that pied-​ piping does not save (32d) nor (34d) (Johnson, 1991; Thráinsson, 2001): (i) a. * I will put [PP with the book]i up ti. b. * Jón talaði [PP við Maríu]i ekki ti. Jon spoke [PP with Mary not (Thráinsson, 2001, p. 151, (6c)) We assume that short movement of PPs must be rightward for independent reasons (Kyle Johnson, personal communication). We note that this restriction also holds of CPs: (ii) a. John pointed out the problem to the dean. b. John pointed the problemi out ti to the dean. c. John pointed out that Bill had left to the dean. d. * John pointed [CP that Bill had left]i out ti to the dean. (iii) Icelandic: a. Jón sagði ekki að María hefði farið. Jon said not that Mary had left b. * Jón sagði [CP að María hefði farið] ekki ti. (Thráinsson, 2001, p. 164, (38i–​j)) We assume that the facts in (i)  are of a piece with the facts in (ii) and (iii), without committing to any particular analysis. Also, note that just because PP movement is ruled out in (i) doesn’t entail that movement of a DP out of a PP should be similarly ruled out; PP-​movement is also impossible in pseudo-​passives (*On the bed was slept), but that does not prevent P-​stranding movement of the DP from being grammatical in those cases. 19 Standard examples of Object Shift cross negation, which is often assumed to occur above vP. Our explanation would thus seem to incorrectly predict that Object Shift across negation should allow P-​stranding, contrary to (34d). However, negation may be lower than is typically thought. For instance, Thráinsson (2001, p. 153) notes that “long” Object Shift across a postverbal subject cannot occur, even though Object Shift does occur across negation. (Though some theories of negation put it lower than Spec,vP; see Champollion (2011) for a theory of this sort.) But also problematic are examples of Object Shift crossing stranded subject-​modifying quantifiers (Holmberg, 1986, p. 165; Thráinsson, 2001, p. 198), which are presumably left behind in Spec,vP following movement of the subject. Explaining such issues may require further refinement of which movements count as “short” and which as “long,” a complication which we leave aside here.

References Abe, J., & Hoshi, H. (1997). Gapping and P-​ stranding. Journal of East Asian Linguistics, 6, 101–​136.

34  Michael Wilson and Rong Yin Al Khalaf, E. (2018). Remarks on the syntax and semantics of so-​called comitative coordination. Linguistic Research, 35(2), 253–​273. Beck, S., & Johnson, K. (2004). Double objects again. Linguistic Inquiry, 35(1), 97–​124. Borer, H. (2005). In Name Only, Vol. 1 of Structuring Sense. New  York:  Oxford University Press. Brucart, J.M., & MacDonald, J.E. (2012). Empty categories and ellipsis. In J.I. Hualde, A. Olarrea & E. O’Rourke (eds.), The Handbook of Hispanic Linguistics, 579–​601. Malden, MA: Wiley-​Blackwell. Bruening, B. (2013). By-​phrases in passives and nominals. Syntax, 16(1), 1–​41. Champollion, L. (2011). Quantification and negation in event semantics. The Baltic International Yearbook of Cognition, Logic and Communication, 6, 1–​23. Cinque, G., & Rizzi, L. (eds.) (2010). Mapping Spatial PPs, Vol. 6 of The Cartography of Syntactic Structures of Oxford Studies in Comparative Syntax. New York: Oxford University Press. Collins, C. (2005). A smuggling approach to the passive in English. Syntax, 8(2), 81–​120. Fábregas, A. (2007). An exhaustive lexicalization account of directional complements. In M. Bašić, M. Pantcheva, M. Son & P. Svenonius (eds.), Nordlyd: Tromsø Working Papers on Language and Linguistics, Vol. 34, Special issue on Space, Motion, and Result, 165–​199. Tromsø: CASTL. Fernández-​Soriano, O., & Mendikoetxea, A. (2013). Non selected dative arguments in Spanish anticausative constructions:  Exploring subjecthood. In I. Seržant & L. Kulikov (eds.), The Diachronic Typology of Non-​Canonical Subjects, Vol. 140 of Studies in Language Companion Series, 3–​34. Amsterdam: John Benjamins. García-​ Pardo, A. (2015). Decomposing the VP:  Aspect, argument structure and the instrumental-​ subject alternation. Abstract of paper presented at the 46th Annual Meeting of the North East Linguistics Society, 16–​18 October, Concordia University. Harley, H. (2002). Possession and the double object construction. In P. Pica & J. Rooryck (eds.), Linguistic Variation Yearbook 2002, Vol. 2, 31–​70. Amsterdam: John Benjamins. Harley, H. (2011). A minimalist approach to argument structure. In C. Boeckx (ed.), The Oxford Handbook of Linguistic Minimalism, 426–​447. Oxford:  Oxford University Press. Heine, B., & Kuteva, T. (2006). The Changing Languages of Europe. New York: Oxford University Press. Hirsch, A. (2017). An inflexible semantics for cross-​categorial operators (Doctoral dissertation). MIT. Holmberg, A. (1986). Word order and syntactic features in the Scandinavian languages and English (Doctoral dissertation). University of Stockholm. Jerro, K. (2017). The causative-​instrumental syncretism. Journal of Linguistics, 53, 751–​788. Johnson, K. (1991). Object positions. Natural Language and Linguistic Theory, 9(4), 577–​636. Johnson, K. (2009). Gapping is not (VP-​) ellipsis. Linguistic Inquiry, 40(2), 289–​328. Johnson, K. (2014). Gapping (Unpublished manuscript). University of Massachusetts Amherst.

Argument and adjunct coordination in Spanish  35 Johnson, K. (2019). Intermediate agents and expletive voice:  The Creek causative. Presentation at University of Massachusetts Amherst Syntax Workshop, February 8. Johnson, Kyle (2018). Gapping and stripping. In T. Temmerman & J. van Craenenbroeck (eds.), The Oxford Handbook of Ellipsis. Oxford: Oxford University Press. Jónsson, J.G. (2008). Preposition reduplication in Icelandic. In S. Barbiers, O. Koeneman, M. Lekakou & M. van der Ham (eds.), Microvariation in Syntactic Doubling, vol. 36 of Syntax and Semantics, 403–​417. Bingley:  Emerald Group Publishing Limited. Koenig, J.P., Mauner, G., Bienvenue, B., & Conklin, K. (2007). What with? The anatomy of a (proto)-​role. Journal of Semantics, 25, 175–​220. Kratzer, A. (1996). Severing the external argument from its verb. In J. Rooryck & L. Zaring (eds.), Phrase Structure and the Lexicon, 109–​137. Dordrecht:  Kluwer Academic Publishers. Kuno, S. (1976). Gapping: A functional analysis. Linguistic Inquiry, 7, 300–​318. Landau, I. (2010). The Locative Syntax of Experiencers, Vol. 51 of Linguistic Inquiry Monographs. Cambridge, MA: MIT Press. Link, G. (1983). The logical analysis of plural and mass nouns: A lattice theoretical approach. In R. Bäuerle, C. Schwarze & A. von Stechow (eds.), Meaning, Use and Interpretation of Language, 302–​323. Berlin: de Gruyter. Maling, J., & Zaenen, A. (1985). Preposition-​stranding and passive. Nordic Journal of Linguistics, 8, 197–​209. Maling, J., & Zaenen, A. (1990). Preposition-​stranding and passive. In J. Maling & A. Zaenen (Eds.), Modern Icelandic Syntax, Vol. 24 of Syntax and Semantics, 153–​ 164. New York: Academic Press. Mari, A. (2006). What do the notions of instrumentality and of manner have in common? A channel theoretic model for causality as dependence. In S.D. Patrick (Ed.), Syntax and Semantics of Prepositions, Vol. 29 of Text, Speech and Language Technology, 263–​287. Dordrecht: Springer. Partee, B., & Rooth, M. (1983). Generalized conjunction and type ambiguity. In R. Bäuerle, C. Schwarze & A. von Stechow (eds.), Meaning, Use and Interpretation of Language, 361–​383. Berlin: de Gruyter. Pesetsky, D. (1982). Paths and categories (Doctoral dissertation). MIT. Pesetsky, D. (1995). Zero Syntax:  Experiencers and Cascades. Current Studies in Linguistics. Cambridge, MA: MIT Press. Radkevich, N.V. (2010). On location: The structure of case and adpositions (Doctoral dissertation). University of Connecticut. Ramchand, G.C. (2008). Verb Meaning and the Lexicon:  A First Phase Syntax. Cambridge: Cambridge University Press. Ramchand, G.C. (2018). Situations and Syntactic Structures:  Rethinking Auxiliaries and Order in English, Vol. 77 of Linguistic Inquiry Monographs. Cambridge, MA: MIT Press. Rissman, L. (2013). Event participant representations and the instrumental role:  A cross-​linguistic study (Doctoral dissertation). Johns Hopkins University. Roy, I., & Svenonius, P. (2009). Complex prepositions: In J. François, E. Gilbert, C. Guimier & M. Krause (eds.), Autour de la preposition: Actes du Colloque International de Caen (20–​22 septembre 2007), 105–​116. Caen: Presses Universitaires de Caen. Schwarz, B. (1999). On the syntax of either…or. Natural Language and Linguistic Theory, 19, 339–​370.

36  Michael Wilson and Rong Yin Steedman, M.J. (1990). Gapping as constituent coordination. Linguistics and Philosophy, 13(2), 207–​263. Svenonius, P. (2007). Projections of P (Unpublished manuscript). CASTL, University of Tromsø. Svenonius, P. (2010). Spatial P in English. In G. Cinque & L. Rizzi (eds.), Mapping Spatial PPs, Vol. 6 of The Cartography of Syntactic Structures of Oxford Studies in Comparative Syntax, 127–​160. New York: Oxford University Press. Thráinsson, H. (2001). Object shift and scrambling. In M. Baltin & C. Collins (eds.), The Handbook of Contemporary Syntactic Theory, 148–​202. Malden, MA: Blackwell. Thráinsson, H. (2007). The Syntax of Icelandic. New York: Cambridge University  Press. Vanden Wyngaerd, G. (2007). Gapping constituents (Unpublished manuscript). Humboldt-​Universität zu Berlin, CRISSP, & Katholieke Universiteit te Leuven. Williams, A. (2015). Arguments in Syntax and Semantics. Key Topics in Syntax. Cambridge: Cambridge University Press.

2  The interaction of parasitic gaps and adjunct control in Spanish Katie VanDyne

1.  Introduction Since the first mention in Ross (1967) and subsequent discussions in Taraldsen (1981), Chomsky (1982), and Engdahl (1983), among others, parasitic gaps have been characterized as a gap that is only licensed when there is also A-​bar movement in the structure, on which the gap is said to be parasitic. Observe the distribution of parasitic gaps in (1) from Engdahl (1983) where wh-​movement and topicalization license an object gap in the adjunct clauses in (1a, b). (1) a. [Which article]i did you file ti without reading pgi?1 b. [This kind of food]i you must cook ti before you eat pgi. c. *John filed [the articles]i without reading pg1. d. *[These articles]i were filed ti without reading pgi.

(wh-​movement) (topicalization) (no movement) (passive movement)

In (1c), a gap in the object position is not licensed, and in order for the sentence to be grammatical, an overt pronoun must be added in that position. In addition to the contrast between A-​bar movement in (1a, b) and the lack of movement in (1c), the requirement for A-​bar movement is also exemplified in passive structures, which do not license a parasitic gap in (1d). This limitation of only being licensed by wh-​moved and topicalized antecedents does, however, seem to be subject to some cross-​linguistic variation. In Spanish, there are parasitic gaps licensed in two additional environments, in both passive structures (Sheehan, 2015) as well as clitic structures (Campos, 1991; García Mayo, 1995). (2) a. ¿ [Qué  libro]i archivaste ti sin   leer pgi?   (wh-​movement) What  book  file.2S G    without read. I N F “What book did you file without reading?” b. [El documento]i fue  archivado ti sin     abrir pg1.    (passive) The document  was  filed      without open. I N F   (Sheehan, 2015) “The document was filed without opening it.”

38  Katie VanDyne c. Loi archivaron ti sin  leer pg1.              (clitic) It   filed.3PL   without read. I NF          (Campos, 1991) “They filed it without reading it.” d. *Archivaron [el documento]i sin abrir pgi.  (no movement) Filed.3PL the document without open. I N F “They filed the document without opening it.” While parasitic gaps in Spanish thus appear to have a wider distribution, in that they are not strictly limited to having only wh-​moved and topicalized antecedents, there are some additional limitations to these constructions and questions that arise regarding how they are licensed. First, while passive subjects can license a parasitic gap, as in (2b), active subjects are still unable to license a gap (3). (3)  *[El árbol]i bloqueó la acera sin talar pgi. The tree blocked the sidewalk without cut down.I N F “The tree blocked the sidewalk without it being cut down.” Further, passive and clitic licensed parasitic gaps in Spanish also display restrictions on the animacy of the antecedent. As has been described in Campos (1991) and García Mayo (1995), only inanimate, and not animate, clitics can license a parasitic gap. Observe this asymmetry in (4). (4) a.

Loi usé      sin     citar pgi. (inanimate, clitic movement) It  used.1 SG  without cite.I N F “I used it without citing it.” b. *Usé     [el artículo]i sin     citar pgi. (inanimate, no clitic movement) Used.1 SG   the article  without  cite. I N F “I used the article without citing it.” c. *Loi vi      sin     saludar pgi. (animate, clitic movement) Him saw.1 SG  without greet.I N F “I saw him without greeting him.”

I observe that this same pattern is also found in passive licensed parasitic gaps. As shown in (5), only inanimate passive subjects can license a gap. (5) a. [El  artículo]i fue usado sin citar pgi. The article was used without cite.I N F “The article was used without citing it.” b. *[La chica]i fue vista sin saludar pgi. The girl was seen without greet.I N F “The girl was seen without greeting her.”

(inanimate) (animate)

Animacy restrictions are not found in other types of parasitic gaps and they have yet to be accounted for in clitic or passive parasitic gaps.

Interaction of parasitic gaps and adjunct control  39 In this chapter, I  address these two main questions regarding the distribution of clitic and passive licensed parasitic gaps in Spanish. First, why are these parasitic gaps limited to only being licensed by passive subjects and not active subjects? Second, why are these structures further limited to only being licensed by inanimate antecedents, when this is not a restriction that occurs with parasitic gaps licensed by wh-​movement or topicalization? I  propose that these questions can be answered by looking at the role of adjunct control in these examples. Notably, in all of these structures there are two empty arguments that need to be licensed: PRO and the parasitic gap. I show how by looking at the controller of PRO, the parasitic gap patterns can be accounted for. Crucially, when the controller of PRO and the parasitic gap antecedent are the same element, a parasitic gap is not grammatical.

2.  Passive subjects but no active subjects 2.1  Passive parasitic gaps In this section, I focus on the asymmetry between passive and active subjects and show how the concurrent relation of adjunct control in these structures helps to explain their distribution. First, however, it is worth mentioning that parasitic gaps in Spanish are unique in that passive subjects, which are not traditionally considered to be A-​bar movement and do not license parasitic gaps in English, can license a parasitic gap. The example from (5a) is repeated below in (6), where a gap licensed by a passive subject in (a) is allowed. (6) a. [El  artículo]i fue usado sin citar pgi. The article was used without cite.I N F “The article was used without citing it.” English structures (like (1d)) that are parallel to (6) are ruled out on the premise of lack of A-​bar movement, a traditional requirement of parasitic gaps (Engdahl, 1983; Manzini, 1994; Culicover, 2001; among others). The grammaticality of (6) then raises many questions about both its underlying structure as well as the licensing requirements of parasitic gaps. Based on the previous literature claiming that parasitic gaps must be licensed by A-​bar movement, Sheehan (2015), uses the passive parasitic gap in (2b) to argue for A-​bar positioned preverbal subjects. The preverbal subject, being moved from a low, postverbal A-​position to a higher A-​bar position, is then a potential licenser of a parasitic gap, just like wh-​movement or topicalization. However, parasitic gap antecedents are not limited to only preverbal passive subjects. Postverbal passive subjects in both the periphrastic passive and se passive structures can also license a parasitic gap, shown in (b)  and (c) respectively.

40  Katie VanDyne b. Fue  usado  [el artículo]i sin   citar pgi. Was  used   an article   without  cite.I N F “The article was used without citing it.” c. Se compraron aceitunasi sin   probar pgi. S E   bought    olives   without  try. I N F “Olives were bought without trying them.” The availability of both preverbal and postverbal structures is left unmentioned in Sheehan’s analysis. Moreover, Sheehan’s analysis of Spanish subjects claims that postverbal subjects remain in a base-​ generated A-​ position. Therefore, this analysis would not predict a parasitic gap to be licensed by postverbal passive subjects, since the analysis of parasitic gaps still crucially relies on the movement of the subject to an A-​bar position. This prediction turns out to be empirically incorrect, as shown in (6b, c) where a parasitic gap is grammatical with an in-​situ subject. Finally, this analysis also does not provide an explanation for why parasitic gaps would be limited to only passive, and not preverbal active subjects, when under this analysis they too would be analyzed as moving to an A-​bar position.2 While it appears that movement may not ultimately be necessary to license a parasitic gap in Spanish, I leave a more in-​depth discussion of how passive parasitic gaps are licensed aside for this chapter. Whatever the correct theoretical analysis of these structures ends up being, it is crucial that it can also account for the active/​passive subject asymmetry. 2.2  Lack of active subject licensed parasitic gaps As shown earlier in (3), repeated below in (7)  with an additional example, there is a clear asymmetry between active voice subjects, which cannot license a parasitic gap, and passive subjects which can (6). (7) a. * [El árbol]i bloqueó la acera   después de talar pgi. The tree blocked the sidewalk after of  cut down.INF “The tree blocked the sidewalk after cutting it down.” b. * [El artista]i pintó su autorretrato sin ver pgi en el   espejo. The artist painted his self-​portrait without see.INF in the  mirror “The artist painted his self-​portrait without seeing himself in the mirror.”

A closer look at the role of adjunct control in the structures in (6)  and (7) leads us towards an explanation for this asymmetry. Common to all the parasitic gap examples we have seen for Spanish thus far is the presence of PRO as the adjunct subject.3 What controls PRO in each example, however, does not appear to be the same.4 Focusing first on the grammatical inanimate passive example in (6), there are two potential controllers of PRO: the thematic subject, el artículo, and the implicit agent. The interpretation of the sentence is that whoever used the article also did not cite it. Having this

Interaction of parasitic gaps and adjunct control  41 interpretation, we can conclude that PRO is controlled by the implicit agent, represented as pro in (8). (8) [El artículo]i fue usado proj sin PROj citar pgi. Although the thematic subject, el artículo, is in T, and under Landau’s theory of control would thus be in a suitable structural position to control PRO, it is instead the implicit argument that controls PRO. While this interpretation of (6) is clear, it raises the question of why the implicit agent is chosen as the controller of PRO, rather than the overt thematic subject. Landau (2017) suggests a potential solution to this kind of question. While Landau focuses on explaining how the same adjunct can display two different types of control, like in (9), I propose that the same analysis can be extended to help explain the differences between the parasitic gaps in (6) and (7). Both structures in (9) contain the same adjunct, before PRO entering the basement, but the controller of PRO is different in each sentence. (9) a. [The rain]i washed the stairs before PROi entering the basement. b. The stairs were washed before PRO entering the basement. (Landau, 2017) Landau (2017) suggests that this ability for the same adjunct clause to have different controllers ultimately arises from them displaying two different types of control that each have their own structures. Landau claims the structure in (a)  is an obligatory control (OC) structure, where the salient interpretation is that the rain both washed the stairs and then entered the basement.5 On the other hand, Landau claims the structure in (b) is a non-​obligatory control (NOC) structure. Landau (2017) following Williams (1985) presents arguments that implicit agents cannot actually control adjuncts and that instances where the implicit agent appears to control, like (9b), ultimately fall under the category of NOC whose reference is determined logophorically. Landau proposes that OC and NOC have different underlying structures, as shown in (10), which explain the differences in control in (9). (10)  a. OC adjunct: [PP before [FinP Fin [TP…]]] b. NOC adjunct: [PP before [CP C+log [FinP Fin [TP…]]]] To explain the distribution of OC vs NOC adjuncts, Landau appeals to the Economy of Projection principle, following Grimshaw (1994), Bošković (1996) and Speas (2006). Because the NOC structure is considered more complex structurally (since it includes an extra projection) following the Economy of Projection, Landau claims that, if everything else is equal, the OC structure

42  Katie VanDyne in (10a) will be favored over the NOC structure (10b), whenever both generate coherent interpretations. The NOC structure in (10b) is only chosen when the OC structure produces a semantic deviance. This accounts for why (9b), although having an overt controller that could display OC, displays NOC. If (9b) were to display OC, the stairs would control PRO, leading to the interpretation that “the stairs were washed before the stairs entered the basement.” This is what Landau would consider to be a semantic deviance, and thus the NOC structure which is also available, where PRO is not controlled by the matrix subject, prevails in this case. This analysis follows smoothly to explain the control patterns in parasitic gap structures like (6). If the structure in (6a) were to display OC, the more economical option, the thematic subject el artículo would control PRO and there would be a semantic deviance. The argument structure of the verb citar requires an external argument that is an agent. El artículo, being inanimate and unlikely to cite anything, does not fit the verb’s requirement for an agent and would produce a semantic deviance if interpreted as the controller of PRO. Under Landau’s analysis, this is the situation that would lead to choosing the NOC structure, where the implicit agent is interpreted as controlling PRO.6 Returning to the difference between the active and passive subject parasitic gaps, the patterns of control in the active voice sentences play out differently. Under Landau’s approach, if (7a) were to display OC, like (6), it would also cause a semantic deviance. The argument structure of talar also requires an agent. El árbol, being inanimate and unlikely to cut down anything, would not be a coherent agent of the adjunct clause. In this situation, Landau’s approach would predict an NOC structure to prevail. However, unlike the passive example in (6), in (7) there is no implicit agent or other pragmatically salient referent to control PRO. Lacking this implicit agent, no NOC structure would be available. The structure in (7b) differs slightly in that if it were to display OC, there would be no semantic deviance. The argument structure of the verb ver also requires an agent. Unlike the example of the passive, having el artista be the agent would be a coherent interpretation. Without a semantic deviance, according to Landau, there would be no reason to choose the NOC structure over OC for (7b). It therefore appears that both examples in (7) could only display OC. The OC structure obtained in (7) is what seems to generate problems for the parasitic gap reading. The controller of PRO, the matrix subject el artista in (7b), would also be the intended parasitic gap antecedent. The matrix subject would then control both the external argument of the adjunct (PRO) as well as the internal argument of the adjunct (the parasitic gap), as in (11). (11) *[El artista]i pintó su autorretrato sin PROi ver pgi en el espejo. This structure seems problematic for the intended parasitic gap reading of the sentence. In (11), two arguments of the same predicate are coreferential. This should result in a reflexive predicate, following Reinhart and Reuland (1993), giving the interpretation that “the artist painted his self-​ portrait

Interaction of parasitic gaps and adjunct control  43 without (him) seeing himself in the mirror.” In Spanish, when an external argument is coreferential with an internal argument of the same predicate (i.e. there is a reflexive structure), an overt se is needed (Otero, 1999). When se is added to the construction in (11) to mark the reflexive, the structure does become grammatical, as shown in (12). (12) El  artista pintó su autorretrato sin ver-​se en el espejo. The artist painted his self-​portrait without seeing-​S E refl in the mirror “The artist painted his self-​portrait without seeing himself in the mirror.”

In this structure, the active subject still controls both PRO and the internal argument. However, in such a structure, only a reflexive reading, with an overt se is grammatical, and a parasitic gap reading is not.7 In sum, the lack of separate antecedents for both PRO and the parasitic gap is what seems to limit parasitic gaps to being licensed by only passive subject antecedents, and not active subjects, in Spanish. The passive subject structures display NOC, where the matrix subject does not control PRO, and a parasitic gap can be licensed. On the other hand, in the active structures, where only OC is available, the matrix subject both controls PRO and is the parasitic gap antecedent. When the controller of PRO and the parasitic gap antecedent are the same element, only a reflexive reading is possible.

3.  Animacy effects In this section, I  proceed to the second asymmetry observed in non-​A-​bar licensed parasitic gaps. In both passive and clitic licensed parasitic gaps, the antecedent must be inanimate for the sentence to be grammatical. No animate antecedents are grammatical. In Section 3.1 I focus first on the animacy effects in passive parasitic gaps followed by a discussion of these effects in the clitic licensed gaps in Section 3.2. In 3.3 I show why no such animacy effects are found in parasitic gaps licensed by wh-​movement or topicalization. 3.1  Animacy effects in passive parasitic gaps Similar to the active subject examples, I propose that animacy effects in parasitic gaps can also be accounted for by considering the role of adjunct control. Recall that only inanimate passive parasitic gaps are grammatical. A pair of examples similar to (5) displaying this asymmetry is shown in (13). (13) a.

[El artículo]i fue criticado sin leer pgi. The article was criticized without read. I N F “The article was criticized without reading it.” b. * [La mujer]i fue besada sin saludar pgi. The woman was kissed without greet.I N F “The woman was kissed without greeting her.”

(inanimate) (animate)

44  Katie VanDyne Recall from the previous section that in the passive structures, there are two potential controllers for PRO: the passive subject (OC) and the implicit agent (NOC). The reading of (13a) is that whoever criticized the article (the implicit agent) did not read the article. Given this interpretation, the controller of PRO is not the thematic subject, but instead the implicit agent (14). Therefore, under Landau’s (2017) account, this structure would display NOC. (14) [El artículo]i fue criticado proj sin PROj leer pgi. On the other hand, the animate passive structures display different control patterns. In the animate example in (13b), while it is ungrammatical, there is a preference for the overt thematic subject controlling PRO rather than the implicit agent. This is demonstrated in (15) where a similar structure is grammatical, although without a parasitic gap, and PRO displays OC and is controlled by the thematic subject. Unlike in (13a), the overt DP, la mujer, in (15) controls PRO and is interpreted as the agent of saludar. (15) [La mujer]i fue besada después de PROi saludar a su esposo. The woman was kissed after of PRO greet.IN F D OM her husband. “The woman was kissed after greeting her husband.”

The interpretation of (15) is that the woman was kissed after the woman greeted her husband. In contrast, if it were the implicit agent in (15) controlling PRO, the interpretation would be that the person who kissed the woman did so after greeting their own spouse. This is not the interpretation that (15) has. The question then arises of why two structures with the same type of temporal adjunct, and both with passive matrix clauses, would have different controllers of adjunct PRO that display two different types of control. I suggest that this can also be accounted for using Landau’s (2017) analysis discussed in Section 2. The inanimate example, where the implicit agent is interpreted as controlling PRO, displays NOC. The animate example, where the matrix subject is interpreted as controlling PRO, displays OC. Landau claims that NOC is only chosen when the OC structure produces semantic deviance. Looking at the inanimate/​animate distinction, this analysis follows straightforwardly. In the inanimate structures, OC by the thematic subject would produce a semantic deviance since un artículo, being inanimate, as the agent of leer would not generate a coherent interpretation. Therefore, we would predict that the NOC structure instead prevails in that case. The NOC reading is the interpretation we get for the inanimate examples, following the analysis that implicit agent control is NOC. However, in the animate structures, OC by the thematic (animate) subject does not produce any semantic deviance. La mujer is capable of satisfying the agent role, as required for the verb saludar. This predicts that the structure will display OC. Like in the active subject examples, animate passive parasitic gaps that display OC, where the thematic subject controls PRO, is what I propose creates

Interaction of parasitic gaps and adjunct control  45 a problem for the parasitic gap interpretation. La mujer, in (13b) would not only be the passive subject, but would also control the adjunct external argument (PRO) and be the antecedent of the internal argument (the parasitic gap). Like with the active subject parasitic gaps from Section 2, this structure would then predict the possibility of a reflexive interpretation for the animate passive parasitic gaps. As predicted, when the reflexive se is added, the structure becomes grammatical, as shown in (16). (6) [La mujer]i fue besada antes de PROi The woman was kissed before of PRO “The woman was kissed before bathing (herself).”

bañar-​se. bathe.I N F -​SE refl

Thus, in the passive parasitic gaps with animate antecedents, the adjunct subject is controlled by the same constituent that is also the intended parasitic gap antecedent. In this situation, the adjunct verb is reflexive and as such needs to be marked with se, and a parasitic gap is therefore not possible. The inanimate examples that display NOC do not have this problem. The implicit agent controls PRO while the thematic subject is the parasitic gap antecedent. Once again, for a parasitic gap to occur, it is crucial that the controller of PRO and the parasitic gap antecedent be different constituents. 3.2  Animacy effects in clitic licensed parasitic gaps 3.2.1  Clitic licensed parasitic gaps Like the passive parasitic gaps, clitic licensed parasitic gaps also differ from the traditional parasitic gaps that are licensed by wh-​movement and topicalization. Under the analysis following Kayne (1989, 1991) that clitics are located in T, which is not an A-​bar position, like the passive parasitic gaps, how these structures license a parasitic gap can be questioned. Campos (1991), who first notes that parasitic gaps in Spanish can be licensed by accusative clitics, analyzes these structures as being a null operator that is bound by a silent (null) topic, located in an A-​bar position, a topic phrase. While this analysis easily accommodates the clitic parasitic gap data into the previous analyses, it raises further questions regarding why a null subject cannot license a parasitic gap. Following Frascarelli (2007), null subjects have been argued to be licensed by a higher, null aboutness topic. If the clitic structures can license a parasitic gap via a null topic, null subjects would also be expected to license a parasitic gap. However, they cannot. Observe (17). (17) a. ¿ Qué pasó con [el  árbol]i? What happened with the tree “What happened with the tree?” b. *Proi bloqueó la acera después de pro blocked the sidewalk after of “It blocked the sidewalk after being cut down.”

talar pgi. cut down

46  Katie VanDyne Leaving aside the ultimately correct analysis for these structures, like the passive parasitic gaps, any analysis will have to be able to also account for the animacy restrictions that are found in these structures. In the rest of this section I focus on these animacy effects. 3.2.2  Only inanimate clitics can license a parasitic gap Additional examples of the animacy effects in clitic licensed parasitic gaps are shown in (18), from García Mayo (1995), who also observes the same animacy restriction with dative clitics, as shown in (18c,d). (18) a. Loi puse    sin   estirar  pgi bien.           (inanimate) It   put on.1 S G  without stretch.I N F    well “I put it on without stretching it well.” b. *Loi   vestí       sin    bañar     pgi antes.           (animate) Him  dressed.1 S G  without bathe.I N F  before “I dressed him without bathing him beforehand.” c. Siempre  lei   cambio   la rueda ti  antes  de poner  más  aceite pgi. Always  CL.3SG  change. 1SG the tire    before  of  put. INF  more oil. (inanimate) “I always change the tire before putting in more oil.” d. *Lei     hablé    por teléfono ti antes  de  mandar pgi la  carta.   (animate) CL. 3 S G  talked.1 S G   for telephone before  of  send. I NF   the letter.  “I talked to her on the phone before sending her the letter.”     (García Mayo, 1995)

After seeing similar patterns in the passive parasitic gap data, the prediction would be that the ungrammaticality of the animate examples is similarly caused by the controller of PRO in the clitic licensed parasitic gaps being the same element as the parasitic gap antecedent, leading to only a reflexive interpretation. Turning first to the inanimate clitic structures, the interpretation of the inanimate example in (18a) is that the subject of puse is also the subject of estirar. Since the matrix subject binds and controls PRO, the sentence displays OC. There is no semantic deviance caused by this structure since the adjunct verb, estirar, requires an agent as its controller and the matrix external argument pro can easily function as an agent. Therefore, OC is expected to (and does) prevail. In this case, the OC structure leads to a grammatical parasitic gap. This follows smoothly from the previous analysis. The matrix subject pro controls the adjunct subject, PRO. The parasitic gap antecedent is the clitic lo. Because the parasitic gap antecedent does not also control PRO, the structure is grammatical, as predicted. However, the animate clitic examples are more puzzling. The matrix subject (pro) would be expected to control the adjunct subject PRO, like in (18a). It is in the correct syntactic position to control and it would not produce a semantic deviance since the null subject pro would be a suitable agent for the verb bañar. In this case, since the clitic is the parasitic gap antecedent, there is not the same problem that occurs in the active subject and animate passive

Interaction of parasitic gaps and adjunct control  47 parasitic gaps. The same argument would not be both controlling PRO and the antecedent of the parasitic gap. If the matrix subject is controlling PRO, it is unclear why the parasitic gap is ungrammatical. There is, however, a second possibility for the controller of PRO:  the accusative clitic. Following a movement approach to clitics (Kayne, 1989, 1991; among others), the clitics in (18) would be moved to and located in T. The clitic’s features would then be in T at the time when adjunct control takes place, allowing the clitic to be in a suitable syntactic position to agree with and thus control PRO. Further, being animate, and thus a potential agent, if the clitic controlled PRO, there would not be a semantic deviance. Lo, meaning him, would also be an appropriate agent for bañar. This prediction that the clitic could (obligatorily) control PRO holds out empirically. In contrast with (18a), where the matrix subject controls PRO, there are also examples of similar structures where a clitic controls PRO in a similar temporal adjunct, as seen in (19). (19) Lai besé después de PROi poner-​se celosa. Her kissed.1S G after of PRO become.I N F -​SE jealous “I kissed her after she got jealous.” While (19) does not contain a parasitic gap, it does show that in a similar type of adjunct, it is possible for a clitic to control the adjunct subject. It could be that in the animate clitic parasitic gap examples, when the direct object clitic is animate and can be interpreted as an agent, the clitic wins out for the controller of PRO.8 If the clitic were to control PRO, and also be the parasitic gap antecedent, we would not predict the structure to be grammatical, which aligns with the data in (18b, d). Yet, in such a case, one would expect that adding the reflexive se would repair the structure, like in the active and animate passive structures. However, when the reflexive se is added to these structures, shown in (20), the grammaticality is not improved as it is expected to be. (20) *Loi vestí sin PROi bañar-​se antes. Him dressed.1S G without PRO bathe.I N F -​SE refl before “I dressed him without him bathing himself beforehand.” There may be an independent reason why (20) is not grammatical, but I leave this question open for further research. 3.3  Lack of animacy effects in A-​bar parasitic gaps This analysis of the animacy effects in parasitic gaps also explains why other types of parasitic gaps, specifically those licensed by wh-​ movement and topicalization, do not display any restrictions on the animacy of the gap’s

48  Katie VanDyne antecedent. Observe the examples in (21), from García Mayo (1995), where animate antecedents are grammatical. (21) a. ¿[A   quién]i viste   ti sin    saludar pgi?   (wh-​movement) D O M   who   see.2S G  without greet.I N F “Who did you see without greeting?” b. [La chica]i que viste  ti sin  saludar pgi. (topicalization) The girl  that see.2S G  without greet.I N F “The girl that you saw without greeting.” While García Mayo (1995) shows that animate parasitic gaps with wh-​ moved and topicalized antecedents are possible, but animate clitic licensed gaps are not, why these animacy effects are not found in all parasitic gaps has yet to be explained. However, following the analysis I present in this chapter, these results are expected. In the examples in (21) the matrix external argument agrees with T and controls PRO. Crucially, this controller of PRO, the matrix subject, is always distinct from the parasitic gap antecedent, which is moved to the left periphery. This is shown below in (22). (22) a.  ¿[A quién]i proj viste ti sin PROj saludar pgi? b.  [La chica]i que proj viste sin PROj saludar pgi. In both examples in (22), the matrix subject, pro, controls the adjunct subject PRO. The parasitic gap antecedent is moved to the left periphery, via wh-​ movement in (22a) and topicalization in (22b). Thus, there is never a situation where the controller of PRO is the same element as the A-​bar moved parasitic gap antecedent. Therefore, no animacy restrictions would be expected to arise in these structures.

4.  Conclusion In this chapter I  have presented an approach for accounting for previously unexplained characteristics of non A-​bar licensed Spanish parasitic gaps. Spanish appears distinct from other languages in that clitics and passive subjects can license a parasitic gap. However, with this wider distribution of parasitic gaps come further limitations. While passive subjects are able to license a gap, no active subjects can. Further, while passive subjects and clitics can license a parasitic gap, the antecedents of these structures must be inanimate. No animate antecedents are grammatical in those types of parasitic gaps.

Interaction of parasitic gaps and adjunct control  49 I have explained these patterns by looking at another empty argument in these structures:  PRO. By focusing on the controller of PRO and its interaction with the parasitic gap structures, we can better understand why these limitations arise. Crucially, in order for a parasitic gap to be licensed, the controller of PRO and the parasitic gap antecedent must be different elements. In the case of active subjects and animate passive subjects, the intended parasitic gap antecedent is also the constituent that controls PRO. In these structures, only a reflexive structure with an overt se is grammatical, and not a parasitic gap. While there is still more work to do regarding the exact analysis of clitic and passive licensed parasitic gaps in Spanish, the asymmetries discussed in this chapter are crucial to account for in any analysis of these structures.

Notes 1 Throughout this chapter I use pg to represent the parasitic gap, and t to represent the trace/​deleted copy of its antecedent. 2 This observation that postverbal subjects in passive constructions license parasitic gaps in Spanish raises the question of how the antecedent can c-​commanded the parasitic gap. This structure can be accounted for under a sidewards movement approach to deriving parasitic gaps following Hornstein (2001) and Hornstein and Nunes (2002) where the parasitic gap can sidewards move out of the adjunct to the matrix clause. I set a more detailed analysis aside for this chapter, however. 3 García Mayo (1994) discusses in detail a tensed/​non-​tensed asymmetry between English and Romance parasitic gaps. While English parasitic gaps can have a non-​ tensed (ia) or tensed (ib) verb in the adjunct, Romance parasitic gaps are only grammatical with non-​tensed adjuncts (ii). i. a. [What articles] did you file after reading pgi? b. [What articles] did you file after you read pgi? ii. a.  ¿[Qué artículos]i archivaste después de leer pgi? b. *¿[Qué artículos]i archivaste después de que leíste pgi? This requirement of having a non-​ tensed adjunct in Spanish leads to the presence of PRO in the adjunct subject of all of these examples. 4 For concreteness I adopt a Landau-​style approach to control (specifically following Landau, 2015, 2017), however the general patterns I  propose seem to be independent of any particular approach to adjunct control. 5 Although the definition can vary within each approach to control, obligatory control typically refers to structures where PRO is bound by a local antecedent and interpreted as a bound variable, while in non-​obligatory control PRO may be a free variable and does not need to be bound by another element. It can have a logophoric or arbitrary/​generic interpretation. (Landau, 2013). 6 While it seems that it is the implicit argument that controls in these cases, rather than PRO receiving an arbitrary interpretation, either interpretation would fall under the category of NOC. 7 As an anonymous reviewer points out, this analysis also gives further support towards the conclusion that these structures do indeed involve parasitic gaps,

50  Katie VanDyne rather than a type of ellipsis or null pronoun. If the structures were thought to be involving a null pronoun, following Condition B it would need to be free in its domain. However, we see that the parasitic gap is necessarily bound by PRO which is why adding the reflexive makes it grammatical. 8 It is not clear why the clitic controls instead of the matrix subject and, while interesting, I leave such a discussion aside for the current chapter. Crucially, there do appear to exist instances of an animate clitic in the matrix clause controlling adjunct PRO.

References Bošković, Ž. (1996). Selection and the categorial status of infinitival complements. Natural Language & Linguistic Theory, 14, 269–​304. Campos, H. (1991). Silent objects and subjects in Spanish. In H. Campos & F. Martínez-​Gil (eds.), Current Studies in Spanish Linguistics, 117–​142. Washington, DC: Georgetown University Press. Chomsky, N. (1982). Some Concepts and Consequences of the Theory of Government and Binding, Vol. 6. Cambridge, MA: MIT Press. Culicover, P.W. (2001). Parasitic gaps: A history. Current Studies in Linguistics Series, 35,  3–​68. Engdahl, E. (1983). Parasitic gaps. Linguistics and Philosophy, 6(1), 5–​34. Frascarelli, M. (2007). Subjects, topics and the interpretation of referential pro. Natural Language & Linguistic Theory, 25(4), 691–​734. García Mayo, M.P. (1994). Operators, clause-​boundedness and finiteness in English vs. Romance parasitic gap constructions. Revista de Filologıa de la Universidad de La Laguna, 13, 145–​152. García Mayo, M.P. (1995). The licensing of parasitic gaps. Miscelánea: A journal of English and American Studies, 16, 125–​140. Grimshaw, J. (1994). Minimal projection and clause structure. In B. Lust, M. Suñer & J. Whitman (eds.), Syntactic Theory and First Language Acquisition:  Cross Linguistic Perspectives, Volume I:  Heads, Projections and Learnability, 75–​83. Hillsdale, NJ: Erlbaum. Hornstein, N. (2001). Move! A  Minimalist Theory of Construal, Vol. 10. Malden, MA: Wiley-​Blackwell. Hornstein, N., & Nunes, J. (2002). On asymmetries between parasitic gap and across-​ the-​board Constructions. Syntax, 5(1), 26–​54. Kayne, R.S. (1989). Null subjects and clitic climbing. In O. Jaeggli & K.J. Safir (eds.), The Null Subject Parameter, 239–​261. Dordrecht: Springer. Kayne, R.S. (1991). Romance clitics, verb movement, and PRO. Linguistic Inquiry, 22(4), 647–​686. Landau, I. (2013). Control in Generative Grammar:  A Research Companion. New York: Cambridge University Press. Landau, I. (2015). A Two-​Tiered Theory of Control. Cambridge, MA: MIT Press. Landau, I. (2017). Adjunct control depends on voice. In C. Halpert, H. Kotek & C. van Urk (eds.), A Pesky Set: Papers for David Pesetsky, 93–​102. Cambridge, MA: MIT Working Papers in Linguistics. Manzini, M.R. (1994). Locality, minimalism, and parasitic gaps. Linguistic Inquiry, 25, 481–​508.

Interaction of parasitic gaps and adjunct control  51 Otero, C.P. (1999). Pronombres reflexivos y recíprocos. In I. Bosque & V. Demonte (eds.), Gramática descriptiva de la lengua española, 1, 1427–​1517. Madrid: Espasa-​Calpe. Reinhart, T., & Reuland, E. (1993). Reflexivity. Linguistic Inquiry, 24(4), 657–​720. Ross, J.R. (1967). Constraints on variables in syntax (Doctoral dissertation). MIT. Sheehan, M. (2015). Subjects, null-​subjects and expletives in Romance. In S. Fischer & C. Gabriel (eds.), Manual of Grammatical Interfaces in Romance, 329–​362. Boston: de Gruyter. Speas, M. (2006). Economy, agreement, and the representation of null arguments. In P. Ackema, P. Brandt, M. Schoorlemmer & F. Weerman (eds.), Arguments and Agreement, 35–​75. Oxford: Oxford University Press. Taraldsen, K.T. (1981). The theoretical interpretation of a class of marked extractions. In A. Belleti, L. Brandi & L. Rizzi (eds.), Theory of Markedness in Generative Grammar, 475–​516. Pisa: Scuola Normale Superiore. Williams, E. (1985). PRO and subject of NP. Natural Language & Linguistic Theory, 3(3), 297–​315.

3  “¿Qué traes güey?” An analysis of Spanish clausal possession Christian Ruvalcaba

1. Introduction1 In Spanish, the most productive way to express clausal possession is by using the stative verb tener (1a) where the possessor is the notional subject and the possessum is the object. In the Principles and Parameters framework, these verbs have been analyzed as copulas with an incorporated element; the incorporated element is argued to be a preposition (Freeze, 1992; Kayne, 1993; Hornstein et al., 1995; Belvin & den Dikken, 1997; Harley, 1995, 2002), an applicative head (Boneh & Sichel, 2010; Kyumin, 2012), a case head (Boneh & Sichel, 2010; Jayaseelan, 2007), to name a few proposals. The emerging pattern is that tener-​like verbs are a derived subtype of copulas. The reasoning behind this claim is that many languages lack have-​like verbs but nevertheless allow semantically equivalent possession constructions with copulas or verbs of existence.2 Spanish allows different copular forms (i.e., haber, estar, ser) for syntactic contexts where English can only employ to be. Likewise, Spanish also demonstrates variation in the form of its possessive verbs, namely tener (1a) and traer (1b-c), where English only employs have. Building on insights from Ruvalcaba (2018), I show how several semantic differences between these possessive verbs (tener, traer) correlate with the semantic differences between ser and estar, namely the former expresses a permanent relation and the latter expresses a temporary relation. In accordance with the idea that have/​tener is a copula with an incorporated element, I explain how Spanish’s two copular forms, ser and estar, derive tener and traer respectively. Constructions like (1d) are analyzed as decomposed instances of traer. (1) a. Adrian tiene una pluma. Adrian has   a  pen “Adrian has a pen.” b. Ellos traen plumas. They bring hunger/​ pens “They are hungry.”/​“They have pens.” c. Ellos   traen  hambre. They   bring  hunger “They bring hungry.”.

An analysis of Spanish clausal possession  53 d. Están con  hambre. Are  with hunger “They’re hungry.” Working within a Distributed Morphology approach (Embick & Noyer, 2007), I argue that the stative possessive verb tener is spelled out in contexts where a comitative preposition con incorporates into a stative little v (vBE) similar to Avelar’s (2004, 2009b) treatment of possessive ter (“to have”) in BP and Levinson’s (2011) treatment of English have constructions. Likewise, Spanish’s stative possessive verb traer decomposes into a comitative preposition and the perfective form estar, which I analyze as a complex element along the lines of Gallego and Uriagereka (2016).3 Simplified versions of the rules of exponence for tener and traer are illustrated in (2): (2) a. /​tener/​ ↔ [SER] [CON] b. /​traer/​ ↔ [ESTAR] [CON] In short, both verbs of possession in Spanish, tener and traer, are argued to result from incorporation of a preposition (i.e., con) into a copula, and both verbs retain the semantics of the copulas they derive from. Section 2 provides an account of tener as a complex copula derived from vBE and a comitative preposition. Section 3 provides an overview of the semantic characteristics of possessive traer constructions. Section 4 points out parallels between possessive traer constructions and possessive estar con (“to be with”) constructions. Building on observations made throughout this chapter, section 5 provides an analysis of tener constructions, traer constructions, and estar con constructions. Section 6 concludes the analysis.

2.  Tener as a complex copula Verbs like Spanish tener (and its English counterpart have) denote a wide range of relations grouped together under the descriptive “possession” label (3a–d). For instance, (3a) can express canonical alienable possession whereby the subject, Lupe, has ownership of the cars. In addition, (3a) expresses that Lupe is wealthy (not in temporary control of someone else’s money), which associates tener with permanent relations.4 The rest of the examples demonstrate that tener can express kinship relations (3b), abstract relations (3c), and part-​whole relations (3d). (3) a. Lupe tiene 2 carros/​dinero. Lupe has 2 cars/​money “Lupe has 2 cars/​money.”

54  Christian Ruvalcaba b. Lupe tiene hermanas/​hijos. Lupe has sisters/​kids “Lupe has sisters/​kids.” c. Lupe tiene dolor de cabeza. Lupe has pain of head “Lupe has a headache.” d. Lupe tiene 2 piernas. Lupe has  2 legs “Lupe has 2 legs.” As discussed in the introduction, verbs of ownership where the possessor is the subject (like tener) are typically analyzed as a copula with an incorporated element (Harley, 1995, 2002; Levinson, 2011; Kyumin, 2012; Kempchinsky, 1996; Kayne, 1993; Freeze, 1992; among others).5 A  general illustration of these approaches is illustrated in (4). In (4), “Z” stands for the element that incorporates into the stative verb vBE to derive tener. I should note that the diagram in (4) does not intend to make a claim about the position of preverbal subjects in Spanish (hence the XP instead of a TP), a debated issue. The exact position of preverbal subjects is discussed more in detail in section 5. (4) General illustration of the derivation of tener-​like verbs.

The current analysis adopts an analysis of tener-​like verbs proposed by Avelar (2004, 2009) for Portuguese and Levinson (2011) for English. These analyses identify the “Z” in (4)  as a comitative preposition, con. In other words, “Z” in (4) is identified as con. Levinson argues that the comitative preposition that generates verbs of possession is forced to move because it is affixal; it must incorporate to a higher head. In this case, the higher head is a stative verb.6 This results in a derivation like the one shown in (5a) where tener is inserted into the node containing the verb and the incorporated preposition at PF (5a–b):

An analysis of Spanish clausal possession  55 (5) a. Avelar/​Levinson analysis of possessive verbs

b. /​tener/​ ↔ [vBE] [CON] Levinson, unlike Avelar, also claims that the possessor is introduced by a V, not the P itself, as illustrated in (6). The reasoning behind this is that P cannot introduce an argument. The head that introduces arguments in the prepositional domain is small p (Svenonius, 2010), similar to v or Voice in the verbal domain. Because there is no small p in (6), Levinson (2011) claims a V introduces the possessor argument. (6) a.

Levinson argues that the possessive semantics in English have constructions are not encoded into the preposition. Instead, she claims that the combination of the accompaniment semantics of the preposition with and the stative semantics of to be are responsible for the implicature of possession.7 The analysis in (6) includes the additional head that introduces a possessor argument. It shows that both the P and the V incorporate into vBE to yield tener. The analysis for possessive traer and estar con constructions will build on the derivation proposed for tener (6).

3.  Overview of possessive traer Although the stative (possessive) denotation of traer (7a) has not received much attention in formal syntax,8 it follows typical diachronic patterns of semantic erosion from dynamic eventualities to stative ones. Heine (1997) notes that

56  Christian Ruvalcaba verbs like tener and its English counterpart have (1b) have dynamic past lives; they are “action” verbs which have undergone semantic bleaching. In fact, the Spanish verb tener denoted a “holding” activity until it eventually replaced haber (to have/​to exist) as a possessive verb in the 15th century. Throughout this “bleaching” process, these verbs are ambiguous between their dynamic reading and their static possession reading (Heine, 1997). While the dynamic reading has been completely lost for tener, it is still available for traer. In (7b), this reading is enforced by the clitic and the PP. In this sense, the verb expresses a notion of transfer and directionality. If one were to remove the benefactive clitic nos and the locative phrase “a la escuela,” the sentence could then express a possession relation where “Juan has a lunch.” In the possessive reading, the sentence expresses a transient relation between the subject and the object.9 (7) a. Demetrio trae dinero. Demetrio brings money “Demetrio has money (on him).” #“Demetrio is wealthy.” b. Demetrio nos trae lonche Demetrio cl.1.pl bring lunch “Demetrio brings our lunch to school.”

a to

la the

escuela. school.

Since traer is commonly omitted in possession studies, this section will provide a descriptive overview of its properties. It will also highlight similarities and dissimilarities it shares with the verb tener. An interesting difference between traer and the other possessive verb, tener, is that traer is restricted to non-​permanent or temporary relations, while tener prefers permanent possession. This is illustrated in (8). The sentence with traer in (7a) can express a reading where Juan had temporary control of some money, but it cannot express a reading that Juan is wealthy, which would be a more permanent relation. Both of these readings can be expressed in (8b) with tener. It is important to mention, however, that tener tends to denote ownership. This is illustrated with the data in (8c) where the phrase “but it’s not his” cancels out the ownership reading. As (8c) shows, this phrase is compatible with traer but marginal with tener. (8) a. Juan trae dinero. Juan brings money “Juan has money (on him).” #“John is wealthy.” b. Juan tiene dinero. Juan has money “Juan has money on him.” “Juan is wealthy.” c. Juan trae/​ ??tiene dinero pero Juan brings/​ has money but “Juan has money, but it isn’t his.”

no not

es is

de of

él him

An analysis of Spanish clausal possession  57 The data in (8)  suggests that traer, although stative, denotes a temporary possessive relation. It also shows that tener tends to be strongly associated with ownership or some kind of permanent relation. The aspectual contrasts of tener and traer are emphasized when the sentence includes an object that is typically “owned” rather than temporarily controlled. For example, someone could “have a car on/​with them,” a borrowed car, at utterance time, but typically people cannot “have a house on/​with them” in sense that they have temporary access to someone else’s house. This is illustrated in (9), where traer can denote possession with “car” objects (9a), but not with “house” objects (9b). The sentence in (9a) also carries an implicature that Alex is the owner, but it can be canceled out with traer verbs as demonstrated in (9c). In contrast, this implicature cannot be canceled out when the verb is tener (9d). This shows that traer cannot express permanent alienable possession, while tener cannot express temporary alienable possession.10 (9) a. Alex tiene/​ trae carro. Alex has/​ brings car “Alex has a car.” b. Alex tiene/​ *trae casa. Alex has/​ *brings house “Alex has a house.” c. Alex trae carro, pero Alex brings car but “Alex has a car but it isn’t his.” d. *Alex tiene carro, pero Alex has car but “Alex has a car but it isn’t his.”

no not

es is

de of

él. him

no not

es is

de of

él. him

Beyond alienable possession, the same contrast exists with abstract possession: constructions with objects that express abstract possessums (e.g., ailments, experiences etc.). Temporary ailments like headaches, colds, hunger, (feelings of) cold, etc., are compatible as objects of both possessive verbs (10a). Long-​term ailments like diabetes, however, are only compatible with tener, which tends to denote more permanent or imperfective relations. Since traer is restricted to temporary, perfective relations, it cannot have abstract objects that denote permanent or long-​term ailments (10b). (10) a. Pedro tiene/​ trae dolor de cabeza/​resfriado. Pedro has/​ brings pain of head /​ cold “Pedro has a headache/​cold.” b. Pedro tiene/​ * trae diabetes. Pedro has/​ brings diabetes “Pedro has diabetes.”

58  Christian Ruvalcaba As the data in (8–​10) show, the possesive verb traer can only express temporary control or temporary possession. This explains why it cannot express kinship relations or part–whole relations with animate subjects (11a), relations which tend to be permanent. These, however, are perfectly fine with tener. A similar pattern emerges from part–whole relations with inanimate subjects (11b). (11) a. Tengo/​ * traigo dos hermanos. /​dos manos. Have/​ bring two brothers /​two hands “I have two brothers/​two hands.” b. La  casa  tiene/ ​ *trae?? muchos cuartos. The  house has/  ​brings  many   rooms “The house has a lot of rooms.” The data in (11) and the discussion so far supports the general observation that traer denotes a temporary relation. This section has shown that traer is a productive verb of possession in Spanish. It is a stative verb that is restricted to temporary relations. On the other hand, tener seems to disallow a “temporary control” reading, especially when it comes to alienable possession (8–​9). Interestingly, Avelar (2009b) points out a similar contrast between ter (“to have”) and tá com (“be with”) constructions in BP. Both are possessive constructions, but the ter construction (12a) employs a possessive verb, and the tá com construction (12b) has a copula. The verb ter is essentially the BP equivalent of Spanish tener and English to have. Avelar (2009b, p. 167) points out that the copular construction tá com (12a) “must be taken as a transitory or recently acquired possession, expressing that Pedro has the money now, at this moment.” In contrast, the construction with the possessive verb ter (12b) is “normally taken as a more permanent or enduring possession. In other words [it] is easily interpreted as Pedro is rich.” (12) a. O Pedro tá The Pedro is “Pedro has money.” b. O Pedro tem The Pedro has “Pedro has money.”

com with

dinheiro. money

dinheiro. money

Avelar’s (2009b) analysis shows us that the distinction between ter and tá com constructions in BP is quite similar to the distinction between tener and traer in Spanish. Indeed, this parallelism points the way towards an analysis of traer. In the next section, I illustrate some of the parallels between the traer constructions and Spanish estar con constructions.

An analysis of Spanish clausal possession  59

4.  Parallels between estar con and traer This section demonstrates several parallels between traer and estar con constructions in Spanish. Like the Brazilian Portuguese constructions analyzed in Avelar (2009a, 2009b), Spanish can also express possession with a copula and a comitative preposition. As it turns out, estar con constructions are similar to equivalent constructions in BP and to traer constructions in that they are restricted to temporary control or possession, as discussed below. The Spanish estar con constructions appear to be less commonly used in alienable possession contexts when compared to BP. However, occasionally they do occur in alienable possession contexts such as the ones in (13a–b). In these cases, the reading is also temporary. For instance, in (13a), the permanent reading would mean that the commissioners are wealthy. However, the entire sentence states that they have to spend the money they currently have access to, not that they are permanently wealthy. (13) a. Los comisarios están con dinero y tienen The commisioners are with money and have to que gastarlo.11 that spend.it “The commissioners have money and need to spend it.” b. No  sabemos qué  pasa     con  él,  si ha Not know   what is.happening with him if has comido,  si  está  con  ropa.12 eaten     if  is   with  clothes “We don’t know what his situation is, if he has eaten, if he has clothes.” The second sentence (13b) is uttered by a woman whose son recently disappeared. She is speculating about his current state of being, not whether he is the owner of clothes. Thus, both of these uses suggest that estar con constructions denote temporary alienable possession.13 Although alienable possession seems limited with the estar con construction, it is productive with other types of possession relations, such as abstract possession. A  closer look at these constructions more readily reveals the perfective, temporary reading of estar con constructions. For example, both estar con and traer can express temporary ailments like “hunger” or “fever” (14a–b). However, when it comes to more long-​term ailments, neither traer (14d) nor estar con (14c) constructions are allowed (only tener is allowed, 10b). Likewise, inalienable and part–whole relations, which tend to be permanent, are disallowed with both estar con and traer constructions (14e–f). The same is true with kinship relations. As mentioned in the previous section, tener has no problem expressing these types of relations (11a–b).

60  Christian Ruvalcaba (14) a. Los niños   están con  mucha hambre/​ fiebre. The children are  with much  hunger/​ fever “The children are very hungry.” “The children have a very high fever.” b. Los niños   traen mucha hambre/​ fever. The children bring much  hunger/​ fever “The children are hungry.” “The children have a very high fever” c. *Los  niños    están  con  diabetes. The  children are  with diabetes “The children have diabetes.” d. *Los  niños   traen  diabetes. The   children  have  diabetes “The children have diabetes.” e. *La casa  está con  muchos cuartos. The house is  with many    rooms “The house has a lot of rooms.” f. *La casa  trae     muchos cuartos. The house brings     many    rooms ‘The house has a lot of rooms.’ g. *Está con  muchos hijos/​ sisters. Is   with  many   sons/​sisters. “He/​she has a lot of kids/​sisters.” h. *Trae   muchos  hijos/​hermanas Brings  many   sons/​sisters “He/​she has many kids/​sisters.” As the data in (14) show, traer and estar con are limited to the same temporary possessive relation. This is similar to the tá com constructions in BP (Avelar, 2009b). Moreover, these two constructions contrast sharply with the tener constructions which allow permanent possession readings. The next section builds on these observations to form an analysis of traer.

5.  Deriving Spanish’s possessive constructions As was just observed, the verb tener in Spanish is characterized by the fact that it can be used to express permanent relations (8c, 9c–d). In contrast, the verb traer (like estar con constructions) tends to be restricted to relations of temporary possession (14). Section 2 provided an analysis of tener that builds on Avelar’s (2004, 2009a, 2009b) work on BP and Levinson’s (2011)

An analysis of Spanish clausal possession  61 work on Icelandic, English, and German. These researchers both analyze tener-​like verbs in other languages as complex copulas; they are copulas with a comitative preposition incorporated into them. I extended this analysis to Spanish (7a), illustrated again below as (15). The base structure verb in (15a) is vBE, which would otherwise be spelled out as ser, a copular form that is simplex and known to express imperfective, permanent relations (Gallego & Uriagereka, 2016). However, in the cases being examined here, the comitative preposition con incorporates into vBE. Consequently, tener is inserted at PF into the node containing vBE and con, following the rule of exponence in (15b). (15) a. Lupe tiene carro. Lupe  has   car “Lupe has a car.”

b. /​tener/​ ↔ [vBE] [V]‌[CON] The diagram (15) shows the DP Lupe in a phrase labeled XP. This is because the preverbal subject position in Spanish has been argued to be above the specifier of TP (Ordóñez & Treviño, 1999). I assume that preverbal subjects in Spanish are Clitic Left-​Dislocated, base-​generated topics (CLLD). In this case, the specifier of the TP would be occupied by a co-​referent pro (Ordóñez & Treviño, 1999; Benincà & Poletto, 2004; Sheehan, 2006). Thus, a more complete illustration of the derivation I am proposing is included in (16). (16) a. Lupe tiene carro. Lupe  has  car “Lupe has a car.”

62  Christian Ruvalcaba

Aside from the position of the possessor, the derivation in (16) is the same as the one illustrated in (15). As proposed by Levinson (2011) the “possessor,” in this case a pro that co-​refers with the CLLD subject Lupe, is introduced by a V. This V has merged with a PP headed by an impoverished14 P, con. The preposition con is affixal and must incorporate into a higher head. In this case, the higher heads are V and vBE, as illustrated in (16). As discussed above, the result is the insertion of the Vocabulary Item tener. Previous sections showed that the verb traer expresses a stative possession relation but, unlike tener, it is restricted to a temporary relation (see sections 3 and 4). In fact, it patterns more closely with possessive estar con constructions than with tener constructions. Given the parallelism between traer and estar con constructions (14), I  propose that traer is essentially the result of con incorporating into estar rather than ser. A  simplified version of this derivation is illustrated in diagram (17): (17) a. Lupe trae  carro. Lupe brings car ‘Lupe has a car.’

An analysis of Spanish clausal possession  63

b. /​traer/​ ↔ [ESTAR] [V]‌[CON] This is essentially the same derivation of tener (16) except that now the vP is headed by estar rather than vBE or simplex ser. The rest of the process is the same as proposed for tener. As discussed for tener, the comitative preposition con incorporates into heads into V and vBE (here spelled out as estar) yielding traer (17a–b). An obvious question is why vBE is sometimes labeled as estar (to derive traer) and as ser (to derive tener). I adopt Gallego & Uriagereka’s (2016) idea that estar is a derived element. They argue that ser is a simplex copula (the spellout of a bare vBE), but if a spatio-​temporal element incorporates into it, then the spellout is estar. The rule of exponence for estar is illustrated in (18b), where “X” is a heuristic label for some spatio-​temporal element. The spatio-​temporal element contextually grounds the relation and, as a consequence, the relation gets a perfective, temporary reading. In contrast, a bare vBE expresses a permanent, imperfect relation because it is not contextually grounded (18a). (18) a. /​ser/​ ↔ [vBE] b. /​estar/​ ↔ [vBE] [X]‌ As mentioned above, tener is derived from a bare vBE plus con, and this explains why it is preferable for permanent control (alienable) possession constructions, inalienable possession, and permanent abstract possession. It follows, then, that if the comitative preposition instead incorporates into a node containing vBE plus an additional spatio-​temporal element (“X”), namely the more temporal estar, then the spellout would be the temporal possessive verb traer.

64  Christian Ruvalcaba As discussed above, traer is restricted to a transient relation, similar to estar con constructions. One of the key points of this analysis is that this aspectual characteristic of traer is due to the fact that it is derived from the copula estar which also tends to express temporary eventualities. If we accept that estar is complex (a vBE plus a spatio-​temporal “X”), then a more detailed derivation for traer would look something like the diagram in (19). (19) a. Lupe  trae   carro. Lupe brings car “Lupe has a car (on/​with her).”

b. /​traer/​ ↔ [vBE] [X]‌[V] [CON] The derivation above essentially builds on two ideas. The first is that a comitative preposition incorporates into a copula to form a possessive verb where the possessor c-​commands the possessum (Avelar, 2004, 2009b; Levinson, 2011). The second is that the temporal copula estar in Spanish is a complex element derived from ser (vBE) and a spatio-​temporal element, “X,” that contextually grounds the relation to a particular time and/​or place. The semantics of the elements that end up forming estar in ordinary, non-​possessive clauses (Lupe está dormida/​en la tienda) carry over into the possessive verb traer. For traer to be inserted at PF, however, both the spatio-​temporal element, “X,” and the comitative preposition, con, must be present (19b). An analysis like this explains the patterns of temporary possession discussed above. It also explains the parallelisms between traer and possessive constructions which overtly contain estar con (14). Finally, there is the question of how estar con constructions are possible if con always incorporates into a higher head. To account for this, I  resort

An analysis of Spanish clausal possession  65 to Levinson’s (2011) analysis of “possessive” attributive constructions (the man with the beard/​the new car) by claiming that the preposition incorporates into a small p (20). This prevents con from moving into the verbal domain. The possessive reading in estar con constructions stems chiefly from the accompaniment semantics of the preposition con (similar to tener and traer constructions) and the “possessor” is introduced by the little p which denotes an underspecified relation, as shown in (20). Levinson (2011) calls this semantically underspecified light preposition pdative. She proposes that it values the uninterpretable case features of the prepositional complement (hambre).15 Above the pP in (20), one can see that an “X” incorporates into vBE to derive estar. (20) a. Lupe está con hambre. Lupe  is    with  hunger “Lupe is hungry/​Lupe has an appetite.”

An analysis like (20) thus accounts for why estar con constructions are possible; a small p is present and this small p blocks incorporation of con into estar. In other words, (20) is a decomposed version of traer constructions. The only difference is that estar con constructions have a pP layer above the PP and traer constructions do not.

6.  Additional syntactic parallels between tener, traer and estar con This section focuses on syntactic patterns between tener/​traer/​estar con constructions to complement the primarily semantic evidence discussed above.16 First, I  show that, in addition to a small p, a negation head can also block incorporation of con into the verb, yielding copular possession

66  Christian Ruvalcaba sentences that pattern closely with traer. This explains the parallels between no trae and está sin constructions discussed in detail below. There are co-​ indexation restrictions between the subject and a silent element in the predicate in tener/​traer constructions that patterns with a similar restriction in various constructions with comitative prepositions, similar to the data discussed in Avelar (2009b). As mentioned above, estar con constructions are less acceptable than verbal possession constructions when it comes to alienable possession. For instance, estar con constructions like (21a) are less acceptable than their verbal counterparts (21a–b). I argue that this is because con is a weak preposition that must incorporate into a higher head in Spanish as argued for English by Levinson (2011), especially in alienable possession constructions.17 Thus (21a) is less acceptable because the preposition needs to incorporate into some other head. The sentence in (21c) is a more acceptable because, as argued above, the preposition has incorporated into the verb. (21) a. ??Adrian está con  cartera/​ pantalones. Adrian  is  with wallet /​ pants “Adrian has a wallet.”/​“Adrian has pants.” b. Adrian tiene cartera/​ pantalones. Adrian  has   wallet /​pants “Adrian has a wallet.” /​“Adrian has pants.” c. Adrian trae  cartera/​ pantalones. Adrian  brings  wallet /​pants “Adrian has a wallet.” /​“Adrian has pants.” Interestingly, the negated counterpart of (21a), shown below as (22a), is more acceptable and it maintains the same possessive reading as the verbal possession constructions in (22b–c). The data in (22) also aligns with the semantic patterns discussed in previous sections; the sentence with tener (22b) suggests that Adrian does not own pants, while the sentences with estar sin (22a) and traer (22c) suggest that Adrian is not wearing pants. Furthermore, the data in (22) suggests that estar sin is the same as estar con except that the former involves negation. (22) a. Adrian  está sin   cartera/​ pantalones. Adrian is  without wallet /​ pants “Adrian doesn’t have a wallet.”/​“Adrian doesn’t have pants.” b. Adrian no tiene cartera/​ pantalones. Adrian  not  has   wallet /​pants “Adrian doesn’t have a wallet.” /​“Adrian doesn’t have pants.” c. Adrian no  trae   cartera/​ pantalones. Adrian not brings wallet /​ pants “Adrian doesn’t have a wallet.” /​“Adrian doesn’t have pants.”

An analysis of Spanish clausal possession  67 As proposed above, the preposition con is affixal and must incorporate into a higher head, especially in alienable possession constructions. One can argue, then, that sin in (22a) started out as con but incorporated into a negation head, as shown in (23a). The Vocabulary Item sin is inserted post-​syntactically in accordance with the rule of exponence in (23b). This incorporation into Neg0 satisfies the need for con to incorporate into another head and also captures why the preposition is spelled out independently in alienable possession constructions. (23) a. [vP estar [VP Adrian [V0 [NegP Neg0 +coni [PP coni [NP pantalones]]]]]] b. /​sin/​ ↔ [NEG] [CON] The contrast between (21a) and (22a) is rooted in the morphosyntactic properties of con and its surrounding syntactic environment, namely the presence or absence of Neg0. As mentioned in previous sections, a small p can also block incorporation of con into the verb. However, these light prepositions seem to be more restricted to contexts with abstract predicates (e.g., “hunger,” “cold,” “headache”). Other syntactic patterns unifying verbs of possession and possessive comitative constructions are discussed in (Avelar, 2009b) for BP. He discusses a “need for the grammatical subject of possessive sentences to be interpreted as the possessor of some element that is within the possessive predicate” (p.  147). In contrast, the grammatical subject of verbs of perception (e.g., ver “to see”) do not have this co-​indexation restriction. This suggests “there are invisible features of the preposition com within the structure of ter-​ constructions” (p.  145), the same argument developed in this chapter. The data in (24) replicate the data provided by Avelar (2009b) for BP and show that Spanish also requires co-​indexation between the subject of the sentence and some element (labeled “ec” by Avelar) within the predicate. In the sentence (24a), for instance, the “bag” or “pocket” necessarily belongs to Pedro. The same is true for estar con constructions (24b). In contrast, the verb ver in (24c) allows a reading where the “bag” or “pocket” can belong to anyone. The data in (24d–f) show the same effect, but here the silent element co-​indexed with Pedro is attached to the figure (i.e., el dinero, “the money”) of the locative phrase. This means that the “money” belongs to Pedro in (24d) and (24e), but it can belong to anyone when there is neither a comitative preposition or possessive verb present. (24) a. Pedroi tiene/​ trae  el  dinero de Maria en [la Pedro has /​ brings the money of Maria in  the bolsa eci/​*j.] bag/​pocket “Pedro has Maria’s money in his bag/​pocket.”

68  Christian Ruvalcaba b. Pedroi está con  el  dinero  de  Maria en  [la  bolsa eci/​*j]. Pedro  is   with the money  of  Maria in  the  bag/​pocket “Pedro has Maria’s money in his bag/​pocket.” c. Pedroi ve  el  dinero  de Maria en [la bolsa eci/j] Pedro  sees the money of Maria in  the bag/pocket “Pedro sees Maria’s money in the bag/​pocket.” d. Pedroi  tiene/​trae   [el   dinero eci/​*j] en la  bolsa     de Pedro  has /​ brings the money   in  the bag/​ pocket of María. Maria “Pedro has money in Maria’s bag/​pocket.” e. Pedroi está con  [el  dinero eci/​*j] en la  bolsa    de Pedro is   with the money   in  the bag/​ pocket of María. Maria “Pedro has money in Maria’s bag/​pocket.” f. Pedro  ve    [el  dinero i/​j] en la  bolsa    de María. Pedro sees the money  in  the bag/​ pocket of Maria “Pedro sees the money in Maria’s pocket.” Avelar (2009b) also points out that this co-​indexation requirement exists at the nominal level or in non-​sentential contexts. The only difference, then, between (25) and (24) is that a copula merges with the predicate in (24).18 (25) a. Aquél hombre con la foto de María en [la bolsa. ec i]. That man with the picture of Maria in the pocket “That man with Maria’s picture in his pocket.” b. Aquél hombre con [el dinero ei] en la  bolsa de María. That man with [the money ei] in the  bag  of Maria “That man with money in Maria’s bag/​pocket.”

In short, there is a required co-​indexation restriction between the grammatical subject and an element in the predicate for verbal possession constructions and copular-​ comitative possession constructions, but it is absent with other verbs like ver. This restriction also exists in non-​verbal domains (25) with comitative prepositions. This syntactic pattern suggests that verbs of possession are derived from a process where a comitative preposition incorporates into a copula in Spanish like in BP.19

7.  Conclusion This analysis shows that Spanish expresses clausal possession in multiple ways. These include tener (26a), traer (26b), and estar con constructions (26c).

An analysis of Spanish clausal possession  69 I argued that all of these constructions are derived from a similar base structure which contains a stative copula and a comitative preposition, con. (26) a. Luz tiene carro. Luz has car “Luz has a car.” b. Luz trae carro. Luz brings car “Luz has a car (currently).” c. Luz está con hambre. Luz is with hunger “Lupe is hungry/​Lupe has an appetite.” Tener constructions tend to express ownership or permanent possession. This is because they derive from the incorporation of con into ser (vBE). The accompaniment semantics of con and the stative semantics of the copula together produce a possessive reading. Traer verbs, on the other hand, are formed with estar, not simplex ser. Building on Gallego and Uriagereka’s idea that estar is formed when a contextually grounding element (“X”) incorporates into vBE, I proposed that traer is inserted when “X” is present (27b). This explains the temporary possession reading of possessive traer verbs (i.e., it inherits them from estar). Finally, estar con constructions are formed when a light preposition merges with the PP headed by con. The P incorporates into p, spelling out con (27c), and preventing it from incorporating into the verb. (27)  a. /​tener/​ ↔ [vBE] [CON] b. /​traer/​ ↔ [vBE] [X]‌[CON] c. /​con/​ ↔ [pdative] [CON] The current analysis thus accounts for three inter-​ related possession constructions, including the lesser studied traer construction. In future work, the current approach can investigate other verbs of transfer, like llevar, which seem to express a possessive-​like relation. This can be seen in sentences like (28a–b). (28) a. Llevan   mucha  hambre. Take.pl.3 much  hunger “They are very hungry (currently).” b. Las ventanas del  baño   no  llevaban mosquitera. The windows of.the bathroom not take.imp screen “The bathroom windows didn’t have screens.”20

70  Christian Ruvalcaba c. ¿ Qué  traes  güey? What bring dude “What’s with you dude?” Although they do not discuss possession, Fernández-​ Soriano and Rigau (2009) find interesting parallels in the verbs llevar and tener when they are used as temporal light verbs (in sentences like llevo/​tengo tres horas esperándote ~ I have 3 hours here waiting for you). They argue that llevar verbs are formed with a directional (to) preposition whereas tener verbs are formed with a abstract static preposition of central coincidence, along the lines of Kayne (1993). Interestingly, these can also be rephrased with estoy con tres días sin dormer ~ I’m with 3 days without sleeping. It is possible that these types of constructions are related to the possessive constructions discussed so far. Lastly, future work can look into experiencer constructions to see whether there are any verbs that show the ser-​ tener/​estar-​traer contrast. Possession constructions and experiencer constructions tend to exhibit similar syntax (Landau, 2009) and several papers have claimed they derive from similar base structure (Harley, 2002; Ruvalcaba, 2018). The verb traer seems to express a state of “bother” or annoyance. For instance, in (28c), the question essentially translates into “what’s wrong with you?” or “what’s bothering you?” Interestingly, the English version uses a copula and a comitative preposition with to express an equivalent meaning. This fits with the idea that traer is a copula plus a comitative preposition, as argued above. In short, this analysis sets a point of departure for many new interesting lines of inquiry.

Notes 1 Thank you to the two anonymous reviewers for their feedback and to the editors of this volume. I  also want to express my gratitude to Antxon Olarrea for his help with this idea in its earlier stages. Correspondence concerning this chapter should be addressed to Christian Ruvalcaba, Confluencenter for Creative Inquiry, University of Arizona, 1133 E.  Helen St. Tucson, AZ 85721. Contact: [email protected] 2 As noted by many researchers, some languages express clausal possession with a copular construction (a-​c). In addition, have in British English patterns with to be in that, like to be and modal verbs, have can take a negation contraction (I haven’t a clue). a. U  menja  byla  sestra.                     (Russian, in Freeze, 1992) at 1sg.gen was sister.nom “I had a sister.” b. Liisa-​lla    on  mies                      (Finnish, in Freeze, 1992) Lisa-​ adessive be man “Lisa has a man.” c. Hún   er með bækurnar    fimm.       (Icelandic, in Levinson, 2011) she.nom  is with books-​the.acc five  “She has five books.”

An analysis of Spanish clausal possession  71 3 Gallego and Uriagereka (2016) argue that estar is essentially ser plus a contextually confining element, X (a). They argue that this additional element confines the relation to a particular context. This confinement is connected to the “transient” or “perfective” meaning of the estar verb. Thus a more exact rule of exponence for traer is illustrated in (b). a. /​estar/​ ↔ [vBE] [X]‌ b. /​traer/​ ↔ [vBE] [X]‌[CON] 4 A reviewer points out that the aspectual nature of tener may be dictated solely by the object. The nature of the object is relevant but does not capture the fact that tener is unable to express a transitory relation when it comes to alienable possession (see ­example 9d). 5 For a different account of tener-​like verbs, see Myler (2016). 6 In other cases, the P may incorporate into a p (Svenonius, 2010; Levinson, 2011) similar to how a V incorporates into a v in the verbal domain. The question of whether all Ps are affixal is an open one. Ruvalcaba (2018) proposes that some locative and dative prepositions are also weak prepositions that must incorporate into a higher head. 7 The nature of the subject may also play an important role in bringing about the possessive reading. For Freeze (1992), animacy is what turns a “locative” role into a “possessor” role. Avelar 2009a, 2009b finds constructions identical to (6) in BP (with the verb ter, “to have”) that express an existential reading when the subject is a PP and a possessive reading when the subject is a DP. 8 See Rábago Tánori and Alarcón Neve (2015) for a Construction Grammar analysis of stative traer. 9 A reviewer points out that traer implies movement and that the object must be “carried” or be present on the subject’s person at utterance time. The “movement” reading, however, appears to be absent in other constructions with traer (a–c). The verb traer in (a–c) is functioning syntactically and semantically like the verb tener. The only difference is an aspectual one; traer denotes a more transient relation. Nevertheless, it is possible that traer could be derived from a pseudo-​copulative element like andar (anda con mucho dinero ~ he has a lot of money on him) which may encode a “motion” reading. Gallego and Uriagereka (2016) argue that the element (called “X”) that incorporates into the copula to form estar spatio-​temporally confines the otherwise imperfective stative copula. It may be that different spatio-​ temporal elements ground the eventuality in different ways (some may associate the relation to a particular time, others to a particular location or circumstance etc.). The result would be a verb that has unique implicatures stemming from this “X” element. One could argue that a prolative preposition like por (“along”) incorporates into ser to the pseudo-​copulative form andar. The verb andar would inherit the meaning of por which may implicate a type of motion. If a comitative preposition were to incorporate into andar, then we would end up with traer, a possessive verb that implicates a type of movement. I leave this possibility open for future research. (a) Trae   la  cara  sucia Brings the face dirty “His/​her face is dirty (right now),” candidato trae a todo el estado (b) El The candidate brings to all the state “The candidates has got the whole state excited.”

alborotado excited

72  Christian Ruvalcaba (c) Me     traes  muy  preocupado Cl.1.sg bring very worried “You’ve got me very worried.” 10 A reviewer notes that terms like “temporary” are not specific aspectual terms. Throughout this chapter, I use the categories of semantic possession put forth in Heine (1997). Heine (1997) argues that there are two subcategories of alienable possession, namely temporary and permanent possession. The latter is defined as something akin to ownership. This distinction is similar to Tenny’s (1994) idea that certain verbs have a terminus or endpoint encoded and this endpoint can optionally be identified by a complement. When it comes to alienable possession, tener appears to have no inherent endpoint or terminus encoded in it, while traer does. 11 This sentence was retrieved from the following article: https://​peru21.pe/​politica/​ procuraduria-​denuncia-​inspector-​general-​policia-​corrupcion-​116721 12 Retrieved from www.eldeber.com.bo/​bolivia/​Ordenan-​detener-​a-​dos-​activistas-​y-​ crece-​la-​grieta-​en-​la-​oposicion-​20181217-​9086.html 13 Estar con constructions can actually be highly productive but in their negated form: están sin dinero/​ropa. 14 Decompositional approaches to prepositions have proposed that Ps are complex. According to Svenonius (2010), a small p can merge with a P, and it is this small p that “introduces a figure in neo-​Davidsonian fashion (parallel to Kratzer’s 1996 voice head in the verb phrase)…[and] is the natural locus of relational notions of containment, attachment, and support, which are commonly expressed by prepositions such as in and on and their counterparts cross-​linguistically” (Svenonius, 2010). If a p were present in (16), the preposition con would incorporate into it. Consequently, the preposition would be spelled out independently of the verb. Since con lacks a p layer in (16), the P con ends up incorporating into a higher head, namely V and vBE. 15 In fact, the name pdative comes from the fact that, in German, similar constructions mark the prepositional complement (the possessum) with dative case. 16 An anonymous reviewer asked for more syntactic evidence in favor of the incorporation analysis proposed here. The reviewer suggested including diagnostics used by Avelar (2009b) in his treatment of Brazilian Portuguese ter. 17 I leave the question of why estar con constructions (where the preposition is spelled out independently of the verb) are more acceptable with abstract predicates (i.e., hambre, frío etc.) than alienably possessed predicates for future research. For now, I assume that it has to do with the nature of the light p that merges with con in estar con constructions. The semantic nature of this small p may be incompatible with alienably possessed predicates. 18 This is further supported by the fact that estar in estar con constructions can be replaced by other verbs and still denote a possessive relation. (a) Juan durmió /​viajó con el dinero Juan slept /​traveled with the money “Juan slept/​traveled with Maria’s money in his bag.” /​viajó con el dinero (b) Juan llegó Juan arrived /​traveled with the money “Juan traveled with his money in Maria’s bag.”

de María en of Maria in en la in the

la bolsa the bag

bolsa de bag of

María Maria

An analysis of Spanish clausal possession  73 19 Avelar (2009) also points out that ter can be used in existential constructions, thus supporting the idea that this verb of possession is essentially a copula. The existential use of ter is traced back to the lack of covert subjects in BP (a property that is itself traced to BP’s impoverished agreement system). Since Spanish does allow covert subjects, the verb ter is unable to express an existential reading and thus patterns more closely with European Portuguese’s ter. Interestingly, Torrego (2002) argues that the existential verb haber in Spanish (previously a possessive verb) is derived from the incorporation of a non-​directional reference marker a into a copula. A closer look into how these two languages’ diachronic trajectories converge and contrast is needed. 20 This sentence was taken (with some modifications) from the following online thread: www.autocaravanas.es/​foro/​tema/​por-​que-​la-​ventana-​de-​los-​banos-​no-​ llevan-mosquitera/​#

References Avelar, J. (2004). Dinâmicas morfossintáticas com ter, ser e estar em português brasileiro (Master’s thesis). Universidade Estadual de Campinas. Avelar, J. (2009a). On the emergence of TER as an existential verb in Brazilian Portuguese. In P. Crisma & G. Longobardi (eds.), Historical Syntax and Linguistic Theory, 158–​175. Oxford: Oxford University Press. Avelar, J. (2009b). The comitative-​copular basis of possessive-​existential constructions in Brazilian Portuguese. In J. Nunes (ed.), Minimalist Essays on Brazilian Portuguese Syntax, 139–​160. Amsterdam: John Benjamins. Belvin, R., & den Dikken, M. (1997). There, happens, to, be, have. Lingua, 101, 151–​183. Benincà, P., & Poletto, C. (2004). Topic, focus and V2: Defining the CP sublayers. In L. Rizzi (ed.), The Structure of CP and IP, 52–​75. Oxford: Oxford University Press. Boneh, N., & Sichel, I. (2010). Deconstructing possession. Natural Language and Linguistic Theory, 28(1), 1–​40. Embick, D., & Noyer, R. (2007). Distributed morphology and the syntax-​morphology interface. In G. Ramchand & C. Reiss (Eds.), The Oxford Handbook of Linguistic Interfaces, 289–​324. Oxford: Oxford University Press. Fernández-Soriano, O., & Rigau, G. (2009). On certain light verbs in Spanish:  The case of temporal tener and llevar. Syntax, 12(2), 135–​157. Freeze, R.A. (1992). Existentials and other locatives. Language, 68, 553–​595. Gallego, Á., & Uriagereka, J. (2016). Estar = Ser + X. Borealis, 5(1), 123–​156. Harley, H. (1995). Subjects, events and licensing (Doctoral dissertation). MIT. Harley, H. (2002). Possession and the double object construction. Linguistic Variation Yearbook, 2, 31–​70. Heine, B. (1997). Cognitive Foundations of Grammar. Oxford: Oxford University Press. Hornstein, N., Rosen S.T., & Uriagereka, J. (1995). Integral predication. In J. Camacho, L. Choueiri & M. Watanabe (eds.), Proceedings of the XIV Annual Meeting of the West Coast Conference on Formal Linguistics, 169–​184. Stanford, CA: CSLI. Jayaseelan, K.A. (2007). The argument structure of the dative construction. In E. Reuland, T. Battacharya & G. Spatha (eds.), Argument Structure, 37–​48. Amsterdam: John Benjamins.

74  Christian Ruvalcaba Kayne, R.S. (1993). Toward a modular theory of auxiliary selection. Studia Linguistica, 47,  3–​31. Kempchinsky, P. (1996). Perfective auxiliaries, possession and existence. In K. Zagona (ed.), Grammatical Theory and Romance Languages, 135–​144. Amsterdam:  John Benjamins. Kratzer, A. (1996). Severing the external argument from the verb. In J. Rooryck & L Zaring(eds.), Phrase Structure and the Lexicon, 109–​137). Dordrecht: Kluwer. Kyumin, K. (2012). Argument structure licensing and English have. Journal of Linguistics, 1,1–​35. Landau, I. (2009). The Locative Syntax of Experiencers. Cambridge, MA: MIT Press. Levinson, L. (2011). Possessive WITH in Germanic: HAVE and the role of P. Syntax, 14, 355–​393. Myler, N. (2016). Building and Interpreting Possession Sentences. Cambridge, MA: MIT Press. Ordóñez, F., & Treviño, T. (1999). Left dislocated subjects and the prodrop parameter: A case study of Spanish. Lingua, 107, 39–​68. Rábago Tánori, Á., & Alarcón Neve, L. (2015). Traer estativo en español de México: su estructura sintáctica y propiedades semánticas. Cuadernos de Lingüística Hispánica, 26,  23–​47. Ruvalcaba, C. (2018). Deriving possession from location, accompaniment, and paths to nowhere: A microcomparative analysis of clausal possession in English and Spanish (Doctoral dissertation). University of Arizona. Sheehan, M. (2006). The EPP and null subjects in Romance (Doctoral dissertation). Newcastle University. Svenonius, P. (2010). Spatial P in English. In G. Cinque & L. Rizzi (Eds.), Mapping Spatial PPs:  The Cartography of Syntactic Structures, Vol. 6, 127–​ 160. New York: Oxford University Press. Tenny, C.L. (1994). Aspectual Roles and the Syntax-​Semantics Interface. Dordrecht: Kluwer. Torrego, E. (2002). Aspect in the prepositional system of Romance. In T. Satterfield, C. Tortora & D. Cresti (eds.), Current Issues in Romance Languages: Selected Papers from the XXIXth Linguistic Symposium on Romance Languages (LSRL), 337–​357. Amsterdam: John Benjamins.

4  Invariable qué-​questions (IQQs) in Spanish Javier Fernández Sánchez and Alfredo García Pardo

1. Introduction1 In this chapter we focus on a type of interrogative construction in Spanish which we will refer to as Invariable Qué-​Questions (IQQs). IQQs consist of two parts, which are clearly separated by a prosodic break: the first part resembles a wh-​question, but the wh-​word is invariably qué “what,” hence the name of the construction. This qué immediately precedes the lexical verb. The second part introduces the tag, i.e., a piece of information which requires confirmation. (1) Qué estás, en la What are.2sg in the “Are you in the library?”

biblioteca?2 library?

IQQs are interpreted as biased, total questions: more in particular, (1) would be uttered in a context where it is very likely that the speaker knows that his addressee is in the library. Thus, (1) could be thought of an indirect evidential in the sense of Izvorski (1997), i.e., an epistemic modal of sorts. This chapter, however, focuses on the syntactic side of the construction. IQQs have received some attention in the literature, see López-​Cortina (2007:  241) for references, and our section 4 for a brief review of previous proposals. Our goal in this chapter is threefold:  first, we present some syntactic properties of IQQs, some of which, to our knowledge, have not been discussed previously, (ii) we reject earlier analyses and (iii) we sketch a new proposal. We will argue that the qué in IQQs is not really a wh-​phrase, but rather a complementizer of sorts base-​generated in the left periphery, with no direct derivational link with the tag. Our chapter is structured as follows:  Section 2 discusses the main syntactic properties of IQQs. We will not deal with suprasegmental phonology or pragmatics, and although we believe that a pragmatic account of IQQs may shed some light on the phenomenon, this is a task we leave for further research. In section 3 we further show that despite the prima face similarities,

76  Fernández Sánchez and García Pardo IQQs cannot be analyzed as a kind of split question in the sense of Arregi (2010), given that the two phenomena are subject to important asymmetries. We address, albeit briefly, previous accounts of IQQs in section 4.  Despite their merits, they all present important empirical shortcomings. We sketch our proposal in section 5, before presenting the main conclusions. A word of caution is in order. To the best of our knowledge, no detailed dialectological study of this construction exists (but see López-​Cortina, 2009). The variety of Spanish that this chapter is based on is Castilian Spanish. However, it may very well be that other varieties of Spanish conflict with the data analyzed here (see section 6 for further discussion), or that some varieties lack this construction altogether. We now proceed with the chapter, having restricted the dialectal scope of our data.

2.  The phenomenon 2.1  Nature of the tag Let us first focus on the properties of the tag. Any phrase can be a tag, independently of its form and function. This is illustrated in (2): (2) a. Qué  saludaste,   a  Pedro? what greeted.2sg to Pedro “Did you greet Pedro?”              (Direct object, a-​DP3) b. El  libro  qué  se  lo  dio,     a   Pedro? the book what him it  gave.3sg to Pedro “Whom did you give the book to?”           (Indirect object, PP) c. Qué  estudiáis, en la  biblioteca? what study.2pl in  the library “Are you studying in the library?”             (Adjunct, PP) d. Qué  está,  enferma? what is.3sg sick “Is she sick?”                 (Subject predicative, AP) e. El  pescado qué  lo comen, crudo? the fish   what it  eat.3pl  raw “Do they eat the fish raw?”           (Object predicative, AP) f. Qué  fueron,   a   París? what went.3pl to Paris “Did they go to Paris?”           (Locative complement, PP) g. Las bebidas qué  las  compra, Susana? The drinks   what them buy.3sg   Susana “Is Susana buying the drinks?”              (Subject, NP)

Even some verbal periphrases can be split. In these cases, the tag is formed, minimally, by the non-​finite verb.

Invariable qué-​questions (IQQs) in Spanish  77 (3) Qué estás, viendo la what are.2sg watching the “Are you watching the TV?”

tele? TV

(4) Qué tenéis, que estudiar? what have.2pl that study.inf “Do you have to study?” (5) Qué debo, llamar a la policía? what must.1sg call.inf to the police “Should I call the police?” 2.2  Invariable qué As one can observe from the previous examples, qué (what) is invariable across the board. This means that even when the corresponding tag is a PP, qué cannot be headed by that preposition: (6) a. Pedro (*de) qué  es,  de  Málaga? Pedro from what is.3sg from Málaga “Is Pedro from Málaga?” b. (*A) qué le has dado el regalo, a tu hermana? to what her have.2sg given the gift to your sister “Have you given the present to your sister?” c. (*A) qué vas, a París? to what go.2sg to Paris “Are you going to Paris?” 2.3  Embedding IQQs are main clause phenomena (Hooper & Thompson, 1973; among others), which means that they cannot be embedded (7a), contrary to regular questions (7b): (7) a. *Me    pregunto qué  has   ido,   a  Londres? to.me wonder  what have.2sg gone to  London (Intended: “I wonder if you’ve been to London.”) b. Me  pregunto dónde has    ido. to me wonder  where  have.2sg gone “I wonder where have you been.”

78  Fernández Sánchez and García Pardo 2.4  Strict adjacency qué-​V Qué and the finite verb must be strictly adjacent. Thus, for example, these two elements cannot be separated by a subject: (8)  *Qué tú vas, en tren? what you go.2sg by train “Are you going by train?” One could hypothesize that the ungrammaticality of (8) is simply a corollary of the fact that subjects cannot appear between a regular wh-​word and a finite verb in European Spanish: (9)  *Dónde tú  vas? where  you go.2sg “Where are you going?” We submit that the reason why (8) is ungrammatical must be different from whatever bans the particular linear order in (9): first, some varieties of Spanish (like Caribbean Spanish, see Ordóñez & Olarrea, 2006, and references therein) do allow subjects to intervene between the wh-​operator and the verb (cf. 9). Crucially, however, these speakers reject (8), although they accept it if the subject is pro.4 Second, in regular wh-​questions in European Spanish, the wh-​word and the verb can be separated by negation (10a) or a parenthetical (10b). IQQs disallow any of these options, as illustrated in (11): (10) a. Dónde  no  has    ido? where   not have.2sg gone “Where haven’t you gone?” b. Dónde, según   Juan, has   ido? where  according Juan  have.2sg gone “Where, according to Juan, have you gone?” (11) a. *Qué  no  has   ido,   a   Tailandia? what not have.2sg gone to Thailand “Haven’t you been to Thailand?” b. *Qué, según   Juan, has   ido,  a Tailandia? what according Juan  have.2sg gone to Thailand “According to Juan, haven’t you been to Thailand?” An anonymous reviewer points out that the datum in (11a), contrary to the example in (11b), may not be syntactically relevant, in that the incompatibility of IQQs with negation may receive a semantic or pragmatic account. In

Invariable qué-​questions (IQQs) in Spanish  79 particular, (s)he hypothesizes that the ungrammaticality of (11a) may result from a clash between the biased nature of the question and the semantics of the negative word. We do not really see how this could be the case: as we will show in the following section, split questions, which are yet another kind of non-​standard, biased, interrogative sentence in Spanish, are fully compatible with negation. We thus conclude that the ill-​formedness of (11a) must be directly related to structural, i.e., syntactic, factors. Further evidence that this is correct is that, in fact, no adverb can be placed between qué and the verb, as illustrated with other adverbs in (12):5 (12) a. *Qué siempre viaja,   en avión? what always  travel.3sg in  plane “Does he always travel by plane?” b. *Qué  ya   has   ido,  a  Tailandia? what already have.2sg gone to Thailand “Have you already been to Thailand?” To summarize: partial interrogatives in Spanish are generally subject to a constraint which forces the wh-​word to be adjacent to the finite verb. However, there are some exceptions to that rule. First, in some varieties of Spanish like Caribbean Spanish, subjects can linearly intervene between the wh-​operator and the verb. In other varieties of Spanish, like European Spanish, the wh-​ word may be separated from the finite verb by negation, by some frequency adverbs under certain conditions (see fn.4), or by some kinds of parentheticals. Importantly, the adjacency condition is way stricter in IQQs:  in this phenomenon, invariable qué and the verb cannot, under any circumstances, be separated. As we will argue, we take this to be an argument against a treatment of qué in IQQs as a regular wh-​word undergoing regular wh-​movement. 2.5  Material between the V and the tag In Castilian Spanish, IQQs disallow having extra material between the verb and the tag. That is the reason that the examples in (13) are ungrammatical: in (13a), we have a subject between the V and the tag, whereas in (13b) we have a locative complement and in (13c) we have temporal and locative adjuncts.6 (13)  a. *Qué  ha ido tu  madre,  al   hospital? what went  your mother to the hospital “Has your mother gone to the hospital?” b. *Qué  ha   ido  al   hospital, con  tu  hermana? what has gone to the hospital with your sister “Has (s)he gone to the hospital with your sister?” c. * Qué  saludaste hoy  por la  tarde   en el  parque, a  Pedro? what greeted  today in  the afternoon in the park,  to Pedro “Did you greet Pedro today in the park?”

80  Fernández Sánchez and García Pardo 2.6  Islands Given what we discussed in the previous section, testing island sensitivity becomes methodologically difficult, as the relevant examples will necessarily involve material between the verb and the tag (i.e., minimally, the island itself). Unsurprisingly, the examples are ungrammatical, as shown in (14) with various kinds of islands: (14a) involves an NP-​island, (14b) a wh-​island and (14c) is an instance of an adjunct island: (14) a. *Qué te    molesta [  el  hecho de  que compre t],  con  tarjeta? what  to you  bother.3sg the fact    of  that  buy.1sg   with  card “Does it bother you the fact that I use my card for shopping?” b. *Qué  te   preguntabas  [quién vendría t],    en tren? what to you wondered.2sg who  would come.3sg in  train “Did you wonder who would come by train?” c. *Qué  fue    a  trabajar [ después de llegar t], de  Tailandia? what went.3sg to work   after   of to arrive from Thailand “Did he go to work after arriving from Thailand?”

Of course, we submit that the reason why the examples in (14) are ungrammatical is independent of the presence of an island. Instead, we claim that the ill-​formedness of these sequences stems from the fact that the condition stated in section 2.5, namely that the verb must be adjacent to the tag, is violated. In this respect, compare (14) to the examples in (15), where some non-​island, clausal material is placed between the verb and the tag, and the resulting string is equally ungrammatical. Obviously, the ungrammaticality of such cases cannot result from any island constraint: (15) a. * Qué te molesta [ que compre t], con tarjeta? what to you bother.3sg that buy.1sg with card “Does it bother you that I do the shopping by card?” b. * Qué te dijo [que acababa    de llegar t], de Tailandia? what to you told.3sg that finished.3sg of  to arrive from Thailand “Did he tell you that he had just arrived from Thailand?”

2.7  Multiple tags Finally, IQQs allow for the tag to be composed of multiple phrases: (16) a. Qué comes, en restaurantes todos los días? what eat.2sg in restaurants all the days “Do you eat in restaurants every day?” b. Qué estáis, en Austin toda la  semana? what are.2pl in Austin all  the week “Are you in Austin the whole week?”

Invariable qué-​questions (IQQs) in Spanish  81 c. Qué visteis, a Juan  en  el  dentista? what saw.2pl to Juan in  the dentist “Did you see Juan in the dentist?” d. Qué fuisteis, a la playa el  domingo? what went.2pl to the beach  the  Sunday “Did you go to the beach on Sunday?” An anonymous reviewer wonders if the relative order of the tags alters the grammaticality of the sequences in (16). Although the judgements are not crystal clear, to our ear the order of the tags does not have an impact on grammaticality in (16a,b), but it does in (16c,d): (17) a. Qué comes, todos los días en restaurantes? b. Qué estáis, toda la semana en Austin? c. *Qué visteis, en el dentista a Juan? d. *Qué fuisteis, el domingo a la playa?

(cf. 16a) (cf. 16b) (cf. 16c) (cf. 16d)

As far as we have been able to test, the order in which the tags appear correlates with the natural order in which these constituents surface in a declarative, pragmatically unmarked sentence in Spanish (i.e., as an answer to a what’s going on sort of question): (18) a. b. c. d.

Comes {todos los días en restaurantes/​en restaurantes todos los días}. Estáis {en Austin toda la semana/​toda la semana en Austin}. Visteis {a Juan en el dentista/​#en el dentista a Juan}. Fuisteis {a la playa el domingo/​#el domingo a la playa}.

A more thorough examination of the data is, however, required. If our correlation is on the right track, we believe our analysis can predict that. We return to this issue in section 5.

3.  IQQs are not split questions At first blush, IQQs look very similar to split questions (SQs), which are studied in depth by Arregi (2010). SQs are indeed identical to IQQs modulo qué, which in the former is replaced by a wh-​operator which matches the tag morphosemantically. Thus, compare the examples in (2)  with the following ones (the wh-​word is underlined to highlight the differences between the two phenomena): (19) a. A quién  saludaste,  a  Pedro? To who  greeted.2sg to Pedro “Who did you greet, Pedro?”

(Direct object, a-​DP)

82  Fernández Sánchez and García Pardo b. El  libro   a  quién  se   lo  dio,     a  Pedro? the book to who   him it  gave.3sg to Pedro “Whom did you give the book to, Pedro?” (Indirect object, PP) c. Dónde estudiáis, en la  biblioteca? what   study.2pl  in  the  library “Where are you studying, in the library?” (Adjunct, PP) d. Cómo  está,  enferma? what  is.3sg  sick “How is she, sick?” (Subject predicative, AP) e. El  pescado  cómo lo comen, crudo? the fish   what   it  eat.3pl  raw “How do they eat the fish, raw?” (Object predicative, AP) f. Dónde  fueron,   a  París? where   went.3pl to Paris “Where did they go to, Paris?” (Locative complement, PP) g. Las bebidas quién las  compra, Susana? The  drinks   who   them  buy.3sg   Susana “Who’s buying the drinks, Susan?” (Subject, NP) Arregi (2010) defends a biclausal analysis for split questions, where the tag is the remnant of a clausal ellipsis operation licensed under semantic identity with respect to an antecedent clause (cf. Merchant, 2001), in this case the partial interrogative which precedes the tag. Thus, (19a) would be analyzed as in (20), where ellipsis is indicated in strikethrough text: (20) [CP A quién saludaste]?

[CP Saludaste a Pedro]?

Crucially for the biclausal analysis, the two sentences involved are grammatically independent (21). This is an important difference between SQs and IQQs: IQQs are ungrammatical without the tag (22), which strongly suggests that we are dealing with a monoclausal phenomenon: (21)  a. A quién saludaste? b. Saludaste a Pedro? (22) Qué saludaste, *(a Pedro)? Arregi’s analysis makes an interesting prediction. Given that clausal ellipsis is a general mechanism which is found crosslinguistically, we expect SQs to be available in all languages. Even though we cannot confirm this prediction, SQs are indeed attested in a large number of languages, and no claims to the contrary have been made in the literature. IQQs, on the other hand, are not frequent. They have been described in Catalan (Contreras & Roca, 2007) and in a variety of Basque (Irurtzun, 2017), but they are unavailable in other languages like English.

Invariable qué-​questions (IQQs) in Spanish  83 In this section we are going to capitalize on the differences between the two phenomena. However, a methodological caveat is in order. IQQs and SQs cannot be teased apart when the tag would respond to the wh-​word qué. In such cases, both a SQ and an IQQ string would obtain. Therefore, we refrain from using such cases in the rest of this chapter. 3.1  Nature of the tag As we can see from the examples in (19), the tag in SQs can be a phrase of any category and fulfil any function. In this respect, IQQs and SQs behave identically. 3.2  Embedding Another similarity between SQs and IQQs is that they constitute main clause phenomena, i.e., they cannot be embedded. (23) * Me pregunto dónde has ido, a Londres? to me wonder.1sg where have.2sg gone to London “*I wonder where have you been to, to London?” 3.3  Prepositional tags With prepositional tags, the wh-​operator must be headed by the same preposition as the tag. In this respect, compare the SQ examples in (24) to the IQQ cases in (6), where qué cannot be headed by any preposition whatsoever: (24) a. Pedro *(de)  dónde  es,   de   Málaga? Pedro from where  is.3sg from Málaga “Where is Pedro from, from Málaga?” b. *(A) quién le  has   dado el  regalo, a  tu  hermana? to who  her  have.2sg  given the  gift   to  your  sister “Have you given the present to your sister?” These data are expected under the assumption that the first part of a SQ is a regular partial question, as the wh-​word corresponds to a bona fide prepositional verbal dependent. 3.4  Strict adjacency The first part before the intonation break in a SQ is a partial interrogative phrase (cf. 21a). Thus, the adjacency restrictions between the wh-​operator and the finite verb should be the same ones as in regular interrogatives, a prediction which turns out to be correct:  on the one hand, the wh-​operator and

84  Fernández Sánchez and García Pardo the V cannot be separated by a subject, at least in European Spanish (25a). As we saw before in (9), this is a property of partial questions in this variety of Spanish.7 On the other hand, however, the wh-​word and the V can be separated by sentential negation (25b), a parenthetical (2c) or an adverb of frequency like ya (25d), see fn.4. In this respect, SQs pattern, unsurprisingly, with regular interrogatives. Crucially, as we saw before, the IQQ counterparts of (25b-​d) are completely ungrammatical. (25)  a. *Dónde  tú has      ido,  a  Tailandia? where    you have.2sg  gone  to  Thailand “Where haven’t you been to, Thailand?” b. Cuándo  no   estudias,   los  domingos? when  not study.2sg the Sundays “When is it that you don’t study, on Sundays?” c. Dónde, según   Juan, has    ido,  a  Tailandia? where  according Juan  have.2sg gone to Thailand “Where, according to Juan, haven’t you been to, Thailand?” d. Dónde ya    has   estado, en Tailandia? where  already have.2sg been   in  Thailand “Where have you already been to, Thailand?” 3.5  Between V and the tag In SQs, we can find material between the verb and the tag. This is unlike IQQs, which, as we saw in section 2.5 for Castilian Spanish, disallow any kind of material between the verb and the tag. See, thus, the contrast between an IQQ (26a) and a SQ (26b): (26) a. *Qué  vino    ayer     María,  en  tren? what came.3sg  yesterday María   in  train “Did Mary come by train yesterday?” b. Cómo vino    ayer   María, en tren? how   came.3sg yesterday  María  in  train “How did Mary come yesterday, by train?” 3.6  Islands The first part of a split question is a regular wh-​question, and as such it will be sensitive to island constraints. Contrary to IQQs, adjacency between the verb and the tag in SQ is not an issue, as we point out in section 3.5, and therefore SQs do not pose any methodological problem when it comes to test islandhood. Expectedly, the following examples, which are the SQ counterpart of the strings in (14), are ungrammatical:

Invariable qué-​questions (IQQs) in Spanish  85 (27) a. * Cómo te     molesta [  el  hecho de que compre t], con  tarjeta? how  to you bother.3sg the fact    of that buy.1sg   with card “*How does it bother you the fact that I buy stuff, by card?” b. * Cómo te   preguntabas [ quién vendría t],  en tren? what   to you   wondered.2sg  who   would come  in train “*How did you wonder who would come, by train?” c. * De dónde fue    a   trabajar [ después de llegar t],  de   Tailandia? of   where  went.3sg to work    after    of to arrive from Thailand “*Where did he go to work after arriving, from Thailand?”

Note that contrary to what happens with IQQs, we do observe a contrast between the cases in (27) and the examples in (28)—​which are the SQ counterpart of the IQQs in (15). In these cases, even though the verb is separated from the tag, no locality constraint is violated and therefore the examples are completely grammatical: (28) a. Cómo  te    molesta [  que  compre t], con  tarjeta? how    to you bother.3sg that buy.1sg  with card “How does it bother you that I do the shopping, by card?” b. De  dónde te  dijo [ que acababa  de llegar t], de   Tailandia? from  where  you  told.3sg that finished.3sg of to arrive from Thailand “From where did he tell you that he had just arrived, from Thailand?”

3.7  Multiple tags Importantly, SQs do not allow for multiple tags (29), whereas IQQs do allow for them (16). The ungrammaticality of SQs with multiple tags does not hinge on the order of these multiple tags, as confirmed by (30): (29) a. * Dónde comes, en restaurantes todos los días? where  eat.2sg  in  restaurants  all    the  days  “*Where do you eat, in restaurants every day?” b. * Dónde estáis,  en Austin toda la  semana? where  are.2pl  in  Austin  all   the  week “*Where are you, in Austin the whole week?” c. * A quién visteis,  a  Juan en el  dentista? to who    saw.2pl to Juan in the dentist “*Who did you see, Juan, at the dentist’s?” d. *Dónde  fuisteis,  a   la   playa  el   domingo? where  went.2pl to the beach the Sunday “*Where did you go, to the beach on Sunday?”

86  Fernández Sánchez and García Pardo (30) a. b. c. d.

*Dónde comes, todos los días en restaurantes? *Dónde estáis, toda la semana en Austin? *A quién visteis, en el dentista a Juan? *Dónde fuisteis, el domingo a la playa?

In (29) we have used the wh-​word which corresponds to the leftmost tag, and we have maintained the same wh-​word in (30). For completeness, however, note that using the tag corresponding to the rightmost tag in (29) yields an ungrammatical output in whatever order the tags surface: (31) a. *Cuándo comes, en restaurantes todos los  días? When   eat.2sg  in  restaurants  all    the  days “*When do you eat, in restaurants every day?” b. *Cuándo estáis,  en Austin toda la  semana? when   are.2pl in  Austin all    the week “*When are you, in Austin the whole week?” c. * Dónde (lo)  visteis,  a  Juan en el  dentista? where  him saw.2pl to Juan in  the dentist “*Where did you see (him), Juan at the dentist’s?” d. *Cuándo  fuisteis,  a  la   playa  el   domingo? when   went.2pl to the beach the Sunday “*When did you go (there), to the beach on Sunday?” (32) a. b. c. d.

*  Cuándo comes, todos los días en restaurantes? *  Cuándo estáis, toda la semana en Austin? *  Dónde (lo) visteis, en el dentista a Juan? *  Cuándo fuisteis, el domingo a la playa?

3.8  Summary of the differences The following table summarizes the differences between the two phenomena that have been discussed in this section:

Nature of the tag Strict adjacency between qué and the V Embedding Material between the V and the tag Islands Multiple tags Prepositional tag

IQQs

SQs

Any Yes No No -​8 Yes No

Any No No Yes Yes No Yes

Invariable qué-​questions (IQQs) in Spanish  87

4.  Previous analyses In this section we review various existing proposals on the syntax of IQQs. For space constraints, we cannot provide a very detailed discussion, so the reader is referred to the original papers for a more careful examination of the proposals. However, we hope to convince the reader that these analyses raise a number of insurmountable challenges, which means that a novel approach is required. For reasons of simplicity, we will illustrate each of the analyses with the following example. (33) Qué vio, a María? what saw to María “Did he see María?” Importantly, (33) can only be analyzed as an IQQ, and not as a SQ. This is so because a SQ would require the wh-​word quién (who) headed in turn by the differential object marking a: 4.1  Camacho (2002) Camacho (2002) proposes that qué is a wh-​word which adjoins to the tag. Because of its wh-​nature, qué must undergo A-​bar movement to spec,CP: (34)

His analysis presents some inconveniences. First, it is unclear what the semantic contribution of the adjunct is. Second, the fact that qué is morphologically invariable is left unexplained. Finally, the asymmetric behavior of qué in IQQs and the wh-​operator in regular wh-​questions with respect to the adjacency facts reviewed in section 2.4 (i.e., the fact that qué in IQQs must be strictly adjacent to the finite verb) casts doubts as to whether both morphemes can be given a uniform syntactic analysis. However, it should be noted that his examples are actually ambiguous between a SQ and an IQQ analysis, so we do not know what phenomenon he is dealing with.

88  Fernández Sánchez and García Pardo 4.2  The spec-​head approach Most previous analyses of IQQs contend that qué and the tag are generated in a specifier-​complement configuration as part of the same projection. The differences lie in the nature of such a projection: Contreras and Roca (2007) defend that it is a BigDP, López-​Cortina (2007) refers to it as Confirmation Phrase, and Irurtzun (2017) argues that it is a verbless copulative structure named RP, in line with den Dikken’s (2006) analysis of copulative structures. For reasons of simplicity, we will call this projection XP. In all of these proposals, XP generates in the lexical layer of the clause and is later split throughout the derivation.9 (35)

Leaving aside the particular issues which have to do with the specific projection that each author defends, the spec-​head approach faces some general problematic aspects. First, these authors claim that qué undergoes A-​bar movement, just like any wh-​word would. But then, just as in Camacho’s analysis, there is no principled way to explain the asymmetry between regular wh-​ words and qué with respect to adjacency. Second, the structure in (35) cannot account for cases in which the tag is an adjunct, as in (36). In these cases, XP is generated as an adjunct, which means that the moment that either the tag or qué vacate the XP, the structure will necessarily involve a violation of the Adjunct Condition. These accounts thus predict that adjunct tags are not possible, contrary to fact: (36) Qué lloraste, por su what cried.2sg for his “Did you cry for his loss?”

pérdida? loss

4.3  Lorenzo (1994) Lorenzo (1994) defends that qué is an expletive wh-​word which is base-​ generated in spec,CP. Qué is coindexed with the tag, which is base generated

Invariable qué-​questions (IQQs) in Spanish  89 as an adjunct to the CP. By means of this coindexation, qué is able to set the relevant value (i.e., person, place, time…) to an abstract interrogative operator located in C0, which in turn binds a variable within the TP which is co-​ indexed with the tag. (37)

The analysis in (37) makes use of various mechanisms which have been abandoned in syntactic theory. But leaving these aside, in our view, Lorenzo’s analysis has to face an important drawback. While we agree that qué is generated in the CP domain, we disagree with the claim that the tag is base-​ generated as an adjunct of the CP. This is so because constituents which are generated directly in the CP area are typically not case marked, and tags in IQQs are. We can show this to be true in Spanish: even though NPs lack morphological case marking in Spanish, we can treat differential object marking (DOM) as some sort of case-​marking mechanism (see fn. 3). Animate, direct objects in this language must be headed by a preposition-​like element a (38a). Hanging topics, however, cannot be case-​marked, even when they refer to an animate, direct object (López, 2009; among others), as shown in (38b). This is frequently attributed to the fact that hanging topics are not generated in the lexical layer of the clause, where differential object marking is assigned (Ormazábal & Romero, 2013): (38) a. Juan  ha     Visto  *(a) María. Juan  has.3sg  seen    to  María “Juan has seen María.” b. (*A) María, Juan ha     visto  a esta  chica  un par     de veces. to  María  Juan has.3sg seen to this girl  a  couple of times “Mary, John has seen this girl a couple of times.”

Importantly, when the tag in an IQQ corresponds to an animate direct object, it must be headed by a, which is unexpected if the tag had been generated in the CP. (39) * Qué  vieron,  María? what  saw.3pl  María “Did they see María?”

90  Fernández Sánchez and García Pardo

5.  Towards an analysis A successful analysis of IQQs, we argue, needs to capture three important aspects: 1

2

3

Qué and the tag are not derivationally or thematically linked, given the existence of adjunct tags, lack of pied-​piping with prepositional tags and the mismatch between the morphology of the wh-​element and the semantics of the tag. IQQs cannot be biclausal with ellipsis, because the presence of the tag is mandatory. For an ellipsis account to work, one would have to assume that deletion must be obligatory, which is an uncommon assumption. Further, the systematic asymmetries between IQQs and bona fide biclausal structures with ellipsis like split questions also constitute arguments against such a view. Qué has a different syntactic distribution from wh-​operators in run-​of-​ the-​mill partial questions. In particular, we have shown that qué in IQQs must be strictly adjacent to the finite verb, a condition which is more relaxed in regular interrogatives (see section 2.4).

For an IQQ like (40a), we propose the analysis in (40b): (40) a. Qué vienes, en bicicleta? What come.2sg in bicycle “Are you coming by bike?” b.

In our analysis, qué base-​generates in the specifier of a projection in the left periphery of the clause. This projection is immediately above TP (i.e., its head is in a sister relation to TP), given the impossibility of having any material

Invariable qué-​questions (IQQs) in Spanish  91 between the verb and the tag.10 The verb undergoes the usual V-​to-​T-​to-​C movement of interrogatives. In turn, the whole VP moves to the specifier of a low focus projection sitting between TP and VP (Belletti, 2004), where it is interpreted as the element requiring confirmation. An important question is how our analysis would differ from a regular wh-​ question, given that most analyses thereof assume V-​to-​T-​to-​C movement.11 Our suggestion, which we cannot fully flesh out here for space reasons, is that the wh-​phrase in regular questions moves to a focus projection (as in Rizzi’s 1997 FocP), whereas in IQQs qué is base-​generated in this lower projection in the left periphery, which also hosts the inflected verb. We can tentatively assume this projection that we call CP in (40b) is FinP, the lowest projection in Rizzi’s (1997) cartography. The syntactic gap between the wh-​phrase and the inflected verb in regular questions explains the possibility of having additional material between them (e.g., adverbs, negation and parentheticals, see (10) and fn. 4). Our analysis stands in contrast to previous ones which posit wh-​movement in IQQs (with the exception of Lorenzo, 1994). Therefore, it would be worthwhile to discuss what arguments have been adduced in favor of the movement approach. The first one concerns island sensitivity. As shown in (14), IQQs are apparently island sensitive. As we have discussed in section 2.6, we believe that the ungrammaticality of such examples is not due to the usual movement constraints, but rather to the independent fact that there cannot be any syntactic material between the verb and the tag. We provided examples in section 2.5 (e.g., (15)) that did not involve islands and were just as bad as purported island violations. The second argument in favor of movement of qué is provided by Arregi (2010) for IQQs in Basque, although it can be easily replicated in Spanish. As extensively discussed in Torrego (1984:  103–​109), wh-​movement triggers subject-​verb inversion (41). In cases of long-​distance, successive-​cyclic wh-​ movement, subject-​verb inversion is visible in all the embedded clauses (42). (41) a.

Qué querían esos dos? what wanted.3pl these two “What did these two want?” b. * Qué esos dos querían?

(42) a.

Qué pensaba Juan [ que le había dicho Pedro [ que había publicado what thought.3sg Juan that him had.3sg told Pedro that had.3sg published la revista]]? the journal “What did Juan think that Pedro had told him that the journal had published?” b. * Qué Juan pensaba [que Pedro le había dicho [que la revista había publicado]]?12

If qué in IQQs undergoes wh-​movement, we should be able to observe subject-​verb inversion. The short-​distance cases are irrelevant, because we have seen that qué must be strictly adjacent to the verb, so even though the

92  Fernández Sánchez and García Pardo verb precedes the subject, it may not be due to subject-​verb inversion. Thus, the relevant examples will involve successive-​cyclic movement. Unfortunately, however, these examples involve again material between the verb and the tag. Indeed, what we observe is that they are ungrammatical both with subject-​ verb inversion (43a) and without (43b): (43)

a. * Qué what

dice says.3sg

Juan Juan

que that

ha has.3sg

ido gone

su his

hermana, sister

a to

Londres? London

“(Int.) Does Juan say that his sister has gone to London?” b. * Qué dice Juan que su hermana ha ido, a Londres?

For completeness, note that the same is not true of split questions. If qué in (43) is replaced by dónde (“where”), thereby creating a split question, the verb must expectedly precede the subject in the embedded clause, as a result of regular wh-​movement: (44) a. Dónde dice Juan que ha ido su hermana, a Londres? b. *Dónde dice Juan que su hermana ha ido, a Londres? Therefore, the inversion test used by Irurtzun for IQQs in Basque cannot successfully apply in Spanish, as the relevant examples are ungrammatical for independent reasons. In our approach, qué in IQQs is not thematically connected to the tag or a phrase within the tag. It is indeed not the kind of wh-​phrase that we find in wh-​questions, but rather a complementizer of sorts (its concrete syntactico-​ pragmatic status yet to be determined). This provides an explanation for two observations. First, we account for why qué is morphologically invariable. Second, the existence of adjunct tags is no longer a problem: recall that on virtually any movement-​based account, such adjunct cases must necessarily involve a violation of the adjunct condition. This is of course not an issue under the analysis we defend as qué and the tag are not derivationally connected. Further, our account integrates IQQs within a monoclausal spine, layered by the standard projections in our field, namely CP>TP>VP. The systematic differences between IQQs and split questions are therefore expected. Finally, IQQs allow for multiple phrases in its tag. Our analysis predicts this, with the possibility of having a VP with a complement and one or more adjuncts move to the specifier of the Low Focus phrase. So technically, “multiple tags,” albeit descriptively adequate, is actually a misnomer, as these cases do not involve multiple tags, but rather just one remnant VP which contains the tag, along with other clause-​internal constituents. In section 2.7 we showed that “multiple tags” in IQQs are subject to a restriction which states that their order must reflect the order they surface in an otherwise pragmatically neutral, declarative sentence. We believe that this

Invariable qué-​questions (IQQs) in Spanish  93 is actually what we predict, as the VP containing the tag(s) moves to a FocP in one block, without other movement operations having taken place.

6.  Concluding remarks This chapter has analyzed IQQs in Spanish. We have analyzed their properties, some of them previously unobserved. We have discussed previous work on this construction and have shown how no account fully captures the properties of IQQs. We also dedicated a section to argue that IQQs are monocausal and very distinct from SQs, which are biclausal. Finally, we have presented a novel analysis that captures this set of facts, namely the lack of a thematic link between qué and the tag, monoclausality, multiple tags and the strict adjacency between qué and the verb. As mentioned in the introduction, our study has focused on Castilian Spanish. However, IQQs seem to behave differently across varieties of Spanish in non-​trivial respects. For instance, Contreras and Roca (2007: 46) provide the examples in (45) for Catalonian Spanish, where we find an adverb and a subject (of an unaccusative verb) between the verb and the tag: (45) a. Qué  vino    ayer    María,  en  tren? what came.3sg yesterday María  in  train “Did María come by train yesterday?” b. Qué  vino    María  ayer,     en  tren? what came.3sg María yesterday in  train “Did María come by train yesterday?” (Contreras & Roca, 2007: 46) However, a co-​author of this chapter and native speaker of Catalonian Spanish disagrees with Contreras and Roca’s (2007) data. For him, the data in (45) are strongly deviant. He only (marginally) accepts material between the verb and the tag in the case of objects: (46) ? Qué comió patatas, Pedro? what ate.3sg potatoes Pedro “Did Pedro eat potatoes?” In any case, these data stand in contrast to the Castilian Spanish data discussed in this chapter, which disallow any kind of material between the verb and the tag (i.e., (45) and (46) are both ungrammatical in that variety, see (26)). We intuit that this is due to a complex cross-​dialectal interplay of interface conditions, which will only be elucidated with future dialectal and experimental studies.

94  Fernández Sánchez and García Pardo

Notes 1 We would like to thank the audiences of the 28th Colloquium on Generative Grammar (Tarragona, Spain, June 2018), the Hispanic Linguistic Symposium (Austin, TX, October 2018) and the 6th Linguistics Beyond and Within conference (Lublin, Poland, October 2018), especially José Camacho, Olga Fernández-​Soriano and Aritz Irurtzun for their excellent feedback. We further thank two anonymous reviewers whose comments, suggestions and criticism have contributed to improve the original draft. Needless to say, all the possible errors in this chapter are entirely our own. 2 Throughout the chapter, we have decided to translate all IQQs as regular total questions in English. We are aware that the biased reading of IQQs is lost, but English lacks these kinds of questions. Instead, to convey a biased reading, English would most likely resort to the use of a split question (Arregi, 2010), as in where are you, in the library? Note that split questions, which also exist in Spanish, look similar to IQQs, except that the wh-​word is not invariably what. Despite the similarities, the two phenomena are radically distinct, as we will discuss, whence our avoidance to use split questions to translate IQQs. 3 We take the morpheme a to be an instance of Direct Object Marking, or DOM (Luján, 1978). 4 Our informant is a native speaker of San Juan (Puerto Rico). 5 The question is whether these adverbs can appear to the immediate left of a wh-​ word in a regular question. As noted by Ordóñez and Olarrea (2006), the adverb ya (“already”) can appear in such contexts. Note, in relation to (12b), that a regular interrogative counterpart would be perfectly grammatical: (i) Dónde ya has estado? Where already have been “Where have you already been?” With respect to adverbs like siempre (“always”), Ordóñez and Olarrea (2006) contend that they may only intervene between the wh-​word and the finite verb if they receive focus (ii). Importantly, a sentence like (12a) is completely ungrammatical independently of how focused the adverb is (iii): (ii)

(iii)

A quién {* siempre/​?SI EM P RE } besas? To whom always Kiss “Whom do you always kiss?” * Qué S I EM P RE viaja, en avión? (cf. 12a)

These remarks are crucial for us, as they reveal systematic asymmetries between IQQs and regular wh-​interrogatives. 6 But see section 6, where we discuss dialectal differences regarding these data. 7 The expectation is, of course, that (25a) will be grammatical in Caribbean Spanish, but we haven’t been able to confirm this datum. 8 It is impossible to determine whether or not IQQs are island sensitive. Effectively, at least from an observational point of view IQQs are disallowed in island contexts. However, such contexts violate a more general constraint which forces the verb to be adjacent to the tag (see section 2.6).

Invariable qué-​questions (IQQs) in Spanish  95 9 The aforementioned analyses differ along various other aspects. We cannot offer a detailed review of each of them, so the reader is referred to the original papers for more information. For the purposes of our chapter, however, we believe that this simplification is enough to illustrate their limitations. 10 Note also that we can have topicalized subjects to the left of qué, which suggests that the functional projection that hosts qué and the inflected verb is lower than TopP. (i) Tú qué vienes, en bici? you what come.2sg in bike “Are you coming by bike?” 11 We thank an anonymous reviewer for raising this important question. 12 An anonymous reviewer indicates that not inverting the subject and the verb in embedded clauses in cases of long-​distance wh-​movement does not yield an ungrammatical output for some speakers. Indeed, the stark ungrammaticality of (42b) stems from the fact that the matrix subject (Juan) intervenes between the matrix verb and the wh-​operator, which is impossible in European Spanish (see section 2.4). Consequently, dropping the matrix subject substantially improves (39b), even though we still find the inversion cases more natural.

References Arregi, K. (2010). Ellipsis in split questions. Natural Language and Linguistic Theory, 28, 539–​592: Belletti, A. (2004). Aspects of the low IP area. In L. Rizzi (ed.), The Structure of CP and IP: The Cartography of Syntactic Structures, 16–​51. Oxford: Oxford University Press. Camacho, J. (2002). Wh-​doubling:  Implications for the syntax of wh-​movement. Linguistic Inquiry, 33, 157–​164. Contreras, J.M., & Roca, F. (2007). Un tipus especial d’oracions interrogatives:  les interrogatives escindides. Caplletra, 42, 145–​184. den Dikken, M. (2006). Relators and Linkers:  The Syntax of Predication, Predicate Inversion, and Copulas. Cambridge, MA: MIT Press. Hooper, J.B., & Thompson, S.A. (1973). On the applicability of root transformations. Linguistic Inquiry, 4, 465–​497. Irurtzun, A. (2017). On the nature and distribution of split wh-​questions in Basque. Paper presented at the The Syntax-​Discourse interface: Approaches, phenomena and variation conference, Universitat Autònoma de Barcelona, June 2017. Izvorski, R. (1997). The present perfect as an existential modal. In A. Lawson (ed.), SALT VII, 222–​239. Ithaca, NY: Cornell University Press. López, L. (2009). A Derivational Syntax for Information Structure. Oxford:  Oxford University Press. López-​Cortina, J. (2007). The Spanish left periphery: Questions and answers (Doctoral dissertation). Georgetown University. López-​ Cortina, J. (2009). Split questions, extended projections, and dialectal variations. In J. Collentine, M. García & B. Lafford (eds.), Selected Proceedings of the 11th Hispanic Linguistics Symposium, 219–​230. Somerville, MA: Cascadilla Proceedings Project.

96  Fernández Sánchez and García Pardo Lorenzo, G. (1994). Qué expletivo en preguntas dislocadas. Archivum, 44, 423–​446. Luján, M. (1978). Direct object nouns and the preposition “a” in Spanish. Texas Linguistic Forum, 10, 30–​52. Merchant, J. (2001). The Syntax of Silence. Oxford: Oxford University Press. Ordóñez, F., & Olarrea, A. (2006). Microvariation in Caribbean/​ non Caribbean Spanish interrogatives. Probus, 18, 59–​96. Ormazábal, J., & Romero, J. (2013). Differential object marking, case and agreement. Borealis: An International Journal of Hispanic Linguistics, 2(2), 221–​239. Rizzi, L. (1997). The fine structure of the left periphery. In L. Haegeman (ed.), Elements of Grammar, 281–​337. Dordrecht: Kluwer. Torrego, E. (1984). On inversion in Spanish and some of its effects. Linguistic Inquiry, 15(1), 103–129.

5  Verbless DP interrogative constructions and enclisis in Galician Brian M. Gravely, Jr. and Timothy M. Gupton

1.  Introduction Amongst the natural languages of the world, several systematically employ verbless null-​copula constructions in declarative sentences. (1) a. He a expert. “He is an expert.” b. Viktor glup Victor stupid “Victor is stupid.” c. dani   nexmad  ad meod Dani  nice     very “Dani is very nice.”

(Bender, 2000, p. 84) (Comrie, 1987, p. 342) (Doron, 1986, p. 314)

In African American Vernacular English (1a), Russian (1b), and Hebrew (1c), phonologically null predicates have the same semantic meaning as sentences with overt copula in other languages. Galician boasts a syntactically fixed interrogative construction (2a) that is minimally comprised of the locative interrogative U (“where”) and a third-​person determiner clitic varying in number and grammatical gender: lo, la, los, las. (2) a. U-​lo? where-​C L .AC C. 3S G. MA S C “Where is it?” b. Où ça? where that “Where is that?” In the following section, we examine the characteristics of U-​lo in Galician, comparing them with similar constructions in other Romance languages such as French (2b). We show that, despite cosmetic similarities, Galician U-​lo constructions behave differently from a syntactic perspective and thus require a different analysis. Given their differences, we refer to these forms as verbless

98  Brian M. Gravely, Jr. and Timothy M. Gupton DP interrogative constructions (VDICs) or simply U-​lo. Subsequently, we examine the explanatory adequacy of a standard null-​copula account, an imperative account, and a small clause account for VDICs. Analyses of all three of these structures bear important cross-​linguistic structural similarities and differences when compared to VDICs, and we adopt many of the ideas present in their associated syntactic models in order to account for these structures in Galician.

2.  Descriptive generalizations: the characteristics of U-​lo Descriptions of U-​lo in traditional Galician grammars have been relatively scarce. Among depictions by grammarians, no two analyses are quite the same. Carballo Calero (1979: 328) calls it “half adverb and half pronoun.” Freixeiro Mato (2006: 503) classifies it as an interrogative adverbial phrase, noting that, apart from the locative adverbial onde (“where”), “the medieval form u also exists, from the Latin ubi, which survives in modern Galician and is always followed by an article or an accusative atonic personal pronoun” (this and subsequent translations from Galician grammars are ours). Only Álvarez, Monteagudo and Regueira (1986:  431) calls this construction an “interrogative without a verb.” Unlike normal wh-​interrogatives, U is not found in any other construction apart from U-​lo.1 Unlike the more common locative adverbial onde, it may not be used with an overt copula verb (3a), a transitive verb (3b), be embedded (3c), or remain in situ (3d). (3) a. Onde/​ *U está    a  chave que  che       dera? where    be.P RS.3 SG   the  key   that  CL . DAT. 2SG  give.PST. 1SG “Where is the key that I gave you?” b. Onde o/​*U-​lo deixaches? where-​C L.AC C.3 SG.M leave.P ST.2 SG “Where did you leave it for me?” c. Vimos    onde/​ *u estaba     ó saír     da festa. see.P S T. 3 P L   where   be.I M P FV.3 SG  to-​ the leave.IN F  from.the.party “We saw where he was when he left the party.” d. Dis     que estivestes  onde/​ *u? say.PRS. 2SG  that be.P ST.2 P L  where “You say that you were where?”

U may not appear without a determiner clitic, ruling out its appearance alone (4a), with an indefinite DP (4b), or with a first-​or second-​person clitic pronoun (4c). We provide details regarding the phonologically identical clitic and determiner paradigms, which we refer to as “determiner clitics,” in section 5.

Verbless DP interrogative constructions and enclisis  99 (4) a. *U? where “Where?” b. *U   unha perruquería por aquí? where  a    salon    by  here “Where is a hair salon around here?” c. *U-​me? where-​C L.1 S G “Where am I?” The interrogative U mandatorily combines with a DP headed by a definite determiner in a variety of sizes: a lone clitic (5a), a full DP (5b), a quantified DP (5c), or a relative clause with a null or overt nominal (5d). (5) a. U-​lo? where-​C L .ACC. 3S G. M “Where is it?” b. U-​ las       nenas? where-​C L .ACC. 3PL. F  girls “Where are the girls?” c. U-​ los         demais (papeis)? where-​C L .ACC. 3PL. M others (papers) “Where are the rest (of the papers)?” d. U-​ la         (bolboreta) que  colliches     onte? where-​C L .ACC. 3S G. F  (butterfly)  that catch.P ST. 2 SG yesterday “Where is (the butterfly) that you caught yesterday?” There are a few important observations regarding the use and interpretation of U-​lo in the examples above. First, it always receives a present-​tense interpretation, a common characteristic of many null-​copula expressions cross-​ linguistically. Second, owing to its prosodic nature, it does not permit pauses or intonational breaks between it and other constituents. Third, the fact that the obligatory DP must be definite is indicative of its strong deictic and temporal nature, and therefore always refers to something or someone that has been recently mentioned in the discourse and is necessarily interpreted as “here-​and-​now.” 2.1  U-​lo is not like où French interrogative où appears in a null-​copula construction similar to U-​lo; however, there are clear differences between French où and U-​lo. First, unlike French où (6a), U-​lo may not appear with a strong pronoun (6b).

100  Brian M. Gravely, Jr. and Timothy M. Gupton (6) a. Où ça? where that “Where is it/​that?” b. *U iso? where that “Where is it/​that?” The French wh-​element may be the head of a relative clause (7a). It may also remain in-​situ in a main clause (8a) and in a subordinate clause (9a). None of these options (cf. 7b, 8b, 9b) are possible for U(-​lo) in Galician. (7) a. On est allé jusq’où one be.PRS. 3S G go.PART until-​where b. *Fomos  até  u   estaba. go.P S T.1PL  until  where be.I MPFV. 3 SG “We went where he was.”

il était. he be.I M P F V. 3 SG

(8) a. Tu   es     allé   où? you be.PRS. 2S G  go.PART  where b. *Fostes   u? go.P S T.2PL  where “You went where?” (9) a. Je veux      partir   en vacances, mais je ne I  want.PRS.1SG leave.I N F   in vacations  but  I  not sais pas où. know.PRS.1SG N EG where b. *Quero   ir   de vacacións, pero non sei     u. want.PRS.1SG go.I N F  of vacations  but   NEG know.PRS.1SG where “I want to go on vacation, but I don’t know where.”

Therefore, based on these important structural differences, we claim that U(-​lo) is a distinct interrogative element from où. 2.2  U-​lo is not like cadê Although ostensibly very similar to U-​lo, the Brazilian Portuguese interrogative cadê also behaves differently. Cadê can be paired with a strong pronoun (10, cf. 6), but it cannot head a relative clause (11, cf. 7). It cannot remain in situ in a root (12, cf. 8), or subordinate clause (13, cf. 9).

Verbless DP interrogative constructions and enclisis  101 (10) Cadê isso?2 where that “Where is that?” (11) Fomos até *cadê/​onde ele estava. go.P S T.3P L until where he be.I MP F V. 3 SG “We went where he was.” (12) Você foi *cadê/​onde? you go.P S T. 3S G where “You went where?” *cadê/​onde. (13) Quero sair de férias, mas não sei want.P RS. 1 S G leave.I N F of vacations but N EG know.PRS. 1SG where “I want to go on vacation, but I don’t know where.”

At first glance, sentences like (14), with cadê in a subordinate clause appear to be counterexamples to our claim above. (14) Estou    procurando       o livro,   mas  não  sei          cadê. be. AUX. 1 S G look-​for. P ROG the book but not know.PRS. 1SG where “I’m looking for the book, but I don’t know where it is.”

However, the behavior of cadê here captures an important difference in comparison with U-​lo: cadê is not an interrogative adverb; rather, we claim it to be a verbal interrogative phrase, as illustrated in (15).3 (15) a. [Cadê] as chaves? where the keys b. [Onde estão] as chaves? where be.PRS. 3PL the keys c. *Onde as  chaves? where the keys “Where are the keys?” Examples in (15) strongly suggest that cadê is equivalent to onde and the relevant conjugation of the predicate estar (“to be”), similar to the Galician examples in (5). Indeed, in (14) cadê may only be substituted by onde plus the verb estar, not by onde alone; therefore, despite clear differences regarding the

102  Brian M. Gravely, Jr. and Timothy M. Gupton syntactic behavior of cadê, its phrasal nature as well as the apparent incorporation of a null copula estar represent similarities to the Galician VDIC, an important property moving forward. 2.3  Interim summary In this section we have provided descriptive generalizations regarding the syntactic behavior of Galician VDICs. We have also examined the similarities and differences between U-​lo and two other Romance interrogative constructions that, at first glance, appear to behave similarly; however, we have shown that, upon further investigation, they do not. French où, unlike U-​lo, is a quintessential wh-​word that may be embedded, paired with a strong pronoun, and remain in situ. Its null-​copula construction (6a) is a case of verb deletion and, as we shall show, differs structurally from U-​lo. The Brazilian Portuguese cadê, on the other hand, is similar to U-​lo in that it exhibits a phrase-​like behavior in that both the [iQ] feature and the verbal construal are found within a single lexical item. The principal difference lies in the fact that cadê may be used in a subordinate clause, which entails that this wh-​word is base-​generated low in the phrase marker and assumedly moves to [Spec,CP] in cases such as (10) and (15a). In section 3 we examine the explanatory adequacy of three promising formal syntactic accounts. The most important aspects observed here that we shall carry forward in our analysis of VDICs are the lack of movement and the apparent incorporation of a null predicate phrase.

3.  Syntactic accounts of verbless constructions Generative syntactic accounts of Galician VDICs are in short supply, having only been examined by Silva-​Villar (1996, 1999, 2001), who attempted a diachronic classification of a group of null-​copula constructions across medieval Western Romance. In his (1996) analysis, U is base-​generated in [Spec, TP], checking the [uD] edge/​EPP-​feature on T, prior to movement to C, presumably to check [+Q] features (16). (16)

He claims that clitic forms like lo have “verb-​like functions” (Silva-​Villar, 1999:  572), thus serving as copula and tense marker in the head of T.4 In

Verbless DP interrogative constructions and enclisis  103 his (2001) proposal, U-​lo is base-​generated in C as the complex categorial [C+D]. In summary, Silva-​Villar’s analyses place U in either the head or specifier position of the C projection and suggest that U-​lo engendered the construction onde está…? (“Where is…?”), despite the fact that this is empirically unattested. Our main objection to Silva-​Villar’s three analyses is that the clitic lo appears separately from its associated DP and does not appear to participate in any sort of agreement operation with it over the course of the derivation. His analyses avoid explicit discussion of this φ-​feature (e.g., [person]/​ [number]/​[gender]) agreement process in T, and how to avoid overgeneration of sequences like *U-​lo nenas. This is of particular importance since T is not a typical locus of [gender]-​checking in Romance. Our approach will account for this relation. We also attempt to clarify the locus of interrogative generation in VDICs since Silva-​Villar’s position fluctuates between different heads and specifiers of T and C. Since the U portion of VDICs never appears in a lower in situ position or in embedded clauses in Galician, we adopt the idea of base-​generation in the higher functional field. We consider possibilities regarding the locus of base-​ generation in section 3.1. 3.1  Null copula Hebrew and the Arabic dialects have motivated a number of null-​copula proposals in the literature. Some of these involve null verbal projections (e.g., Eid, 1991; Fassi Fehri, 1993), while others, like Benmamoun (2000, 2008), propose that there is no need for a vP projection. In his proposal for Hebrew and Arabic, null-​copula sentences (17a) involve a functional category T that selects a NP, AP, or PP complement: (17) a. Omar f-​d-​dar. Omar in-​the-​house “Omar (is) in the house.” b.

(Arabic)

Benmamoun claims that the lack of a verbal projection in (17b) is because, in null-​copula constructions, T lacks [+V]. Instead, T has [+D], [+Present], and a bundle of unvalued nominal ϕ-​features (number, gender). Null-​copula clauses differ from those with a tensed verbal projection (i.e., those in the past or future tense) in that the latter have a categorial [+V] feature, whilst the former do not.

104  Brian M. Gravely, Jr. and Timothy M. Gupton (18) a. Past/​Future: [+V], [+D] b. Present: [+D]

(Benmamoun, 2008, p. 115)

Adopting the typology in (18) and the analysis in (17b) for Galician appears to be a promising avenue in that it captures the fact that VDICs are always present tense. Further examination of Benmamoun’s proposal, however, reveals a problem for the explanatory adequacy of this account for our data:  in Semitic languages, a verbless clause can be subordinated under a tensed clause (19). (19) a. qal bəlli Omar f-​d-​dar say.P S T.3S G that Omar in-​the-​house “You said that Omar is in the house.” b. qul-​ti bəlli Omar naʕəs say.P S T.2S G that Omar sleep.PROG “You said that Omar is sleeping.” As we saw, U-​lo cannot be embedded under any circumstance; therefore, adopting Benmamoun’s account would not rule out embedded VDICs in Galician and incorrectly generate ungrammatical structures. Assuming that the presence of T allows clausal embedding in Semitic null-​copula constructions, we are left with the possibility that U is base-​generated in the CP field. But is C capable of encoding a necessarily present-​tense reading? As we pointed out at the end of section 2, the Galician VDIC has a distinct, temporally deictic nature in that it is obligatorily interpreted as involving the “here and now.” Early descriptions of the clausal left periphery in Uriagereka’s (1995a) Functional Projection (FP) and Rizzi’s (1997) expanded CP describe precisely this: speaker-​related and discourse-​related phenomena. We feel that this is a promising possibility for capturing the fact that the questions in (5) have to do with the whereabouts of the DPs in question at utterance time and not at any moment prior or subsequent to that moment. In section 3.2.1, we examine independent justification of the notion that tenseless constituents are bound to speech time. Imperatives are similarly temporally bound to utterance time, therefore suggesting an analogous tenseless nature. 3.2  An imperative account Here we examine Platzack and Rosengren’s (1997) syntactic analysis of German imperatives and its application to VDICs. Following their account, true imperatives are morphologically impoverished (20): (20) Declarative and imperative forms in German and Galician (true imperatives in bold)

Verbless DP interrogative constructions and enclisis  105

2sg 2pl

declarative German Galician kaufst mercas kauft mercades

imperative German kauf(e) kauft

Galician merca mercade

Furthermore, true imperatives cannot be embedded and have no subject. The syntactic analysis of imperatives that they propose lacks FinP, MoodP, and TP (21b).5 (21) a. Kauf das Buch! buy.IM P.2S the book “Buy that book!” b.

The lack of FinP in imperative clauses means that these forms do not express a proposition anchored in time and space (see, e.g., Bianchi, 2003), which allows them to apply to the “here-​and-​now” of utterance time. Both Platzack & Rosengren and den Dikken & Blasco (2007) propose that, in true imperatives, there is no TP projection; the verb moves directly from the vP shell to the head of Rizzi’s (1997) ForceP, the highest projection within the expanded CP, thus capturing the empirical fact that no lexical material may precede true imperatives. In Galician, true imperatives (20, 22a) are replaced by (comparatively) morphologically rich subjunctive forms under negation (22b) and in non-​ second-​person forms (22c) (see, e.g., den Dikken & Blasco, 2007 for more on Romance commands). (22) a. Fai-​no! do.IMP.3S G -​C L. AC C. 3S G. M “Do it!” b. Non  o            fagas! NEG  C L .ACC. 3S G. M  do.PRS. S BJ V. 2S G “Don’t do it!”

106  Brian M. Gravely, Jr. and Timothy M. Gupton c. Fagámo-​lo! do.P R S.S BJ V. 1PL-​C L. AC C. 3S G. M “Let’s do it!” Subjunctive forms (22b-​c), however, are morphologically more robust and assumed to pass through TP (see, e.g., Kempchinsky, 1998, 2009 for a Spanish account that we assume to be applicable here). Consider the following model for (22a): (23)

The only difference between German (21b) and Galician (23) is the projection FP. Following the account for clitic movement proposed by Raposo and Uriagereka (2005), clitics end up in a morphologically “active” left peripheral projection (Uriagereka, 1995a, b).6 We assume this analysis to be operative for imperative structures, returning to FP in section 5. Examples (22b) and (22c), which are not true command forms, are proposed to exhibit a different derivational structure. Both of these verb forms are the same used in the subjunctive, which entails that they raise at least as high as T (and likely further to MoodP, FinP as in, e.g., Kempchinsky, 1998, 2009). We follow Raposo and Uriagereka’s (2005) in assuming “active” left-​peripheral FP to be the locus for clitics in Western Iberian. In (24) the clitic o is hosted by negation (non), allowing the verb to remain in T.7 In affirmative suppletive commands (25), we assume that the verb moves to f after checking its features on T in order to serve as the host for the clitic, perhaps as Last Resort. In short, the main difference between true imperatives and suppletive subjunctive imperatives in the above syntactic structures is that true imperative verbs move to Force and lack T. (24)

Verbless DP interrogative constructions and enclisis  107 (25)

Having examined an account of imperatives for Galician, in the next subsection we attempt to adapt this proposal for VDICs in Galician. We claim that VDICs are similar to true imperative forms owing to their discourse-​rooted nature, and, therefore, require a similar structural analysis.

4.  What’s below U? Small clauses and locative copula in Romance Intuitively, a question starting with “U-​lo…?” communicates the covert command “Tell me…” prior to asking “Where is…?”. For this reason, we claim that the interpretation of tense is deictic, interpreted “here and now.’ ”8 This is another reason why we claim that it appears in ForceP; however, unlike true imperative forms, we propose that U is base-​generated in ForceP. As we saw in section 2, U does not bear the hallmarks of movement typical of other wh-​elements. U is always sentence initial and under no circumstance may be preceded; therefore, we claim that U is externally merged in the head of ForceP and has an interpretable [+Q] feature, which is similar to Bošković’s (2015) proposal that in situ wh-​elements have interpretable [+Q]. In order to account for the DPs that may appear enclitically on U, we take a brief excursus on locative copular sentences in section 4.1. 4.1  The small clause of a featureless v Accounting for copular derivations in minimalist theory has proven to be problematic, in particular for Romance varieties that boast twin copulas (ser and estar) with similar predication behaviors.9 A thorough review of the literature on copula is well beyond the scope of this study. We therefore limit ourselves to an examination of concepts as they pertain to the current problem at hand, which is how to account for an apparent null verb estar in Galician U-​lo structures and Brazilian Portuguese cadê structures. Gallego and Uriagereka’s (2016) understanding of the copular verbs ser and estar can be summarized as estar = ser + X. They propose that the subject and predicate adjective of both ser and estar are base-​generated within a small clause (SC) complement of v, with estar having an extra projection “XP” between v and SC to account for its prepositional nature. In essence, in the following structures from Gallego and Uriagereka (2016: 133), X head-​ adjoins to ser in v, thus begetting estar.

108  Brian M. Gravely, Jr. and Timothy M. Gupton (26) a. SER

b. ESTAR

While much of this investigation deals with details not immediately relevant to the current study, the SC captures ideas about copulas dating to Abney (1987: 64) as having no theta grid and selecting no arguments; only an SC ‘complement’ (see also Bosque, 2001). From this standpoint, copulas are purely functional elements that provide tense. Our interpretation of Gallego and Uriagereka’s vP projection for U-​lo is as in (27). (27)

The XP between vP and the SC captures the null stative matrix of the predicate that combines with null v, as in (26b). We maintain the SC configuration for control purposes, holding that Gallego and Uriagereka’s sister of the SC subject DP is a locative controller in order to account for U’s lack of movement. As Juan Uriagereka (p.c.) points out, this thematic relation between the two daughters of the SC enables the same interpretation as in a normal copular sentence such as Onde está o neno? (“Where is the boy?”), which would have onde base-​generated as the sister of the DP and would move leftward to check the uninterpretable wh-​feature; however, if U is a base-​generated in situ interrogative as we propose, we must account for this relation elsewhere. 4.2  Subject case assignment in copulas and null-​copulas vs. U-​lo Following Benmamoun (2000, 2008), the locus of nominative case assignment for a null and overt copula is T. This poses a problem for our account of VDICs since we have proposed that, as in the case of imperatives, its derivation lacks

Verbless DP interrogative constructions and enclisis  109 all functional structure (i.e., TP, NegP, FinP) apart from the FP required for clitics. To account for the lack of case assignment, we claim that the DP in VDICs receives default case. Below we apply Schütze’s (2001) various tests to determine default case for Galician. 4.3  Default case Below we examine results of Schütze’s (2001) tests for default case in Galician and English vis-​à-​vis the pronouns (in bold) appearing in (28–​32). We avoid glosses here to highlight the fact that only first-​person-​singular nominative and accusative personal pronouns (eu/​min) differ in Galician. (28)  Left dislocation/​Apposition a. Eu/​*Min, quero ir á feira. b. Me/​*I, I want to go to the fair. (29)  Ellipsis Who wants to try it? a. Eu/​*Min!; Só eu/​*min!; Eu/​*Min non!; Eu/​*Min tampouco! b. Me/​*I!; Just me/​*I!; Not me/​*I!; Me/​*I neither! (30)  Gapping a. Non podedes comer vieiras e eu/​*min raxo. b. You guys can’t eat scallops and me/​*I fried chicken. (31)  DP coordination a. Ti e máis eu/​*min imos ter problemas. b. You and them/​*they are going to have problems. (32)  Modified pronouns a. O eu/​*min que ninguén coñece. b. The me/​*I nobody knows. The Galician pronouns above appear in the nominative case (28–32a) in a systematic fashion, unlike their accusative English counterparts (28b–32b), thus suggesting that nominative is the default case in Galician, as in many other Romance varieties (see Schütze, 2001: 222–​226 for apparent default case variation in Romance). We propose another test unique to Galician with the impersonal expression oxalá (‘God willing’).

110  Brian M. Gravely, Jr. and Timothy M. Gupton (33) a. Oxalá chova mañá, que nos fai falla. “God willing it will rain tomorrow because we need it.” b. E eu/​*min, amigo. (lit.) “Me, too, friend.” Admittedly, it is an unorthodox rhetorical turn to demonstrate that Galician has default nominative case when preceding descriptions of lo in VDIC constructions have suggested that the pronoun is accusative. We clarify this matter in the following section, claiming that this ersatz accusative pronoun is a determiner detached from a nominative-​case DP.

5.  Fusion of U and lo At this point, our derivation seems simple and straightforward, with U base generated in the head of ForceP and a small-​clause configuration selected by a phonologically null estar complex. (34)

Although the analysis in (34) seems to be on the right track, we believe that one step remains. Raposo and Uriagereka (2005: 663) proposed a series of clitic placement patterns for Western Iberian Romance that result in determiner-​clitics forming part of a prosodic word at PF (see also Uriagereka, 1995a, b; Raposo, 2010, 2018). (35) a. If there is a right-​adjacent head, this is the target for the clitic. b. Otherwise, a left-​adjacent head is the target. c. Otherwise, either optimal clitic placement or displacement of an appropriate host provides a target for fusion of the clitic. Following (35), we assume that early rightward fusion of a clitic obeys morphological constraints (i.e., it only targets a head), while late leftward fusion obeys prosodic constraints (i.e. it can target a head or a phrase). Although these constraints are proposed to account for all determiner clitics across Western Iberian variants, in section 5.1, we examine evidence suggesting that this may not be as uniform as proposed.10

Verbless DP interrogative constructions and enclisis  111 5.1  Determiner clitic behavior in Galician Galician determiners have two sets of morphological forms, typically called “first form” and “second form” in traditional grammars. The second forms are suppletive, occurring obligatorily when acting as third-​person accusative clitics (36) and optionally when heading a full DP (37) when a verbal host’s inflectional class ends in –​r or –​s.11 (36) Comémo-​las./​*Comemos as. eat.P R S.1P L -​CL. AC C. 3PL. F “We eat them.” (37) Comémo-​las almeixas./​ Comemos as eat.P R S.1P L -​D ET/​C L. 3PL. F clams eat.PRS. 1 P L the “We eat the clams.”

almeixas. clams

Uriagereka (1996) called the phenomenon in (37) determiner clitic placement. Although this may seem like a purely morphophonological phenomenon, Uriagereka proposed that determiners heading full DPs in Galician are only able to cliticize onto “structural governors,” which are heads in a structurally adjacent relationship that c-​command a determiner clitic with which it shares φ-​features and/​or case (38).12 (38)

Here, X c-​commands its complement YP and assigns it accusative case, showing a clear structural governor relationship as laid out above. The importance of these stipulations becomes apparent when we see that determiner cliticization in Galician may not occur with conjunctions (38), adverbs (39), bare-​quantifier nominals (40) or verbs with which the determiner shares no ϕ-​features or case, as in instances of reconstruction (41). (38) Non  me        peta          moito,  mais neg  C L .DAT. 1S G  appeal.PR S. 3S G  much  but o/​*mai-​lo que prefiras the/​but-​D E T/​C L. 3S G. M that prefer.PRS. S B JV. 2 SG “I don’t like it, but whatever you prefer.”

112  Brian M. Gravely, Jr. and Timothy M. Gupton (39) Iñantes a/​*Iñante-​la cousa before the/​before-​D ET/​C L. 3S G. F thing máis  sinxelo more simple “Before, things were much simpler.”

era be. I M P F V. 3 SG

moito much

(40) Saben todos a/​*todo-​la canción xa, non é? know.P R S. 3PL all the/​ALL-​DET/​CL.3SG.F song already N E G  be.P RS. 3 SG “They all know the song by now, right?” (41) Fixemos cantar os/​ *canta-​los dous make.P R S. 1PL sing.I NF the/​sing.I NF-​D E T. C L . 3 P L . M two a  Xabier to Xabier “The two of us made Xabier sing.” As we can see upon examining (40) in (42), neither todos nor a canción share case or a φ-​set. Although todos c-​commands a and they are structurally adjacent, todos does not govern a in any way, ruling out cliticization. (42)

Uriagereka formalized observations by Álvarez-Cáccamo (1989), showing that determiner cliticization occurs obligatorily in a number of fixed cases, such as when the preposition por “for” (43) or the quantifier ambos “both” (44) selects a full DP. The constraints mentioned above hold in these examples (45). (43) Fai-​no  polo/​*por                           o  ben do.IMP -​C L . AC C. 3S G. M for-​D ET. CL. 3S G. M /​for the well de todos. of all “Do it for the good of everyone.”

Verbless DP interrogative constructions and enclisis  113 (44) Ámbo-​los  /​*Ambos      os  dous  viron both-​D E T.C L. 3PL. M both of two  see.P RT. 3 P L o  que  fixera. the that do.PS T. PFV. 3S G “Both of them saw what he had done.” (45)

As we can see, the notion of structural governor is pertinent to our investigation of U-​lo given that, as we have shown, the determiner always takes this form and cliticizes to U.13 Although these forms are “frozen” for historical reasons not relevant to a synchronic account, even in fixed forms, the determiner clitic host is a structural governor that heads its own projection. 5.2  Where do determiner clitics go? Uriagereka (1996:  288, ex. 48)  models an argument for determiner clitics attracted to f, “piggybacking” onto a mobile dative clitic moving across the accusative DP. We update this structure in (46). (46) Lembróu-​vo-​la súa dor (“He reminded you of his pain”)

An intriguing idea left open for further investigation was that determiner clitics without an accompanying dative clitic (36) are attracted to F, leaving their DP. We leave an exhaustive study of this phenomenon for further research since it is orthogonal to the issue of VDICs, but we adopt the basics of this possibility for U-​lo in section 5.2.1.14

114  Brian M. Gravely, Jr. and Timothy M. Gupton 5.2.1  U-​lo and FP Uriagereka’s (1995a) original proposal for FP as the locus for clitic movement dealt with a separate set of morphological ϕ-​features that attracted the clitic to the head of F(=f).15 We also assume that f is the landing site of enclitic pronouns on finite predicates and in the case of U-​lo, as well. Our motivations for doing so are based in part on the facts in (4–​5), showing that U requires a definite determiner to its immediate right. In a loose sense, we may think of U as a clitic-​like element that requires a specified/​pre-​selected functional morpheme to appear alongside it. Following Raposo and Uriagereka (2005), the PF requirements for Galician clitics are constrained by structural locality such that a clitic in F may only be hosted by a leftward-​adjacent specifier ([Spec, FP]) or the next higher head. From the data provided in this section, we have two questions to answer: (1) How can U be considered a structural governor if it shares no features with lo? and (2) What rules out a structural model in which the head of ForceP simply selects a DP complement? The answer to the first question is tricky. Thus far, we have only postulated that U has an [iQ] feature, which has no relevance for the DP it selects: U only selects lo in a narrow sense. Crucial to the generative analysis of clitics was the proposal (e.g., Chomsky, 1995: 249; Uriagereka, 1995b: 113) that clitics are minimal-​maximal constituents, base-​generated as DPs and then moving as D heads. We propose that, much like clitics in Western Iberian, U also has a morphological requirement [Morph] that must be met for the derivation to converge. Considering its meager morphophonological character, this clitic-​ like nature appears to be a concomitant property. We know that this relationship must be a local, head-​to-​head relation, owing to the determiner clitic morphology (47).16 (47) U-​lo can? (“Where is the dog?”)17,18

Upon moving to f, the determiner clitic checks morphological φ-​features on f, forming the clitic cluster [o, f], which satisfies morphological requirements on U.19

Verbless DP interrogative constructions and enclisis  115 The second question is answered by the solution to the first. If U has a morphological requirement, then selecting a full DP would not solve the problem. Setting f aside as part of the solution, not even the determiner alone would satisfy this requirement if it remained within the DP. Recall (35a): (35) a. If there is a right-​adjacent head, this is the target for the clitic. We can easily imagine that if the determiner remains in the DP it heads, its morphological requirements need not be met elsewhere. While we recognize the circularity of this statement, the data in (38–​41) and (43, 44) speak for themselves and beg an explanation.20 Although a deeper investigation of this phenomenon is due, we believe that this represents a solid first step toward an explanatorily adequate account of determiner clitic movement in Galician. 5.2.2  Types of cliticization: Specifier or head? The last topic that we wish to touch on briefly deals with the position of U in the head of ForceP and its direct effect on the clitic that it hosts. The requirement that U be in the head of ForceP and not the specifier is twofold. First, let’s hypothesize that U is a functional wh-​element that is base-​generated in [Spec, ForceP]. This is problematic for several reasons, one of them being why we need a spec-​head layout for an element that inherently carries an [iQ] feature and does not move. This concept is optimal for a Probe-​Goal configuration, as in wh-​phrases that are known to move throughout the derivation, but it is unclear based on the data presented in this chapter what this would buy us theoretically. Although base-​generated items are rare in heads throughout the Left Periphery, U seems to exhibit all of the hallmarks of an in situ interrogative. A reviewer brings to our attention the fact that Bošković uses the [iQ] feature in order to stop wh-​elements from moving and that merging a lexical item with said feature is predicated on accurate justification; however, the reviewer claims that this “crucially excludes the C-​layer since from there you cannot satisfy other checking/​licensing requirements of the item” such as case or theta roles. We wish to point out that U being a non-​argumental wh-​element does in fact permit us to posit its base-​generation in the C-​layer; moreover, the parallels to Bošković’s examples are typologically different. Just as all wh-​elements in languages such as Chinese show permanence further down in the phrase marker, Romance typically fronts their wh-​elements. Possibly the clearest reason comes from the type of cliticization that we see with U-​lo. Unlike other Romance languages, cliticization in Galician may be found in two configurations: attraction to a phrase or to a head. It is clear that only head-​to-​head constructions may produce PF modifications at Spell-​ Out (see fn. 16); therefore, due to the fact that the determiner o is externalized as lo when paired with U in this construction, we can theoretically relate this to a head-​to-​head configuration. Although we know that phrases in a local

116  Brian M. Gravely, Jr. and Timothy M. Gupton specifier position may serve as a host for clitics in Western Iberian, these constructions do not allow for morphophonological modification at PF (see, e.g., Gupton, 2014: 190) for more on locality constraints in Western Iberian).

6.  Conclusion In this chapter, we have proposed an analysis for Galician U-​lo…? constructions, which we call verbless DP interrogative constructions (VDICs), owing to their unique behavior compared to ostensibly similar structures in Romance. We have shown that U does not have an in situ counterpart low in the clause structure, which suggests that it is not a product of classic wh-​movement (e.g., from within the proposed small clause structure). Were this the case, or were we to propose that U is base-​generated in a vP adjunct position, we expect it to appear in a left-​peripheral specifier position (i.e., Spec, FocP, following, e.g., Rizzi, 1997) and, crucially, not a head position; however, as we have stated in section 5.2.2, U appearing as a specifier would preclude phonological PF processes and predict unattested and ungrammatical forms like *U o can. We have shown that VDICs do not behave like Semitic null copula in Benmamoun (2000, 2008) in that they may not be embedded. We showed that they bear similarities to Platzack and Rosengren’s (1997) “true imperative forms,” owing to their clause-​initial behavior, inability to be embedded, and their discourse-​rooted nature. The combination of these studies suggests that the clausal analysis of VDICs minimally lacks TP and FinP, which leaves them without temporal anchoring, thus correctly predicting their “here and now” behavior. We have shown that Gallego and Uriagereka’s (2016) analysis of overt copula is promising in that their analysis can account for a null estar matrix. This, combined with previous observations, led us to propose that the U portion of VDICs is base-​generated in the head of ForceP, which merges with a vP endowed with a null estar XP projection. We have posited a small clause account for the DP in VDICs with a locative feature serving as a control relation between the DP and the locative interpretation in the sentence. Finally, we have claimed that the definite article determiner clitic may behave as a true clitic in Western Iberian, attracted to F/​f owing to morphological and syntactic considerations. We feel that this account improves upon previous proposals in that it provides greater empirical coverage and explanatory adequacy. Further research on determiner-​clitic constructions in Galician and other languages will falsify the predictions of this analysis.

Acknowledgements We would like to thank the audience at the HLS22 in Austin, TX, our two anonymous reviewers for their helpful comments, and our language consultants for French (Debbie Bell) and Brazilian Portuguese (Camila Emidio and Juliano Saccomani). We also thank Juan Uriagereka for his insights and scintillating discussion of several aspects of this phenomenon. Most of all,

Verbless DP interrogative constructions and enclisis  117 we would like to thank Fernando Torreiro for his invaluable native intutions and for providing the catalyst for interest in pursuing a formal account of this phenomenon. We dedicate this chapter to him. All remaining errors are the responsibility of the authors.

Notes 1 In medieval Galician, U had the syntactic flexibility of its Latin predecessor ǔbī (Ferreiro, 1999), boasting locative, spatial, temporal, and discursive uses (cf. Gravely, 2017), as in (i). (i) …e non esperes a cras, mais pon-​llo u x’ ant’ estaba. “…and don’t wait until tomorrow but put it where it was before”

(Gravely 2017, p. 72)

2 Although cadê is widely thought to have evolved from que (é feito/​há) de…? ‘What happened to…?’ (e.g., da Silva Menon, 2014), its current usage clearly targets a locative reply; therefore, we gloss it as “where.” 3 A reviewer asks why we deem cadê a verbal interrogative phrase. Although an extensive answer is beyond the scope of this chapter, we hypothesize that cadê must have a categorial [+V] feature in addition to its locative interpretation and [iQ] feature, which comes as no surprise considering its origins (see also fn. 2). Calling it a phrase simply clusters characteristics that are typically dispersed across multiple lexical items in other well-​known locative expressions in Romance. 4 This is, admittedly, a simplification of “clitics-​ for-​ copulas,” stemming from Doron’s (1986) analysis of Hebrew null-​copula constructions, the particulars of which are not relevant here. We refer the curious reader to her original proposal and Benmamoun’s (2000, 2008) analysis of it in minimalist terms. 5 We modify Platzack and Rosengren’s (1997: 193, ex. 13) structure trivially in (20b), substituting vP for AgrSP. We also omit the XP that they insert to the left of VP, an abbreviated structure capturing “functional projections needed for the checking of features associated with the object DP (AgrOP, AspP, etc.)” in order to avoid confusion with our eventual proposal. We also include their imperative null subject (ImpNP). 6 Elements appearing in F are there not for reasons of feature-​checking-​related narrow syntax but for those related to the clitic appearing in a well-​formed prosodic word at PF. The imperative is proposed to check [IMP] features in Platzack and Rosengren’s account and, from the head of ForceP, it meets the locality constraints for hosting clitics in Galician (e.g., Raposo & Uriagereka 2005); therefore, given that FP is not proposed to involve narrow syntactic features per se, we propose that the verb proceeds directly to Forceº without violating the Head Movement Constraint (e.g., Travis, 1984). 7 See Fernández-​Rubiera (2013) and Gupton (2014) for different explanations for how this hosting process happens. If it is the case that F is implicated in true imperatives and suppletive-​ form imperatives for clitic hosting and assuming that Platzack and Rosengren’s comments on the unavailability of FinP in true imperatives is correct, it would appear that FP and FinP are not the same syntactic projection, against Gupton’s (2014) claim.

118  Brian M. Gravely, Jr. and Timothy M. Gupton 8 A reviewer suggests that a present tense is the default, resulting from a tenseless T. Pursuing such an analysis would take us in a direction that ignores striking similarities to commands, which in our opinion offer a more convincing explanation for deictic tense. 9 Given the extensive generative literature on ser/​estar in Spanish and the fact that these verbs appear to behave identically in Galician and Spanish, we follow formal accounts of Spanish. For a descriptive account of Galician ser/​estar, see Freixeiro Mato (2006: 127–​129). 10 Uriagereka (1996) and Raposo and Uriagereka (2005) utilize the term “determiner clitic” but with different meanings. Uriagereka (1996) referred to determiner clitics as we do in section 5.1: determiners that cliticize to a structural host to its left while heading a full DP. Raposo and Uriagereka (2005) used this term to refer to all determiners, claiming that all determiners are essentially clitics on some material, whether a part of minimal-​maximal projections (e.g., third-​person accusative clitics) or determiners heading a full nominal DP. 11 For Freixeiro (2006), the inflectional classes are 1st-​person plural, 2nd-​person singular and plural, and infinitives. Incidentally, this cliticization may occur onto a plural dative clitic that ends in –​s (e.g. 46). These are the 1st-​person plural nos, the 2nd-​person plural vos, and the 3rd-​person plural lles. 12 We adopt the term structural governor throughout the remainder of the chapter. Structural adjacency (i) was first formalized by Marantz (1988, 1989): (i) A head X is structurally adjacent to a head Y if: i.) X c-​commands  Y ii.) There is no head Z that a.  is c-​commanded by X and b.  c-​commands  Y 13 As with the prosodic considerations in fn. 6, we recognize the role of a morphophonological component. Nevertheless, we follow Uriagereka’s (1996) observation that syntactic constraints feed the morphology and phonology, which compels us to account for cliticization within the syntactic model. 14 U-​lo may emerge as unique in the overall scope of cross-​categorial determiner cliticization in Galician, but our intuition is that the general motivation will be shown to be the same. 15 Uriagereka’s (1995a) entire argument for FP goes well beyond just clitic movement and, in turn, the scope of this chapter. We focus solely on formal aspects related to clitic movement in Galician. 16 Another limitation is that U does not meet the morphological requirement of ending in –​r or –​s, a characteristic present in every preceding instance of determiner cliticization examined; however, we may consider postulating a morphological requirement for U as a replacement for its lack of overt agreement morphology. The XP-​YP head-​to-​head relationship is in line with examples we have seen in which determiner cliticization takes place; therefore, a morphological requirement for U-​lo keeps this phenomenon uniform. 17 While the overt movement of the determiner clitic brings o to U’s side, the externalized product U-​lo is formed at PF as the result of the morphophonological underspecification of U’s morphology; moreover, the same is expected in (22), (24), (35–​37), and (43–​44), as these structural governors show morphological underspecification in ending in –​s or –​r.

Verbless DP interrogative constructions and enclisis  119 18 It should be noted that there is no Agree relation or overt feature needed to account for U’s morphological requirements. Although a deeper analysis of these details is beyond the scope of this investigation, what Raposo and Uriagereka (2005) proposed as morphological ϕ-​features on accusative clitics in Western Iberian should extend to all definite determiners, as well as U. In a morphological relation such as the one posited between f and clitics in Western Iberian, only the functional null-​head is endowed with ϕ-​features. This is because, conceptually, it is part of their lexical constitution (see Gallego, 2016 for approaches to abstract feature bundles in lexical items from both a lexicalist and Distributed Morphology approach). 19 Uriagereka’s (1995a) proposal regards F/​f as a clitic, which makes clitic movement to F complementarily satisfactory: f needs the morphologically overt clitic to lean on, while the clitic needs f to appear in a full prosodic word at PF. See also e.g., Bošković (2005) for null enclitic C in English. 20 A reviewer asks us to detail our explanation behind the circularity of this statement. In brief, we may assume that Galician differs from other Romance languages pace the claim by Raposo and Uriagereka (2005) in (34) in that all of its definite determiners lean left due to morphophonological constraints, yet only certain conditions permit them to leave the DP in order to satisfy said constraints. When this morphological relation is not present, we find that the determiner leans right on the noun phrase. Were this not the case, it would be impossible to explain the behavior of the determiner in U-​lo and the other cases of determiner cliticization provided above. For a more complete analysis of this from a phonological standpoint regarding word boundaries and the syntax behind them, see Dubert-​ García (2014).

References Abney, S. (1987). The English noun phrase in its sentential aspect (Doctoral dissertation). MIT. Álvarez, R., Monteagudo, H., & Regueira, X.L. (1986). Gramática galega. Vigo: Galaxia. Álvarez-​Cáccamo, C. (1989). Variaçom lingüistica e o factor social na Galiza. Hispanic Linguistics, 2(2), 253–​298. Bender, E. (2000). Syntactic variation and linguistic competence:  The case of AAVE copula absence (Doctoral dissertation). Stanford University. Benmamoun, E. (2000). The Feature Structure of Functional Categories: A Comparative Study of Arabic Dialects. Oxford: Oxford University Press. Benmamoun, E. (2008). Clause structure and the syntax of verbless sentences. In R. Freidin, C.P. Otero & M.L. Zubizarreta (eds.), Foundational Issues in Linguistic Theory: Essays in Honor of Jean-​Roger Vergnaud, 105–​132. Cambridge, MA: MIT Press. Bianchi, V. (2003). On finiteness as logophoric anchoring. In J. Guéron & L. Tasmovski (eds.), Temps et point de vue/​Tense and Point of View, 213–​246. Nanterre: Université Paris X. Bošković, Ž. (2005). Null C in English as an enclitic. Lingvisticki Vidici, 34, 39–​57. Bošković, Ž. (2015). Phrasal movement:  Wh-​ movement. In A. Fábregas, J. Mateu & M. Putnam (eds.), Contemporary Linguistic Parameters, 251–​280. New York: Bloomsbury.

120  Brian M. Gravely, Jr. and Timothy M. Gupton Bosque, I. (2001). On the weight of light predicates. In J. Herschenson, K. Zagona & E. Mallén (eds.), Features and Interfaces in Romance, 23–​38. Amsterdam: John Benjamins. Carballo Calero, R. (1979). Gramática elemental del gallego común. Vigo: Galaxia. Chomsky, N. (1995). The Minimalist Program. Cambridge, MA: MIT Press. Comrie, B. (1987). Russian. In B. Comrie (Ed.), The World’s Major Languages, 329–​ 347. Oxford: Oxford University Press. da Silva Menon, O.P. (2014). Cadê e variantes: gramaticalização em língua portuguesa. Caligrama: Revista de Estudos Românicos, 19(2), 99–​130. den Dikken, M., & Blasco, M. (2007). Clitic climbing in Spanish imperatives. In W. van der Wurff (Ed.), Imperative Clauses in Generative Grammar:  Studies in honour of Frits Beukema, 135–​152. Amsterdam: John Benjamins. Doron, E. (1986). The pronominal copula as agreement clitic in the syntax of pronominal clitics. Syntax and Semantics, 19, 313–​332. Dubert-​García, F. (2014). Syntax and word-​specific phonetics:  the origins of the allomorphs of the Galician definite article. Loquens, 1(2), 1–​13. Eid, M. (1991). Verbless sentences in Arabic and Hebrew. In B. Comrie & M. Eid (eds.), Perspectives on Arabic Linguistics, Vol. 3, 31–​61. Amsterdam: John Benjamins. Fassi Fehri, A. (1993). Issues in the Structure of Arabic Clauses and Words. Dordrecht: Kluwer. Fernández-​ Rubiera, F. (2013). Clisis revisited:  Root and embedded contexts in Western Iberian. In C.M. Salvesen & H.P. Helland (eds.), Challenging Clitics, 55–​86. Amsterdam: John Benjamins. Ferreiro, M. (1999). Gramática histórica galega. Vol. I:  Fonética e Morfosintaxe. Santiago de Compostela: Laiovento. Freixeiro Mato, X.R. (2006). Gramática da lingua galega. Vol. II: Morfoloxía. Vigo: A Nosa Terra. Gallego, Á. (2016). Lexical items and feature bundling:  Consequences for microparametric approaches to variation. In L. Eguren, O. Fernández-​Soriano & A. Mendikoetxea (eds.), Rethinking Parameters, 133–​ 169. Oxford:  Oxford University Press. Gallego, Á., & Uriagereka, J. (2016). Estar = Ser + X. Borealis:  An International Journal of Hispanic Linguistics, 5(1), 123–​156. Gravely, B. (2017). Unha mirada cara a u e onde n’As Cantigas de Santa María. Revista Galega de Filoloxía, 18, 69–​78. Gupton, T. (2014). The Syntax-​information Structure Interface: Clausal Word Order and the Left Periphery in Galician. Berlin: De Gruyter. Kempchinsky, P. (1998). Mood phrase, case checking and obviation. In A. Schwegler, B. Tranel & M. Uribe-​Etxebarria (eds.), Selected Papers from the 27th Linguistic Symposium on Romance Languages (LSRL XXVII), Irvine, 20–​22 February 1997, 143–​154. Amsterdam: John Benjamins. Kempchinsky, P. (2009). What can the subjunctive disjoint reference effect tell us about the subjunctive? Lingua, 119, 1788–​1810. Marantz, A. (1988). Clitics, morphological merger, and the mapping to phonological structure. In M. Hammond & M. Noonan (eds.), Theoretical Morphology, 253–​ 270. New York: Academic Press. Marantz, A. (1989). Clitics and phrase structure. In A. Kroch & M. Baltin (eds.), Alternative Conceptions of Phrase Structure, 99–​116. Chicago, IL:  University of Chicago Press.

Verbless DP interrogative constructions and enclisis  121 Platzack, C., & Rosengren, I. (1997). On the subject of imperatives:  A minimalist account of the imperative clause. Journal of Comparative Germanic Linguistics, 1(3), 177–​224. Raposo, E. (2010). Clitic placement in Romance:  A phase-​ theoretic approach. In D.B. Gerdts, J.C. Moore & M. Polinsky (eds.), Hypothesis A-​Hypothesis B: Linguistic Explorations in Honor of David M. Perlmutter, 401–​426. Cambridge, MA: MIT Press. Raposo, E. (2018). Nominal ellipsis and prepositional modifiers in Portuguese:  A phase-​theoretic approach. Probus, 30(2), 277–​303. Raposo, E., & Uriagereka, J. (2005). Clitic placement in Western Iberian: A minimalist view. In G. Cinque & R. Kayne (eds.), The Oxford Handbook of Comparative Syntax, 639–​697. Oxford: Oxford University Press. Rizzi, L. (1997). On the fine structure of the left periphery. In A. Belletti & L. Rizzi (Eds.), Elements of Grammar, 281–​337. Dordrecht: Kluwer. Schütze, C. (2001). On the nature of default case. Syntax, 4(3), 205–​238. Silva-​Villar, L. (1996). Enclisis in northwestern Iberian languages: A diachronic theory (Doctoral dissertation). University of California, Los Angeles. Silva-​Villar, L. (1999). Diachronic qualitative studies on minimalist operations: Verbless locative inversion in NILs. In J. Gutiérrez-​Rexach & F. Martínez-​Gil (eds.), Advances in Hispanic Linguistics:  Papers from the 2nd Hispanic Linguistics Symposium, Vol. 2, 562–​578. Somerville, MA: Cascadilla Press. Silva-​Villar, L. (2001). Verbless derivations in historical syntax:  A case study of Northwestern Iberian languages. In J. Gutiérrez-​Rexach, & L. Silva-​Villar (Eds.), Current Issues in Spanish Syntax and Semantics, 309–​346. New York: Mouton de Gruyter. Travis, L. (1984). Parameters and effects of word order variation (Doctoral dissertation). MIT. Uriagereka, J. (1995a). An F position in Western Romance. In K. Kiss (Ed.), Discourse Configurational Languages, 153–​175. New York: Oxford University Press. Uriagereka, J. (1995b). Aspects of the syntax of clitic placement in Western Romance. Linguistic Inquiry, 26(1), 79–​123. Uriagereka, J. (1996). Determiner clitic placement. In R. Freiden (Ed.), Current Issues in Comparative Grammar: Studies in Natural Language and Linguistic Theory, Vol. 35 (pp. 257–​295). Dordrecht: Springer.

6  Leísmo One le or two? Adolfo Ausín and Francisco J. Fernández-​Rubiera

1.  Introduction Most Spanish speakers use le for Indirect Objects as in (1), but some speakers use le to refer to Direct Objects as in (2), an example of what is standardly called leísmo. (1) a. A Luis,  le    entregué un regalo. to Luis  himDAT.CL gave1SG   a  present “To Luis, I gave him a present.” b. A Laura,  le    dije    la  verdad. to Laura  herDAT.CL  told1SG   the  truth “To Laura, I told her the truth.” (2) A Juan, le    vimos. to Juan himDAT.CL saw1PL “Juan, we saw him.” The question this study would like to answer is whether the two les that we see in these examples are the same element or different; that is, whether these two different instantiations of les should be analyzed as the same morphosyntactic element, or whether they should be analyzed as different elements that just happen to be pronounced the same. In this study, we will argue for the second option (that is, two different les) based on some restrictions found on the use of le instead of les, what is often called “le-​for-​les.” This chapter is structured as follows. First, we briefly discuss leísmo. Then, we review some arguments that have been offered regarding the existence of one le or two. Next, we introduce “le-​for-​les” and show how this phenomenon can be used to argue for the existence of two different les. A  brief review of “le-​for-​les” and how it can be used as an argument for the existence of two les follows, supported by data illustrating so-​called “fake leísmo” (Fernández-​Ordóñez,  1999).

Leísmo: one le or two?  123

2.  The Spanish clitic system and leísmo Let’s start by reviewing the Spanish clitic system, which appears summarized in Table 6.1. Table 6.1  Spanish clitic system  Masculine

Feminine

Lo

Accusative

La

(3) A Juan, lo    vimos. to Juan  himACC.CL  saw1PL “Juan, we saw him.” (4) El libro, lo    compré   en la tienda. the book  itACC.CL   bought1SG  at the store “The book, I bought it at the store.”

(5) A Marta, la    conozco. to Marta  herACC.CL  know1SG “Marta, I know her.” (6) La paella, la    hice     yo. the paella  itACC.CL made1SG I “The paella, I made it.”

Le

Dative

(7) A Luis,  le     entregué un regalo. to Luis  himDAT.CL  gave1SG   a present “To Luis, I gave him a present.” (8) A Laura,  le      dije   la  verdad. to Laura  herDAT.CL  told1SG  the truth “To Laura, I told her the truth.”

This table reflects a standard dialect of Spanish, where le is used only for datives, and accusative is expressed with lo or la. It is well known that there is a considerable amount of dialectal variation with respect to clitics. Fernández-​ Ordóñez (1994) identifies around ten different systems, only within European Spanish. Here we reproduce three of them. System (a)

Singular

Accusative

Countable

Dative

Le Le

Plural

Masculine

Uncountable

Masculine

Feminine

Lo Lo ~ le

Les Les

Las Les

Feminine La La

System (d) Singular Accusative

Dative

Plural

Countable

Uncountable

Masculine

Feminine

La

Les ~ los

Las

La ~ le

Les

Las

Masculine

Feminine

Masculine

Feminine

Le ~ lo

La

Lo

Le

La

Lo ~ le

124  Adolfo Ausín and Francisco J. Fernández-​Rubiera System (h)

Singular

Accusative

Countable

Dative

Plural Uncountable

Masculine

Feminine

La

Lo

Los

Las

Le

Le

Les

Las

Masculine

Feminine

Le Le

One of the features that we see repeatedly in many of these systems is leísmo; that is, the use of le instead of some other clitic. The clitic being replaced depends on the dialect. In some dialects le will be used to refer to DOs that are masculine and human as in (9): (9) A Juan, le vimos. to Juan himDAT.CL saw1PL “Juan, we saw him.”

(cf. (3) A Juan, lo to Juan himACC.CL

vimos.) saw1PL

In other dialects, le will also be used to refer to inanimate masculine DOs as in (10), and in some other dialects it can also be used with feminine DOs, as in (11). (10) El libro,  le     compré   en la  tienda.  (cf. (4)  El libro,  lo    compré   en la tienda.) the book  itDAT.CL bought1SG  at the  store      the book itACC.CL bought1SG  at the store “The book, I bought it at the store.”

(11) A Marta, le   conozco.  to Marta  herDAT.CL  know1SG “Marta, I know her.”

(cf. (5) A Marta,  la        conozco.) to Marta  herACC.CL know1SG

Crucially, for the purposes of our study, we are not very interested in the different types of leísmo nor in its geographical distribution, but rather in the correct characterization of le when it is used to refer to DOs in leísta dialects. The key question we are trying to answer is the following: Is the le that appears in the leísta examples (9)–(11) the same le that appears in (7) and (8), repeated below? That is, do we have one le or two? We review previous accounts and tackle this issue in the next section. (7) A Luis, le    entregué un regalo. to Luis  himDAT.CL gave1SG  a  present “To Luis, I gave him a present.” (8) A Laura, le    dije  la verdad. to Laura  herDAT.CL  told1SG  the truth “To Laura, I told her the truth.”

Leísmo: one le or two?  125

3.  One le or two? A review of previous accounts As far as we know, the clearer implementation of the idea that the Indirect Object le in all dialects and the Direct Object le in leísta dialects are actually the same element is Bleam (1999), who coins the term “One le Hypothesis.” Bleam considers at least three types of evidence in favor of the One le Hypothesis:  Morphophonological similarities, the possibility of doubling, and event participant restrictions on doubling, which we review next. First, regarding morphophonological similarities, Bleam notes that “the facts are that accusative le has the same form as dative le. If we believe that this similarity is more than coincidental, we are led to posit a single analysis for the two clitics. Call this ‘the One le Hypothesis’ ” (Bleam, 1999: 54). Elsewhere, Bleam writes: “A priori, an analysis that takes these elements to be the same seems superior to one that takes them to be accidentally homophonous” (Bleam, 1999: 50–​51). Second, regarding doubling, Bleam observes that “the possibility of clitic doubling in leísta Spanish is tied to the form of the clitic and not to the grammatical function” (Bleam, 1999: 51). This can be seen in the contrast between (12a) and (13).1 (12) a. (*La)   vi   a María. (*herACC.CL) saw1SG to Maria “I saw Maria.” b. A María, la vi. to Maria herACC.CL saw1SG “Maria, I saw her.” c. La vi. herACC.CL saw1SG “I saw her/​it.” (13) Le vi a herDAT.CL saw1SG to “I saw Maria.”

María. Maria

Third, regarding event participant, Bleam observes that “direct object and indirect object doubling do share some interpretive properties: in both cases, the entity represented by the NP must be conceived of as an event participant” (Bleam, 1999:  150) She discusses the following examples to illustrate her point. (14) a. Marta (*le) envió un libro a Nueva York. Marta LE sent3SG a book to New York b. *Marta sent New York a book.

126  Adolfo Ausín and Francisco J. Fernández-​Rubiera (15) Le di una patada a la mesa. itDAT.CL gave1SG a kick to the table “I gave the table a kick.”          

(Bleam, 1999: 10)

This proposal of there being a single le also underlines characterizations of leísmo like the one found in Ordóñez (2012), where leísmo is described as dativization of the direct object. Gallego, Roca and Colomina (2017) also describe leísmo as dativization of the accusative clitic, which sometimes they formalize as in (16). They take this as an instantiation of the case displacement process that can be found in languages such as Basque. (16) LEÍSMO: DPACC → DPOBL

(Gallego et al., 2017)

However, there are also counterarguments to the “One le Hypothesis,” some of which are discussed by Ormazabal and Romero (2007), who list the following differences between dative le and accusative le (Ormazabal & Romero 2007: 320–​321). First, they note that “dative le, but never accusative animate le, is compatible with passive constructions,” as illustrated in (17). (17) a. El  libro le     fue entregado. The book him/​ herDAT.CL was given “The book was given to him/​her.” b. *Le     fue  visto. him/​herDAT.CL was seen “S/​he was seen.” They also note that “while dative le can—in some contexts must—​double, some leísta dialects do not allow doubling of the clitic le when it corresponds to an animate accusative clitic.” This point is illustrated in (18). (18) a. Le    di   una  manzana  a  Mateo. himDAT.CL gave1SG an   apple    to Mateo “I gave Mateo an apple.” b. (*Le)  vi   a Mateo. himDAT.CL saw1SG a Mateo “I saw Mateo.” Finally, Ormazabal and Romero point out that, “unlike accusative le, dative le is not selective with respect to animacy, and in fact may refer to inanimate entities,” as the following example shows:

Leísmo: one le or two?  127 (19) Le puse azúcar al pastel. itDAT.CL put1SG sugar to-​the cake “I put sugar on the cake.” From all of this, Ormazabal and Romero conclude that “[leísta] le must be analyzed as an accusative third person clitic.”2 After reviewing the arguments for and against the “One le Hypothesis,” we would like to discuss the phenomenon generally known as “le-​for-​les” to show that its inclusion in this debate brings another piece of evidence to claim for the existence of two different les. We provide next an overview of this phenomenon.

4.  “le-​for-​les” and its role in the one/​two le debate At least since Cuervo (1955), it has been observed that in dative clitic-​doubling constructions, we can frequently find lack of agreement between the clitic and its postverbal double DP, as illustrated in (20). (20) Le dice adiós a las garzas que pasan. leSG says bye to the herons that fly-​by “S/​he says goodbye to the herons that fly-​by.” (Cuervo, 1955) To show how widespread the phenomenon was, Cuervo noted the example in (21) from the Spanish Royal Academy (RAE). The intended purpose of this example was to illustrate cacophony. The unintended effect was to illustrate the use of “le-​for-​les” even among the most educated speakers, such as the members of the Royal Academy. (21) Dale las lilas a las niñas. give-​leSG the lilacs to the girls “Give the lilacs to the girls.” (RAE, 1885: 287) The use of singular le instead of the expected plural les has been acknowledged, documented and discussed in a number of places, such as Cuervo (1955), Casares (1918:107–​120), Sturgis (1927), Rini (1988), DeMello (1992), Roca (1992, 1996), Fernández-​Soriano (1999), Boeckx and Jeong (2004), Huerta Flores (2005), RAE & AALE (2009:  2664), Ausín and Fernández-​Rubiera (2017).3 One of the features that is discussed in these sources is that “le-​for-​ les” is acceptable mostly with postverbal dative phrases, relevant examples in (22).

128  Adolfo Ausín and Francisco J. Fernández-​Rubiera (22) “Le-​for-​les” contexts a. Le  dice  adiós    á  las  garzas  que pasan. leSG say3SG goodbye to the herons that fly-​by “S/​he says goodbye to the herons that fly-​by.” b. Yo no  le   tengo  miedo á  las  balas. I  not leSG have1SG fear  to the bullets “I am not scared of bullets.” c. Le   dice   a   todos que vengan. leSG say3SG to all   that come3PL “S/​he tells everyone to come.”         (Cuervo, 1955, §309) d. Todo se   acabaría     si le  pegaran cuatro tiros  a  unos  cuantos granujas. all   rflCL  would-​end3SG if leSG shoot3PL four   shots to some few   crooks “It would all be over if they shot dead a few crooks.” (Casares, 1918: 108)

The examples in (22), which exhibit postverbal dative phrases, contrast sharply with those instances in which the dative phrase is in preverbal position, when it is absent, or when the doubled element is a pronoun. In these syntactic environments, illustrated in (23), obligatory agreement arises (i.e., obligatory les): (23) Ungrammatical “le-​for-​les” contexts /​obligatory agreement les contexts: a. á las garzas {*le/​ les} dice  adiós. [cf. 22a] to  the herons  LE/​LES  say3SG goodbye “To the herons, s/​he says goodbye.” b. á  las  balas  yo no  {*le/​les} tengo  miedo.  [cf. 22b)] to the bullets I  not LE/​LES have1SG fear “Bullets, I am not scared of them.” c. a  todos {*le/​les} dice   que  vengan.  [cf. 22c] to all   LE/​LES  say3SG that come3PL “Everyone, s/​he tells them to come.” d. Aquí hay dos caballeros que  desean  ver   al señorito.  ¿Qué {*le/​ les} digo  proPL? here  are  two gentlemen that wish3PL seeINF  to-​the master  what  LE/​LES  say1SG  pro “Here are two gentlemen who wish to see the master. What should I tell them?” e. Juan {???le/​les}    entregó un libro  a  ellos.4 Juan    LE/​LES  gave3SG  a  book to them “Juan gave them a book.”

Interestingly, the contexts that require agreement in the dative (shown in (23) are the same syntactic contexts where accusative clitics are required: (24) Accusative clitic doubling contexts (cf. (23)): a. A Marta Juan  *(la)  vio.   [cf. 23a, 23b and 23c] to Marta Juan  herCL saw3SG “Marta, Juan saw her.” b. Juan  *(la)  vio    pro[FEM, SING].  [cf. 23d] Juan herCL saw3SG  pro “Juan saw her.” c. Juan *(la)  vio   a ella. [cf. 23e]5 Juan herCL saw3SG  to her “Juan saw her.”

Leísmo: one le or two?  129 Table 6.2 summarizes the parallelism between agreeing les and the presence of accusative clitic. In previous work, we capitalized on this parallelism to provide a formal account of “le-​for-​les”, which we’ll review later. Now what we would like to point out is that “le-​for-​les” is not possible with accusative les. Although the relevant contrast is harder to find due to the fact that doubling (a prerequisite for “le-​for-​les”) is more limited with accusatives than with datives, the contrast is clear. Leísta speakers who accept clitic doubling with accusatives will accept sentences like those in (25), which showcase a doubling construction with les, but reject the use of the non-​agreeing le in the same environment (as in (26)). (25) Obligatory agreement with accusative les–leísta dialects (cf. (22)): a. (?)Les  vimos  a  los  chicos en  el  parque.   lesPL saw1PL to the kids  in the  park    “We saw the kids in the park.” b. (?)Les  conocen  a  los  famosos.   lesPL know3PL to the famous    “They know the famous people.” c. (?)La   policía  les   capturó    a  los  terroristas.   the  police  lesPL captured3SG  to the  terrorists    “The police captured the terrorists.” (26) Ungrammatical le with accusative–leísta dialects (cf. (22) and (25)): a. *Le   vimos  a los chicos en el parque. leSG saw1PL  to  the  kids   in  the park “We saw the kids in the park.” b. *Le  conocen  a  los  famosos. leSG know3PL to the famous “They know the famous people.” c. *La policía le  capturó   a los terroristas. the  police  leSG captured3SG  to the terrorists “The police captured the terrorists.”

The conclusion that we draw from these examples is that “le-​for-​les” is possible with datives (as in (22)), but it is not possible with accusatives (see (25a) and (26a)). If the One le account were on the right track, the different behavior that we find with respect to “le-​for-​les” would remain unexplained. We believe that these facts support the idea that there are two different les: accusative le and dative le. That is, the le in accusative constructions in leísta Spanish, as in (25), is different from the le that we find in dative constructions, as those in (22). We are taking the ungrammaticality of (26) as evidence that “le-​for-​les” is not possible with accusative le. However, examples like these in (27) could be interpreted at face value as a counterexample to this generalization since in all of them we see “le-​for-​les” in constructions that could be described as leístas.

newgenrtpdf

Agreeing les required (i.e., “le-​for-​les” not possible) Pro

YES

a. Les     regalé un libro proPL.    themDAT.CL   gave1SG  a  book “I gave them a book.” Left dislocated YES b. A los chicos, les     regalé   un libro. to the guys   themDAT.CL  gave1SG  a  book “The guys, I gave them a book.” Full pronoun YES? c. Les      regalé  un libro a  ellos. themDAT.CL gave1SG  a  book to them “I gave a book to them.” Full DP Optional d. Le(s)  regalé  un libro  a  los  chicos. LE(s) gave1SG  a  book to the guys “I gave the guys a book.”

Presence of accusative clitic required YES e. Los     vi    proPL. themACC.CL  saw1SG  “I saw them.” YES f.  A los chicos,  los    vi. to the guys    themACC.CL  saw1SG “The guys, I saw them.” YES g.  Los    vi   a ellos. themACC.CL  saw1SG  to them “I saw them.” Depends on the dialect h.  (Los)    vi   a los chicos. themACC.CL  saw1SG  to  the guys “I saw the guys.”

130  Adolfo Ausín and Francisco J. Fernández-​Rubiera

Table 6.2  Parallelism between agreeing “les” and accusative clitic doubling 

newgenrtpdf

Leísmo: one le or two?  131

(27) Potential counterexamples of “le-​for-​les” with accusative le a. …no debe   de asustarle   a  los  pasajeros  que tienen   todo en orden… not  should of frighten.leSG to  the passengers that have3PL  all   in order “…it should not frighten those passengers that have everything under control…” b. Esta situación en general le   asusta      a   los  hermanos  mayores. this  situation  in general leSG frighten3SG to the  siblings  older “In general, this situation frightens older siblings.” c. …de  tener   la  oportunidad de servirle  a  los  estudiantes,  maestros  y   padres   de… of   haveINF  the opportunity  of serve.leSG  to the students   teachers  and parents of “…to have the opportunity to serve the students, the teachers, and the parents at…” d. El  gobierno   se  compromete  a  pagarle  a  los  jubilados. the  government rflCL commits   to payINF.leSG  to the retirees “The government is committed to pay the retirees.” e. ¿Hay  que pagarle  a  los  buitres? must3SG that payINF.leSG to the vultures “Do we have to pay the vulture (funds)?” f. Por  qué  no  hay    que  pegarle  a  los  niños  nunca. for  what no must3SG that hitINF.leSG to the kids never “Why you should never hit children.” g. Pegarle  a   los  niños  no  es efectivo. hitINF.leSG  to  the  kids  not is effective “Hitting your children is not effective.” h. …para (…)  ayudarle  a  los  vecinos  en sus  traslados. to     helpINF.leSG to the neighbors in their moving “…to help the neighbors in their moving.” i. …se  atrevió  a  retar    a  Dios para ayudarle  a  los  humanos. rflCL dared3SG  to challenge to God to  helpINF.leSG to the humans “…dared to challenge God to help humans.”

132  Adolfo Ausín and Francisco J. Fernández-​Rubiera In what follows, we argue that these examples should not be taken as counterexamples to the previous generalization. In fact, we argue that they provide further support to the idea that “le-​for-​les” is only possible with dative le. To do that, we contend that these examples do not instantiate “real leísmo”, but instead they showcase what Fernández-​Ordóñez (1999) describes as “fake leísmo” (“leísmo aparente”).

5.  “le-​for-​les”, “fake leísmo”, and the one/​two le debate Starting with asustar (“to frighten”), this verb can appear with le or with lo, as shown in (28). At face value, it could be argued that the examples with le are instances of leísmo. If so, the examples in (29) with asustar and a non-​ agreeing le in would be counterexamples to our claim that “le-​for-​les” is not possible with leísmo. (28) a. Lo asustan. himACC.CL frighten3PL b. Le asustan. himDAT.CL frighten3PL “They frighten him.” (29) a. …no debe  de asustarle  a  los  pasajeros  que  tienen  todo en orden… not  should of frighten.leSG to the passengers that have3PL  all  in order “…it should not frighten those passengers that have everything under control…” b. Esta situación en general le  asusta    a  los  hermanos  mayores. this  situation  in general leSG frighten3SG  to the siblings  older “In general, this situation frightens older siblings.”

However, as noted by many scholars, including Fernández-​Ordóñez (1999), the examples in (28) do not mean the same, since the example with lo is more likely to have an animate, agentive subject. The proposal that has been made is that the version with lo has an accusative complement whereas the version with le has a dative complement. If so, the examples in (29) with the non-​ agreeing le are not really a counterexample to our claim that “le-​for-​les” is not possible with leísmo since here we are dealing with dative le, which again brings further evidence for the claim that there are two different les. Thus, these examples are just instances of regular “le-​for-​les” with a dative. Let’s turn now to verbs like servir and pagar (“to serve” and “to pay” respectively), as in (30) and (31). (30) a. A Juan le sirvieron. to Juan himDAT.CL served3PL b. A Juan lo sirvieron. to Juan himACC.CL served3PL “Juan, they served him.”

Leísmo: one le or two?  133 (31) a. A Juan le pagaron. to Juan himDAT.CL paid3PL b. A Juan lo pagaron. to Juan himACC.CL paid3PL “Juan, they paid him.” It could be argued that we are dealing with leísmo, since it looks as if le in the (a) examples is replacing the lo we find in the (b) examples. If this were the case, then the examples in (32) would be counterexamples to our claim that there is no “le-​for-​les” with leísmo. (32) a. …de tener  la  oportunidad de servirle  a  los  estudiantes,  maestros of haveINF  the opportunity  of serve.leSG to the students    teachers y  padres… and parents “…to have the opportunity to serve the students, the teachers, and the parents…” b. El gobierno  se  compromete a pagarle  a  los  jubilados. the government rflCL commits   to payINF.leSG  to the retirees “The government is committed to pay the retirees.” “The government is committed to pay the retirees.” c. ¿Hay   que  pagarle   a  los  buitres? must3SG that payINF leSG  to the vultures “Do we have to pay the vulture (funds)?”

However, Fernández-​Ordóñez (1999) argues that the le that we have in these examples is not an accusative le but a dative le. She proposes that the direct object is omitted in this type of construction and that the remaining object is the indirect object as in (33). (33) a. Aunque habían pedido las cervezasj hacía media hora, todavía tardaron otro rato en servirles Øj. Although had3PL ordered the beers since half hour, still lasted another bit in serving.themDAT.CL “Although they had ordered the beers half an hour before, it still took them a while to be served.” b. A Juani lei /​lo*i pagué Øj. to Juan himDAT.CL/​himACC.CL paid1SG “Juan, I paid him.” If as Fernández-​Ordóñez says the object in this example is dative, then there is no leísmo and the examples in (32) are not a counterexample to our claim that “le-​for-​les” is not possible with leísmo. Similar remarks could be made regarding pegar “to push/​hit/​stick.” The examples in (34) could be interpreted as leísmo. If we consider the examples

134  Adolfo Ausín and Francisco J. Fernández-​Rubiera in (34) as instances of leísmo, the presence of “le-​for-​les” that we find in examples like (35) would be unexpected under our proposal. (34) a. A Juan lo pegaron. to Juan himACC.CL hit/​stick3PL b. A Juan le pegaron. to Juan himDAT.CL hit/​stick3PL “Juan, they hit him.” (35) a. Por qué  no  hay  que  pegarle   a   los  niños  nunca. for what not must that hitINF.leSG to the kids  never “Why kids should never be hit.” b. Pegarle  a  los  niños no  es efectivo. hitINF.leSG to the kids  not is  effective “Hitting children is not effective.” However, Fernández-​Ordóñez notes that there is a difference in meaning that correlates with the type of argument being selected. If accusative is selected, it means “to push” or “to stick”; if dative is selected, it means something like “to hit.” In the second interpretation, and not in the first one, an implicit golpes (“beats”) could be made explicit, as in (36b). (36) a. A Pedro lo agarré y   lo    pegué (*golpes) contra la to Pedro him caught and  himACC.CL  sticked  *beats   against  the pared (~stick). wall “I grabbed Pedro and I sticked him to the wall.” b. A Pedro lo agarré    y   le      pegué  (golpes)  hasta  dejarlo to Pedro him caught  and  himDAT.CL  hit   beats    until  leaving.him sin   sentido (~hit). without sense “I grabbed Pedro and I hit him until I left him unconscious.”

Interestingly, the examples in (35) have the “hit” interpretation, which means, according to Fernández-​Ordóñez, that the argument is a dative. If so, the presence of “le-​for-​les” (35) is expected under our generalization. Furthermore, when verbs like pegar are passivized, the more natural interpretation is the one associated with the accusative clitic (~to stick) rather than the one associated with the dative clitic (~to hit), as illustrated in (37), where the continuation contra la pared (“against the wall”) which is compatible with the “to stick” interpretation feels more natural than the continuation hasta dejarlos sin sentido (“until they were unconscious”), which favors the “to hit” interpretation.6

Leísmo: one le or two?  135 (37) Fueron  pegados  (contra la pared / ​??hasta dejarlos   sin   sentido) were   hit/​stuck  against the wall / ​till    leave.them without sense “They were hit against the wall /​They were beaten senseless.”

Finally, let’s consider cases with ayudar (“to help”). If the presence of le in (38a) is described as an instance of leísmo, then the presence of “le-​for-​les” in (39) would be problematic for our generalization. (38) a. A Marta, no le ayudé. to Marta not herDAT.CL helped1SG b. A Marta, no la ayudé. to Marta not herACC.CL helped1SG “Marta, I didn’t help her.” (39) a. …para (…)  ayudarle    a  los  vecinos    en sus  traslados.  to    helpINF.leSG to the neighbors in their moving “…to help the neighbors in their moving.” b. …se  atrevió  a  retar    a  Dios para  ayudarle   a  los  humanos.  rflCL dared3SG  to challenge  to God to   helpINF.leSG  to the humans ‘…dared to challenge God to help humans.”

However, Fernández-​Ordóñez points out that verbs like ayudar (“to help”) have historically selected a dative complement, and thus the le in (38a) is not an instance of leísmo but a regular dative. The more recent innovation is the use of the accusative with verbs like ayudar as in (38b). If the le that appears with ayudar is a real dative, then it is not a surprise that there can be “le-​for-​les.” Fernández-​Ordóñez discusses an additional case of fake leísmo which does not seem to follow the same pattern we have been discussing. It is the fake leísmo that we find with passive-​reflexive/​impersonal se. Consider the examples in (40). (40a) instantiates a transitive sentence with a CLLDed [+human] DO. (40b) is the same sentence but with an impersonal se. In these types of examples there is dialectal variation regarding the DO clitic. Some dialects prefer los, but some others prefer les. (40) a. A sus  hijos,  sus  padres  los       quieren mucho. to their  kids   their  parents  themACC.CL love3PL  much “Their kids, parents love them a lot.” b. A los hijos,  se         los/​les         quiere  mucho. to the kids  SEPASS/​IMP.CL themACC.CL/​DAT.CL  love3SG much “Kids are loved very much.”

136  Adolfo Ausín and Francisco J. Fernández-​Rubiera Fernández-​Ordóñez claims that the variant with les in (40b) is not an instance of real leísmo, but another case of fake leísmo.7 Remember that cases of fake leísmo are real datives and therefore, under our analysis we expect to find “le-​for-​les.” The prediction is partially fulfilled because we do find some cases of “le-​for-​les” in this type of construction as illustrated in (41). However, some of the examples in (41) seem to us to be less natural than corresponding examples with the agreeing les like those in (42). In fact, a Google search reveals that, at least for the sequence se le/​les quiere a los, the option with les is more common than the option with le, which is the opposite that we find in typical “le-​for-​les” contexts. (41) a. se     le    quiere  a   los artistas. SEPASS/​IMP.CL leSG love3SG to the artists “Artists are (much) loved.” b. Por que  no  se         le  quiere  a  los Musulmanes? for what not SEPASS/​IMP.CL leSG love3SG  to  the Muslims “Why aren’t Muslims loved?” c. así se     le   quiere  a  los amigos. so SEPASS/​IMP.CL leSG love3SG to the friends “So are friends loved.” (42) a. se     les  quiere  a  los dos por  igual. SEPASS/​IMP.CL lesPL love3SG to the two by same “Both of them are equally loved.” b. Se les quiere a los muchachos de Fútbol Total. SEPASS/​IMP.CL lesPL love3SG to the guys from Fútbol Total “Guys from Fútbol Total are (much) loved.”

At the current stage in our research, we do not have a definite proposal to account for the facts in (41)–(42). If Fernández-​Ordóñez is right and these are all instances of fake leísmo, it has to be the case that there is an additional factor that favors the agreeing option in (42). Alternatively, these are all cases of true leísmo, and the non-​agreeing examples in (41) are just statistical noise. Putting aside the examples in (41)–(42), we have shown in this chapter how “le-​for-​les” can be used to tease apart two different grammatical proposals, namely whether dative and accusative le should be treated as the same or as two different elements. We have seen that “le-​for-​les” is only possible with dative le, but ungrammatical with accusative le. We have used this contrast to argue against the one le hypothesis, as it will predict a similar behavior contrary to the facts discussed. We would like to conclude this chapter with a brief overview of a proposal that accounts for “le-​for-​ les”, and show how it can be extended to account for the examples that we have been discussing.

Leísmo: one le or two?  137

6.  A formal analysis of “le-​for-​les”, leísmo and the one/​two le debate As we discussed earlier, there is a parallelism between the contexts that do not allow “le-​for-​les” and the contexts that require the presence of an accusative clitic—​see Table 6.2. In this section, we review and summarize the proposal that we make in Ausín and Fernández-​Rubiera (2017) to account for “le-​for-​les” and its parallelism with accusative clitic doubling. In Ausín and Fernández-​Rubiera (2017), we assume that accusative clitics are essentially agreement morphemes, as in Suñer (1988), Sportiche (1996), among many others. Thus, the underlying structure we assume for (43a) is the one shown in (43b). (43) a. A Marta Juan *(la) vio. to Marta Juan herACC.CL saw3SG “Marta, Juan saw her.” b.

As for dative clitics, we proposed that they should be lexically decomposed into an agreement morpheme and an applicative morpheme (see Cuervo, 2003, 2010, and Demonte, 1995 for a similar intuition regarding applied complements in these instances, and Martín 2012 for a decomposition analysis of dative clitics in Catalan). In those constructions where agreement is required, the agreeing les will appear, as in (44). (44) a. A las garzas les dice adiós. to the herons LES say3SG goodbye “The herons, s/​he says goodbye to them.” b.

138  Adolfo Ausín and Francisco J. Fernández-​Rubiera When the non-​agreeing le shows up as in Cuervo’s (45a), there is no agreement morpheme and the applicative morpheme is materialized as an invariant le (in Roca’s 1996 words, “defective”), as in Roca’s (45b). (45) a. Le dice adiós á las garzas que pasan. LE say3SG goodbye to the herons that fly-​by “S/​he says goodbye to the herons that fly-​by.” b.

Thus, the way we capture the parallelism between the presence of the accusative clitics and the presence of the agreeing dative clitic is by proposing the same element (the abstract agreement) as the source behind both phenomena. The presence of the agreement morpheme in (43b) is the source of the accusative clitic in (43a), whereas the presence of the agreement morpheme in (44b) is the source of the agreeing dative clitic in (44a). A further question is how the common peculiarities of accusative clitics and agreeing datives are derived. Here we cannot offer an answer to this question due to space limitations but see Ausín and Fernández-​Rubiera (2019) for an attempt to provide an answer. Going back to the examples that we have been discussing in this study such as (46), we propose that their structure should be the one in (47), irrelevant details omitted. The accusative agreement morpheme is materialized as the dative les or the accusative los depending on the dialect (i.e., les in leísta dialects, and los everywhere else), as in (47a), and the agreement morpheme is present in the structure. As for the structure showing the invariant le, as in no le pegues a los niños, we argued for the structure in (47b), where there is an applicative structure which is the source of the dative clitic. Since there is no agreement morpheme, the non-​agreeing le shows up. (46) Les/​ Los     vimos (a los niños) vs. no le  pegues (golpes) a los niños themDAT/​ACC saw1PL  (to the kids)     not  leSG  hit    (beats)   to the kids

Leísmo: one le or two?  139 (47) a. cf (43b)

b. cf. (45b)

The analysis in (47) follows the proposals that we made in Ausín and Fernández-​Rubiera (2017), where clitics are considered agreement morphemes attached to T. Following a reviewer’s suggestion, we would like to flesh out this proposal in more current terms adopting Kramer’s (2014: 596) analysis of clitics as object agreement under a minimalist framework. Kramer’s proposal appears in (48):  the functional head v has unvalued phi-​features, which are valued under agreement with the valued phi-​features of the DP. (48)

Assuming Kramer’s proposal in (48) and the idea that clitics should be analyzed as agreement morphemes,8 the structure of (46) (Les/​Los vimos (a los niños)) would be as in (49): the unvalued phi-​feature in (49a) gets valued under agreement as in (49b), and those valued phi-​features in v are spelled out as an accusative clitic (les or los, depending on the dialect). (49)

140  Adolfo Ausín and Francisco J. Fernández-​Rubiera When there is agreement with the dative complement as in (50), then the structure would be as in (51):  the phi-​features of v agree with the DP and combined with the incorporated applicative morpheme le it is spelled out as the agreeing dative les. (50) no les pegues (golpes) a los niños not lepl hit (beats) to the kids (51)

In turn, when the dative clitic shows no agreement, a relevant example in (52), we contend that there are no phi-​features on v for agree to take place, and thus the clitic we find, le, instantiates exclusively the applicative marker we are advocating for, as in (53). (52) no le pegues (golpes) a los niños not leSG hit (beats) to the kids (53)

Leísmo: one le or two?  141

7.  Conclusion This chapter tackled the question whether there should be a single le or two (one accusative and one dative). We supported the idea that there should be two different les based on the facts of “le-​for-​les,” which we extended to leísta dialects and we saw the defective le is possible only with dative le, but crucially not with accusative le. The analysis entertained supports the claim by Fernández-​Ordóñez that some data attributed to leísmo are in fact instances of “fake leísmo.” Differently from these data in which “le-​for-​les” is found and grammatical, true instances of leísmo do not exhibit “le-​for-​les.” If we were to entertain the idea of a single le, it is not clear to us how this contrast could be explained. Assuming, as we have shown, the existence of two different les—​ i.e., dative le and accusative le, this contrast can be naturally explained.

Notes 1 Bleam focuses on the leísta dialect spoken in the north of Spain, which allows the use of le not only with masculine animate referents, but also with feminine or inanimate referents. Thus, the contrast in (12)–(13) is only relevant in this dialect. Bleam’s remark does not hold in dialects where (12) is acceptable. However, we believe that Bleam’s point is still relevant in those dialects that allow both le and lo for [+human] DO since in those dialects there seems to be a preference for le in doubling constructions: (i) Les  vimos a  los  chicos en el parque. CLDAT saw1PL  to  the  boys   at  the  park (ii) ??Los vimos a  los chicos en el parque. CLACC  saw1PL to the boys  at the park “We saw the boys at the park.” 2 To be fair with Bleam, she addresses in her dissertation some of the issues that Ormazabal and Romero raise in ways that we will not discuss here for reasons of space. 3 Similar facts are attested in Catalan, as discussed in Pineda (2018): (i) Li[sg] vaig dir als  clients[pl] que tancaríem. (spontaneous talk, central Catalan, young female, Barcelona) ClDAT go1SG  say  to.the clients   that  would.close1PL “I told the clients that we would close.” 4 See note 5. 5 The parallelism between the obligatory use of the clitic with the accusative and the preference of agreement with the dative is not as strong as in the other cases presented. However, the contrast has been documented by Soler Arechalde (1992) who concludes that when the double is a pronoun the agreeing les tends to be kept.

142  Adolfo Ausín and Francisco J. Fernández-​Rubiera 6 We would like to thank an anonymous reviewer for bringing up this issue. The data presented seem to support the idea that there are two different lexical entries for “pegar”. For a recent discussion of constraints on passives, see Ambridge et  al. (2015) and Sánchez López (2015). For further discussion of how these accusative /​ dative alternations are related to semantic considerations such as affectedness, see Pineda (2020) and Ausín (2021). 7 Her claim is based on the observation that the dative option was the required one originally, before leísmo spread. 8 This goes against Kramer’s main proposal, since she argues that clitics are not agreement morphemes but determiners. See Ausín and Fernández-​Rubiera (2019) for an analysis of “le-​for-​les” under the assumption that clitics are determiners.

References Ambridge, B., Bidgood, A., Pine, J.M., Rowland, C.F., & Freudenthal, D. (2015). Is passive syntax semantically constrained? Evidence from adult grammaticality judgment and comprehension studies. Cognitive Science, 40(6), 1435–​ 1459. doi: 10.1111/​cogs.12277 Ausín, A. (2021). What’s courteous about leísmo? In J. J. Colomina-Almiñana & S. Sessarego (eds.), Language Patterns in Spanish and Beyond: Structure, Context and Development (pp. 147–165). London: Routledge. Ausín, A., & Fernández-​Rubiera, F.J. (2017). Laísmo and “le-​for-​les”:  To agree or not to agree. In S. Perpiñán, D. Heap, I. Moreno-​Villamar & A. Soto-​Corominas (eds.), Romance Languages and Linguistic Theory 11: Selected Papers from the 44th Linguistic Symposium on Romance Languages (LSRL), London, Ontario, 101–​125. Amsterdam: John Benjamins. Ausín, A., & Fernández-​Rubiera, F.J. (2019). Towards a uniform account of accusative and dative clitic doubling. Talk delivered at the 49th Linguistic Symposium on Romance Languages, University of Georgia, Athens, GA, May 1–​4. Bleam, T. (1999). Leísta Spanish and the syntax of clitic doubling (Doctoral dissertation). University of Delaware. Boeckx, C., & Jeong, Y. (2004). The fine structure of intervention in syntax. In C. Kwon & W. Lee (eds.), Issues in Current Linguistic Theory: A Festschrift for Hong Bae Lee, 83–​116. Seoul: Kyungchin. Casares, J. (1918). Crítica efímera. Madrid: Saturnino Calleja S.A. Cuervo, M.C. (2003). Datives at large (Doctoral dissertation). MIT. Cuervo, M.C. (2010). Against ditransitivity. Probus, 22(2), 151–​180. Cuervo, R.J. (1955). Apuntaciones críticas sobre el lenguaje bogotano, con frecuente referencia al de los países de Hispano-​América (9th ed.). Bogotá: Instituto Caro y Cuervo. DeMello, G. (1992). ‘Le’ for ‘les’ in the spoken educated Spanish of eleven cities, Canadian Journal of Linguistics, 37(4), 407–​430. Demonte, V. (1995). Dative alternation in Spanish. Probus, 7(1), 5–​30. Fernández-​Ordóñez, I. (1994). Isoglosas internas del castellano: El sistema referencial del pronombre átono de tercera persona. Revista de Filología Española, 74(1), 71–​125. Fernández-​ Ordóñez, I. (1999). Leísmo, laísmo y loísmo. In I. Bosque & V. Demonte (eds.), Gramática descriptiva de la lengua española (pp. 1317–​ 1398). Madrid: Espasa Calpe.

Leísmo: one le or two?  143 Fernández-​ Soriano, O. (1999). El pronombre personal. Formas y distribuciones. Pronombres átonos y tónicos. In I. Bosque & V. Demonte (eds.), Gramática descriptiva de la lengua española, 1209–​1274. Madrid: Espasa Calpe. Gallego, Á., Roca, F., & Colomina, M.P. (2017). Micro-​parameters of case displacement:  Leísmo across Spanish dialects. Paper presented at International workshop: Dative structures and beyond. Universitat Autònoma de Barcelona, Spain. January 26–​ 27. Retrieved from Huerta Flores, N. (2005). Gramaticalización y concordancia objetiva en el español: Despronominalización del clítico dativo plural. Verba, 32, 165–​190. Kramer, R. (2014). Clitic doubling or object agreement:  the view from Amharic. Natural Language & Linguistic Theory, 32, 593–​634. Martín, F.J. (2012). Deconstructing Catalan objects (Doctoral dissertation). New York University. Ordóñez, F. (2012). Clitics in Spanish. In J.I. Hualde, A. Olarrea & E. O’Rourke (eds.), The Handbook of Hispanic Linguistics, 423–​451. Chichester: Wiley-​Blackwell. Ormazabal, J., & Romero, J. (2007). The object agreement constraint. Natural Language & Linguistic Theory, 25(2), 315–​347. Pineda, A. (2018). Li-​per-​els[i]‌o la despronominalització del clític datiu en català: un fenomen incipient. Quaderns de Filologia:  Estudis Lingüístics, 23, 247–​ 281. doi: 10.7203/​qf.23.13530 Pineda, A. (2020). From dative to accusative. An ongoing syntactic change in Romance. Probus, 32(1), 129–173. doi: https://doi.org/10.1515/probus-2019-0001. RAE. (1885). Gramática de la lengua castellana. Madrid: G. Hernando. RAE, & AALE. (2009). Nueva gramática de la lengua española. Madrid: Espasa Libros. Rini, J. (1988). A new perspective on the origin of le for les. Journal of Hispanic Philology, 12(3), 207–​219. Roca, F. (1992). Object clitics in Spanish and Catalan. Catalan Working Papers in Linguistics, 2, 245–​280. Roca, F. (1996). Morfemas objetivos y determinantes. Verba, 23, 83–​119. Sánchez López, C. (2015) Pasividad e impersonalidad. In J. Gutiérrez-​Rexach (ed.), Enciclopedia de Lingüística Hispánica, 773–​784. London: Routledge. Soler Arechalde, M.Á. (1992). LE/​LES> LE con duplicación de objeto indirecto y sin ella. Estudios de Lingüística Aplicada, 15/​16,  57–​68. Sportiche, D. (1996). Clitic constructions. In J. Rooryck & L. Zaring (eds.), Phrase Structure and the Lexicon, 213–​276. Dordrecht: Kluwer. Sturgis, C. (1927). Uso de le por les. Hispania, 10(4), 251–​254. Suñer, M. (1988). The role of agreement in clitic doubled constructions. Natural Language and Linguistic Theory, 6(3), 391–​434.

Part II

Pragmatic patterns

7  What’s courteous about leísmo? Adolfo Ausín

1.  Introduction The main goal of this study is to provide a formal account of what is frequently called leísmo de cortesía (henceforth “courteous leísmo”) following Pineda (2012a)’s proposals regarding accusative /​dative alternations. Courteous leísmo is a type of leísmo that is found in many dialects, and its defining feature is that le is used with 2nd-​person formal usted in contexts where we normally would expect lo (or la) as illustrated in (1). The use of le in (1a) with usted contrasts with the use of lo with Juan in (1b). (1) a. Y  a usted ¿  quién estará esperándole  cuando  llegue a casa? And  to you,   who will.be waiting.himDAT  when    arrive to home “And you sir, who will be waiting for you when you go home?” b. Cuando  Juan llegue  a casa,  nadie   estará   esperándolo. When  Juan arrives  to home nobody will.be waiting.himACC “When Juan goes home, nobody will be waiting for him.”

This chapter is organized as follows. First, I will briefly discuss leísmo in general. Second, I will review some of the descriptions of courteous leísmo that can be found in the literature, as well as some of the explanations that have been offered. Then I will discuss a proposal made by Pineda (2012a) regarding the accusative /​dative alternation that appears exemplified in (2). (2) a. Le robaron. himDAT robbed “They robbed himDAT” b. Lo robaron. himACC robbed “They robbed himACC” Finally, I  will present a configurational account of courteous leísmo. In a nutshell, I will argue that the two structures that Pineda proposes for (2a) and

148  Adolfo Ausín Table 7.1 3rd-​person clitic paradigm in Spanish: etymological system SINGULAR ACCUSATIVE DATIVE

masculine lo(s) le(s)

PLURAL feminine la(s)

(2b) are approximately the same two structures that underlie the possibility of having courteous leísmo.

2.  Leísmo Table  7.1 summarizes the most widely spread system of 3rd-​person pronominal clitics in Spanish. This system is sometimes labelled “etymological” because it is the one that most closely resembles the Latin usage. Relevant examples appear in (3)–(5). The key feature of this system is that le is used only for datives as in (5). (3) a. A Juan, lo vimos. to Juan himACC saw “Juan, we saw him.” b. A usted,  lo   vimos. to you.fml  himACC saw “Sir, we saw you.fml.” (4) a. A Marta, la   conozco. to Marta  himACC know “Marta, I know her.” b. A usted,  la   conozco. to you.fml  herACC know “Ma’am, I know you.fml.” (5) a. A Luis, le   entregué un regalo. to Luis  himDAT  gave    a   present “To Luis, I gave him a present.” b. A usted,  le   entregué un regalo. to you.fml  himDAT gave    a  present “To you {Sir/​Ma’am}, I gave you.fml a present.” It will be important for the entire chapter to remember that in Spanish 3rd-​ person clitics and 3rd-​person verb agreement are used to refer both to 3rd

What’s courteous about leísmo?  149 person and formal /​deferential 2nd-​person usted. In (3)–(5), the a-​examples are for 3rd-​person referents, and the b-​examples are for formal, deferential 2nd-​person referents. When relevant, I will be using “you.fml” (you formal) to translate usted and related 3rd-​person clitics. As is well known, there is a great deal of dialectal variation within the Spanish pronominal clitic system. One type of variation that is found is leísmo: the use of le for other clitics (Fernández-​Ordóñez, 1999, among many others). What type of clitic is being replaced is going to depend on the type of leísmo under consideration. In the most extended type of leísmo, le replaces lo when it refers to masculine human referent as in (6). This is sometimes called leísmo de persona. Less frequently, le can also be used to replace [-​human] lo as in (7) (leísmo de cosa, “inanimate leísmo”). Even less frequently, le is used for personal feminine referents as in (8) (leísmo femenino, “feminine leísmo”). (6) ¿ Dónde viste a Juan?  Le     vimos en el parque. where  saw  to Juan himACC  saw   at the park “Where did you see Juan? We saw him at the park.” (7) ¿ Dónde  compraste  el libro?  Le    compré  en la FNAC. where   bought   the book  himACC  bought  at the FNAC “Where did you buy the book? I bought it at the FNAC.” (8) ¿ Dónde  viste  a Marta?  Le    vimos  en la universidad. where    saw   to Marta  herACC  saw   at the university “Where did you see Marta? We saw her at the university.” There are many other possible variations within the pronominal clitic system such as laísmo (the use of la for feminine dative), loísmo (the use of lo for masculine dative), as well as differentiation between mass and count nouns.1 Instead, I would like to focus on the so-​called leísmo de cortesía. In this type of leísmo, accusative lo (and less frequently la) is being replaced by le only in some specific contexts, mostly with usted.

3.  Leísmo de cortesía In this section, I review in detail some of the descriptions of courteous leísmo. Lorenzo Ramos (1981) coined the term “leísmo de cortesía” when discussing potential cases of leísmo in Canary Islands Spanish, a non-​leísta dialect, such as (9). He claims that a speaker would use an example like (9a) to address a superior. If we were to describe the same scenario with 3rd-​person singular, (9b) would be used.2

150  Adolfo Ausín (9) a. Le    llaman  por  teléfono  [a  usted]. HimDAT  call     by   phone   to you.fml “They are calling you.fml on the phone.” b. Lo   llamaron por teléfono [a él]. HimACC called     by  phone   to  him “They are calling him on the phone.” He also considers examples like the following: (10) a. No le    oyó  (a usted). Not himDAT heard to you.fml “S/​he did not hear you.fml” b. No  lo    oyó  (a él). Not himACC  heard  to him “S/​he did not hear him.” (11) a. Si le   molesta (a usted),  la  apago. if himDAT bothers to you.fml itACC  turn off “If it is bothering you.fml, I can turn it off.” b. No lo   molestes (a él).3 Not   himACC  bother  to him “Don’t bother him.” Lorenzo Ramos notes that in all these cases le is referring to usted and for this reason he decided to identify this type of uses as “leísmo de cortesía.” The facts that Lorenzo Ramos initially discussed were impressionistic. However, a few years later, he presented the results of a questionnaire that was distributed among high school students. Some of the test sentences included in the questionnaire appear in (12). (12) a. (A Juan)  el gobernador  _​_​_​_​_​ recibió  en el  despacho. to Juan  the governor        received in   the office “(Juan) the governor received him in the office.” b. Espere  usted   un  momento   que enseguida   _​_​_​_​_​ reciben. Wait  you.fml a  moment   that immediately     receive “Wait a moment, somebody will receive you immediately.”

According to Lorenzo Ramos (1984), the results from the questionnaire showed that courteous leísmo was used 50 percent of the time. Even though Lorenzo Ramos named the phenomenon, he was not the first one to describe it. García (1975, p.  338–​339), among others, had described how some speakers use le instead of lo with usted. After noticing that a sentence like (13) is ambiguous between a 3rd-​person singular interpretation and

What’s courteous about leísmo?  151 a formal 2nd-​person singular interpretation, she goes on to add that many speakers use le instead of lo to eliminate the ambiguity as in (14). (13) Ayer   lo   vi  en el parque. Yesterday himACC  saw  at  the park “Yesterday I saw him/​you(formal) at the park.” (14) a. Ayer lo vi en el parque → a él “Yesterday, I saw him at the park” b. Ayer le vi en el parque → a usted. “Yesterday, I saw you.fml at the park” The choice of le for usted is due, according to García, “to the coherence between the meaning of le, which implies a ‘higher’ non-​agent than lo, and the respect shown to the addressee” (García, 1975: 338). This explanation is based on García’s activeness scale in (15), where dative le is said to be more active than accusative lo or la. The subject is the focused constituent which typically is the most active constituent. (15) García’s activeness scale Focus (typically, most active): subject Less active: le Least active: lo /​la García (1975, ­chapter 7) uses this activeness scale to account for the choice of le vs. lo in many different contexts. A paradigmatic case would be the choice of clitic in ditransitive constructions, what she calls “three-​participant situations.” In García’s theory, the choice of clitic in these constructions would not be based on the traditional categories of direct and indirect object, or dative and accusative, but on the activeness scale in (15). Thus, the choice of le for the traditional indirect object in a ditransitive construction such as (16) is due to the fact that it refers to a participant that is more active than “the book.” (16) Le compré el libro. himDAT bought the book “I bought the book for/​from him.” Going back to the alternation of le /​lo in (14), the use of le with usted in (14b) would follow from the desire to assign a more active role to the addressee. Carmona Yanes (2010) presents the results of a survey designed to evaluate the factors that determine the use of leísmo among young people in Seville. The results that she obtained for the overall relevance of the distinction between 2nd-​person usted reference and 3rd-​person reference appear in (17).

152  Adolfo Ausín As can be seen, when the reference was masculine usted, le was used 52 percent of the time, but only 37 percent when the reference was masculine 3rd person. When the reference was feminine usted, le was used 35  percent of the time, but only 23 percent of the time when it was feminine 3rd person. Therefore, usted triggers a higher frequency of le both in the masculine and in the feminine. (17) % of le vs. lo or other possibilities a. Masculine usted: 52% b. Masculine 3rd-​person sg.: 37% c. Feminine usted: 35% d. Feminine 3rd-​person sg.: 23% To conclude this literature review section, let me briefly introduce Dumitrescu and Branza (2012).4 They were also trying to determine how widespread courteous leísmo is and created a questionnaire with sentences like the following, (18) a. Usted ‘La enfermera se me acercó y me preguntó: -​“¿Y a usted, señor López, quién…examinó la vez pasada?” ‘The nurse approached me and asked me: “Mr. López, who examined you last week?” ’ b. Él Nadie se acordaba exactamente del señor López y por eso ya no se sabía si la última vez…había examinado el doctor Soto o el doctor Fernández. “Nobody remembered Mr. López exactly, and nobody remembered if Dr. Soto or Dr. Fernández had examined him last time.” What they found was that in all but one of the cities that they surveyed, the percentages of le versus lo were higher with usted than with 3rd person, with different degrees of statistical significance. A summary of their results appears in Table 7.2. The le /​lo difference was only statistically significant for the last four cities in Table 7.2. A comparison of the different studies that we have just reviewed reveals considerable variation among them. This is probably due to the different methodologies and also to the variation intrinsic to the phenomenon under consideration. However, I believe that we can conclude that, at least for some speakers, there is a preference to use le over lo with usted. Now the question that we need to ask is why. The literature discusses several possible factors. It has sometimes been suggested that this type of leísmo with usted should be related to the leísmo in Standard European Spanish. For instance, Cantero Sandoval (1979) claims

What’s courteous about leísmo?  153 Table 7.2 Usted vs. él. Distribution of le and lo (Dumitrescu & Branza, 2012) Cities

Buenos Aires Bogotá San José Santiago La Habana Quito Vigo Sevilla

Usted

Él

Abs. f. le

Abs. f. lo

Rel. f. le

Abs. f. le

Abs. f. lo

Rel. f. le

1 8 11 5 5 34 26 28

55 59 39 47 39 34 21 6

2% 12% 22% 10% 11% 50% 55% 82%

0 2 2 2 5 11 20 16

54 65 36 31 30 58 25 10

0% 3% 5% 6% 14% 16% 44% 62%

that the reason le has a courteous, formal flavor is because it relates to the allegedly prestigious educated European Spanish. It has also been suggested that le is more respectful than lo because le, as opposed to lo, is rarely used to refer to things (DeMello, 2002). As we already saw earlier (see the discussion of the examples in (14)), García (1975) describes courteous leísmo as a strategy of disambiguation. Since lo could refer in principle to 3rd person or to formal 2nd person, the use of le for formal 2nd person eliminates the potential ambiguity. Fernández-​Ordóñez (1999) defends similar views. Both García and Fernández-​Ordóñez based their views on the observation that in a system without courteous leísmo a sentence like (13) (repeated below) is ambiguous between a 3rd-​person him interpretation and a formal 2nd-​person you /​usted interpretation. In a system with courteous leísmo as in (14) (also repeated below), the ambiguity disappears since 3rd person is expressed with accusative lo as in (14a) and formal 2nd person is expressed with dative le as in (14b). (13) Ayer   lo   vi   en el  parque. Yesterday himACC  saw at  the park “Yesterday I saw him/​you(formal) at the park.” (14) a. Ayer lo vi en el parque → a él “Yesterday, I saw him at the park.” b. Ayer le vi en el parque → a usted. “Yesterday, I saw you.fml at the park.” Both Fernández-​Ordóñez (1999) and García (1975), as well as others, argue that the choice of le for usted is a way of assigning the addressee a more prominent and active role.5 In the rest of this chapter, I provide a configurational account of Fernández-​Ordóñez and García’s views. In their views, courteous leísmo is the result of the dative clitic having a more active interpretation.

154  Adolfo Ausín I propose that the source of courteous leísmo lies in the syntactic configuration that underlies the choice between accusative and dative clitics.6

4.  Accusative /​dative alternations The proposed configurational account of courteous leísmo is based on Pineda (2012a)’s account of some dative /​accusative alternations that we find in Spanish and Catalan in verbs like the following:7 i verbs of transfer of communication (telefonear “phone,” escribir “write,” contestar “answer”); ii verbs of transfer of possession (pagar “pay,” robar “steal/​rob”); iii verbs of contact (pegar “hit,” disparar “shoot,” seguir “follow”); iv verbs of saying, thinking and social interaction (servir “serve,” entender “understand,” mentir “lie,” suplicar “beg,” silbar “whistle,” aplaudir “applaud”). Thus, a verb like telefonear can alternate between the accusative and the dative. (19) a. La telefoneó. herACC phoned b. Le telefoneó. herDAT phoned “S/​he phoned her.” At face value, this alternation could be interpreted as an instance of leísmo. However, Pineda points out several reasons that this cannot be the case. First, the same alternations are found in Catalan, where there is no leísmo. (20) a. La Maria la the Maria herACC b. La Maria li the Maria herDAT “Maria phones her.”

telefona. phones telefona. phones

Pineda also notes that in some cases the alternation between accusative and dative carries a difference in meaning in some dialects. The general tendency is that when there is an accusative /​dative alternation, the accusative alternative yields a more affected interpretation. One of the verbs she discusses is robar. Robar is ambiguous and it can mean either “steal” or “kidnap.” Under the “kidnap” interpretation it always selects accusative. However, under the “steal” interpretation it normally selects dative. Pineda claims that, at least in some dialects, it can alternate between accusative and dative, and that in the version with accusative as in (21) the object is more affected.

What’s courteous about leísmo?  155 (21) a. Lo    robaron y  le   devolvieron todo:    increíble. himACC robbed and himDAT gave.back  everything  incredible “They robbed himACC and gave everything back to him: incredible.” b. Se quedó con   dinero  robado, pero después lo      robaron a él. se  took  with  money  stolen  but  then    himACC robbed to him “He took stolen money, but then they robbed himACC.”

To corroborate her views that the accusative version implies a more affected interpretation, she quotes Fernández-​Ordóñez (1999), who claims that the version with the accusative “parece implicar que el acto de sustracción tuvo lugar en presencia del individuo afectado y con notable perjuicio del mismo” (seems to imply that the event took place in the presence of the person affected and caused significant harm to him/​her). As an additional example where there is a change of meaning between the accusative and the dative version, Pineda also considers pagar “to pay”: (22) a. A Juan le pagué el alquiler. “Juan, I paid himDAT the rent.” b. A Juan le pagué con creces. “Juan, I paid himDAT by far.” c. A Juan lo pagué con creces. “Juan, I paid himACC by far.” She claims that the accusative in (22c) implies a more affected interpretation. She supports her views with comments by García about similar examples. García (1975:  318) claims that a sentence like los pagó can convey “a final (contemptuous) paying off of employees who will have no further claim to the payer” whereas les pagó “is appropriate in reference to creditors, regular employees, or anybody who is treated properly.” (23) a. Los pagó. themACC paid b. Les pagó. themDAT paid “S/​he paid them.” A more dramatic example of the different interpretation associated with the accusative /​dative alternation can be found with disparar “to shoot.” According to Pineda (2014), which attributes this observation to Juan Romero and Javier Ormazabal (personal communication), disparar “to shoot” has different meanings with accusative lo and with dative le. This is shown in the following examples from Pineda (2014: 136):

156  Adolfo Ausín (24) a. Le     dispararon pero no  le    dieron. himDAT  shot   but not himDAT  got “They shot (at) him but they didn’t get him.” b. Lo   han   disparado, está muerto. himACC  have  shot    is   dead “They shot him, he’s dead.” When disparar appears with accusative, the more natural interpretation is that the person was actually shot. Whereas with the dative, it is possible that no bullets hit the target. Again, Pineda (2014) relates these different interpretations to the notion of affectedness.8 With accusative there is more affectedness. The contrast that we find in the accusative /​dative version of disparar is related to the contrast in English between to shoot and to shoot at as implied in the translations. We will return to this later. Now, let’s turn to Pineda’s (2012a) proposal to account for the accusative /​ dative alternations. The departing point is Torrego’s proposal regarding transitive verbs like contratar “to hire.” Torrego (2010) claims that a verb like contratar should be analyzed as if it were hacer o dar un contrato a alguien “to give a contract to somebody.” In other words, both contratar and hacer un contrato a alguien would share a common syntactic structure, namely the one in (26). Here we have a low applicative phrase that relates the noun contrato with the DP a alguien.9 If contrato “contract” incorporates into the light verb, we obtain contratar “to hire”; otherwise we get dar un contrato a alguien “to give a contract to somebody.” (25)  a. Él contrató a  alguien. he hired   to someone “He hired someone.” b. Él dio un contrato a  alguien. he gave a contract to someone “He gave someone a contract.” (26)

Elaborating on Torrego, Pineda proposes two different structures to account for the accusative /​dative alternation. For the dative variant in (27),

What’s courteous about leísmo?  157 she follows Torrego in assuming that the internal argument appears in the specifier position of an applicative phrase, where it receives dative case. In the accusative variant in (28), the goal argument cannot receive dative case, and it moves up to get assigned accusative case.10 This movement to the accusative marked position is what triggers the more affected interpretation. (27)

(28)

If we apply this analysis to the disparar “to shoot” examples that we were discussing earlier, we would have something like (27’) and (28’). The extra movement of lo is what triggers the more affected interpretation. (27’)

158  Adolfo Ausín (28’)

It is well known that there is a close relationship between accusative marked direct objects and affectedness. We already saw some clear examples of it when we looked at the English translations of (24). Furthermore, if we consider the sentences in (29), the sandwich is more affected, in the sense of being completely eaten, in (29a) where it is the direct object, compared to (29b), where it is just the object of the proposition. (29) a. The forlorn diner ate his sandwich. b. The forlorn diner ate at his sandwich.

(Beavers, 2006)

Similarly, in (30) both in the Spanish examples as well as in the English translation, el maletero “the trunk” receives a more affected interpretation, in the sense of being full in the a-​example where it is the accusative marked direct object, than in the b-​example, where it is the object of the preposition. (See Pinker, 2007, and references therein for discussion of this type of alternation.) (30) a. Cargué  el maletero   con las   maletas. loaded  the trunk   with the  suitcases “I loaded the trunk with the suitcases.” b. Cargué las maletas en el maletero. loaded  the suitcases  in   the trunk “I loaded the suitcases in the trunk.” As a summary of Pineda’s account, remember that she identifies a number of verbs that show the accusative /​dative alternation. To account for this alternation, she proposes two different structures: (27) and (28). The key difference between these two structures is that in (28) there is an additional movement of the goal argument to an accusative marked position, and that this movement is associated with the more affected interpretation.

5.  A formal account of courteous leísmo If we apply Pineda’s analysis to the examples that we have been considering, the structure of the courteous leísmo example in (31a) would be (31b), and the structure of (32a) would be (32b).

What’s courteous about leísmo?  159 (31) a. Y  a usted   ¿ qué doctor le   examinó? and to you.fml what doctor himDAT  examined “Sir, who examined you.fml?” b.

(32) a. A Juan, lo examinó el Dr. Pérez. to Juan himACC examined the Dr. Pérez “Juan, Dr. Pérez examined him.” b.

The crucial difference between (31b) and (32b) is that the applicative head assigns dative case to its specifier in (31b) but not in (32b) (see fn. 10), which causes the goal argument to move up to the accusative case marked position. If this is on the right track, I  believe there is a straightforward account of courteous leísmo, especially when we are dealing with Pineda-​type verbs that show dative /​accusative alternation. I would like to propose that speakers have both structures at their disposal, and that they can choose the dative option, instead of the accusative one, to prevent the addressee DP from being assigned a more affected role. It is this avoidance of the more affected position what yields the effect of the more courteous flavor of the structure. In other words, imagine that a speaker wants to say the Spanish equivalent of (33). What I am proposing is that this speaker has two different possibilities

160  Adolfo Ausín to convey (33), namely (34a) and (34b), and that each of these sentences corresponds roughly to the structures in (31b) and (32b), respectively. (33) Sir, what doctor examined you? (34) a. Señor, Sir, b. Señor, Sir,

¿ qué médico le what doctor himDAT ¿ qué médico lo what doctor himACC

examinó? examined examinó? examined

In (34b) the addressee is being referred to with accusative lo which conveys a more affected interpretation. In (34a), on the other hand, the addressee is being referred to with dative le, which conveys a less affected interpretation, yielding the deferential flavor typically associated to courteous leísmo. The proposed account of courteous leísmo relies on Pineda’s account of dative /​ accusative alternations. In her account, both variants share a common argument structure, and they differ in the fact that in the alternative with the accusative clitic there is a further movement of the argument to the accusative marked position. There is a simpler, more naïve version of Pineda’s (2012a) approach, which is to assume that whenever you have accusative you have a simpler transitive construction as in (35), and only when there is dative, is there an underlying applicative structure, as in (36).11 (35) [vP Agent [v’ v [VP V ObjectAcc]]]

(36) [vP Agent [v’ v [VP V [ApplP ObjectDat [Appl’ Appl N]]]

What’s courteous about leísmo?  161 This is a bit problematic from the point of view of argument structure since different syntactic structures are being postulated for the same argument structure. But it seems more transparent from the point of view of case assignment, and for the relationship between case assignment and affectedness, since the accusative marked argument appears in the position typically associated with such arguments. At any rate, independently of the exact formalism, the idea behind courteous leísmo would be that speakers have two constructions at their disposal and that they can choose the dative variant as a deferential treatment towards the addressee to prevent the addressee DP from appearing in a more affected position. As Pineda (2020) states, “given the existence of a case alternation, the speaker decides to take advantage of it and use the two encoding options to express a semantic contrast.” To conclude my account of courteous leísmo, I would like to discuss two further issues. First, whether courteous leísmo should be considered a true instance of leísmo, and second, an account of the fact that courteous leísmo is more frequent with masculine referents than with feminine ones. Fernández-​Ordóñez (1999) established the distinction between true leísmo (leísmo real) vs. fake leísmo (leísmo aparente). Real leísmo would be the use of le instead of the accusative lo as in the examples in (6)–(8) discussed earlier in this chapter. Fake leísmo is not leísmo at all because the relevant argument is not an accusative direct object but a dative indirect object. For Fernández-​ Ordóñez, an example of fake leísmo would be examples like in (37) since in her analysis the dative le is not replacing an accusative la but is actually selected by ayudar “to help.”12 (37) María está muy atareada, pero Juan le   está ayudando. María  is   very busy,    but  Juan  herDAT  is helping “María is very busy, but Juan is helping herDAT” Fernández-​Ordóñez seems to include courteous leísmo under the label of fake leísmo, without much justification. The proposal made in this study would provide such a justification, since the dative clitics that appear in examples of courteous leísmo are actual dative arguments. Finally, I would like to address the issue regarding courteous leísmo and gender. It has been noted that courteous leísmo is less frequent in the feminine. The survey in Carmona Yanes (2010) summarized in (17) clearly reflects this.13 I  believe that the fact that feminine courteous leísmo is less common than masculine courteous leísmo is expected under the current proposal. Under the current account, speakers can choose between a dative and an accusative alternative. The dative option is more respectful because it prevents the DP that refers to the addressee from appearing in the more affected position. On the other hand, opting for the dative option means losing gender specification. This is not a problem with masculine referents, since masculine is the default gender in Spanish. However, it might be a problem with feminine referents because a marked gender feature that could be present, is not. In other words,

162  Adolfo Ausín I am claiming that speakers are more likely to make the change in (38a) than in (38b) because the change in (38a) means the loss of an unmarked feature whereas the change in (38b) means the loss of a marked feature. (38) a. Lo     ayudo. →  Le    ayudo. himACC help   himDAT help “I help him.” b. La   ayudo.  →  Le     ayudo. herACC help        himDAT   help “I help her.” Admittedly, the proposal I am making to account for the lesser frequency of feminine courteous leísmo is rather tentative and lacks independent support. I am hoping to find support for it in future research.

6.  Conclusion To conclude, in this chapter I have reviewed courteous leísmo and proposed a formal account based on Pineda’s account of accusative /​dative alternation. Pineda argues that there are two different structures for each option, and that the dative option yields a less affected interpretation of the argument. I propose that speakers can choose between the structures and in the courteous leísmo cases speakers are choosing the dative variant with le to convey a less affected interpretation and to show a deferential treatment towards the addressee.

Notes 1 For further discussion and references, see the work by Inés Fernández-​Ordóñez, in particular, Fernández-​Ordóñez (1994, 1999). 2 Throughout the chapter I will be glossing le as dative him, and lo as accusative him. However, the reader should keep in mind that identifying le as dative is part of the analysis and that the scholar who provided those examples might not share that analysis. 3 As pointed out by a reviewer, the examples in (11) are not minimal pairs since in (11a) the subject of molestar is an inanimate subject (the implicit light), whereas in (11b) the subject is an animate agent. Mostly likely, the two examples in (11) have different argument structures. See Mendívil Giró (2005) for detailed discussion on Spanish molestar and Pineda and Royo (2017) for its Catalan counterpart. Closer minimal pairs with molestar would have been the ones below, which I believe show the same point that Lorenzo Ramos was trying to make: (i) A Juan, no quiero molestarlo. To Juan not want bother.himACC ‘Juan, I don’t want to bother him.’

What’s courteous about leísmo?  163 no quiero molestarle. (ii) A usted, To you.fml not want bother.himDAT ‘Sir, I don’t want to bother you.’ 4 For further discussion of courteous leísmo see García (1990), Flores Cervantes (2002), Mejía-​ Gómez (2008), Real Academia Española and Asociación de Academias de la Lengua Española (2005, 2009). 5 See the discussion around (15) for more details about García’s proposal. 6 Although I  argue for a configurational, syntactic account of courteous leísmo, I  do not deny the possibility that courteous leísmo might also be influenced by other factors discussed in the literature. 7 Pineda has recently developed further her analysis in Pineda (2020). I  will be using her earlier account making reference to her more recent proposals when appropriate. 8 For more discussion and references on the notions of affectedness and transitivity in Spanish, Catalan and Basque see Pineda (2012a, 2012b). 9 See Cuervo (2003, 2010) and references therein for further discussion of applicative constructions. 10 Pineda (2012a) proposes that the ability of the Applicative head to assign Dative case is subject to parametric variation, and that languages like Catalan and Spanish that have an accusative /​dative alternation are exploiting this variation. For more details see Pineda (2012a: §4.5) among other places. See Pineda (2020) for further refinements. 11 This would be stage III proposed in Pineda (2020). Pineda (2020) proposes four stages to characterize the syntactic change that accusative /​dative alternating verbs undergo. In stage I, these verbs only take dative. In stage II, there are two different applicative heads that result in accusative or dative. This stage is the one assumed in Pineda (2012a) and assumed in the text. In stage III, the dative option would correspond to an applicative head, whereas the accusative option would correspond to a plain transitive structure. Finally, in stage IV the transitive option would generalize resulting in the disappearance of the dative /​accusative alternation in favor of the accusative option. See Pineda’s work for more details. 12 See Fernández Ordóñez (1999) for more details, and Sáez (2009) for a different view on ayudar type verbs. 13 A reviewer suggests that the reason for this is that we treat males more formally than females. A similar view was expressed by García (1975: 330), who, in the context of the choice between le, la, lo with ayudar “to help” type verbs, points out that “within Spanish culture (if not universally), males are socially more highly valued than females.” Verifying these proposals falls beyond the scope of this study. At any rate, if true, they would be compatible with the proposal sketched in the text.

References Beavers, J.T. (2006). Argument/​oblique alternations and the structure of lexical meaning (Doctoral dissertation). Stanford University. Cantero Sandoval, G. (1979). Casos de leísmo en México. Anuario de Letras, 17, 305–​308.

164  Adolfo Ausín Carmona Yanes, E. (2010). Selección del caso de los pronombres átonos de tercera persona en el habla juvenil de Sevilla. Interlingüística, 20. Retrieved from Cuervo, M.C. (2003). Datives at large (Doctoral dissertation). MIT. Cuervo, M.C. (2010). Against ditransitivity. Probus, 22(2), 151–​180. DeMello, G. (2002). Leísmo in contemporary Spanish American educated speech. Linguistics 40(2): 261–​283. Dumitrescu, D., & Branza, M.D. (2012). Sobre el llamado “leísmo de cortesía” en Hispanoamérica. In J. Escamilla Morales & G. Henry Vega (eds.), Miradas multidisciplinares a los fenómenos de cortesía y descortesía en el mundo hispánico, 669–​692. Barranquilla: Edice. Fernández-​Ordóñez, I. (1994). Isoglosas internas del castellano: El sistema referencial del pronombre átono de tercera persona. Revista de Filología Española, 74(1), 71–​125. Fernández-​Ordóñez, I. (1999). Leísmo, laísmo y loísmo. In I. Bosque & V. Demonte (eds.), Gramática descriptiva de la lengua española, 1317–​1398). Madrid: Espasa Calpe. Flores Cervantes, M. (2002). Leísmo, Laísmo y Loísmo. Sus Orígenes y Evolución. México, DF: Universidad Nacional Autónoma de México. García, E.C. (1975). The Role of Theory in Linguistic Analysis: The Spanish Pronoun System. North-​Holland Linguistic Series; 19. Amsterdam: North-​Holland. García, E.C. (1990). Bilingüismo e interferencia sintáctica. Lexis, 14(2), 151–​195. Lorenzo Ramos, A. (1981). Algunos datos sobre el leísmo en el español de Canarias. In M. Alvar (ed.), I Simposio Internacional de Lengua Española [1978], 175–​180. Las Palmas: Ed. del Excmo. Cabildo Insular de Gran Canaria. Lorenzo Ramos, A. (1984). Observaciones sobre el uso de los pronombres en el español de Canarias. In M. Alvar (ed.), II Simposio Internacional de Lengua Española [1981], 253–​263. Las Palmas: Ediciones del Excmo. Cabildo Insular de Gran Canaria. Mejía-​ Gómez, M. (2008). Ándale, Apágale. ¡Órale!:  La (socio) pragmática de da construcción le en el español mexicano (Doctoral dissertation). Ohio State University. Mendívil Giró, J.L. (2005). El comportamiento variable de molestar: “A Luisa le molesta que la molesten”. In G. Wotjak & J. Cuartero (eds.), Entre semántica léxica, teoría del léxico y sintaxis, 261–272. Frankfurt: Peter Lang. Pineda, A. (2012a). Double object constructions and dative /​accusative alternations in Spanish and Catalan: A unified account. Borealis – An International Journal of Hispanic Linguistics, 2(1), 57–​115. Pineda, A. (2012b). Transitividad y afectación en el entorno lingüístico romance y eusquérico. In X. Viejo (ed.), Estudios sobre variación sintáctica peninsular, 31–​73. Oviedo: Trabe. Pineda, A. (2014). What lies behind dative/​accusative alternations in Romance. In K. Lahousse & S. Marzo (eds.), Romance Languages and Linguistic Theory, Vol. 6, 123–​140. Amsterdam: John Benjamins. Pineda, A. (2020). From dative to accusative. An ongoing syntactic change in Romance. Probus 31(1), 129–173. doi: https://doi.org/10.1515/probus-2019-0001. Pineda, A., & Royo, C. (2017). Differential indirect object marking in Romance (and how to get rid of it). Revue Roumaine de Linguistique, 62(4), 445–​462.

What’s courteous about leísmo?  165 Pinker, S. (2007). The Stuff of Thought: Language as a Window into Human Nature. New York: Penguin. Real Academia Española, & Asociación de Academias de la Lengua Española. (2005). Diccionario panhispánico de dudas. Retrieved from Real Academia Española, & Asociación de Academias de la Lengua Española. (2009). Nueva Gramática de La Lengua Española. Madrid: Espasa Libros. Sáez, L. (2009). Applicative phrases hosting accusative clitics. In R.P. Leow, H. Campos & D. Lardiere (eds.), Little Words:  Their History, Phonology, Syntax, Semantics, Pragmatics, and Acquisition, 61–​ 73. Washington, DC:  Georgetown University Press. Torrego, E. (2010). Variability in the case patterns of causative formation in Romance and its implications. Linguistic Inquiry, 41(3), 445–​470.

8  Verbless exclamatives in Spanish beyond the syntax-​semantics interface A pragmatic account Melvin González-​Rivera

1.  Introduction Non-​verbal or verbless utterances posit a great deal of challenge to any linguistic theory. Despite their frequency and productiveness among many languages (e.g., Arabic, Chinese, Finnish, Guaraní, Haitian Creole, Hdi, Hebrew, Hungarian, Irish, Korean, Mauritian Creole, Mina, Northern Kurdish, Romandalusí, Russian, Samoan, Turkish, Yucatec Maya, etc.), these clauses have so far received relatively little attention from most theoretical frameworks (Abdel-​Ghafer, 2003; Al-​Horais, 2006; Armstrong, 2017; Attia, 2008; Avgustinova and Uszkoreit, 2003; Bakir, 1980; Benmamoun, 2000, 2008; Bentley, 2017; Berman, 1978; Butts, 2006; Carnie, 1995, 1997; Doron, 1983; Fassi Fehri, 1982, 1993; González-​Rivera, 2011; Hazout, 2010; Markman, 2008; Mouchaweh, 1986; Moutaouakil, 1987; Nordlinger and Sadler, 2006; Rapoport, 1987; Revell, 1989; Rosén, 1996; Shin, 2018; Wilson, 2018). The study of such clauses in general raises many interesting questions, since they appear to involve main-​clause structure without overt verbs. Some of the questions that arise when dealing with verbless constructions are the following: (i) are these clauses a projection of T(ense), or some sort of another functional category, (ii) do verbless clauses have an overt or null verbal head, (iii) are verbless clauses small clauses, and (iv) can verbless clauses be interpreted as propositions or statements that are either true or false. This chapter focuses on a special type of verbless construction in Spanish, namely, Predicative Noun Phrase (henceforth Spanish PredNP). Spanish PredNPs are non-​verbal or verbless constructions that exhibit clausal properties: each of the sentences in (1) may be interpreted as containing a semantically empty verb; in other words, these sentences are characterized by the absence of an overt verbal copula (cf. Alonso-​Cortés, 1999; González-​Rivera, 2011; Guillemin-​ Flescher, 2005; Gutiérrez-​Rexach and González-​Rivera, 2014; Laurens, 2008; Lefeuvre, 1999; Munaro, 2006, 2016; Paul, 2008; Paul and Stainton, 2006; Hernanz and Suñer, 1999; Tanguy, 2013; Vinet, 1991). (1) a. [AP muy inteligente] el profesor este very intelligent this professor “This professor I S very intelligent.”

Verbless exclamatives: a pragmatic account  167 b. [NP un verdadero idiota] aquel camarero a real idiot that waiter (over there) “The waiter over there I S an idiot.” c. [PP con toda la cara] esa estafa with a lot of nerve that swindle “That swindle I S (done) with a lot of nerve.” Spanish PredNPs consist of two expressions: an initial XP predicate and a XP—​that is a DP interpreted as the subject of that expression. The predicate or head phrase of the expression can either be an AP (1a), a NP (1b), or a PP (1c) (Laurens, 2008). There are other similar constructions in Spanish but here I will focus my attention on structures like the ones instantiated in (1), especially examples (1a) and (1b). There are other verbless utterances that I will not consider in detail in the chapter, such as the ones illustrated in (2). (2) a. Rápidamente el tiro a tercera “Very quickly the throw to third base”” b. A qué hora el bus “At what time the bus?” c. ¡ Qué bella la tía esta! “What a beauty, this girl!”

(Adverbial Phrase) (wh-​question) (wh-​ exclamative)

Following an insight by Paul (2006), I will argue that these clauses involve predicate inversion à la den Dikken (2006) (see also Vinet, 1991 for French). Thus, the internal structure of Spanish PredNP comprises an asymmetric small clause with a RELATOR head. Contrary to Paul (2006), however, I will show that the clause head is not a missing copula, but rather the feature T(ense). Finally, it will be argued that these clauses should be considered a subtype of exclamatives in keeping with Vinet (1991), and contra Alonso-​Cortés (1999). The remainder of this chapter is organized as follows: in sections 2 and 3 I discuss the grammar of non-​verbal or verbless constructions and the grammar of Spanish PredNPs, respectively. In section 4 I elaborate a syntactic analysis for these clauses based on predicate inversion, as previously stated. Finally, in Section 5 I present evidence in favor of considering PredNPs as pertaining to the set of Spanish exclamatives (contra e.g. Alonso Cortés, 1999; cf. Paul and Stainton, 2006).

2.  The grammar of non-​verbal or verbless constructions Within the framework of the minimalist program (Chomsky, 1995, 2000, 2001), the locus of study of verbless clauses has been whether these clauses contain a functional projection like TP or CP. From a minimalist point of view, these functional projections are held responsible for agreement, Case, the distribution of expletives, and some cases of displacement such as NP-​and

168  Melvin González-Rivera wh-​movement (Benmamoun, 2000, 2008). Verbless clauses have raised another question, namely whether these clauses contain a (silent) copula. Benmamoun (2000, 2008), based on data from Arabic, argues against this second view, while suggesting that verbless clauses do indeed have a functional projection. A similar argument against abstract auxiliaries, which parallel the one against a silent copula, was previously raised by Loporcaro (1998) based on data from acquisitional varieties of Romance languages. These proposals keep with den Dikken’s hypothesis that between the subject and the predicate lies a functional projection or RELATOR . Thus, according to Benmamoun, verbless clauses have the following syntactic representation (cf. Bakir, 1980; Fassi Fehri, 1982, 1993):

Syntactic representation 1.  A functional projection account of verbless constructions

Benmamoun’s analysis is motivated by the following:  in Arabic verbless clauses can contain a temporal adverb, which indicates that they must be anchored by tense (Eisele, 1988), as shown in (3). (3) Omar f-​d-​dar daba Omar in-​the-​house now “Omar is in the house now.” Furthermore, verbless clauses embedded under a matrix sentence need not have the same temporal reference as the verb in the matrix sentence—​that is, in (4) the matrix sentence has a past tense interpretation, while the embedded or verbless clause has a present tense interpretation. This is strong evidence against treating verbless clauses as small clauses without a functional projection since tenseless small clauses depend on the temporal reference of the matrix sentence. (4) qal belli Omar f-​d-​dar say.Past.3ms that Omar in-​the-​house “He said that Omar is in the house.” Notice that the verbless clause in (4)  is dominated by the complementizer “belli,” which selects tensed clauses (Benmamoun, 2000). The general assumption is that the presence of C signals the presence of T. Another argument for positing a functional projection in verbless clauses has to do with Case assignment: the subject of verbless clauses in Arabic bears the nominative

Verbless exclamatives: a pragmatic account  169 Case—​that is, the subject is assigned nominative Case, and the only head that can be assigned or checked nominative Case is T, as shown in (5). (5) t-​taalib-​u fii l-​maktabati the-​student-​Nom in the library “The student is in the library.” Finally, the presence of NPIs and sentential negation suggests that there is functional projection headed by T (Benmamoun, 2000, 2008). Thus, this syntactic evidence demonstrates that, at least in Arabic, verbless sentences contain a functional projection, and this functional projection is headed by T. On the other hand, some have assumed a small clause analysis for verbless clauses (Mouchaweh, 1986, Rapoport, 1987)—​that is, these authors have suggested that verbless clauses are small clauses and have posited that there is no functional projection above the lexical projection. According to this approach, the subject and the predicate of verbless clauses are contained within a small clause, and this small clause can be a projection of AP, NP or PP.

Syntactic representation 2. Small clause account of verbless constructions

As per the previous discussion and facts from Arabic this account is not tenable. Finally, others suggest that verbless clauses contain a silent or null copula (Bakir, 1980; Fassi Fehri, 1982, 1993). Thus, we are left with the following syntactic representation in which there is a null or silent copula between TP and the main predicate—​that is, there is a lexical category VP between TP and the main predicate projection.

Syntactic representation 3.  Null copula account of verbless constructions

Examples from Arabic prove this to be false since, for instance, when the copula is overt in this language it assigns accusative Case to the predicate (6a);

170  Melvin González-Rivera however, when there is no copula the predicate carries nominative Case (6b) (Benmamoun, 2008). (6) a. kana l-​waladu be.Past.3ms the-​boy “The boy was sick.” b. l-​waladu mariid-​un the-​boy sick-​Nom “The boy is sick.”

mariid-​an sick-​Acc

Thus, it is not clear why the overt copula assigns a different Case (Déchaine, 1993). Another problem for the null copula analysis is the fact that in most languages the copula is absent in the present tense only, as we can see in Arabic, Hebrew, and Russian.1 If we assume the null copula analysis, we will be forced to posit the presence of a copula in the present tense that becomes deleted in the course of the syntactic derivation. In other words, a deletion rule must be assumed, one that deletes the copula only in the present tense. This fact however does not follow from any property of the present tense. Therefore, one can conclude that a verbal copula is not present in verbless clauses, and that these types of clauses are indeed configurationally verbless—​that is, verbless sentences are full-​fledged clauses that exhibit the same properties of tense clauses (Benmamoun, 2008). In sum, in mainstream generative grammar the predominant assumption has been that verbless clauses contain a functional projection that may be specified for tense (TP) but need not occur with a verbal projection or a copula. This is strong evidence against the view that tense needs to co-​occur with a verbal head (Benmamoun, 2008). This proposal departs from Chomsky’s (2001), in which the category tense may be specified for categorial verbal or nominal features. In general, verbless clauses may be considered TPs that dominate a non-​verbal predicate.2 In the following section I turn my attention to the syntax, semantics and discourse properties of Spanish PredNPs. I will not be assuming the presence of a null copula, but rather the existence of a R E LAT O R /​L IN K ER head—​that is, an element that mediates the relationship between a predicate and its subject in the base representation of predication structures (den Dikken, 2006). I  will assume furthermore that this element may host the tense feature. On the other hand, the interpretation of the copula in (1a) and (1b), repeated here as (7a) and (7b) for convenience’s sake, is a by-​product of the context, and (2) due to the fact that only the copula can be elided in such a context because it does not assign a thematic role. (7) a. [AP muy inteligente] el profesor este very intelligent this professor “This professor I S very intelligent.” b. [NP un verdadero idiota] aquel camarero a real idiot that waiter (over there) “The waiter over there I S an idiot.”

Verbless exclamatives: a pragmatic account  171 Following den Dikken (2006), the subject-​predicate relationship can be instantiated as in (8a), with (8b) having the syntactic representation of (8c). In this syntactic configuration the YP projected by “beautiful” serves as the complement of the R ELATOR and “the poem” is seated in the specifier position of the R ELATOR’ S projection (8d). (8) a. [RP [XP SUBJECT] [R’ RELATOR [YP PREDICATE]]] b. The poem is beautiful. c. [RP [The poem] [RELATOR=be [beautiful]]] d.

3.  The grammar of Spanish PredNPs Spanish PredNPs are verbless utterances which are not discourse fragments and display clausal properties (Vinet, 1991; Laurens, 2008). For example, each of the expressions in (7)  may be interpreted as full sentences, as the paraphrases in (9) show. (9) a. El profesor este es muy inteligente. “This professor is very intelligent.” b. Aquel camarero es un verdadero idiota. “That waiter (over there) is a real idiot.” As (9)  shows, these clauses involve predication and they actually denote a complete proposition—​that is, they are expression of type in Mon­ tague’s terminology. Their clausal content type can be easily demonstrated by the fact that the Spanish PredNPs can be coordinated with full verbal clauses: only clauses with the same or similar content type may be coordinated (Bosque and Gutiérrez-​Rexach, 2009), as shown in (10). (10) Un verdadero idiota, el camarero este, aunque es un buen empleado. “A real idiot, that waiter, even though he is a good employee.” 3.1  General facts As example (11) shows, Spanish PredNPs behave like those in other well-​ known Romance languages: Italian, French, Spanish and Brazilian Portuguese (Munaro, 2016; Hernanz and Suñer, 1999; Vinet, 1991; etc.).

172  Melvin González-Rivera (11) a. Triste, questa storia! “Sad, this story!” b. Un génie, ce Charlie. “A genius, Charlie.” c. Muito sábio, o presidente. “Very wise, the president.”

(Italian) (French) (Brazilian Portuguese)

Similar constructions are found in other languages, such as Serbian, and English (Paul and Stainton, 2006), as we can see in (12). (12) a. Fina žena, tvoja majka. (Serbian) “Nice lady, your mom.” b. Really intriguing, that PredNP. (English) In examples (12a, 12b), there is a predicate (mostly with an evaluative/​ appreciative reading) followed by a DP with a deictic determiner. These utterances have several distinct properties, such as: i There is a subject-​predicate relationship:  the property denoted by the adjective or noun phrase is predicated of the DP-​subject. ii These clauses typically have a predicate with an evaluative reading interpretation followed by a DP with a deictic determiner (Hernanz and Suñer, 1999). iii They express a categorical predication—​that is, the subject is singled out from the event itself and the predicate ascribes a property to this subject (Basilico, 2003). iv Verbless clauses such as Spanish PredNPs cannot be constructed with stage-​level predicates, but rather with individual-​level predicates. v The subject must be a definite and a specific nominal. Thus, there is a referentiality constraint affecting the subject of the clause (e.g., Arabic, Mina and Hdi). 3.1.1  The XP-​Predicate The DP/​AP predicate of this clause imposes some restrictions, which have been discussed by Vinet (1991) for French, Hernanz and Suñer (1999) for Spanish, and Munaro for Italian (2016), among others. First, the predicate may be interpreted as a copulative verb “be.” In other words, verbs that do not assign a thematic role may be interpreted in Spanish PredNPs. Thus, the following Spanish sentence in (13a) cannot be interpreted as in (13b). (13) a. Un idiota el presidente “An idiot the president” b. ≠ El presidente se comporta como un idiota. “The president acts like an idiot.”

Verbless exclamatives: a pragmatic account  173 Or may be interpreted as containing the verb “have,” as in (14) from French (Vinet, 1991: 87). (14) Un nez extraordinaire, cette Cléopâtre! “An extraordinary nose, this Cleopatra (has).” Second, only predicative adjectives are allowed in these constructions. This restriction explains why equitative clauses are not permitted in Spanish PredNPs: they cannot be used predicatively (15, 16). (15) a. *Solar este sistema. “Solar, this system.” b. *Numerable este conjunto. “Countable, this set.” (15) a. *Pepe Mario. = Pepe es Mario. ‘Pepe is Mario.’ “Pepe, Mario.” b. *Ella María. = Ella es María. ‘She is María.’ “María, she.” Third, non-​evaluative adjectives are excluded (e.g., professional names, relational adjectives, nouns of relationship, etc.). Thus, it is a general fact about Spanish PredNP constructions that the predicative XP must be evaluative, as in (16): (16) a. *Padre mi hermano. “Father, my brother.” b. *Chileno este vino. “Chilean, this wine.” This same restriction applies to complements since they are classifying, referential and are not modifiers, as in (17). (17) a. *De Hillary esta derrota. “Hillary’s, this defeat.” b. *De Estados Unidos el embargo a Cuba. “United States, Cuba’s embargo.” Vinet (1991:77) previously noted that in French the predicate must bear a personal evaluation of the speaker (cf. Guillemin-​Flescher, 2005; Munaro, 2016), as in (18). (18) a. Étonnante, cette histoire! “Astonishing, this story.” b. Un imbécile, ce Rodolphe! “An imbecile, this Rodolphe.”

174  Melvin González-Rivera Fourth, whereas individual-​ level predicates are allowed, stage-​ level predicates are not. However, adverbs of quantification or Q-​adverbs such as “always” and “never” license stage-​level predicates (19, 20). (19) a. *Cansado Pepe. “Tired Pepe.” b. Siempre cansado Pepe. “Always tired, Pepe.” (20) a. *Disponible el profesor. “Available, the professor.” b. Nunca disponible el profesor. “Never available, the professor.” Following Kratzer (1988/​ 1995), this restriction is usually explained as follows:  individual-​level predicates do not project an event variable—​that is, they are not associated with a spatio-​temporal variable; while stage-​level predicates do have a variable ranging over variables (Jäger, 2001 for an alternative account). If individual-​level predicates do not project an event variable, and Q-​adverbs quantify over situations or occasions, then there will not be an appropriate variable for the adverb to bind. In the case of stage-​level predicates, the insertion of a Q-​adverb gives the predicate a generic or universal reading, and the predicate can be interpreted as characterizing or homogenizing (Carlson, 1977/​1980). Finally, predicates in Spanish PredNPs must correspond to an opinion. Usually, the speaker mentions something outstanding about some individual or event. It follows then that adjectives in these clauses are not merely evaluative adjectives, but also appreciative ones:  the XP predicate must denote something extraordinary about the subject (Vinet, 1991; Munaro, 2016), as in (21). (21) a. ??Normal este vino. “Average, this wine.” b. Muy normalito este vino. “Very average, this wine.” c. Riquísimo este vino. “Delicious, this wine.” d. Muy rico este vino. “Very delicious, this wine.” e. Espectacular la victoria de Obama. “Spectacular, Obama’s victory.”

Verbless exclamatives: a pragmatic account  175 For example, (21a) is odd if we are just stating that the wine is of standard quality. On the other hand, muy normalito “very average” in (21b) indicates that the quality of the wine is below what was expected. It follows, then, that adjectives in these clauses are not merely evaluative-​ affective adjectives, but also ones that place such evaluation at a high or extreme point in scale—​that is, the XP-​predicate must denote something extraordinary about the subject (Gutiérrez-​ Rexach, 1999; Lasersohn, 2005; Villalba, 2004). In sum, the predicate in Spanish PredNPs is a modifier with an evaluative/​appreciative interpretation. Also, individual-​level predicates are preferred in this construction. Recall that individual-​level predicates attribute properties to individuals:  they express properties of individuals that are permanent or tendentially stable (Chierchia, 1995). In this sense, Spanish PredNPs denote properties or states of affairs that are individualizing or characterizing and are also dependent on the subjective evaluation of a contextually established judge (Gutiérrez-​Rexach and González-​Rivera, 2014). 3.1.2  The DP-​Subject Like the XP predicate, the DP subject in these constructions imposes some restrictions as well. First, the subject must be definite or strong in Milsark’s (1974, 1977) terminology. This could be a direct consequence of the XP predicate: individual-​level predicates select strong determiners—​that is, weak subjects cannot have a state predication. Thus, nonspecific or specific indefinite DPs and bare plurals are ruled out, as in (22). (22) a. Muy simpática está tía. “Very nice, this girl.” b. *Muy simpática una tía cualquiera. “Very nice, any girl.” c. *Muy simpática alguna chica. “Very nice, some girl.” d. *Muy simpáticas chicas. “Very nice, girls.” Second, most quantified DPs are disallowed from the subject position, namely, monotone decreasing quantifiers (Gutiérrez-​Rexach, 2012). In this sense, the DP subject of Spanish PredNPs behaves like right-​ dislocated subjects (Vallduví, 1992; Laurens, 2008), as in (23, 24). (23) a. *Un genio ningún estudiante. “A genius, none of the students.” b. *Unos genios pocos estudiantes. “Genius, few students.”

176  Melvin González-Rivera c. *Un genio todo estudiante. “A genius, all student.” d. *Unos genios todos los estudiantes. “Genius, all of the students.” (24) a. Muy sabio, el presidente. “Very clever, the president.” b. Interesante la solución al problema. “Interesting, the problem’s solution.” However, sentences like (23) are improved if the quantified DP is modified (Paul and Stainton, 2006). This is so because the DP-​subject becomes a principal filter quantificationally, rendering the sentence grammatical—​that is, the modifier forces a strong reading of the quantifier (25). (25) a. Unos genios todos los estudiantes de física. “Genius, all students of physics.” b. Unos genios todos los estudiantes allí sentados. “Genius, all students seated over there.” Fourth, in many Spanish PredNPs a deictic determiner is present in the DP-​subject  (26). (26) a. Maravilloso el piso este. “Marvelous, this apartment.” b. Siempre averiado, este maldito ascensor. “Always broken, this damned elevator.” Demonstratives, especially postnominal ones (26a), tend to associate with informational elements and, furthermore, have characteristic affective uses of an appreciative or depreciative nature. In this case, the demonstrative would indicate whether a given topic is presented as subjectively valued by the speaker or not (Bernstein, 1997; Silva-​Villar and Gutiérrez-​Rexach,  2001). Summarizing, the subject DP in Spanish PredNPs must be definite and a specific nominal—​that is, nonspecific DPs, bare plurals and monotone decreasing quantifiers are ruled out. Most of the time, a deictic determiner is present. Notice that the definite/​specific constraint on Spanish PredNPs is not necessarily a language-​specific constraint: non-​verbal or verbless clauses in Arabic may require the subject to be definite. Furthermore, in Mina and Hdi the subject of verbless clauses carries high tone (Frajzyngier and Shay, 2001; Frajzyngier, Johnston and Edwards, 2005). In other words, it must involve referent identifiability (cf. Vinet, 1991; Hernanz and Suñer, 1999; Munaro, 2016).

Verbless exclamatives: a pragmatic account  177 3.1.3  Syntactic structure An important characteristic of Spanish PredNPs is the fact that these constructions license anaphors in predicate position. The subject, however, cannot be co-​indexed with a pronoun in predicate position (Conditions A and B of Chomsky’s Binding Theory). This seems to be strong evidence in favor of considering the XP-​predicate as originating to the right of the DP-​subject. (27)  Orgulloso de sí mismoi este chiquilloi. “Proud of himself this boy.” At the derivational point where binding theory applies—​either at the base generation point or after reconstruction—​the DP has to c-​command the XP, thus suggesting that the overt position of the XP is a derived position. Finally, while positive polarity items (PPIs) are possible in the predicate constituent (28a), a negative subject does not license a negative polarity item (NPI) in that same predicate (28b). (28) a. Bastante incrédulas estas chicas. “Quite incredulous, these girls.” b. *Santo de mi devoción ningún estudiante de pragmática. Lit.: “Saint of my devotion none of the pragmatic students.” “I don’t like any of the pragmatic students.” The ungrammaticality of (28b) could be explained as follows:  ningún estudiante de pragmática “none of the pragmatic students” is a non-​specific or weak DP, henceforth its presence is banned from the subject position of Spanish PredNP. However, this conclusion is inadequate since the partitive de pragmática “of pragmatics” makes the DP specific. Another solution would be based on the monotonicity of the quantifier: monotone decreasing quantifiers are not allowed in Spanish PredNPs clauses (29). (29) a. *Muy sabios, pocos estudiantes. “Very clever, few students.” b. *Muy sabios, menos de cinco estudiantes. “Very clever, less than five students.” c. *Muy sabios, como máximo cinco estudiantes. “Very clever, at most five students.” NPIs occur within arguments of monotonic decreasing functions but not within arguments of monotonic increasing functions (Keenan, 1996; cf. the Ladusaw-​Fauconnier Generalization). If monotone decreasing quantifiers are not allowed in the subject position of Spanish PredNPs, then it follows that NPIs cannot be licensed in the predicate position (Gutiérrez-​Rexach and González-​Rivera,  2014).

178  Melvin González-Rivera 3.1.4  Information structure The DP subject in Spanish PredNPs is a background topic, or presupposed information in Zubizarreta’s (1998) terminology, and the predicate XP is the focus (cf. Munaro, 2016)—​that is, the predicate provides new information and the subject is old information and must be salient (Paul and Stainton 2006). The subject, in other words, must be linked to a topic in the common ground or background context (Lambrecht, 1994; Vallduví, 1992). For example, the following dialogue is felicitous as long as the epithet is linked to the discourse topic. (30) Speaker A: Speaker B:

¿ Qué piensas del tío este? “What do you think of this guy?” Un idiota el tío este. “An idiot, this guy.”

(31) Un idiotafocus el tío estetopic The topic nature of the subject of this clause has important consequences for the strong referential constraint on the subject, namely that the subject must be specific (von Heusinger, 2002). It follows then that when a quantifier shows a specific/​non-​specific alternation, the more specific the better the quantifier will be in the subject position. The contrast in (32) confirms this. (32) *Inteligente todo estudiante. “Intelligent, every student.”

4.  A new proposal: predicate inversion and strong (semantic) feature In order to explain the internal structure of Spanish PredNPs I  will assume that the XP-​predicate moves to the left of the subject (predicate inversion). The predicate moves in order to discharge (or check, in Chomsky’s generative grammar) the strong semantic feature evaluativity (cf. Guillemin-​Flescher, 2005). This feature in turn blocks the occurrence of stage-​level predicates since evaluative adjectives have the strong tendency to be individual-​level predicates. In this case the individual-​and stage-​level distinction in Spanish PredNP can be reduced entirely to the feature evaluativity, which is a property of these constructions. Thus, only those predicates that carry this semantic feature are allowed in these clauses. Notice that in this derivation movement is not triggered by a formal feature (Chomsky, 1995), but rather by a semantic property—​that is, an interpretable feature. Paul (2008), following den Dikken (2006), invokes a similar strategy of predicate inversion for the interpretation of Spanish PredNP, but for a different reason: for her, the predicate moves to Spec,TP in order to license the null copula head. If the copula is overtly

Verbless exclamatives: a pragmatic account  179 realized, then movement is blocked. In Paul’s proposal, (33a) would have the syntactic representation in (33b). (33) a. Beautiful, your girlfriend. b. [TP [PredP [beautiful]]j T [RP [DP your girlfriend] R0 tj]] There are problems with Paul’s account because predicate inversion is possible even when the copula is overtly realized, as in (34). (34) a. b. c. d.

John is the tallest man in this building. The tallest man in this building is John. Mary is a good professor. A good professor is (in fact) Mary.

Predicate inversion will only be possible only when the clause is interpreted predicatively (35). (35) a. b. c. d.

Mary is beautiful. *Beautiful is Mary. Mary is a woman. *A woman is Mary.

Thus, according to Paul (2008) only predicative adjectives are allowed in Spanish PredNP clauses. However, this is not true since stage-​level adjectives can be used predicatively, but still are not allowed in Spanish PredNP. (36) a. *Available, the doctor. b. *Tired, your friend. Furthermore, Paul’s (2008) argument suggests that if the copula were non-​ null then predicate inversion would not apply. This prediction is not borne out. For instance, in Spanish we may find examples like (37). (37) a. Muy inteligente es ese profesor. “Very intelligent is that professor.” b. Es muy inteligente ese profesor. “It He is very intelligent that professor.” Of course, it can be argued that (37a) and (37b) are different kinds of constructions, but even if this is argued, there is still the fact that predicate inversion can happen in the absence of a null copula. In fact, Moro (1991a, 1991b, 1997, 2000) does not link predicate raising to null copula. However, more importantly, the idea that predicate inversion serves precisely to license

180  Melvin González-Rivera the null copula is diametrically opposed to the central argument in den Dikken (2006), which the author follows. According to den Dikken (2006), predicate inversion leads precisely to the obligatory overtness of the RE L AT OR (or copula in this case), as in (38). (38) a. I consider John (to be) the best candidate for the job. b. I consider the best candidate for the job *(to be) John. In turn, it will be assumed that what motivates predicate inversion in Spanish PredNPs clauses is not the absence of the copula, but a strong semantic feature that needs to be discharged in the course of the syntactic derivation. This strong feature is evaluativity and it is represented as [+E]. It will be assumed furthermore that there is a RE L AT OR between the DP-​ subject and the XP-​predicate, but contrary to the small clause analysis, this R E LAT O R may not empty as it can accommodate the tense feature. The default tense in Spanish, and presumably in the Romance languages, is the present tense, but a past tense interpretation is also possible. Benmamoun (2008) has argued that tense is universally projected but does not need to co-​ occur with a verbal head. He provides evidence from Hebrew and Arabic, two well-​known languages with verbless clauses, as previously stated. This is the case for Spanish PredNP: even though there is no verbal head, tense can be interpreted. A sentence like (39a) will have the syntactic configuration in (39b), following a minimalist account: external and internal merge. (39) a. Muy guapa Mariana. “(very/​extremely) attractive, Mariana.” b. [[YP muy guapa+affective]i [RP [XP Mariana][R0] ti]] The phrase has a strong semantic feature [+E], instantiated in the example above as +affective ([+Eaffective]), that needs to be discharged. This feature associates with focus and triggers movements to a left-​peripheral projection. This configuration would derive the desired spellout without a copular element and with the predicate in a preposed position. The presence of a copula would block movement to the left periphery, so the absence of an overt copular element becomes essential in deriving the proper configuration and, henceforth, deriving a verbless clause. Lastly, an analysis of Spanish PredNPs based on predicate inversion explains why the DP-​subject needs to be definite and specific: the DP-​subject is highly topical in predicate fronting constructions.

5.  Spanish PredNPs as exclamatives One last issue remains in the analysis of Spanish PredNPs:  whether these clauses are a subtype of exclamative sentences (cf. Vinet, 1991; Munaro, 2016). Here it is argued, contra Alonso-​Cortés, that Spanish PredNPs may be analyzed as exclamative sentences. The main argument against considering

Verbless exclamatives: a pragmatic account  181 Spanish PredNPs exclamatives has been that these clauses are not associated with a scalar implicature, a property that Zanuttini and Portner (2003) have proposed for exclamative sentences (Paul and Stainton, 2006). Thus, according to these authors, exclamatives convey that something is surprising or noteworthy in some way. In this sense, exclamatives introduce a conventional scalar implicature to the effect that the message they denote lies at the extreme end of some contextually given scale—​that is, exclamatives generate a characteristic conventional scalar implicature by which an individual, for example, is placed on the extreme of a determined scale. For instance, an exclamative such as (40a) would have the implicature in (40b), where d’ is a degree that exceeds the speaker’s expectations of beautifulness (Gutiérrez-​Rexach, 1996). (40) a. How beautiful Marianita is! b. Marianita is beautiful to degree d’. (40b) can be considered an implicature because it goes beyond the sentence’s truth-​conditional meaning. However, the implicature must be a conventional implicature, rather than a conversational one, since the former is non-​defeasible and non-​detachable. There are at least two problems for not considering Spanish PredNPs as exclamatives based only on the scalar implicature property. First, in Spanish there are exclamatives that do not trigger such scalar implicature but are still considered exclamative sentences. A short list of such exclamatives is found in (41) (Gutiérrez-​Rexach, 2001). (41) a. ¡ Por supuesto que sí voy a ir al cine! “Of course, I’ll be going to the movies!” b. ¡ Claro que te va a salir bien el examen! “Of course, you’ll do fine in the exam!” c. ¡ Sí que tienes mala suerte tú! “You really have bad luck!” The second problem that faces the argument against treating Spanish PredNPs as exclamatives is even more important than the first one, namely, these clauses may indeed trigger a scalar implicature. Consider example (42). (42) a. ¡ Extremadamente preciosa Marianita! “Extremely gorgeous Marianita!” b. ¡ Ridículamente preciosa Marianita! “Ridiculously gorgeous Marianita!” When uttering (42) the speaker expresses an attitude (surprise or admiration) toward the fact that Marianita is d-​gorgeous, where d is Marianita’s degree of gorgeousness. Following Gutiérrez-​Rexach (1996), this idea can be formally represented as follows:

182  Melvin González-Rivera EXC(a)(w)(ιp ∃d [p(w) & p = λw’[d = MAX(λd’[gorgeous(w’)(M,d’)])]]) iff ∃P ∈EMOT [P(w)(ιp∃d[p(w) & p = λw’[d = MAX(λd’[gorgeous(w’) (M,d’)])]])(a)] where a is the agent, w a world, p a proposition, and P∈EMOT: the set of emotive properties. Spanish PredNPs not only may trigger a scalar implicature, but they may also express the speaker’s strong feeling or emotive attitude toward something, and this is a property of exclamatives. Spanish PredNPs and exclamatives share other properties: for example, both clauses involve referent identifiability: the entity of whom something is predicated must be identifiable. Thus, in exclamatives and Spanish PredNPs the entity that something has been predicated on must be part of the common ground between hearer and speaker. Lambrecht (1994), for example, defines identifiable referent as a shared representation that exists in the minds of speaker and hearer at speech time. It follows from this definition that this referent must be a pronominal or a definite DP. This analysis resembles Sadock and Zwicky’s (1985) proposal about exclamative sentences in which exclamatives are differentiated from declaratives. For Sadock and Zwicky, exclamations have an expressive force, whereas declaratives are informatives. Both clauses represent a proposition as being true, but in exclamative constructions, the speaker emphasizes his strong emotional reaction or feeling toward what he takes to be a fact, while in a declarative, the speaker emphasizes his intellectual appraisal that the proposition is true. Spanish PredNPs do just that: they are not only informatives but may express a strong emotional reaction toward some fact. Finally, the last property exclamatives and Spanish PredNPs share is that of deixis, both personal and temporal deixis. The notion of personal deixis entails the presence of someone making a judgment, and the speaker is the judge by default, whereas temporal deixis requires that the effect evoked by these clauses must hold at speech time, whether the denoted situation also holds at speech time. These properties seem to hold true not only of exclamatives, but for verbless clauses in general.

6.  Conclusion There are some languages that do not make use of a copula in predicative contexts–​that is, predication in this context is just the reflex of putting together two syntactic functions related to each other without the intervention of a verb (Rosén, 1996). These clauses have received the name of (predicative) verbless clauses in the linguistic literature and are composed of a predicate and the argument which saturates it (García Marchena, 2017). In this chapter the grammatical properties of a subtype of verbless clause in Spanish, namely PredNPs, have been examined, mainly from a generative perspective. Spanish PredNP is a type of non-​verbal or verbless predication with a fixed syntactic order: an initial XP-​predicate precedes the DP-​subject of the clause and there is no verb between the two expressions. The most

Verbless exclamatives: a pragmatic account  183 salient properties of this construction are the following: (i) they typically have a predicate with an evaluative reading interpretation followed by a DP with a deictic determiner; (ii) the predicate constitutes new information (focus); (iii) the subject needs to be definite and specific; and (iv) the subject needs to be interpreted as a background topic. I have argued that Spanish PredNPs are better understood if we assume that the XP-​predicate moves to the left of the DP-​subject to a focus position where the former may check its affective feature (cf. the A FFEC T -​criterion; see Brody, 1990; Haegeman, 1995; Rizzi, 1997 for a detailed discussion). It has been argued also that these clauses can be considered a subtype of exclamatives.

Notes 1 In these languages a pronominal element, or Pron, may appear between the subject and the predicate of verbless constructions. Pron, however, is not to be interpreted necessarily as the realization of the copula (see González-​Rivera 2011 for further discussion). 2 See Nordlinger and Sadler (2006) and Avgustinova and Uszkoreit (2003) for an analysis of verbless clauses in model-​theoretic grammars.

References Abdel-​ Ghafer, Omar. (2003). Copular constructions in Modern Standard Arabic, Modern Hebrew and English (Ph.D. dissertation). University of Kansas. Adger, David and Ramchand, Gillian. (2003). Predication and equation. Linguistic Inquiry 34, 325–​360. Al-​ Horais, Nasser. (2006). Arabic verbless sentences:  Is there a null VP? Pragmalinguistica 14, 101–​116. Alonso-​Cortés, Angel. (1999). Las construcciones exclamativas. La interjección y las expresiones vocativas. In I. Bosque and V. Demonte (eds.), Gramática descriptiva de la lengua española, Vol. 3, 3993–​4050. Madrid: Espasa Calpe. Armstrong, Grant. (2017). The syntax of non-​verbal predication in Yucatec Maya. Cuadernos de Lingüística del Colegio de México 4(2), 137–​212. Attia, Mohammed. (2008). A unified analysis of copula constructions in LFG. In M. Butt and T. Holloway King (eds.), Proceedings of the LFG08 Conference, 89–​ 108. Stanford, CA: CSLI. Avgustinova, Tania and Uszkoreit, Hans. (2003). Reconsidering the relations in constructions with non-​verbal predicates. In P. Kosta, J. Blaszzak, J. Frasek, L. Geist and M. Zygis (eds.), Investigations into Formal Slavic Linguistics. Contributions of the Fourth European Conference on Formal Description of Slavic Languages (FDSL IV) Part II, 483–​498. Frankfurt am Main: Peter Lang Europäischer Verlag der Wissenschaft. Bakir, Murtadha J. (1980). Aspects of clause structure in Arabic (Ph.D. dissertation). Indiana University. Basilico, David. (2003). The topic of small clauses. Linguistic Inquiry 34(1), 1–​35. Benmamoun, Elebbas. (2000). The Feature Structure of Functional Categories:  A Comparative Study of Arabic Dialects. Oxford: Oxford University Press. Benmamoun, Elebbas. (2008). Clause structure and the syntax of verbless sentences. In R. Freidin, C. Peregrín Otero and M.L. Zubizarreta (eds.), Foundational Issues in

184  Melvin González-Rivera Linguistic Theory: Essays in Honor of Jean-​Roger Vergnaud, 105–​131. Cambridge, MA: MIT Press. Bentley, Delia. (2017). Copular and existential constructions. In A. Dufter and E. Stark (eds.), Manual of Romance Morphosyntax and Syntax, 332–​366. (Manuals of Romance Linguistics; No. 17). Berlin: de Gruyter, Walter GmbH & Co. Berman, Ruth A. (1978) Modern Hebrew Structure. Tel Aviv: University Publishing Projects. Bernstein, Judy B. (1997). Demonstratives as reinforcers in Romance and Germanic languages. Lingua 102(2–​3), 87–​113. Bosque, Ignacio and Gutiérrez-​ Rexach, Javier. (2009). Fundamentos de Sintaxis Formal. Madrid: Akal. Brody, Michael. (1990). Remarks on the order of elements in the Hungarian focus field. In I. Kenesei (ed.), Approaches to Hungarian 3:  Structures and Arguments, 95–​122. Szeged: JATE. Butts, Aaron M. (2006). Observations on the verbless clause in the language of Neophyti I. Aramic Studies 4(1), 53–​66. Carlson, Gregory. (1977/​1980). Reference to Kinds in English. New York: Garland. Carnie, Andrew. (1995). Non-​ verbal predication and head movement (Ph.D. dissertation). MIT. Carnie, Andrew. (1997). Two types of non-​ verbal predication in Modern Irish. Canadian Journal of Linguistics 42(1–​2),  57–​73. Chierchia, Gennaro. (1995). Individual-​ level predicates as inherent generics. In G.  Carlson and F.J. Pelletier (eds.), The Generic Book, 176–​ 223. Chicago, IL: University of Chicago Press. Chomsky, Noam. (1995). The Minimalist Program. Cambridge, MA: MIT Press. Chomsky, Noam. (2000). Minimalist inquiries:  The framework. In R. Martin, D. Michaels and J. Uriagereka (eds.), Step by Step:  Essays on Minimalist Syntax in Honor of Howard Lasnik, 89–​156. Cambridge, MA: MIT Press. Chomsky, Noam. (2001). Derivation by phase. In M. Kenstowicz (ed.), Ken Hale: A Life in Language, 1–​52. Cambridge, MA: MIT Press. Citko, Barbara. (2008). Small clauses reconsidered:  Not so small and not all alike. Lingua 118(3), 261–​295. Corriente, Federico. (1997). Poesia dialectal árabe y romance en Alandalús. Madrid: Gredos. Déchaine, Rose-​Marie Anne. (1993). Predicates across categories (Ph.D. dissertation). University of Massachusetts at Amherst. den Dikken, Marcel. (2006). Relators and Linkers. The Syntax of Predication, Predicate Inversion, and Copulas. Cambridge, MA: MIT Press. Diesing, Molly. (1992). Indefinites. Cambridge, MA: MIT Press. Doron, Edith. (1983). Verbless predicates in Hebrew (Ph.D. dissertation). University of Texas at Austin. Eisele, John. (1988). The syntax and semantics of tense, aspect, and time reference in Cairene Arabic (Ph.D. dissertation). University of Chicago. Escandell-​Vidal, Victoria and Leonetti, Manuel. (2002). Coercion and the stage/​individual distinction. In J. Gutiérrez-​Rexach (ed.), From Words to Discourse. Trends in Spanish Semantics and Pragmatics, 159–​179. Oxford: Elsevier Science. Fassi Fehri, Abdelkader. (1982). Linguistique Arabe:  forme et interpretation. Rabat: Publications de la Faculté des Lettres et Sciences Humaines.

Verbless exclamatives: a pragmatic account  185 Fassi Fehri, Abdelkader. (1993). Issues in the Structure of Arabic Clauses and Words. Dordrecht: Kluwer. Frajzyngier, Zygmunt and Shay, Eric. (2001). A Grammar of Hdi. Berlin: Mouton de Gruyter Frajzyngier, Zygmunt, Johnston, Eric and Edwards, Adrian. (2005). A Grammar of Mina. Berlin: Mouton de Gruyter. Gamut, L.T.F. (1991). Intensional Logic and Logical Grammar (Volume 2 of Logic, Language and Meaning). Chicago, IL: University of Chicago Press. García Marchena, Oscar. (2017). The structure of Spanish verbless clauses. EPiC Series in Language and Linguistics 2, 335–​345. González-​Rivera, Melvin. (2011). On the internal structure of Spanish verbless clauses (Ph.D. dissertation). Ohio State University. Guillemin-​Flescher, Jacqueline. (2005). L’énoncé averbal: repérage et subjectivité. In F. Lefeuvre and I. Behr (eds.), Les énoncés averbaux autonomes entre grammaire et discours, 11–​30. Paris: Ophrys. Gutiérrez-​Rexach, Javier. (1996). The semantics of exclamatives. In E. Garret and F. Lee (eds.), UCLA Working Papers in Linguistics, 146–​162. Gutiérrez-​Rexach, Javier. (1999). The structure and interpretation of Spanish neuter degree constructions. Lingua 109(1), 35–​63. Gutiérrez-​Rexach, Javier. (2001). Spanish exclamatives and the semantics of the left periphery. In Johan Rooryck, Y.d. Hulst and J. Schroten (eds.), Selected Papers from Going Romance 99, 167–​194. Amsterdam: John Benjamins. Gutiérrez-​Rexach, Javier. (2012). Quantification. In J.I. Hualde and A. Olarrea (eds.), Handbook of Spanish Linguistics, 307–​332. Oxford: Blackwell. Gutiérrez-​Rexach, Javier and González-​Rivera, Melvin. (2014). Spanish predicative verbless clauses and the left periphery. In A. Dufter and A.S. Octavio de Toledo (eds), Left Sentence Peripheries in Spanish: Diachronic, Variationist and Comparative Perspectives, 101–​124. Amsterdam: John Benjamins. Haegeman, Liliane. (1995). The Syntax of Negation. Cambridge:  Cambridge University Press. Hazout, Ilan. (2004). The syntax of existential constructions. Linguistic Inquiry 35, 393–​430. Hazout, Ilan (2010). Remarks and replies:  Verbless sentences and clause structure. Linguistic Inquiry 41(3):471 – 485. Heggie, Lorie A. (1988a). The syntax of copular structures (Ph.D.  dissertation). University of Southern California. Heggie, Lorie A. (1988b). A unified approach to copular sentences. In H. Borer (ed.), The Proceedings of the Seventh West Coast Conference on Formal Linguistics, 129–​142. Stanford, CA: Stanford Linguistics Association. Heggie, Lorie A. (1989). Constructional focus and equative sentences. In E.J. Fee and K. Hunt (eds.), The Proceedings of the Eighth West Coast Conference on Formal Linguistics, 154–​166. Stanford, CA: Stanford Linguistics Association. Hernanz, Ma. Lluïsa and Suñer, Avellina. (1999). La predicación:  la predicación no copulativa. Las construcciones absolutas. In I. Bosque and V. Demonte (eds.), Gramática descriptiva de la lengua española, Vol. II: Las construcciones sintácticas fundamentales– Relaciones temporales, aspectuales y modales, 2525–​2560. Madrid: Espasa Calpe. Jäger, Gerhard. (2001). Topic-​comment structure and the contrast between stage level and individual level predicates. Journal of Semantics 18(2), 83–​126.

186  Melvin González-Rivera Keenan, Edward. (1996). The semantics of determiner. In S. Lappin (ed.), The Handbook of Contemporary Semantic Theory. Oxford: Blackwell. Kratzer, Angelika. (1988/​ 1995). Stage-​ level and individual-​ level predicates. In G.  Carlson and F.J. Pelletier (eds.), The Generic Book, 125–​175. Chicago, IL: University of Chicago Press. Lambrecht, Knud. (1994). Information Structure and Sentence Form. Cambridge: Cambridge University Press. Lasersohn, Peter. (2005). Context dependence, disagreement, and predicates of personal taste. Linguistic and Philosophy 28(6), 643–​686. Laurens, Frédéric. (2008). French predicative verbless utterances. In S.M. Müller (ed.), Proceedings of the 15th International Conference on Head-​Driven Phrase Structure Grammar, 152–​172. Stanford, CA: CSLI. Lefeuvre, Florence. (1999). La Phrase averbale en français. Paris: L’Harmattan. Loporcaro, Michele. (1998). Sintassi comparata dell’accordo participiale romanzo. Turin: Rosenberg & Sellier. Markman, Vita G. (2008). The case of predicates (revisited): predicate instrumental in Russian and its restrictions. Journal of Slavic Linguistics 16(2), 187–​246. Milsark, Gary. (1974). Existential Sentences in English. Cambridge, MA: MIT Press. Milsark, Gary. (1977). Towards an explanation of certain peculiarities of the existential construction in English. Linguistic Analysis 3, 1–​29. Moro, Andrea. (1991a). The raising of predicates: Copula, expletives and existence. In L. Cheng (ed.), MIT Working Papers in Linguistics 15, 183–​218. Cambridge, MA: MIT Press. Moro, Andrea. (1991b). The Raising of Predicates: Copula, Expletives and Existence, Vol. 15. Cambridge, MA: MIT Press. Moro, Andrea. (1997). The Raising of Predicates: Predicative Noun Phrases and the Theory of Clause Structure. Cambridge: Cambridge University Press. Moro, Andrea. (2000). Dynamic Antisymmetry. Cambridge, MA: MIT Press. Mouchaweh, Lina. (1986). De la syntaxe des petites prépositions (Ph.D. dissertation). Université de Paris VII. Moutaouakil, Ahmed. (1987). min qaDaayaa r-​ raabiT fii l-​ lugha l-​ ’arabiyya. Casablanca: ‘ocaadh. Munaro, Nicola. (2006). Verbless exclamatives across Romance: standard expectations and tentative evaluations. University of Venice Working Papers in Linguistics 16, 185–​209. Munaro, Nicola. (2016). Verbless predicative structures across Romance. Journal of Linguistics 52(3), 609–​637. Nordlinger, Rachel and Sadler, Louisa. (2006). Verbless clauses: Revealing the structure within. In J. Grimshaw, J. Maling, C. Manning, J. Simpson and A. Zaenen (eds.), Architectures, Rules and Preferences: A Festschrift for Joan Bresnan. 139–​160. Stanford, CA: CSLI Publications. Paul, Ileana. (2008). Great coffee, that Maxwell House! In E. Efner and M. Walkow (eds.), Proceedings of the North East Linguistic Society (NELS). 37. Amherst, MA: GLSA Publications. Paul, Ileana and Stainton, Robert. (2006). Really intriguing, that PredNP! Philosophy Publications. Proceedings of the 2006 annual conference of the Canadian Linguistic Association. Rapoport, Tova. (1987). Copular, nominal and small clauses: A study of Israeli Hebrew (Ph.D. dissertation). MIT.

Verbless exclamatives: a pragmatic account  187 Revell, Ernest John. (1989). The conditioning of word order in verbless clauses in Biblical Hebrew. Journal of Semitic Studies 34, 1–​24. Rizzi, Luigi. (1997). The fine structure of the left periphery. In L. Haegeman (ed.), Elements of Grammar, 281–​337. Dordrecht: Kluwer. Rosén, Victoria. (1996). The LFG architecture and “verbless” syntactic constructions. In M. Butt & T.H. King (eds.), Proceedings of the First LFG Conference. Stanford, CA: CSLI Publications. Rothstein, Susan. (2001). Predicates and Their Subjects. Dordrecht: Kluwer. Sadock, Jerrold and Zwicky, Arnold M. (1985). Sentence types. In T. Shopen (ed.), Language Typology and Syntactic Description, Vol. 1: Clause Structure, 155–​196. Cambridge: Cambridge University Press. Shin, Gyu-​Ho. (2018). Event structure composition in Korean verbless constructions by particles and verbal nouns:  Evidence from newspaper headlines. Journal of Language Sciences 25, 403–​425. Silva-​Villar, Luis and Gutiérrez-​Rexach, Javier. (2001). Demonstratives in a feature-​ based theory of syntax. In G.M. Alexandrova and O. Arnaudova (eds.), The Minimalist Parameter. Amsterdam: John Benjamins. Tanguy, Noalig. (2013). La phrase averbale, une phrase sans verbe? L’exemple des constructions binaires à prédicat focalisé. In L. Fesenmeier, A. Grutschus and C. Patzelt (eds.), L’absence au niveau syntagmatique, 219–​ 234. Frankfurt am Main: Klostermann. Vallduví, Enric. (1992). The Informational Component. New York: Garland. Villalba, Xavier. (2004). Exclamatives and Negation. GGT Technical Report (GGT-​ 2004–​02). Bellaterra:  Grup de Gramàtica Teòrica, Universitat Autònoma de Barcelona. Vinet, Marie-​Thérèse. (1991). French non-​verbal exclamative constructions. Probus 3, 77–​100. von Heusinger, Klaus. (2002). Specificity and definiteness in sentence and discourse structure. Journal of Semantics 19(3), 245–​274. Wilson, Daniel Joseph. (2018). Copular and existential sentences in Biblical Hebrew (Ph.D. dissertation). University of the Free State. Zanuttini, Raffaella and Portner, Paul. (2003). Exclamative clauses at the syntax-​ semantics interface. Language 79(1), 39–​81. Zubizarreta, María L. (1998). Prosody, Focus and Word Order. Cambridge, MA: MIT Press.

9  The meaning of y tó Conventional implicatures in Puerto Rican Spanish Nieves Rivera

1.  Introduction Although Puerto Rican Spanish is one of the most studied varieties of Latin American Spanish, it still remains highly understudied within semantic and pragmatic theory. In this chapter I will discuss the pragmatic meaning of the expression y tó, which translates to “and all”/​“and everything,” and provide an accurate characterization of its meaning and uses. This construction is not exclusive to this Spanish variety, and it has the same uses in other dialects. However, for the purposes of this study, I will focus on Puerto Rican Spanish. While there is no formal analysis of y tó in the literature, native speakers have described the uses of y tó, the reduced version of y todo, as something that conveys surprise, something that is exaggerated, a disbelief, or shock. For example, in (1)  taken from Cee_​gonz’s (2018) Twitter account, the speaker conveys surprise and shock that the car is still working even though the check engine notification has been on for a long time. (1) Mi carro dice “check engine” hace tanto tiempo que pa mi que se acostumbró y tó. “My car says ‘check engine’ so long ago that I think it got used to it and all.” Thus, Puerto Rican Spanish speakers make use of y tó to signal unexpectedness or an exceeded degree as a reaction to the current discourse context. Moreover, the meaning that y tó adds to the utterance seems to be one that cannot be cancelled after it is conveyed. Considering this principal feature of y tó, and building upon previous work on implicatures, I will propose and show that this expression should be analyzed as a conventional implicature, by demonstrating several cases of the use of y tó as evidence.The structure of this chapter is the following: in section 2, I will present the felicity conditions of y tó by explaining different types of contexts where it is licensed. In section 3, I will discuss what conventional implicatures are, and its main characteristics. In section 4, I will demonstrate how each of these characteristics applies to the expression y tó. Finally in section 5 I will present my conclusions.

The meaning of y tó  189

2.  Felicity conditions The expression y tó can be licensed in two types of similar contexts: concessive and unexpected contexts. I make this distinction because while both concessive and unexpected contexts have an element of surprise, the unexpectedness in concessive contexts comes from connectives such as pero, and therefore it is compatible with y tó. However, y tó can convey unexpectedness on its own, as I will show below. In the following section, I will present the use of y tó in contexts where a concessive meaning is conveyed. 2.1  Concessive interpretation The expression y tó can be used in contexts that have a concessive interpretation. Following König and Siemund (2000), concessivity conveys a relationship of unexpectedness between propositions. In this sense, y tó is compatible with a pesar de “in spite of/​although,” and can be felicitously used in such contexts. In these concessive contexts, the speakers are trying to convey that there is a mismatch in the proffered content. For example, the following tweet from Saddiemuchacha’s Twitter account, which has been deleted since, shows the mismatch between two properties that are associated to the same speaker: (2) Yo soy chula y tó pero abro la boca y se me sale de lo de Cataño. “I am cute/​nice and all but when I speak the Cataño comes out.” To put things in perspective, Cataño is a small town in Puerto Rico relatively close to the San Juan metropolitan area and Guaynabo. Generally speaking, people from these two towns are perceived as more polished (cf. guaynabito/​a) and belong to the upper class. On the other hand, people from Cataño are considered loud, very direct, and they belong to a lower class. Therefore, they are usually attributed negative characteristics. In the example in (2), the speaker is contrasting two properties that do not go together per usual: the property of being nice/​cute with the property of being from Cataño, i.e. having the negative characteristics of the people who live in this town. In other words, it would be infelicitous to use y tó in (2), if being cute/​nice and being from Cataño held the same properties. In these types of contexts, y tó allows the speaker to intensify the surprise as a reaction to the mismatch of properties. Although y tó is compatible with a concessive connective, the latter is not required to create the unexpectedness associated with y tó as shown in (3a) and (3b). (3) a. Qué puerca, yo estoy enferma y tó y aún así vine a la escuela. “How nasty, I am sick and all, and still I came to school.” b. Enferma y tó, vine a la escuela. “Sick and all, I came to school.”1

190  Nieves Rivera The same idea applies to similar situations where two or more properties do not seem as compatible in the world or culture as we know it. Another example to show that the concessive interpretation is compatible with the use of y tó is shown below in example (4). (4) Context: Penélope knows that César is a budget analyst, and as such he is good at managing his personal budget, but she notices that he’s been having trouble with his finances lately. She’s telling her friend Claudia about this and says: César es analista de presupuestos y tó pero últimamente está teniendo problemas financieros. “César is a budget analyst and all but lately he is having problems with his finances.” In this case, the speaker can reason that someone who works with budgets must be good with their personal finances as well. Since in (4) that is not the case, then the speaker can use y tó and convey surprise as a reaction to the situation. In the previous examples of concessive interpretations we have seen that y tó is compatible when we contrast two properties. In addition, we can also contrast propositions. For example, take the utterance in (5) where the speaker is contrasting two propositions instead of two properties. (5) Context: Celia does not like to generate too much plastic because she is very conscientious about the environment. Her friend tells her to buy a shampoo bar that does not come in a plastic container and that it is made of natural ingredients. Celia asks her friend how much it costs, and her friend says eleven dollars, and Celia responds: Yo seré ambientalista y tó pero no voy a pagar $11 por un shampoo. “I am an environmentalist and all, but I am not going to pay $11 for shampoo.” In (5), Celia is known to be in favor of protecting the environment and avoiding the use of plastic, but her budget does not allow her to buy an $11 shampoo, thus the speaker can use y tó in this context. The main idea in (5) is that being environmentalist is contrasting with paying $11 for the shampoo, as opposed to comparing with properties from the speaker. The use of y tó in the previous example can be infelicitous if we replace pero “but” with y “and” as in (6). (6) #Yo seré ambientalista y tó y no voy a pagar $11 por un shampoo. #“I am an environmentalist and all, and I am not going to pay $11 for shampoo.”

The meaning of y tó  191 The main use of y tó in these examples seems to be licensed in a concessive context, and what is being contrasted is the content of the two propositions. We can also contrast properties as in (2)–(4), but this is not always the case as we saw in (5)–(6) where we can contrast propositions. The important characteristic to consider is the unexpectedness, coming from concessive contexts or from y tó itself. In other words, for the utterance with y tó to be felicitous, it is necessary to create an unexpected contrast, regardless of what is being contrasted. In this section I have shown one of the felicity conditions of y tó. In the following section, I will discuss a second felicity condition for y tó. 2.2  Beyond the expected: y tó + degrees Y tó can also be used with degrees. These contexts require a degree or an expectation that was exceeded by the speaker or another relevant agent in the context of utterance. The degree or expectation must be previously established, implicit or explicit, in the conversational context. In order to show this, let’s examine the following example: (7) Context: Elena and Miguel are going to Miami Beach for the weekend. They’re only going for a few days so they didn’t had to pack too much. However, Miguel always packs too many clothes that he does not need. After their weekend in Miami Beach, Elena is telling her friend about it and she says: Miguel trajo maleta y tó (!) “Miguel brought a suitcase and all.” In the previous situation, the implicit established degree is that both Elena and Miguel could have packed small or medium size bags for their trip since its duration was short. Therefore, since Elena was expecting that Miguel would just pack a few things, as opposed to an entire suitcase, this licenses the use of y tó in this context. The previous example becomes infelicitous if we try to follow it up with a statement like the one below in (8). Adding this follow-​up to the main utterance with y tó creates a contradiction resulting in infelicity. (8) #Miguel trajo maleta y tó, pero era necesario. #“Miguel brought a suitcase and all, but it was necessary.” The previous example showed a case where the expected degree was implicitly established in the context. In the following examples, I  present cases where the expectation was established explicitly in the context. In (9), we can observe that the use of y tó is not compatible with the contexts in (9a) and (9b), but can be felicitously used in (9c):

192  Nieves Rivera (9) Gabriela gave her son Luis a list of 5 items to buy for a dinner party: napkins, chips, wine, cups, and forks. Gabriela is telling a friend who is also organizing the dinner party: (a) Luis only bought 2 of the 5 the items on the list. #Luis compró dos de las cinco cosas y tó. #“Luis bought 2 of the 5 items and all.” (b) Luis bought the 5 things on the list (and nothing else). #Luis compró las cinco cosas y tó. #“Luis bought the 5 items and all.” (c) Luis bought the 5 things on the list (plus he brought beer). Luis compró las cinco cosas (más trajo cerveza) y tó. “Luis bought the 5 items (plus he brought beer) and all.” Y tó does not seem to work in (9a) because it was expected from Luis that he would buy all five items that were on the list. Since he did not meet the requirement, i.e. he only bought two items from the list, using y tó yields an infelicitous proposition. Similarly, in (9b) y tó is also infelicitous when the expected proposition is met. Even though Luis bought all five items that were on the list, which was expected, he did not go beyond that expectation. However, it is felicitous in a context like (9c) where he exceeded the items on the list. It seems that in general, y tó quantifies over the degree to which a property or a predicate obtains. Y tó is also compatible with unexpectedness. Without these elements available in the discourse context, y tó cannot be used felicitously. Now that I have presented the felicity conditions of this expression, we can move on to section 3 where I discuss the main characteristics of conventional implicatures. The next section provides the theoretical framework to analyze y tó, which I will do later in section 4.

3.  Conventional implicatures Implicatures are one of the most studied phenomena in semantics and pragmatics. By using implicatures, speakers can imply one thing by saying something different. There are different types of implicatures, those that are dependent on the conversational context, conversational implicatures, and conventional implicatures (CIs), which are part of the conventional meaning of the sentence, i.e. they are part of what is said. Let’s consider the following example: (10) Even Emma is bored in class. The utterance in (10) is true if and only if Emma is bored in class. In other words, the inference calculated by even is irrelevant to the truth conditions of the sentence in (10). However, the truth condition of Emma is bored in class is

The meaning of y tó  193 not the entire interpretation we get from this example. This example conventionally implicates that Emma is the least likely person to get bored in class. Thus, the resulting interpretation is a CI that results from the use of even. Following Potts (2005), and Amaral et  al. (2007) CIs have the following properties: i They contribute to the meaning of the utterance in a conventional way, and are non-​cancellable. ii They are not an at-​issue content of the utterance. However, they may take parts of the at-​issue content. iii They never take narrow scope with respect to operators in the at-​issue content. CIs are also invariant under plugs to presupposition. iv CIs are generally speaker-​oriented, except in direct quotations. For the purposes of this chapter, I will only discuss in detail properties (i–ii) and (iv) in the next section, but before moving on to that, let’s see how we can diagnose the status of CI of even. Conventional Implicatures properties: i Contribute to the meaning of the utterance in a conventional way, i.e. non-​cancellable: Even Ken knows it’s unethical, but that’s not surprising. ii They are different from at-​issue content; they may be relevant to the at-​ issue content but not necessarily: (Even) Ken knows it’s unethical answers the Question Under Discussion Does Ken knows it’s unethcial? iii They generally do not take narrow scope with respect to operators in the at-​issue content, and are invariant under plugs to presuppositions: Even Ken didn’t know it’s unethical. (Has the same CI). iv They are generally speaker-​oriented: From the speaker’s point of view, Ken is the least unlikely to know what’s ethical. The contradiction in the example in (i) shows that the implicature conveyed by even is not cancellable. In example (ii), (Even) Ken knows it’s unethical answers the Question Under Discussion with or without even, thus the CI associated with even is different from the at-​issue content, where the latter has to at least partially or indirectly answer the Question Under Discussion. Example (iii) shows that even when even interacts with plugs to presuppositions, such as negation, it remains invariant and the implicature is generated. Finally, in (iv) we see that CIs tend to be speaker-​oriented, i.e. they reflect the speaker’s point of view on the at-​issue content. These are considered to be the main properties of CIs. In the next section, I will apply each property to y tó and show that it can be analyzed as a CI.

194  Nieves Rivera

4.  Y tó as a conventional implicature 4.1  Non-​cancellability CIs are considered to be non-​cancellable. In other words, once a speaker uses a trigger word or expression, the inference that can be drawn from it cannot be cancelled or ignored. Looking again at the example in (10), repeated in (11) below, we see that the resulting sentence is infelicitous. (11) #Even Emma is bored in class, but that is not surprising. Following this diagnostic, we can now see if the meaning contributed by y tó is non-​cancellable. Let’s take as example the following context: (12) Linda needs a box from a shelf that she cannot reach, and she asks Luisa to get it for her because she is tall. Luisa goes to the shelf to try to get the box but she cannot reach it either and says to Linda: Soy alta y tó pero tampoco llego a la caja. “I am tall and everything, but I cannot reach the box either.” In the previous example, y tó can be used because it is unexpected that Luisa cannot reach the box because she is tall. If we try to get rid of the unexpectedness, and Linda can reach the box, then the example results infelicitous: (13) #Soy alta y tó pero llego a la caja. #“I am tall and everything, but I can reach the box.” We get the same result in the example (14), repeated from (7): (14) #Miguel trajo maleta y tó, pero era necesario. #“Miguel brought a suitcase and all, but it was necessary.” Example (14) shows that we cannot cancel the unexpectedness contributed by y tó that is required by the context. Therefore, y tó follows one of the main properties of CIs: y tó has a non-​cancellable meaning. 4.2  At-​issue content CIs are not part of the at-​issue content. Following Tonhauser 2012 at-​issue content can be defined as: At-​issueness: A proposition p is at-​issue if and only if the speaker intends to address the Question Under Discussion (QUD) via ?p. An intention to address the QUD via?p is felicitous only if ” ?p is relevant to the QUD The speaker can reasonably expect the addressee to recognize this intention.

The meaning of y tó  195 According to Amaral et al. 2007 “The felicity of an utterance necessitates that its at-​ issue content be relevant to the question under discussion.” Following these definitions, at-​issue content has to answer the relevant QUD, or at least partially answer it (Roberts 1996). In this sense, CIs do not have to answer the QUD, although they could. In the previous example (10), the relevant QUD could be Is Emma bored in class?, in which case just uttering Emma is bored in class, yes or no will suffice to answer the QUD. Similarly, Even Emma is bored in class also answers the QUD, but the additional inference we get from even does not. Therefore, we do not consider the CI to be part of the at-​issue content because the information it contributes is not relevant in the way at-​issue content is. If we apply the same reasoning to y tó, we observe that this expression also follows the same pattern. Let’s consider the examples I have presented throughout the chapter. (15) Yo seré ambientalista (y tó) pero no voy a pagar $11 por un shampoo. QUD: ¿Compraré el shampoo? “I am an environmentalist (and all), but I am not going to pay $11 for shampoo.” QUD: Will I buy the shampoo? (16) Miguel trajo maleta (y tó). QUD: ¿Qué trajo Miguel? “Miguel brought a suitcase (and all).” QUD: What did Miguel bring? (17) Luis compró las cinco cosas (y tó). QUD: ¿Qué compró Luis? “Luis bought the 5 items (and all.).” QUD: What did Luis buy? In all of these cases, the QUD can be addressed without using y tó. Thus, the inference that y tó contributes is not part of the at-​issue content. Instead, it serves as a “comment” on the utterance from the speaker. Notice that even the speaker’s relationship to the at-​issue content can be conveyed using other means such as prosody. The contributions of prosody are worth exploring, but since it does not fall under the scope of this chapter, I will leave that for future research. In the next section I will discuss how CIs, in this case y tó, usually contribute information about the speaker’s relationship/​attitude to the at-​issue content, and more importantly how this relationship does not have to be only from the speaker. 4.3  Speaker-​orientation vs. doxastic-​anchoring According to Potts (2005), CIs are speaker-​oriented, i.e. except for special circumstances only the speaker of the utterance is committed to the inference

196  Nieves Rivera contributed by the CI. This seems to be true of some CIs. However, there are cases where another relevant agent is committed and not the speaker. In the example below, from Amaral et  al. (2007), Sheila is the one committed to Chuck being a sweetheart, instead of the speaker. (18) Sheila believes that Chuck, a sweetheart if she ever met one, is fit to watch the kids. Following this example, I  agree with Amaral et  al. (2007) that speaker-​ orientation does not seem to fully account for this property of the CIs. On the other hand, doxastic anchoring seems to be more accurate for this purpose. Amaral et al. 2007 defines doxastic anchoring as “the belief in the truth of the proposition expressed is being attributed to a specific agent. The speaker is the default, but in some contexts it can be anchored to another agent”. In other words, given the appropriate context we can shift the doxastic ground for some CIs (Amaral et al. 2007). More specifically, CIs are indexed to the attitude of a doxastic agent, whose point of view is salient at the utterance time. In the case of y tó while it is mostly anchored to the speaker in (19), it can also be anchored to another agent as in (20), given the appropriate context. (19) Yo seré ambientalista y tó pero no voy a pagar $11 por un shampoo. “I am an environmentalist and all, but I am not going to pay $11 for shampoo.” (20) No fui a Miami con ellos, pero Elena dijo que Miguel trajo maleta y tó. “I did not go to Miami with them, but Elena said that Miguel brought suitcase and all.” In (20), Elena is the one committed to Miguel’s exaggeration, not the speaker. Thus, doxastic anchoring is a more general term and accounts for both cases, as opposed to speaker-​orientation which only suggests one type of anchoring. Once more, y tó follows another property of CIs. 4.4  Invariant under presupposition plugs One final property of CIs is that they generally do not take narrow scope with respect to operators in the at-​issue content. Thus they are invariant under plugs to presuppositions as in Even Ken didn’t know it’s unethical (has the same CI). In the case of y tó, the same can be observed in (21) and (22) adapted from (7): (21) Miguel trajo maleta y tó? “Did Miguel bring a suitcase and all?”

The meaning of y tó  197 (22) Quizás Miguel trajo maleta y tó. “Maybe Miguel brought a suitcase and all.” In other words, y tó can be embedded under presupposition plugs, such as an interrogative (21) or a modal (22) and the unexpectedness conveyed by y tó can still project. These examples illustrate that y tó behaves as a CI in that the meaning contributed by y tó is invariant under presupposition plugs. In this section, I have discussed the properties of CIs and applied them to the expression y tó, showing that it can be analyzed as a CI. The next section, summarizes the main finding and conclusions.

5.  Conclusion As previously mentioned in the introduction, Puerto Rican Spanish remains understudied within semantics and pragmatics. Thus the goal of this study was to provide an analysis of a Puerto Rican expression that is widely use by native speakers of this dialect. Throughout, the chapter we have seen that y tó conveys a meaning of surprise. In the cases were there is unexpectedness, y tó is compatible when contrasting two or more properties or propositions. Moreover, y tó is also felicitous when an expectation or degree that was established in the context of utterance is exceeded. In this study, I argue that y tó is a CI. I show this by discussing the different properties of CIs, and how each and one of these can be applied to the expression y tó. I showed that the inference contributed by y tó is non-​cancellable, it also not at-​issue content, and while it is mostly speaker-​oriented, it can also be anchored to another agent with a salient point of view in a given context. Overall, this study constitutes the first pragmatic analysis of the expression y tó. While I argue that this expression is a CI, more research needs to be done including other variables such as prosody. A perception study seems the ideal next step to explore speakers’ acceptability of y tó in the felicity conditions described in this chapter.

Note 1 Thanks to the anonymous reviewer for pointing out that this construction also follows the felicity conditions.

References Amaral, P., Roberts, C., & Smith, E.A. (2007). Review of the logic of conventional implicatures by Chris Potts. Linguistics and Philosophy, 30(6), 707–​749. Cee_​gonz. (2018, Oct 15). Mi carro dice “check engine” hace tanto tiempo que pa mi que se acostumbró y to [Twitter post]. Retrieved from

198  Nieves Rivera König, E., & Siemund, P. (2000). Causal and concessive clauses: Formal and semantic relations. Topics in English Linguistics, 33, 341–​360. Potts, C. (2005). The Logic of Conventional Implicatures. New  York:  Oxford University Press. Roberts, C. (1996) Information structure:  Towards an integrated formal theory of pragmatics. In J.H. Yoon & A. Kathol (eds.), OSUWPL, Vol. 49:  Papers in Semantics, 1996. The Ohio State University Department of Linguistics. Available in updated form on the Semantics Archive:  Tonhauser, J. (2012). Diagnosing (not-​) at-​issue content. Proceedings of Semantics of Under-​represented Languages of the Americas (SULA), 6, 239–​254.

10  Exploring extended focus and meaning in Chilean Spanish intonational plateau contours Brandon M.A. Rogers, Rajiv Rao and Matthew Burner

1.  Introduction Over roughly the past two decades, studies on the intonation of Spanish have been on the rise. While much of this research, as well as cross-​linguistic research on intonation, has been conducted on the ramifications of F0 modifications in human speech at both the phonetic and phonological levels using laboratory speech, some researchers (e.g., Hirschberg, 2002; Prieto, 2015) have asserted that when established cross-​linguistic interpretations of intonation have been examined in more natural speech, researchers are frequently presented with data that runs counter to certain long-​held assumptions in previous literature. Hirschberg and Prieto agree that intonational meaning cannot be fully and accurately understood if studies do not take into account contextual and pragmatic data. As a result, more studies have sought to connect intonational behavior and pragmatic meaning (for studies on Romance, see e.g., Hübscher, Borràs-​Comes, & Prieto, 2017; Nadeu & Prieto, 2011; Rao, 2006, 2013; Vanrell et  al., 2013). Additionally, as has been shown for other Spanish linguistic phenomena, such as segments, intonational phenomena and behaviors that are frequently observed in laboratory speech (e.g., downstepping and final lowering), do not always appear in spontaneous speech with the same consistency (e.g., Face, 2003; Rao, 2006). Likewise, some intonational phenomena, such as deaccenting, are more readily observed in spontaneous speech than in controlled conditions (Face, 2003; Rao, 2006). Thus, by studying intonation and its potential meanings with primarily laboratory data, a broader scope of pragmatic uses is potentially obscured. The aim of the present study is to work toward filling the research gaps just outlined by exploring intonational meaning in the spontaneous speech of Chilean Spanish and expanding upon the connection between intonational phonology and pragmatics in Spanish. This understudied variety of Spanish is an intriguing object of inquiry primarily due to the “hat patterns” or “intonational plateaus” (i.e., rise, extended plateau, fall) that occur in spontaneous declaratives (Rogers, 2013). The current study employs spontaneous speech data of 530 intonational plateaus from 40 speakers recorded in Chile

200  Rogers, Rao and Burner to address the pragmatic meaning behind and phonological repercussions of manifestations of the plateaus. We argue that the raised F0 Chilean plateaus, which are characterized by a rapid speech rate, are a form of extended focus through which speakers signal the salience of an entire (often syntactically complex) idea, or strings of prosodic words that together form a phrase-​level unit that is one highlighted entity. This proposed analysis relies primarily on narrow focus in Chilean Spanish (and other varieties) being manifested at a higher F0 level and with words carrying the L+H* pitch accent (Ortiz, Fuentes, & Astruc, 2010), the parameters of Gussenhoven’s (2002) Effort Code related to increased articulatory energy being tied to more F0 excursions, and thus, emphasis (see, e.g., Chen, Gussenhoven, & Rietveld, 2002; House, 2004; Rao, 2013; Vanrell et al., 2013), and Cho and Flemming’s (2015) observations on speech rate effects driving the appearance of high plateaus in some Korean accentual phrases. The remainder of this chapter is divided as follows: Section 2 covers previous literature on Spanish intonation and intonational meaning in general as well as intonational plateaus in Spanish and other languages; Section 3 describes the collection and analysis of spontaneous data for the present study; Section 4 describes the types of plateaus found; Section 5 provides a phonetic/​phonological and pragmatic account of the results in Section 4; and Section 6 concludes the chapter with a summary of the contribution of this research and future avenues of related inquiry.

2.  Literature review 2.1  Background on Spanish intonation and meaning One of the cognitive/​linguistic roles that intonation contributes to in human language is the parsing of an utterance into smaller “chunks” of information (D’Imperio et al., 2005; Ladd, 2008; among others). This division of information is regulated by specific rules (Gussenhoven, 2004; Ladd, 2008). These rules have been used to construct what is known as the Prosodic Hierarchy, as seen in (1), wherein an utterance can be divided into smaller constituents. Each level of the hierarchy governs all the levels below it and is simultaneously governed by those levels above it (Gussenhoven, 2004). Of particular interest to the current investigation are the levels of Intonational Phrases (IPs, or major phrases), Intermediate Phrases (ips, or minor phrases), and Prosodic Words (PWs).1 Studies on Spanish intonation have suggested that the absolute limit for PWs in an ip is four, with the ideal number being two (e.g., Prieto, 2006; Rao 2007). These limits are thought to be determined by cognitive processes and the intended meaning that a speaker attaches to a given utterance (Christophe et  al., 2004). Speech rate also plays a role, as faster speech favors a reduction in the number of minor phrases, resulting in more PWs/​minor phrases (Prieto, 2006). It must be noted, though, that the speech

Exploring extended focus and meaning  201 samples that have led to these ostensible PW limits in Spanish have been largely controlled and produced in laboratory conditions. (1) Prosodic hierarchy IP Intonational Phrase ip Intermediate Phrase PW Prosodic Word F Foot σ Syllable The IP, ip, and PW are the most relevant to the current study. The IP is non-​isomorphic with syntactic structure, is a unit that possesses meaning, and is a terminal point of discourse (Rao, 2009). Studies on Spanish phrasing have shown that boundaries are signaled by audible pauses (Rao, 2010) at their right edge, as well as lengthening effects (D’Imperio et  al., 2005; Elordieta et  al., 2003; Prieto, 2006; Rao, 2009, 2010). IPs usually indicate the end of a complete or more complex thought. The ip is a shorter phrase embedded within an IP that often contains incomplete ideas. Their right edges are cued by F0 rises to the final syllable of words, an F0 rise to a sustained level, final lengthening, a reset in F0 level, and short pauses (D’Imperio et al., 2005; Elordieta et al., 2003; Prieto, 2006; Rao, 2007, 2010). The level of F0 is particularly important to the current analysis; as we will see, drastic reset from a relative low (not always present) to sustained high is the F0 activity that characterizes the plateau patterns in our data. Since ip boundary cues beyond the F0 modifications at this juncture are absent in our data, we expect the plateau to be its own ip and any preceding low to be its own ip. When falls from plateaus are exhibited, they are typically associated with movement toward a low IP boundary rather than forming a separate ip. Furthermore, high tonal ip boundaries communicate a thought that is incomplete, with more information on its way, while the less common low variants are attested in relatively infrequent constructions such as dislocations.2 Below both phrase-​level constituents is the PW, which is the level of the prosodic hierarchy affiliated with perceptual prominence in content words. A word exhibiting an acoustic cue to stress (e.g., accent, duration, intensity) in its lexically specified stressed syllable is considered a PW. 2.2  Chilean Spanish intonation Chilean Spanish intonation has been relatively understudied. Several studies have examined intonational phenomena such as vocatives, orders and petitions, tonal clash, the interaction of intonation and word order, and the pragmatic functions of intonation (Atria, 2009; Cepeda, 1997, 2001; Cid & Ortiz, 1998; Fuentes, 2012; Silva-​Corvalán, 1983; Véliz, 2001). Ortiz, Fuentes, and Astruc (2010) present the most detailed, encompassing, and recent description of Chilean Spanish based on the

202  Rogers, Rao and Burner SP-​ToBI and AM conventions found in Estebas & Prieto (2008). They report the existence of two main monotonal and three bitonal pitch accents in Chilean Spanish namely: L*, H*, L+H*, L+>H*, and H+L*. Their results report that L+H* and L+>H* are the two most common pitch accents and that, depending on the type of utterance, L+H* has three different levels of prominence: L+!H*, L+H*, and L+¡H*. The L+H* realization was found in nuclear position of narrow focus statements of the obvious as well as in a number of different interrogative contexts. The downstepped version was found in neutral statements, such as broad focus declaratives, while the upstepped version manifested itself in contradiction and emphatic statements, as well as in invitational and rhetorical questions. In other words, the upstepped version of the L+H* pitch accent is used to call attention to certain specific details in an utterance, while the downstepped version appears to be the preference for information that speakers do not deem outstanding or more important than the rest of a given utterance. They give no indication as to any specific level of prominence for L+>H* but the authors do report that this specific pitch accent delay was not nearly as phonetically or perceptually salient as it has been found to be in other dialects such as Castilian Spanish. 2.3  Intonational plateaus cross-​linguistically and in Spanish The phonetic forms of intonational plateaus are not unique to any one particular language or dialect of Spanish, although the communicative function of the form varies from one linguistic community to the next. Cross-​ linguistically, plateaus have been documented in the intonation of languages such as English (Ladd, 2008), Guarani (Clopper & Tonhauser, 2011) K’iche’ (Yasavul, 2013), Tyrolean German (Barker, 2005; Féry, 1993), Basque (Elordieta, 2003), and Korean (Cho & Flemming, 2015). The phonetic/​phonological analysis of the final study is particularly noteworthy for the current study because the authors observe that high plateaus within accentual phrases in Korean are a function of time pressure (i.e., faster speech rate; as seen, also a feature of Chilean plateaus), which leads to compression of F0 movements; that is, the Ls of HLH sequences are undershot, resulting in a high plateau. In Spanish, plateau patterns have been reported in mainly questions and cases of truncation (e.g., see chapters in Prieto & Roseano, 2010). In Dominican Spanish (Willis, 2010), Puerto Rican Spanish (Armstrong, 2010), Chilean Spanish (Ortiz, Fuentes, & Astruc, 2010) and Venezuelan Andean Spanish (Astruc, Mora, & Rew, 2010), plateau patterns have been documented in yes-​no questions, while they appear in wh-​questions in Canarian Spanish (Cabrera & Vizcaíno, 2010) and L2 Galician Spanish (Pérez, 2015). Cid and Ortiz (1998) also briefly make mention of similar plateau patterns in varying pragmatic contexts in Chilean Spanish. Despite appearing in different

Exploring extended focus and meaning  203 pragmatic contexts than the plateaus in the aforementioned dialects, they share very similar phonetic forms with initial rises, final falls, and intermediate high portions maintained over a varying quantity of content. For Buenos Aires Spanish, Gabriel, Feldhausen and Pešková (2011) report the presence of brief plateaus in controlled declaratives. These plateaus, however, share very little phonetic similarities with those reported in other studies, as they only are documented as lasting for the duration of the prosodic word that initiates their initial rise. As first described by Rogers (2013), Chilean Spanish declarative plateaus begin with an optional valley in which all content is realized at the same low F0 level, and can be as short as a few syllables or may extend for multiple stressed and unstressed words. The valley portions end on a sudden F0 rise on the final stressed syllable of the final valley-​internal word. The rise continues to a higher F0 level, which can last for as little as two syllables or can extend for multiple function and content words. The contour ends on a final F0 drop. In most cases this drop occurs on either the final stressed syllable or the absolute final syllable of the final plateau-​internal word. In some cases, speakers continue this descent to the end of the utterance through multiple stressed and unstressed syllables and/​or words. In a smaller number of cases, the plateaus end with no drop. Finally, in a number of other cases, speakers jump to an even higher F0 level on the final stressed syllable within the high portions, and initiate the final drop on the following content. Most plateaus occur just before a phrase-​final descent, while speakers occasionally produce them in more internal positions of the phrase. These specific plateau patterns also vary greatly in length, from a few words to entire utterances, and appear to frequently surpass previously established PW limits for Spanish ips. As previously mentioned, studies on Spanish have noted that the allowable number of PWs per ip is between two and four (Prieto, 2006; Rao, 2006, 2007), with four being the maximum and two being ideal in terms of restrictions on prosodic well-​formedness (Nespor & Vogel, 1986; Selkirk, 1984; Truckenbrodt, 1999). However, the excursions analyzed by Rogers (2013) demonstrate that several portions, and especially the plateau, can house between one and six PWs. Some variants within this range of PWs violate this two to four PW per ip norm. It appears that the primary reason for compressing an above-​ normal number of PWs into phrases is speech rate; that is, the plateau portions are realized in a more rapid fashion (in syllables/​second) than both the content in relative F0 lows that precede them and typical speech rates cited in previous literature (e.g., see Goldman-​Eisler, 1961; Kowal, Wiese, & O’Connell, 1983; Pellegrino, Coupé, & Marsico, 2011). Figure 10.1 illustrates key concepts from Rogers’ (2013) study through an example of a short valley followed by a lengthy plateau containing six PWs. Following Figure 10.1 is more context from which the plateau was taken. The plateau portion is in bolded and capitalized.

204  Rogers, Rao and Burner

Figure 10.1 Short valley and prolonged plateau Source: Rogers, 2013

¿ Qué opino de la guerra en Irak? Yo odio la guerra. [pause] Encuentro que es horrible. Que es por intereses económicos [pause]…que hayan guerras. Porque en el FONDO SON INTERESES ECONÓMICOS LOS QUE HACEN LAS GUERRAS “What do I think about the war in Iraq? I hate war [pause]. I think it’s horrible. It’s because of economic interests [pause] that there are wars. Because at the end of the DAY ECONOMIC INTERESTS ARE WHAT CREATE WARS.”

3.  Methodology 3.1  Data collection and plateau identification The data for the current study were gathered from 40 spontaneous sociolinguistic interviews carried out in the regions in and around Concepción and Temuco, Chile. Each interview lasted between 15–​40 minutes. The data from 20 female and 20 male speakers ranging in age from 18 to 55 years were used for the current study. The participants interviewed in Concepción came specifically from downtown Concepción and the nearby neighborhoods of Candelaria, Michaihue, Lomas Coloradas, Hualpén, and La Villa San Pedro. Additionally, one of the participants had recently moved to Concepción from the small town of

Exploring extended focus and meaning  205 Lebu, which is situated about 90 minutes outside of the Concepción. The Concepción interviews took place primarily in the participants’ homes or the homes of friends or family members of the participants. Several were recorded in local church buildings and rooms in a local neighborhood recreation center. All participants from the Concepción region were the family members, friends, or the friends of friends or family members of the first author of the current study. All interviews realized in Temuco were recorded in a quiet room at La Universidad de la Frontera (UFRO). Participants were all university students or employees and grew up in and around Temuco, with the exception of two, who grew up in Santiago but recently moved to Temuco to study at UFRO. 3.2  Types of data analyzed The contours analyzed were extracted using Praat (Boersma & Weenink, 2015) and were examined from the perspective of the Prosodic Hierarchy to investigate the possible conditions driving the different categorizations of the plateau pattern.3 Likewise, Prosodic Word (PW) thresholds were measured in the high portions. Recall that, as mentioned in Section 2.1, the lack of cues to previously cited ip boundaries, outside of changes to F0 level from an optional relative low to a relative high plateau, warrant each relative high and low to constitute its own ip. Also, in the context of the current study, a PW should be understood to be a stressed word (Quilis, 1999).

4.  Results 4.1  Plateau categorizations In all, 530 plateaus were identified and extracted from the interviews analyzed, for an average of just over 13 plateaus per speaker. The corpus used for the current investigation was much more robust than that used by Rogers (2013), and along with the plateau pattern that he describes, the present study identified six additional manifestations of plateaus as produced by the participants. The remainder of this section contains subsections dedicated to a description and illustration of the features of each plateau type. Below each representative F0 contour, we provide contextual information from the sociolinguistic interview from which each contour was extracted, which will help us interpret the communicative reason behind the production of high plateaus. Finally, before moving into details about each plateau type, it is crucial to mention that, as anticipated based on the summary of Chilean plateaus in Section 2, our plateaus were produced at a rapid speech rate, both relative to preceding F0 lows (when present) and the syllables/​second values cited in previous studies (e.g., Kowal, Wiese, & O’Connell, 1983; Pellegrino, Coupé, & Marsico, 2011); that is, the ratio of plateau rate to preceding low rate, when

206  Rogers, Rao and Burner

Figure 10.2 Example of a Type 1 contour

comparable, is approximately 2:1, and rates are typically in the 8–​10 syllable/​ second range, but reach as high as 11 syllables/​second in some cases. 4.1.1  Type 1 Type 1 plateaus were the same patterns described by Rogers (2013) and were the most common realizations, making up 57.2 percent of all productions (303 of 530). As previously stated, Type 1 plateaus may or may not be preceded by a low, tonal valley and begin on a sharp rise that can extend for the duration of one stressed syllable, or for several words, both stressed and unstressed. Upon reaching the target high tonal level, their length varied significantly, sometimes extending for as few as 1–​3 syllables, and as much as 10 unstressed words and a number of unstressed words. The average PW count for the plateau portion of Type 1 contours was 3.276 PWs and the average overall word count for the high portions was 4.961 words. The greatest number of PWs observed in a single Type 1 high portion was ten. Finally, Type 1 plateaus ended in a drop on the final stressed syllable of the corresponding utterance. Figure 10.2 is an example of a Type 1 plateau from the current data set. Following Figure 10.2 is the context from which the plateau was taken with the plateau portion bolded and in caps. Interviewer:  ¿ Cómo era Temuco cuando eras niña? ¿ Era diferente o era—​ Speaker:  Mmm—​ Interviewer:  —​más o menos igual? Speaker:  —​es que no RECUERDO MUY BIEN COMO ERA TEMUCO ‘INTERVIEWER:  What was Temuco like when you were a child? Was it different or was it—​

Exploring extended focus and meaning  207 SPEAKER: Mmm—​ INTERVIEWER:  —​more or less the same? SPEAKER:  —​the thing is that I don’t REMEMBER

VERY WELL WHAT

TEMUCO WAS LIKE 4.1.2  Type 2 Type 2 plateaus were very similar to Type 1 plateaus, with one small difference: both began in the sharp rise that may or may not have originated in a preceding low valley and both contained varying amounts of stressed and unstressed content in their respective high portions. However, Type 2 plateaus maintained the high tonal portion until their absolute final syllable, many times unstressed, instead of dropping on the final stressed syllable. The average PW content for the high portions was 3.168 PWs and the overall average word content was 4.974 words. The greatest amount of PWs observed in a single Type 2 high portion was 7. Type 2 productions only made up 6.6 percent of all tokens analyzed (35 of 530). Figure 10.3 illustrates a Type 2 plateau from the current data. Speaker:  Sacando cosas, sacando cosas y fuimos al centro. [pause] Estaba yo, me quedé en el centro de Concepción [pause] y se me ocurrió ir a Johnson’s [pause] porque ESTABA LLENO DE ZAPATILLAS TIRADO EN LA CALLE ‘SPEAKER:  Taking things, taking things and we went downtown. [pause] I was, I stayed in downtown Concepción [pause] and I decided to go to Johnson’s [pause] because THERE WERE A BUNCH OF SNEAKERS LYING IN THE STREET

Figure 10.3 Example of a Type 2 contour

208  Rogers, Rao and Burner 4.1.3  Type 3 Type 3 plateaus were similar to both Type 1 and Type 2 productions with regards to how they began their high portions. Where Type 3 productions deviated from Type 1 and Type 2 productions is that they ended at a high tonal level and never dropped. Type 3 plateaus were produced more frequently at different utterance-​internal junctures. The average PW count for the high portions was 3.132 PWs and the overall average word content was 4.651 words. The greatest number of PWs observed in a single Type 3 high portion was five. In all, 2.8 percent of all tokens (15 of 530) were classified as Type 3, an example of which is illustrated in Figure 10.4.4 Speaker:  Mis pobres perros. Me dio tanta pena. [laugh] Interviewer:  …pobre perros, si poh. Me dijo que ni siquiera pon-​ponen flores ahí… Speaker:  No, no Interviewer:  …se roban las flores y la-​ Speaker:  …es que, no, no me QUIERO ATREVER LA VERDAD PO porque uno no sabe a veces ‘SPEAKER:  My poor dogs. I felt so bad. [laugh] INTERVIEWER:  …poor dogs. He told me that you don’t even pu-​put flowers there… SPEAKER:  No, no INTERVIEWER:  …they steal the flowers and the—​ SPEAKER:  …the thing is, I  don’t, don’t DARE TO BE HONEST because you never know sometimes’

Figure 10.4 Example of a Type 3 contour

Exploring extended focus and meaning  209

Figure 10.5 Example of a Type 4 contour

4.1.4  Type 4 Type 4 plateaus differed from all plateaus in their final drops. Whereas the final drops of previously described plateaus contained a limited number of syllables from the same word, the final drop on Type 4 plateaus contained up to two stressed words and a varying number of unstressed words, as seen in Figure 10.5. The average PW content for the high portions was 3.633 PWs and the overall average word content was 5.691 words. The greatest amount of PWs observed in a single Type 4 high portion was 11. In all, Type 4 plateaus made up 14.5 percent of all tokens (77 of 530). Speaker:  …una muy buena situación económica a través del fraude, etcétera uno lo desconoce. [pause] Pero uno, yo al menos [pause] ESE ESPÍRITU REVOLUCIONARIO YA QUEDÓ ATRÁS ‘SPEAKER: …a very favorable economic situation through fraud, etcetera, that is unknown. [pause] But you, me at least [pause] THAT REVOLUTIONARY SPIRIT IS A THING OF THE PAST 4.1.5  Type 5 Type 5 plateaus showed a gradual downward slant toward a final drop in the high portion, similar to plateaus documented for German by Féry (1993) and Barker (2005), and made up 3.7 percent (20 of 530) of all tokens. Despite the gradual decline in the intonational contour toward the end of a phrase or utterance, the extensively documented trend of Spanish downstepping (e.g., Face, 2008; Prieto, Shih, & Nibert, 1996; Willis, 2010; among others)

210  Rogers, Rao and Burner

Figure 10.6 Example of a Type 5 contour

did not occur, as shown in Figure 10.6. The average PW content for the high portions was 3.923 PWs and the overall word content average was 5.887 words. The greatest number of PWs observed in a single Type 5 high portion was six. Speaker:  El, lo del deber cívico. Dejó de de respetarse en cosas tan simples. Dejó de construirse la sociedad. Por ejemplo, em, que una persona adulta—​ Interviewer:  mhm Speaker:  de tercera edad se sube a la micro y NADIE ES CAPAZ DE DARLE EL ASIENTO ‘SPEAKER:  The, the whole civic duty thing. We stopped respecting each other in the simplest of ways. Society stopped building itself up. For example, an adult—​ INTERVIEWER: mhm SPEAKER:  —​an elderly adult gets on the bus and NO ONE MAKES AN ATTEMPT TO GIVE THEIR SEAT UP OF HIM/​HER’ 4.1.6  Type 6 Like all of the previously mentioned classifications, Type 6 plateau contours may or may not have been preceded by a valley. However, they stood out from other patterns because of the nature of their final portions. The final stressed syllable of Type 6 contours rose above all of its preceding high tonal content. After the rise on the final stressed syllable, a fall occurred on the remaining

Exploring extended focus and meaning  211 unstressed syllables until the end of the utterance. The average PW content for the high portions was 3.617 PWs while the overall word count average was 5.484 words. The greatest number of PWs observed in a single Type 6 high portion was nine. Figures  10.7 and 10.8 illustrate examples of Type 6

Figure 10.7 Example of a Type 6 contour

Figure 10.8 Example of a Type 6 contour

212  Rogers, Rao and Burner contours. In all, 15.3 percent of all productions (81 of 530) were categorized as Type 6. Speaker:  pero ahí va a ser como demasiado tarde. [pause] Eso, por ESO PUEDE IR DECAYENDO ‘SPEAKER:  …but by then it’ll be like too late. [pause] That, that’s WHY IT CAN GET WORSE’ Speaker:  …o a una hora cuarenta acá de Concepción, en tren. Sí estuve allá viviendo hasta los diecisiete años. Interviewer:  Diecisiete años Speaker:  Después por tema de universiDAD Y TODO ME VINE A VIVIR (NN) YO SOLO ACÁ (EN) CONCE Y MIS PAPÁS QUEDARON  ALLÁ ‘SPEAKER:  …o an hour and forty minutes from here in Concepción, by train. Yeah, I was living there until I was seventeen. INTERVIEWER:  Seventeen years old SPEAKER:  After that because of the universiTY AND EVERYTHING, I  CAME HERE MM TO CONCEPCIÓN TO LIVE ON MY OWN AND MY PARENTS STAYED BEHIND’ 4.1.7  Type 7 Contours that did not occur in utterance-​final position and instead were produced at an utterance-​internal juncture were sub-​categorized as Type 7 patterns. Additionally, Type 7 plateaus were also given a second categorization based on how they ended. Figures 10.9 and 10.10, both followed by the contexts in which they were produced, show Type 7 plateaus with Type 1 endings and the content that follows, to illustrate that they occur at utterance-​ internal junctures. Note that the utterance in Figure 10.9 ends in a plateau as well, as indicated in bold in the context. Speaker:  ee, a ver. CuanDO HAY JUNTA FAMILIAR, eee, siempre son los aniversarios de matrimonio a mis abuelos, o cuando ELLOS ESTÁN DE CUMPLEAÑOS. ‘SPEAKER:  ee, let’s see. WHEN THERE IS A FAMILY GET-​TOGETHER eee it’s always the marriage anniversaries of my grandparents, o when THEY ARE HAVING A BIRTHDAY’ Interviewer:  …pero si alguien dice “weón”, ¿ es mal hablada esa persona? Speaker:  No, no, no, no Interviewer:  No Speaker:  No PORQUE UNA PERSONA CON MÁS EDUCACIÓN IGUAL te puede echar la chuchá de repente. ‘INTERVIEWER:  …but if someone says “weón”, are they considered rude?

Exploring extended focus and meaning  213 SPEAKER:  No, no, no, no INTERVIEWER: No SPEAKER: No, BECAUSE

SOMEONE WITH MORE EDUCATION can

talk shit to you as well

Figure 10.9 Example of a Type 7 subcategorized contour

Figure 10.10 Example of a Type 7 subcategorized contour

214  Rogers, Rao and Burner 4.1.8  Type 8 In several cases, speakers broke up or interrupted the high portion of a contour through various means, such as taking a breath, stuttering, or pausing. However, despite these interruptions, no pitch reset occurred and the speakers continued to maintain themselves at the same high tonal level after the interruption, until finishing the entire idea that they had originally intended to communicate through the high portion of the contour. These types of productions were subcategorized as Type 8 patterns. Additionally, Type 8 subcategorizations were further categorized based on how they ended. In other words, if a broken plateau ended by dropping on the last stressed syllable it was considered a Type 8 contour with a Type 1 ending. If the drop occurred on the absolute final syllable it was categorized as a Type 8 contour with a Type 2 ending, etc. Figure 10.11, the same plateau seen previously in Figure 10.8, shows a Type 8 contour with a Type 4 ending and Figure 10.12 shows another Type 8 with a Type 2 ending. Context for Figure  10.12 is provided as well. Figure 10.11 is the same plateau as Figure 10.8, where the original context is provided as well (see p. 212). SPEAKER: 

Eso responde al nivel de vida que se tiene en Chile. Porque es súper complicado tener acceso a esas cosas yo creo [pause] para la mayorÍA DE LAS FAMILIAS [pause] DE [pause] ESTRATOS MÁS BAJOS ‘SPEAKER:  That reflects the quality of living in Chile. Because getting access to those things is super complicated [pause] for the majorITY OF [pause] OF [pause]LOWER-​INCOME FAMILIES’

Figure 10.11 Example of a Type 8 subcategorized contour with a Type 4 ending

Exploring extended focus and meaning  215

Figure 10.12 Example of a Type 8 subcategorized contour with a Type 2 ending

5.  Discussion: an account of why and how Chilean Spanish plateaus are manifested Given the variation in length and structure of the types of plateaus we have documented, it seems that prosodic phrasing rules (e.g., the constraints related to length outlined in Section 2)  are not the driving force behind Chilean Spanish plateaus. While the majority of plateaus contained 1–​5 PWs, 34 cases had 6 or more PWs in a phrase, indicating that length conditions do not appear to be dictating the overall structure of the plateaus. Likewise, the observed behaviors in the data concord with Prieto (2006) in that speakers show a tendency to reduce the overall number of phrases in cases of more rapid speech. Given these observations, a more satisfactory explanation outside of previously established conventions merits exploration. Due to parallels between our data and those seen in Cho & Flemming’s (2015) analysis of Korean accentual phrases, in particular, a faster speech rate resulting in high plateaus, we adapt aspects of their phonetic/​phonological proposals to account for the hows and whys of our data. Furthermore, based on the types of plateaus summarized and the examples of each one that we have shown, we view the pragmatic function of Chilean Spanish plateaus in our spontaneous data as one of highlighting a string of PWs, which are entire ideas or thoughts produced with extra emphasis and a heightened sense of emotion. If the contexts below each figure in Section 4 were not accompanied by corresponding examples of plateaus, it would be easy to construe all of the contextual information as comprised of broad focus declaratives; that is, there are not obvious reasons to create narrow focus (e.g., contrast)

216  Rogers, Rao and Burner other than speakers choosing to draw attention to a stretch of speech because they consider it particularly meaningful. In the remainder of this section, we will explain how increased speech rate drives the shape and representation of plateaus, as well as why this strategy is an effective and efficient means of conveying focus across a longer stretch of information. To initiate our proposal, we assume that Chilean Spanish declaratives are comprised of a series of L and H phonological targets, as is generally assumed across Spanish, with the narrow focus pitch accent being L+H* (Ortiz, Fuentes, & Astruc, 2010). We also view increases in F0 level as a strategy to communicate narrow focus (see, e.g., Face, 2001, among others), which is reinforced by the upstepped version of L+H* being used in emphatic contexts (Ortiz, Fuentes, & Astruc, 2010). Figure 10.13 provides a schematic of this low–high sequence at a normal speech rate; each L+H would be associated with a content word and the final L would typically be the L percent boundary that generally concludes a declarative (our Type 3 contour would require this final target to be H). The arrows in Figure  10.13 point to modifications to the typical low–high contour that are induced when speech rate increases; the inward-​pointing horizontal lines indicate the compression of F0 movements that would take place if more content were produced in less time, and the upward-​pointing vertical lines allude to effects on the lower end of the F0 spectrum as a result of the compression created by reduced time. Figure 10.14 is provided in order to demonstrate more clearly the effects of time-​induced compression; here, we can see that the contour has shortened and the relative lows are being squeezed upward. We only include this figure to facilitate an understanding of the overall transition from a low–high pattern to a plateau; that is, the figure is not based on some sort of intermediate realization attested in our data. Finally, Figure 10.15 is a schematic of the documented trends in our data, where relative lows are similarly aligned but occur at a higher level (i.e., the L target is not disappearing), yielding a stretch of high F0 plateau. Motivated by Cho & Flemming’s (2015) insights, we ascertain that the change

Figure 10.13 Schematic of typical low-​high trend in Spanish at a normal speech rate. The horizontal arrows point out possible time pressure caused by an increased speech rate, and the vertical arrows indicate the upward movement of valleys resulting from rapid-​speech-​based compression

Exploring extended focus and meaning  217

Figure 10.14 Schematic of an intermediate phase of compression, showing changes away from the shape of Figure  10.13 and toward the plateau in Figure 10.15

Figure 10.15 Schematic of undershoot of medial L targets due to time-​pressure-​ based compression of the series of F0 movements, resulting in a high F0 plateau like those observed for Chilean Spanish

between Figure  10.13 and 10.15 is caused by speakers undershooting L targets, resulting in a series of Hs that form a plateau, with longer plateaus having more cases of undershoot. What our claim of undershoot reveals is that when speakers choose to extend focus, they speed up their speech, and in doing so, they consider it more important to keep their alignment of L targets in order rather than producing the drastically sloped up and down movements required to hit L targets when faced with a decreased window of time within which movements can be realized. The discussion to this point begs the question: Why do speakers use fast speech and undershoot of L targets and resulting high plateaus as a strategy to communicate the discursive importance of larger chunks of information? We suggest that an answer to this question can be formulated through reference to Gussenhoven’s (2002) Effort Code as well as Zipf’s (1949) Principle of Least Effort. From the former, we understand that expanding F0 excursions positively correlate with articulatory effort. Such increases are also tied to increasing the precision of speech and the salience of items in speech. When

218  Rogers, Rao and Burner attempting to extrapolate these insights to our Chilean plateau data, we actually notice that they seem to highlight a longer string of speech in a fashion that runs contrary to the Effort Code; their speech is faster (i.e., less precise), and there are fewer F0 movements within the focal stretch (i.e., it is an F0 plateau). Their increased speech rate triggers F0 compression and undershoot of L, which creates a stretch of high but flat F0, which, based on the Effort Code, points to a reduction of articulatory effort. We argue that, in line with Zipf’s (1949) Principle of Least Effort, this reduction of effort is employed by Chilean Spanish speakers due to its cost effectiveness, or articulatory economy, allowing them to highlight longer portions of speech, or complete thoughts, in an effective yet energy-​efficient manner (i.e., the most bang for their discursive buck). The fact that all speakers in our data set produced plateaus implies that the strategy we have outlined pervades this variety of Spanish (at least in the regions we investigated) and could even be a grammaticalized intonational structure (extending upon Chen, Gussenhoven, & Rietveld, 2004). We cautiously put forth this last point, as further research is needed to shed more light on this issue.

6.  Conclusion The surge in studies on Spanish intonation over the last two decades has resulted in numerous developments and contributions to our current understanding of the role that intonation plays in how Spanish speakers communicate with one another and interact with their surrounding environment. However, despite these advances, there is still much that remains unknown. As previously mentioned, this is partly a consequence of methodology, since the complicated nature of intonation has resulted in much of what we know being the result of tightly controlled mechanisms of data control and analysis. Face (2014) states that the Chilean plateau contours in Rogers (2013) are an example of variation from more natural speech that creates problems for current theorizing on intonational function and meaning. He further postulates that as more intonational studies using more natural speech data are performed, it is likely that more variation will arise that will further challenge current conventions used to describe Spanish intonation. Both Hirschberg (2002) and as recent as Prieto (2015) have also asserted that when established cross-​linguistic interpretations of intonation have been examined in more natural speech, researchers are frequently presented with data that runs counter to certain long-​held assumptions and claims in previous literature. Both agree that intonational meaning cannot be fully and accurately understood if studies do not take into account contextual and pragmatic data. An approach from a purely phonological angle essentially places limits on the ability of speakers to innovate using intonation, and makes the overreaching assumption that established phonological constraints are the principal driving force behind speakers’ communicative use of intonation.

Exploring extended focus and meaning  219 While alternate analyses are certainly possible, the current study is a preliminary analysis that provides support for a more speaker-​centered analysis of these specific intonational patterns. As the data demonstrate, the principal motivation behind the numerous unique behaviors of the plateau contours is not phonological, rather speakers appear to use the structure of these patterns in otherwise declarative contexts to highlight entire ideas or concepts rather than individual words. The use of Gussenhoven’s (2002) Effort Code, as well as the proposed modification that speakers compress phonetic material via undershoot as a result of increased speech rate, explains one way speakers may be able to achieve this. Likewise, it highlights the need to consider speaker innovation when analyzing intonational meaning, which steps outside the parameters of a strictly phonological approach. The next step in understanding these contours and their function within the Chilean Spanish linguistic community is the analysis of perceptual data. Perceptual insights from speakers can help confirm the proposal made that the primary motivators for these patterns are the communicative desires and goals of the speakers, as well as contextually-​based pragmatic factors.

Notes 1 Research on the syntax-​prosody interface refers to the lower level of phrasing as a phonological phrase (PPh). The PPh generally coincides with a major syntactic phrase (XP), for example, a Noun Phrase (NP), a Verb Phrase (VP) or an Adjectival Phrase (AP), where heads dominate lower constituents. In such work, the IP is tied to more complex syntactic clauses (Nespor & Vogel, 1986; Selkirk, 1984; Truckenbrodt, 1999). Since our study is not based on syntactic conditions, we will refer to the second level of phrasing as an ip, and will use it interchangeably with PPh, including when referring to studies that use PPh rather than ip. 2 For a list of less common ip and IP boundary tones, which are not described here, see Aguilar, de-​la-​Mota, and Prieto (2009). 3 While it is beyond the scope of the current study, Rogers (2016) examines the interaction of syntax and prosody at boundaries along the plateau patterns and notes a number of theoretical challenges that the plateaus engender for the relationship between the Prosodic Hierarchy and syntax. 4 An anonymous reviewer pointed out that the presence of a final oxytone, such as verdad in Figure  10.4, could be driving the use of a suspended contour; however, our data shows examples of Type 3 plateaus concluding in both oxtyones and paroxytones.

References Aguilar, L., de-​la-​Mota, C., & Prieto, P. (2009). Sp_​ToBI training materials. Retrieved from . Armstrong, M.E. (2010). Puerto Rican Spanish intonation. In P. Prieto & P. Roseano (eds.), Transcription of Intonation of the Spanish Language, 155–​189. Munich: Lincom Europa.

220  Rogers, Rao and Burner Astruc, L., Mora, E., & Rew, S. (2010). Venezuelan Andean Spanish intonation. In P. Prieto & P. Roseano (eds.), Transcription of Intonation of the Spanish Language, 191–​226. Munich: Lincom Europa. Atria, J. (2009). Estrategias de resolución de choques acentuales en el castellano hablado en Santiago de Chile. Onomázein, 19, 11–​31. Barker, G. (2005). Intonational Patterns in Tyrolean German:  An Autosegmental Metrical Analysis. New York: Peter Lang. Boersma, P., & Weenink, D. (2015). Praat:  Doing phonetics by computer, version 5.3.84 [computer program]. Retrieved from Cabrera, M.A., & Vizcaíno, F.O. (2010). Canarian Spanish intonation. In P. Prieto & P. Roseano (eds.), Transcription of Intonation of the Spanish Language, 87–​121. Munich: Lincom Europa. Cepeda, G. (1997). Las unidades entonacionales del habla de las mujeres de Valdivia. Onomázein, 2, 83–​110. Cepeda, G. (2001). Las unidades de entonación del español de Valdivia, Chile. Onomázein, 6, 31–​51. Chen, A., Gussenhoven, C., & Rietveld, T. (2002). Language-​specific uses of the Effort Code. In B. Bel & I. Marlien (eds.), Proceedings of Speech Prosody 2002, 215–​218. Aix-​en-​Provence: Laboratoire Parole et Langage. Chen, A., Gussenhoven, C., & Rietveld, T. (2004). Language-​specificity in the perception of paralinguistic intonational meaning. Language and Speech, 47(4), 311–​349. Cho, H., & Flemming, E. (2015). Compression and truncation:  The case of Seoul Korean accentual phrase. Studies in Phonetics, Phonology and Morphology, 21(2), 359–​382. Christophe, A., Peperkamp, S., Pallier, C., Block, E., & Mehler, J. (2004). Phonological phrase boundaries constrain lexical access:  Adult data. Journal of Memory and Language, 51, 523–​547. Cid, M.E., & Ortiz, H. (1998). La conducta prosódica del vocativo en el español culto de Santiago de Chile. Onomázein, 3, 143–​162. Clopper, C.G., & Tonhauser, J. (2011). On the prosodic coding of focus in Paraguayan Guarani. In M. Byram (ed.), Proceedings of the 28th West Coast Conference on Formal Linguistics, 249–​257. Somerville, MA: Cascadilla Proceedings Project. D’Imperio, M., Elordieta, G., Frota, S., Prieto, P., & Vigário, M. (2005). Intonational phrasing in Romance:  The role of syntactic and prosodic structure. In S. Frota, M. Vigário, & M.J. Freitas (eds.), Prosodies, 9–​98. Berlin: Mouton de Gruyter. Elordieta, G. (2003). The Spanish intonation of speakers of a Basque pitch-​accent dialect. Catalan Journal of Linguistics, 2, 67–​95. Elordieta, G., Frota, S., Prieto, P., & Vigário, M. (2003). Effects of constituent length and syntactic branching on intonational phrasing in Ibero-​Romance. In M.J. Solé, D. Recasens, & J. Romero (eds.), Proceedings of the 15th International Congress of Phonetic Sciences. 487–​490. Barcelona: Futurgraphic. Estebas, E., & Prieto, P. (2008). La notación prosódica del español: Una revisión de Sp_​ToBI. Estudios de Fontética Experimental XVII, 263–​283. Face, T. (2001). Focus and early peak alignment in Spanish intonation. Probus, 13(2), 223–​246. Face, T. (2003). Intonation in Spanish declaratives: Differences between lab speech and spontaneous speech. Catalan Journal of Linguistics, 2, 115–​131.

Exploring extended focus and meaning  221 Face, T. (2008). The intonation of Castilian Spanish declaratives and absolute interrogatives. Munich: LINCOM. Face, T. (2014). Sp_​ToBI and the phonological analysis of Spanish intonation: A critical perspective. Studies in Hispanic and Lusophone Linguistics, 7(1), 185–​210. Féry, C. (1993). German Intonational Patterns. Tübingen: Niemeyer. Fuentes, M. (2012). Análisis fonético-​acústico de la conducta prosódica de los enunciados del tipo imperativo (petición y orden) del español de Santiago de Chile (Master’s thesis). Universidad Pontificia Católica de Chile. Gabriel, C., Feldhausen, I., & Pešková, A. (2011). Prosodic phrasing in Porteño Spanish. In C. Gabriel & C. Lleó (eds.), Intonational Phrasing in Romance and Germanic: Cross-​Linguistic and Bilingual Studies,153–​182. Amsterdam: John Benjamins. Goldman-​Eisler, F. (1961). The significance of changes in the rate of articulation. Language and Speech, 4(4), 171–​174. Gussenhoven, C. (2002). Intonation and interpretation: Phonetics and phonology. In B. Bel & I. Marlien (Eds.), Proceedings of Speech Prosody 2002, 47–​57. Aix-​en-​ Provence: Laboratoire Parole et Langage. Gussenhoven, C. (2004). The Phonology of Tone and Intonation. Cambridge: Cambridge University Press. Hirschberg, J. (2002). The pragmatics of intonational meaning. In B. Bel & I. Marlien (eds.), Proceedings of Speech Prosody 2002, 65–​68. Aix-​en-​Provence: Laboratoire Parole et Langage. House, D. (2004). Pitch and alignment in the perception of tone and intonation: pragmatic signals and biological codes. In B. Bel & I. Marlien (eds.), Proceedings of the International Symposium on Tonal Aspects of Languages:  Emphasis on Tone Languages, 93–​96. Beijing. Hübscher, I., Borràs-​Comes, J., & Prieto, P. (2017). Prosodic mitigation characterizes Catalan formal speech: The Frequency Code reassessed. Journal of Phonetics, 65, 145–​159. Kowal, S., Wiese, R., & O’Connell, D. (1983). The use of time in storytelling. Language and Speech, 26(4), 377–​392. Ladd, D.R. (2008). Intonational Phonology (2nd ed.). Cambridge:  Cambridge University Press. Nadeu, M., & Prieto, P. (2011). Pitch range, gestural information, and perceived politeness in Catalan. Journal of Pragmatics, 43, 841–​854. Nespor, M., & Vogel, I. (1986). Prosodic Phonology. Dordrecht: Foris. Ortiz, H., Fuentes, M., & Astruc, L. (2010). Chilean Spanish intonation. In P. Prieto & P. Roseano (eds.), Transcription of Intonation of the Spanish Language, 255–​283. Munich: Lincom Europa. Pellegrino, F., Coupé, C., & Marsico, E. (2011). Across-​language perspective on speech information rate. Language, 87(3), 539–​558. Pérez, S. (2015). La entonación del español de Galicia desde una perspectiva sociofonética (Doctoral dissertation). University of Minnesota. Prieto, P. (2006). Phonological phrasing in Spanish. In F. Martínez-​Gil & S. Colina (eds.), Optimality-​theoretic Studies in Spanish Phonology, 39–​60. Amsterdam: John Benjamins. Prieto, P. (2015). Intonational meaning. WIREs Cogn Sci, 6, 371–​381. Prieto, P., & Roseano, P. (2010). Transcription of Intonation of the Spanish Language. Munich: Lincom Europa.

222  Rogers, Rao and Burner Prieto, P., Shih, C., & Nibert, H. (1996). Pitch downtrend in Spanish. Journal of Phonetics, 24, 445–​473. Quilis, A. (1999). Tratado de fonología y fonética españolas (2nd ed.). Madrid: Editorial Gredos. Rao, R. (2006). On intonation’s relationship with pragmatic meaning in Spanish. In T. Face & C. Klee (eds.), Selected Proceedings of the 8th Hispanic Linguistics Symposium, 103–​115. Somerville, MA: Cascadilla Press. Rao, R. (2007). On the phonological phrasing patterns in the Spanish of Lima, Perú. Southwest Journal of Linguistics, 26(1), 81–​111. Rao, R. (2009). Deaccenting in spontaneous speech in Barcelona Spanish. Studies in Hispanic and Lusophone Linguistics, 2(1), 31–​75. Rao, R. (2010). Final lengthening and pause duration in three dialects of Spanish. In M. Ortega-​Llebaria (ed.), Selected Proceedings of the 4th Conference on Laboratory Approaches to Spanish Phonology, 69–​82. Somerville, MA: Cascadilla Proceedings Project. Rao, R. (2013). Intonational variation in third party complaints in Spanish. Journal of Speech Sciences, 3(1), 141–​168. Rogers, B.M.A. (2013). The extent of tonal events: Intonational hat patterns in Chilean Spanish. Estudios de fonética experimental, 22, 171–​192. Rogers, B.M.A. (2016). When theory and reality collide:  Exploring Chilean Spanish intonational plateaus (Doctoral dissertation). University of Minnesota. Selkirk, E. (1984). Phonology and Syntax: The Relation Between Sound and Structure. Cambridge, MA: MIT Press. Silva-​Corvalán, C. (1983). On the interaction of word order and intonation:  Some OV constructions in Spanish. In F. Klein-​Andreu (ed.), Discourse Perspectives on Syntax, 117–​140. New York: Academic Press. Truckenbrodt, H. (1999). On the relation between syntactic phrases and phonological phrases. Linguistic Inquiry, 30, 219–​255. Vanrell, M., Stella, A., Gili Fivela, B., & Prieto, P. (2013). Prosodic manifestations of the Effort Code in Catalan, Italian and Spanish contrastive focus. Journal of the International Phonetic Association, 43(2), 195–​220. Véliz, M. (2001). A contrastive study of English and Spanish post-​nuclear patterns. Onomázein, 6, 53–​68. Willis, E. (2010). Dominican Spanish intonation. In P. Prieto & P. Roseano (eds.), Transcription of Intonation of the Spanish Language, 123–​153. Munich:  Lincom Europa. Yasavul, M. (2013) Prosody of focus and contrastive topic in K’iche’. OSUWPL, 60, 129–​160. Zipf, G.K. (1949). Human Behavior and the Principle of Least Effort. Cambridge, MA: Addison-​Wesley  Press.

11  The intonation of yes-​no questions in Basque Spanish Carolina González and Lara Reglero

1.  Introduction The present study examines the intonation of yes-​no questions in various pragmatic contexts in the Spanish spoken in the Basque Country in Spain (henceforth Basque Spanish). Yes-​no questions, also known as “absolute” questions, can have different word order than statements in Spanish. However, it is intonation that is crucial to distinguish between both sentence types (see for example Hualde & Prieto, 2015) (1). While pragmatically neutral statements typically have a final falling intonational contour, yes-​no questions in many Spanish dialects are realized with a final rise (Navarro Tomás, 1944; Quilis, 1993:  428–​429; Sosa, 1999:  211; Face, 2008; Hualde, 2014, among others).1 In addition, the pitch of pre-​nuclear accents is higher in yes-​no questions than in statements (Sosa, 1999; Face, 2004, 2008; Prieto, 2004; cf. Navarro Tomás, 1944). (1) a. Statement b. Yes-​no question

Está disfrutando de las be.PRS. 3S G enjoy.PROG of the “S/​he is enjoying his/​her vacation.” ¿ Está disfrutando de las be.PRS. 3S G enjoy.PROG of the “Is s/​he enjoying his/​her vacation?”

vacaciones. vacations vacaciones? vacations

However, final rising contours are not universal across Spanish neutral yes-​ no questions, since some dialects display a final fall or rise-​fall instead. Examples include Venezuelan Spanish (Astruc, Mora & Simon, 2010), Caribbean and Canary Island Spanish (Quilis, 1987; Alvord, 2010), Buenos Aires Spanish (Gabriel et al., 2010) and Asturian, Cantabrian and Basque Spanish (López-​ Bobo & Cuevas-​Alonso, 2010; Canellada, 1984; Robles-​Puente, 2011, 2012; Elordieta & Romera, in press). In addition, yes-​no questions as in (1b) can have different pragmatic interpretations besides the “canonical” elicitation of new information (referred to as “information-​seeking” or “neutral” in the literature; Prieto & Roseano, 2010). For example, they can be employed to verify

224  Carolina González and Lara Reglero information (“confirmation yes-​no questions”) or express surprise or disbelief (“surprise yes-​no questions”). They can also be used to make a request, as in (2) (Escandell Vidal, 2013).2 (2) ¿ Puedes darme el día libre? can-​P R S.2S G give-​I NF-​C L.1SG the day free “Could you give me the day off ?” Many of the pragmatic contexts mentioned in the previous paragraph are associated with specific intonational characteristics (Fernández Ramírez, 1959; Escandell Vidal, 1998, 1999). For example, in Castilian Spanish, information-​seeking yes-​no questions are realized primarily with a final rising contour, but pragmatically marked yes-​no questions, including confirmation, counter-​expectational, and some echo questions, tend to be associated with a falling final contour (Navarro Tomás, 1944; Quilis, 1993; Estebas-​Vilaplana & Prieto, 2010). The intonational characteristics associated with specific pragmatic contours have been shown to differ across Spanish dialects, but are, for the most part, still under-​investigated (Hualde & Prieto, 2015). The main goal of the present study is to investigate the intonational characteristics of pragmatically biased yes-​no questions in Basque Spanish. The study of this Spanish variety from Northern Spain, in contact with Basque, is important for several reasons. First of all, its intonation is for the most part under-​investigated. Previous research focuses mainly on statements and information-​seeking yes-​no questions, which reportedly differ in their overall melodic pattern and nuclear pitch alignment. Specifically, while statements show declination and alignment of the nuclear F0 peak with the stressed syllable, yes-​no questions tend to be characterized by declination suspension or up-​step, and F0 alignment on the post-​tonic syllable (Robles-​ Puente, 2011, 2012). In addition, Basque Spanish is one of the Spanish varieties where information-​seeking yes-​no questions have final circumflex (rising-​falling) contours (L+H* HL% or L+¡H* HL%) (Robles-​ Puente, 2011; Elordieta & Romera, in press). Both the Basque language and other North-​Western Peninsular varieties such as Cantabrian and Galician reportedly have rising-​ falling contours in neutral yes-​ no questions (Elordieta & Hualde, 2014; López-​Bobo & Cuevas-​Alonso, 2010 and references therein). While recent studies do not find an influence of bilingualism or specific geographical area in the intonation of Basque Spanish neutral yes-​no questions (Robles-​Puente, 2011; Elordieta & Romera, in press), Elejabeitia, Iribar & Pagola (2005, 2007, 2008) report rising final intonation in some speakers in Bilbao, Donostia-​San Sebastian and Markina-​Xemein, and rising/​circumflex variation occurs in Vitoria-​Gasteiz.3 Thus, the potential impact of bilingualism and geographical area in the intonation of Basque Spanish yes-​no questions remains to be investigated in more detail.4 Thus, a second goal of our study is to examine

The intonation of yes-no questions  225 whether the degree of Basque-​Spanish bilingualism impacts the intonation of yes-​no questions in this dialect. Based on previous studies, we hypothesize that information-​seeking yes-​ no questions will be primarily realized with final circumflex contours, since our participants hail from the Bizkaia region, of which Bilbao is the administrative capital. We also expect to find an effect of bilingualism, whereby more Spanish-​dominant participants will have final rising contours rather than circumflex ones in this context. For non-​neutral pragmatic contexts, we have two competing hypotheses. One is that neutral and non-​neutral contexts in Basque Spanish will have very different final contours. This is the case for example in Castilian Spanish, where final rises (L* HH%) are typical of information-​seeking questions, while circumflex contours characterize counter-​expectational contexts; Sosa 1999, Estebas-​Vilaplana & Prieto, 2010; Face, 2008; Hualde & Prieto, 2015). The competing hypothesis is that circumflex contours will be present regardless of pragmatic intent, as in Puerto Rican Spanish (Armstrong, 2017). The remainder of this chapter is organized as follows. Section 2 describes the methodology used in our study. Section 3 presents the results, while sections 4 and 5 provide a discussion and conclusion, respectively.

2.  Methodology 2.1  Data collection and participants Data for this study was collected in Summer 2015 in the Phonetics Laboratory at the University of Deusto in Spain. A total of 22 participants listened to task scripts with headphones and completed two contextualized experimental tasks: a reading task, and an elicitation task. All interviews were conducted entirely in Spanish and took approximately one hour per participant. All participants were paid for their participation. In this chapter, we report elicitation data from 12 female participants with varied degrees of Spanish-​Basque bilingualism (Table 11.1). All were born and raised in the province of Bizkaia and were 21–​24 years old. Their degree of Basque-​Spanish bilingualism was assessed via the Bilingual Linguistic Profile (BLP) (Birdsong, Gertken & Amengual, 2012), completed after obtaining consent, and before data collection. The BLP provides information about Spanish and Basque language history, use, proficiency and attitudes. It provides a score in Spanish, another in Basque, and an overall dominance score, obtained by subtracting the Basque score from the Spanish one. Positive dominance scores indicate language dominance in Spanish, negative scores point to dominance in Basque, and scores close to 0 signal balanced bilingualism. The relevant information for our participants is provided in Table  11.1; note that they are ranked from highest to lowest according to dominance score.

226  Carolina González and Lara Reglero Table 11.1 Participants Dominance score 15 22 8 1 5 21 4 20 13 14 7 3

168 85 80 76 51 49 38 26 14 −​2 −​5 −​40

History

Use

Spn

Basque

Spn

117 95 87 111 94 98 116 97 97 109 120 101

–​ 51 57 83 87 78 116 98 99 109 120 119

34 31 46 42 42 41 43 35 31 22 21 14

Proficiency

Attitude

Basque

Spn

Basque

Spn

Basque

0 1 3 2 3 5 3 9 15 20 26 32

24 20 22 24 24 24 24 24 24 22 23 24

18 12 19 20 21 23 24 24 24 22 23 24

24 15 22 24 22 15 18 24 24 17 24 20

13 12 18 20 20 23 20 23 24 21 24 24

The examination of history, use, proficiency and attitude scores for Spanish reveals overlap in participants that are “Spanish-​dominant” vs. “Basque-​ dominant.” However, the inspection of scores for Basque history, use and proficiency reveals more clear differences between participants with a higher vs. lower dominance score. We group the first six participants in the table, which score 51–​87 points in the history of Basque portion of the BLP, as “relatively Spanish-​dominant,” and the last six participants, which score 98–​ 120 points in this component, as “relatively Basque-​dominant.” Scores for Basque use and proficiency roughly correspond to this division.5,6 2.2  Target sentences Sixteen yes-​no questions (four per pragmatic type) were recorded and analyzed per participant. All were elicited by means of naturalistic contexts presented aurally and visually via a PowerPoint presentation. Yes-​no questions were randomized among other utterance types (statements and wh-​in-​situ questions). Target sentences were designed to avoid stress clashes, particularly in the final part of the question, to avoid tonal crowding effects (Face, 2002; Ladd, 2008; Arvaniti & Ladd, 2009; Henriksen, 2012). Most yes-​no questions ended in paroxitone words. The only final oxitone words included related to two proper names (Beyoncé, Tamariz) and one app (WhatsApp). The last two target words were pronounced by some participants as paroxitone. All pragmatic contexts investigated included tokens with oxitone final words, except for requests. The intonational differences in oxitone vs. paroxitone words in our data are explored in section 3.1. Examples of elicitation contexts and expected target questions are provided below for each pragmatic type. Note that information-​seeking questions are

The intonation of yes-no questions  227 genuine questions that request new information (3). In contrast, confirmation questions imply an expectation about the answer, which can range from very strong to less confident (4). Surprise questions (also known as counter-​ expectational questions) convey disbelief and/​or wonder (5). Last but not least, requests ask for something in a polite manner (6)  (Escandell Vidal, 2013; Hualde & Prieto, 2015).7 (3) Information-​seeking yes-​no question: Context:

Expected question:

Llegas a la panadería justo antes de que cierren. Pregúntale al panadero si les quedan bollos de mantequilla. “You arrive at the bakery just right before closing time. Ask the baker if they have any sweet rolls left.” ¿ Os quedan bollos de mantequilla? C L.2 P L have-​P RS .2SG pastries of butter “Do you have any sweet rolls left?”

(4) Confirmation yes-​no question:8 Context: Estás estudiando en la biblioteca y ves un libro en el suelo. Crees que es del chico que está sentado al lado pero no estás del todo seguro. Pregúntale si el libro es suyo. “You are studying at the library and you see a book on the floor. You think the book belongs to the guy sitting next to you, but you are not completely sure. Ask him if the book is his.” Expected question: ¿ Este libro es tuyo? this book be-​P RS .3SG yours “The book is yours?” (5) Surprise yes-​no question: Context: Vas a tu restaurante favorito con un amigo y pides langosta, pero el camarero dice que no tienen. Te sorprende mucho porque es uno de los platos típicos del restaurante. Pregúntale al camarero si de verdad no tienen langosta. “You go to your favorite restaurant with a friend, and order lobster, but the waiter tells you that they don’t have any. You are surprised by this given that lobster is one of the typical dishes at the restaurant. Ask the waiter if they don’t really have any lobster.” Expected question: ¿ De verdad no tenéis langosta? of really NEG have-​P RS .2SG lobster “You really don’t have lobster?”

228  Carolina González and Lara Reglero (6)

Request yes-​no question: Context: Tu familia va a venir a visitarte este fin de semana y tu piso esta tan sucio que necesitas ayuda para limpiarlo. Pregúntale a tu compañera de piso si le importaría limpiar el baño. “Your family is coming for a visit over the weekend. Your apartment is so messy that you need help to clean it up. Ask your roommate if she could clean the bathroom.” Expected question: ¿Te importaría limpiar el baño? CL .2S G mind-​C OND. 2 SG clean-​I N F the bathroom “Could you clean the bathroom?”

2.3  Recording, coding and analysis Data was recorded using a Tascam DR-​05 digital recorder with two built-​ in omni-​directional microphones, in 44,000 Hz wav, in mono. A  total of 192 sentences were coded and analyzed (16 questions x 12 participant) using Praat (Boersma & Weenink, 2015). The framework of our study is the Autosegmental-​ Metrical model of intonation analysis, in particular, the Spanish Tones and Breaks Indices (also known as Sp_​ToBI; Beckman, Díaz-​Campos, McGory & Morgan, 2002; Hualde, 2003; Sosa, 2003; Face & Prieto, 2007; Estebas-​Vilaplana & Prieto, 2008; Aguilar, De-​la-​Mota & Prieto, 2009). Trained research assistants transcribed each question orthographically and phonetically, and manually coded the location of nuclear peaks and valleys. Both authors revised the transcription and coding, and independently examined the acoustic realization of pre-​ nuclear accents, nuclear accents, and boundary tones, as well as prosodic phrasing. Any disagreements (which occurred for about 5 percent of the data) were resolved via consensus. Because there are no previous reports on the intonation of pragmatically varied yes-​no questions in Basque Spanish, we examine and report their overall melodic contour, paying particular attention to the nuclear configuration, which tends to be most relevant for question interpretation (Face, 2008). We break down the overall melodic contour in three parts. The initial part comprises the beginning of the question up to the first pre-​nuclear accent. The second part encompasses the end of the first pre-​nuclear accent up to, but not including, the nuclear configuration. The third part contains the nuclear configuration, i.e., the last stressed syllable plus any post-​nuclear syllables, if present. As an example, consider the melodic contour of the information-​ seeking question ¿ Os quedan bollos de mantequilla? “Do you have any sweet rolls left?” from Participant 14 (Figure 11.1).

The intonation of yes-no questions  229

Figure 11.1 Information-​seeking yes-​no question: ¿ Os quedan bollos de mantequilla? “Do you have any sweet rolls left?” Participant 14. Pitch range:  150 Hz-​425  Hz.

There are two-​pre-​nuclear accents in Figure  11.1:  the first is associated with the first syllable of quedan, and the second with the first syllable in bollos. The nuclear accent falls on the last stressed syllable (qui in mantequilla). The phonological phrase os quedan would comprise the initial part of the melodic contour (up to the first pre-​nuclear accent, typically realized as delayed on the first post-​tonic syllable). Bollos de mante would comprise the medial part, and quilla would be the location of the nuclear configuration, comprising the nuclear pitch accent (L+H* in this example) and the final boundary tone (L%). In addition, we examine the specific nuclear contour for each question. As mentioned in section 1, rising-​falling (circumflex) contours are attested in information-​seeking yes-​no questions in Bilbao Spanish (Robles-​Puente, 2011, 2012; Elordieta & Romera, in press), but final rising contours have also been reported in some areas of Bizkaia Spanish (Elejabeitia et al., 2005, 2008). Figures 11.2–​11.4 show examples of confirmation, surprise and request yes-​no questions realized with differing final contours. In Figure  11.2, the nuclear syllable is realized with a low pitch (L*); the final boundary rises to a very high level (HH%). The nuclear configuration is, thus, rising. Figure 11.4 presents another instance of final rising boundary tone, this time preceded by a H+L* nuclear accent. In Figure 11.3, on the other hand, the nuclear syllable displays a rising nuclear pitch (L+H*) and a fall on the final syllable (HL%). The nuclear configuration in this case is rising-​falling (circumflex). Finally, because differences in pitch scaling can distinguish between pragmatic meanings in Spanish and other languages (see for example Ladd, 2008; Borràs-​Comes, Vanrell & Prieto, 2014), we also measured the pitch of the highest nuclear peak. As an example, the nuclear High reaches 408 Hz. in the confirmation question in Figure 11.2, but 362 Hz. in the surprise question in Figure 11.3.

230  Carolina González and Lara Reglero

Figure 11.2 Confirmation yes-​no question:  ¿Esa maleta es suya? “Is that suitcase yours?” Participant 15. Final rising contour. Pitch range: 125 Hz-​575 Hz.

Figure 11.3 Surprise yes-​no question: ¿De verdad no tenéis una baraja? “You really don’t have a deck of cards?” Participant 3. Final circumflex contour. Pitch range: 100 Hz-​425 Hz.

Descriptive statistics are reported for all analyses. In addition, independent measures chi-​ square tests were run to test whether rising vs. circumflex contours correlate with specific pragmatic meanings. Significance is set at p ≤ .05. Nuclear configuration peaks are measured in Hz. Peak differences across pragmatic contexts are also reported in semitones (ST). Following Rietveld and Gussoven (1985), Toledo (2000) and Pàmies-​Bertrán et  al. (2002), if pitch differences in the nuclear peak across two given pragmatic contexts reach or surpass the perceptual threshold of 1.5 ST, they are considered to be prosodically different.

The intonation of yes-no questions  231

Figure 11.4 Request: ¿Te puedo llevar algo? “Can I bring you something?” Participant 1. Final rising contour. Pitch range: 100 Hz-​500 Hz.

3.  Results 3.1  Contours For all speakers analyzed, yes-​no questions begin with a rise up to the first pre-​nuclear accent, realized as delayed. This agrees with findings reported for information-​seeking questions in read speech (Robles-​Puente, 2011) and semi-​spontaneous speech (Elordieta & Romera, in press). Our data shows variability between declination (overall pitch lowering), declination suspension (a flat contour, without melodic ups and downs), and up-​stepping (pitch rising) between the first pre-​nuclear accent and the nuclear configuration. This variation appears to be a matter of individual preference, as reported in Elejabeitia et al. (2005, 2007, 2008) and Robles-​Puente (2012).9 Eight of the 12 participants analyzed show a preference for final falling or circumflex contours, while the remaining four participants display rising contours (Table 11.2). Circumflex contours are typically realized as a final rise beginning in the nuclear syllable and peaking on the final syllable, before a steep fall (L+H* HL%). The High tone is frequently up-​stepped (L+¡H* HL%). This is also the most frequent contour reported for information-​seeking yes-​no questions in Bilbao Spanish (Robles-​Puente, 2011; Elordieta & Romera, in press). Five of the participants analyzed display this circumflex pattern, but a variant is attested in participants 7, 14, involving a final rise on the nuclear syllable followed by a fall (L+H* L%). In addition, participant 22 shows variable circumflex and falling contours, with the contour H+L* L% being most common. The four remaining participants show a preference for a steep rise in the nuclear configuration, with contour L* HH% preferred in participants 15, 4, 20, and H+L* HH% in participant 1.

232  Carolina González and Lara Reglero Table 11.2 Final contours by participant (all pragmatic contexts pooled) Participant

Dominance Predominant final score contour

%

Most common nuclear configuration

15 22 8 1 5 21 4 20 13 14 7 3

168 85 80 76 51 49 38 26 14 −​2 −​5 −​40

94% 86% 100% 81% 93% 100% 71% 60% 62% 100% 81% 100%

L* HH% H+L* L% L+H* HL% H+L* HH% L+H* HL% L+H* HL% L* HH% L* HH% L+H* HL% L+H* L% L+H* L% L+H* HL%

Rising Falling/​circumflex Circumflex Rising Circumflex Circumflex Rising Rising Circumflex Circumflex Circumflex Circumflex

Only four participants display contour differences in paroxitone vs. oxitone words (Table  11.3). Except in participant 13, final oxitone contours in our dataset can be analyzed as tonal truncation. Specifically, participants 8, 21 have monotonal rather than bitonal boundary tones, and participants 8, 22 evidence monotonal rather than bitonal nuclear accents. When data is split according to pragmatic context, we observe that rising-​ falling contours are prevalent across question types, particularly in surprise questions. Results from a chi-​square test shows no relationship between final contour and pragmatic type (χ2 (3, N=182) = 5.83; p=.12). Table 11.5 outlines the dominant contour for each participant in each pragmatic context. Seven participants show (near) identical nuclear configurations across the contexts investigated:  rising for participants 4, 15, circumflex in participants 3, 5, 8 and 21, and falling in participant 14. The remaining five participants show different nuclear configurations in two or more pragmatic contexts. For these participants, a preference for rising contours is observed in information-​seeking contexts, falling or circumflex contours in surprise contexts, and variation between the two in confirmation questions and requests. The analysis of preferred nuclear contours by participant does not appear to indicate a clear pattern between overall language dominance and contour. While participants with negative dominance scores (3, 7, 14) show a prevalence of falling and circumflex contours, these contours are also preferred by other participants classified as “Spanish-​dominant” according to the BLP, such as participants 8, 5 or 21. The comparison of final contours in more Spanish-​dominant participants vs. more Basque-​dominant participants shows a more robust tendency for circumflex contours in the second group in three of the contexts investigated,

The intonation of yes-no questions  233 Table 11.3 Contour differences in paroxitone and oxitone words Participant

Yes-​no contexts

Paroxitone

Oxitone

8 13 21 22

All, except requests

L+H* HL% H* L% L+H* HL% H+L* L%

H* L% L* HH% L+H* L% H* L%

Confirmation

Table 11.4 Final contours (all pragmatic contexts pooled)

Rise+fall Rising

Requests

Information-​ seeking

Confirmation

Surprise

Overall

62% 38%

60% 40%

62% 38%

80% 20%

66% 34%

Table 11.5 Dominant nuclear configurations by pragmatic context10,11 Participants

Information-​ seeking

Confirmation

Requests

Surprise

Relatively Spanish dominant

15 22 8 1 5 21

L* HH% H+L* HH% L*+H HL% H+L* HH% L*+H HL% L*+H HL%

L* HH% H+L* L% L*+H HL% H+L* LH% L*+H HL% L*+H HL%

L* HH% H+L* L% L*+H HL% H+L* HH% L*+H HL% L*+H HL%

L* HH% L+H* L% L*+H HL% L+H* HL% L*+H HL% L*+H HL%

Relatively Basque dominant

4 20

H+L* HH% L* HH%

H+L* HH% L+H* HH%

L* HH% Inconclusive

13

L+H* HL% L* HH% L+H* L% L+H* L% L*+H HL%

L+H* HL%

L* HH% H+L* L% L* HH% H+L* LH%

L+H* L% H+L* L% L*+H HL%

L+H* L% L% L*+H HL%

L+H* L% H+L* L% L*+H HL%

14 7 3

Inconclusive

particularly in requests and information-​seeking questions (Tables 11.6, 11.7). However, this difference is not significant (p=.85). Results from chi-​square statistical analysis show no relationship between final contour and pragmatic type in either participant group (Spanish-​dominant group:  p=.16; Basque-​ dominant group: p=.44).

234  Carolina González and Lara Reglero Table 11.6 Final contours: relatively Spanish-​dominant group

Rise+fall Rise

Requests

Information-​ seeking

Confirmation

Surprise

Overall

52% 48%

50% 50%

63% 37%

79% 21%

61% 39%

Table 11.7 Final contours: relatively Basque-​dominant group

Rise+fall Rise

Requests

Information-​ seeking

Confirmation

Surprise

Overall

71% 29%

70% 30%

61% 39%

84% 16%

71.5% 28.5%

3.2  Nuclear High Nuclear peaks are lowest in requests and highest in surprise contexts for all participants pooled (Table  11.8). The difference between confirmation and request yes-​no questions is 1.56 ST, and between requests and surprise contexts 1.82 ST. Thus, requests can be considered to be prosodically different from both confirmation and surprise yes-​no questions. For relatively Spanish-​dominant participants, the nuclear High is lowest in information-​seeking questions and requests, and highest in surprise questions (Table 11.9). Surprise questions surpass the perceptual threshold compared to requests (1.75 ST) and information-​seeking questions (1.76 ST) in this group. For relatively Basque-​dominant participants, the nuclear High is lowest for requests and highest for confirmation yes-​no questions (Table 11.10). The difference between requests and surprise questions surpass the perceptual threshold (1.8 ST), as does the difference between confirmation questions and requests (2.4 ST). In addition, the difference between requests and information-​seeking questions approaches the perceptual threshold (1.4 ST). The comparison of relatively Spanish-​dominant and relatively Basque-​ dominant participants shows that the former have a more elevated nuclear High in all pragmatic contexts, particularly in surprise yes-​no questions (43 Hz or 2.3 ST higher), and requests (40 Hz. or 2.3 ST higher). Cf. information and confirmation questions (14 Hz or .79 ST, and 12 Hz or .64 ST higher, respectively). Summarizing this section, requests and information-​ seeking questions have overall a less elevated nuclear High, in contrast to surprise and confirmation questions. Both participant groups use pitch scaling to distinguish between requests and surprise contexts. However, while the relatively

The intonation of yes-no questions  235 Table 11.8 Nuclear High (in Hz and ST; all speakers pooled) Requests 297 Hz.

+.63 ST

Information-​seeking

Confirmation

Surprise

308 Hz.

325 Hz.

330 Hz.

+.93 ST

+.26 ST

Table 11.9 Nuclear High (in Hz and ST): relatively Spanish-​dominant group Information-​seeking

Requests

315 Hz.

317 Hz.

+.1 ST

+.75 ST

Confirmation

Surprise

331 Hz.

351 Hz.

+1 ST

Table 11.10 Nuclear High (in Hz and ST): relatively Basque-​dominant group Requests 277 Hz.

+1.4 ST

Information-​seeking

Surprise

301 Hz.

308 Hz.

+.4 ST

Confirmation +.6 ST

319 Hz.

Spanish-​dominant group also uses pitch scaling to distinguish between surprise and information-​ seeking questions, the relatively Basque-​ dominant employs it to distinguish between requests and confirmation questions.

4.  Discussion One of the goals of our study was to investigate the intonational characteristics of four pragmatically different yes-​no questions in Basque Spanish. Our findings show that both neutral and biased yes-​no questions generally show a delayed pre-​nuclear accent and a rising-​falling final contour, most commonly realized as L*+H HL%. Although surprise yes-​no questions have the highest proportion of rising-​falling contours and information-​seeking questions the lowest, there are no significant differences in contour use among these or other pragmatic meanings. Our data suggest that it is pitch scaling rather than final contour what distinguishes among some pragmatic contexts in yes-​no questions in this dialect. Specifically, the nuclear peak is more elevated in confirmation and surprise questions than in neutral questions and requests. Our data shows that the difference in scaling between requests and surprise questions is prosodically relevant in both participant groups. In addition, our findings suggest an

236  Carolina González and Lara Reglero intonational split between requests and information-​seeking questions on the one hand, and confirmation and surprise questions on the other. The former tend to be realized with a less elevated nuclear peak, unlike the latter. We tentatively propose that neutral questions and requests are characterized phonologically by the contour L+H* HL%, while confirmation and surprise questions are characterized by L+¡H* HL%. However, this needs to be confirmed in future studies. Requests and information-​seeking questions appear to have very similar prosodic characteristics. The main difference between the two types of questions appears to be syntactic, with the pragmatic intent of (polite) requests rendered quite obvious through the use of conditional verbal forms such as podrías “could you” or te importaría “would you mind.” Confirmation and surprise yes-​no questions also share similar intonational characteristics (although surprise questions are circumflex more often). Both types of questions might be distinguished by syntactic cues as well. Indeed, in our data, surprise questions often began with the adverbial de verdad “really,” which might be a salient cue in the data to express surprise. Future perceptual studies should be conducted to examine in more detail the interaction between syntactic cues and intonation across pragmatically different yes-​no questions in this dialect. A second goal of our study was to examine if the degree of Basque-​Spanish bilingualism impacts the intonational characteristics of yes-​no questions in Basque Spanish. In Basque, information-​seeking yes-​no questions are typically realized with high-​falling or low-​falling nuclear contours; they also have a higher pitch than declaratives, and display final lengthening (Elordieta & Hualde, 2014).12 In contrast, rising contours are common in this and other pragmatic contexts in Castilian Spanish (Navarro Tomás, 1944; Quilis, 1993; Face, 2008; Estebas-​Vilaplana & Prieto, 2010). Our study did not find a significant difference between the final contours used by more Spanish-​dominant vs. more Basque-​dominant participants, along the lines of Robles-​Puente (2011, 2012) and Elordieta & Romera (in press) (cf. Elejabeitia et  al., 2005, 2007, 2008). However, our analysis of relative language dominance suggests that certain aspects of Basque use impact the prosody of yes-​no questions in Basque Spanish to some extent. Specifically, relatively Basque-​dominant participants do show a more consistent preference for falling contours in information-​ seeking questions and requests compared to relatively Spanish-​dominant speakers. They also have less elevated nuclear peaks overall, and show prosodic differences in peak scaling in confirmation-​seeking questions vs. requests. In contrast, more Spanish-​dominant participants distinguish prosodically between surprise and information-​seeking questions. For these participants, information-​seeking contexts had the least elevated High, whereas surprise yes-​no questions had the highest values for Spanish-​dominant speakers; this is consistent with similar results reported for Castilian Spanish (Estebas-​ Vilaplana & Prieto, 2010).

The intonation of yes-no questions  237 These results are interesting in view of those reported in Elordieta & Romera (in press). Although these authors did not find an effect of bilingualism in the realization of yes-​no questions in Basque Spanish, they show that speakers with a more positive attitude towards Basque exhibited a higher number of final falling or rising-​falling contours. While we agree that the overall prevalence of circumflex contours in Basque Spanish yes-​ no questions can be explained as an general preference for this contour type in North-​West Spain (see for example López-​Bobo & Cuevas-​Alonso, 2010), our data and that of Elordieta & Romera suggest that some aspects of Basque-​Spanish language contact in this region contribute to the intonational characteristics of yes-​no questions in this dialect. A more in-​depth examination of the impact of history, use, attitudes and proficiency in Basque (versus Spanish) in the prosodic realization of yes-​no questions and other sentence types in a larger pool of participants will shed further light on this issue.

5.  Conclusion Our study contributes to the on-​going investigation of the intonation of yes-​ no questions in Basque Spanish. Like Elordieta and Romera (in press), it includes a larger participant pool than previous studies, and provides additional evidence for the extensiveness of the circumflex (rising-​falling) nuclear pattern in information-​seeking questions. To our knowledge, this is also the first study to examine data from non-​information-​seeking contexts (surprise, confirmation, and requests) in this Spanish dialect, and therefore contributes to our understanding of the intonational characteristics of Basque Spanish, and of Spanish dialects more generally. Our study reports results for overall melodic pattern, as well as nuclear configuration contours, and the relative height of the nuclear High tone. Future research should examine additional acoustic cues, including the pitch of the first pre-​nuclear peak, its relative alignment, and the duration and intensity, particularly of the final syllable. A  more detailed investigation of syntactic cues (inversion, and the use of adverbials, for example) across question types will also be needed to explore whether Basque Spanish speakers rely on syntactic cues rather than intonation for some pragmatic contexts. Data analysis from additional participants, and from an additional reading task, will also shed further light on the intonational characteristics of yes-​no questions in this dialect.13

Notes 1 Madrid, Ecuadorian, Mexican and Chilean Spanish are included in this group (Hualde & Prieto, 2015 and references therein). 2 For additional examples, see section 2.2.

238  Carolina González and Lara Reglero 3 Elejabeitia et  al. (2005, 2007, 2008) analyze a total of seven female Basque Spanish speakers:  one Spanish monolingual from Bilbao, one from Donostia-​ San Sebastian, one Basque-​dominant speaker from Markina-​Xemein, and one Spanish-​Basque bilingual and three Spanish monolingual speakers from Vitoria-​ Gasteiz. Robles-​Puente (2011, 2012) analyze six and five participants from the Bilbao area, respectively. Elordieta and Romera’s (in press) study includes 12 participants with various degrees of bilingualism from Bilbao and Donostia-​San Sebastian. 4 For statements, peak alignment is timed in Lekeitio Spanish, showing a potential influence of Basque (Elordieta, 2003), but is delayed in Bilbao, Vitoria-​Gasteiz, Donostia-​San Sebastian, and Goierri Spanish. Delayed peak alignment is also common in Madrid Spanish (Elordieta, 2003; Calleja, 2004; Elordieta & Calleja, 2005; Elordieta & Irurtzun, 2012, 2016). 5 Participant 5 did not answer the section on Basque history in the BLP. However, her scores for Basque use and proficiency and her extreme dominance score support her inclusion in the “relatively Spanish-​dominant” group. 6 We are treating BLP scores here as binary. We leave an analysis of scores as points in a continuum for future analysis of all participants in the dataset. 7 Because of space reasons we cannot include the complete list of yes-​no questions and contexts used in this task. However, we are happy to share them with interested readers upon request. 8 One anonymous reviewer asks whether word order was altered in any of the expected questions. Even though this was an elicitation task, participants used the same word order as in the context statement, with only two exceptions: ¿ es tuya esa maleta? and ¿ es esa maleta suya? (rather than ¿ esta maleta es tuya? “This suitcase is yours?”). As noted by Escandell-​Vidal (1998), lack of inversion is common in yes-​no questions in Peninsular Spanish. Future research should compare inverted and non-​inverted yes-​no questions in this dialect to determine if any intonation differences apply between the two. 9 Cf. Robles-​Puente (2011), which reports a preference for declination suspension in his data. 10 Final rising contours in Basque are less common and are more typically found in some younger speakers who acquired Basque as a second language (possibly under the intonational influence of yes-​no questions in Spanish; Gaminde, Romero, Garay & Etxebarria, 2011; Elordieta & Hualde, 2014; Elordieta & Romera, in press). 11 A single contour in a cell indicates that it occurs in 75 percent to 100 percent of tokens. When two contours are given, each occurs in 50 percent of cases. If no contour occurs in at least 50 percent of cases, this is indicated by “inconclusive.” 12 Note that although participant 15 has the highest dominance score and rising contours in almost all cases, while participant 3 has the lowest dominance score and circumflex contours in all questions, other participants tend to have rising-​ falling contours regardless of dominance score. 13 Thank you to the editors of this volume and two anonymous reviewers for their feedback, and to the audience at HLS 2018 for their comments and questions. All errors are our responsibility. Special thanks to Jon Franco and Alex Iribar and the Phonetics Lab at the University of Deusto for their help with participant recruitment and data recording. This research was funded by a COFRS grant awarded to Lara Reglero in 2014–2015.

The intonation of yes-no questions  239

References Aguilar, L., De-​la-​Mota, C., & Prieto, P. (eds.) (2009). Sp_​ToBI training materials. Retrieved from Alvord, S.M. (2010). Disambiguating declarative and interrogative meaning with intonation in Miami Cuban Spanish. Southwest Journal of Linguistics, 28(2),  21–​66. Armstrong, M. (2017). Accounting for intonational form and function in Puerto Rican Spanish polar questions. Probus, 29, 1–​40. Arvaniti, A., & Ladd, D. (2009). Greek wh-​questions and the phonology of intonation. Phonology, 26, 43–​74. Astruc, L., Mora, E., & Simon, R. (2010). Venezuelan Andean Spanish intonation. In P. Prieto & P. Roseano (eds.), Transcription of Intonation of the Spanish Language,191–​226. Munich: Lincom Europa. Beckman, M., Díaz-​Campos, M., McGory, J.T., & Morgan, T.A. (2002). Intonation across Spanish, in the Tones and Break Indices framework. Probus, 14, 9–​36. Birdsong, D., Gertken, L., & Amengual, M. (2012). Bilingual language profile:  An easy-​to-​use instrument to assess bilingualism. COERLL, University of Texas at Austin. Retrieved from Boersma, P., & Weenink, D. (2015). Praat: Doing phonetics by computer [Computer program]. Version 6.0.15. Retrieved from Borràs-​Comes, J., Vanrell, M. del M., & Prieto, P. (2014). The role of pitch range in establishing intonational contrasts. JIPA, 44(1), 1–​20. Calleja, N. (2004). Alineamiento fonético de acentos tonales en el castellano de Vitoria. Estudios de Fonética Experimental, 13, 39–​63. Canellada, M.J. (1984). Notes de la entonación asturiana. Lletres Asturianes, 10, 23 – 26. Elejabeitia, A., Iribar, A., & Pagola, R.M. (2005). Notas sobre la prosodia del castellano en Bizkaia. Estudios de Fonética Experimental, 14, 247–​272. Elejabeitia, A., Iribar, A., & Pagola, R.M. (2007). Aproximación a la prosodia del castellano de Araba:  oraciones sin y con expansión en el SV. In M. González González, E. Fernández Rei & B. González Rei (eds.), III Congreso Internacional de Fonética Experimental, 249–​269. Santiago de Compostela: Xunta de Galicia. Elejabeitia, A., Iribar, A., & Pagola, R.M. (2008). La prosodia del castellano en tres ciudades vascas: Oraciones con y sin expansión en el SN. Language Design (Special Issue) 2, Experimental Prosody, 147–​154. Elejabeitia, A., Iribar, A., Pagola, R.M., & Feijóo, B. (2007). Rasgos prosódicos del castellano de Araba. In J. Dorta (ed.), La prosodia en el ámbito lingüístico románico, 245 –269. Madrid: La Página Ediciones. Elordieta, G. (2003). The Spanish intonation of speakers of a Basque pitch-​accent dialect. Catalan Journal of Linguistics, 2, 67–​95. Elordieta, G., & Calleja, N. (2005). Microvariation in accentual alignment in Basque Spanish. Language and Speech, 48, 397–​439. Elordieta, G., & Hualde, J.I. (2014). Intonation in Basque. In S-​A. Jun (ed.), Prosodic Typology II: The Phonology of Intonation and Phrasing, 405–​463. Oxford: Oxford University Press. Elordieta, G., & Irurtzun, A. (2012). Estudio comparativo de alineamiento tonal en el castellano de Bilbao. In B. Camus-​Bergareche & S. Gómez-​Seibane (eds.), El castellano del País Vasco, 119–​137. Bilbao: University of the Basque Country.

240  Carolina González and Lara Reglero Elordieta, G., & Irurtzun, A. (2016). Pitch accent tonal alignment in declarative sentences in the Spanish of the Basque Country: A study of language contact. In M. Armstrong, N. Henriksen & M. del M. Vanrell (eds.), Intonational Grammar in Ibero-​Romance:  Approaches across Linguistic Subfields, 25–​44. Amsterdam:  John Benjamins. Elordieta, G., & Romera, M. (in press). The influence of social factors on the prosody of Spanish in contact with Basque. In M. Bouzouita, R. Enghels & C. Vanderschueren (eds.), Different Perspectives on Convergence and Divergence in Ibero-​ Romance:  Language Contact and Contrasting National Varieties. Amsterdam: John Benjamins. Escandell Vidal, V. (1998). Intonation and procedural encoding: The case of Spanish interrogatives. In V. Rouchota & A. Jucker (eds.), Current Issues in Relevance Theory, 169–​203. Amsterdam: John Benjamins. Escandell Vidal, V. (1999). Los enunciados interrogativos:  aspectos semánticos y pragmáticos. NGDLE, 3, 3929–​3991. Escandell Vidal, V. (2013). Speech acts. In J.I. Hualde, A. Olarrea & E. O’Rourke (eds.), The Handbook of Hispanic Linguistics, 629–​651. Malden, MA: Wiley-​Blackwell. Estebas-​Vilaplana, E., & Prieto, P. (2008). La notación prosódica en español. Una revisión del Sp_​ToBI. Estudios de Fonética Experimental (EFE), 18, 263–​283. Estebas-​Vilaplana, E., & Prieto, P. (2010). Castilian Spanish intonation. In P. Prieto & P. Roseano (eds.), Transcription of Intonation of the Spanish Language, 17–​48. Munich: Lincom Europe. Face, T. (2002). When push comes to shove:  Tonal crowding in Madrid Spanish. Linguistic Association of Korea Journal, 10(1), 77–​100. Face, T. (2004). Intonation of absolute interrogatives in Castilian Spanish. Southwest Journal of Linguistics, 23(2), 65–​79. Face, T. (2008). The Intonation of Castilian Spanish Declaratives and Absolute Interrogatives. Munich: Lincom Europa. Face, T., & Prieto, P. (2007). Rising accents in Castilian Spanish: A revision of Sp_​ ToBI. Journal of Portuguese Linguistics, 6, 117–​146. Fernández Ramírez, S. (1959). Oraciones interrogativas españolas. Boletín de la Real Academia Española, 39, 243–​276. Gabriel, C., Feldhausen, I., Pesková, A., Colantoni, L., Lee, S.-​A., Arana, V., & Labastía, L. (2010). Argentinian Spanish intonation. In P. Prieto & P. Roseano (eds.), Transcription of Intonation of the Spanish Language, 285–​317. Munich:  Lincom Europa. Gaminde, I., Romero, A., Garay, U., & Etxebarria, A. (2011). Los tonos de frontera de las oraciones interrogativas absolutas producidas por hablantes bilingües vasco-​ español. Estudios de Lingüística Aplicada, 54, 61–​79. Henriksen, N. (2012). The intonation and signaling of declarative questions in Manchego Peninsular Spanish. Language and Speech, 55(4), 543–​576. Hualde, J.I. (2003). El modelo métrico y autosegmental. In P. Prieto (ed.), Teorías de la entonación, 155–​184. Barcelona: Ariel. Hualde, J.I. (2014). Los sonidos del español. Cambridge: Cambridge University Press. Hualde, J.I., & Prieto, P. (2015). Intonational variation in Spanish:  European and American varieties. In S. Frota & P. Prieto (eds.), Intonation in Romance, 350–​391. Oxford: Oxford University Press. Ladd, D.R. (2008). Intonational Phonology (2nd ed. [1st ed.  1996]). Cambridge: Cambridge University Press.

The intonation of yes-no questions  241 López-​Bobo, M.J., & Cuevas-​Alonso, M. (2010). Cantabrian Spanish intonation. In P. Prieto & P. Roseano (eds.), Transcription of Intonation of the Spanish Language, 49–​58. Munich: Lincom Europe. Navarro Tomás, T. (1944). Manual de pronunciación española. Madrid:  Consejo Superior de Investigaciones Científicas. Pàmies-​ Bertrán, A., Fernández-​ Planas, A.M., Martínez-​ Celdrán, E., Ortega-​ Escandell, A., & Amorós, M.C. (2002). Umbrales tonales en español peninsular. In J. Díaz García (ed.), Actas del II Congreso de Fonética Experimental, 272–​278. Sevilla: Universidad de Sevilla. Prieto, P. (2004). The search for phonological targets in the tonal space: H1 scaling and alignment in five sentence-​types in Peninsular Spanish. In T.L. Face (ed.), Laboratory Approaches to Spanish Phonology, 29–​59. Berlin: Mouton de Gruyter. Prieto, P., & Roseano, P. (2010). Transcription of Intonation of the Spanish Language. Munich: Lincom Europa. Quilis, A. (1987). Entonación dialectal hispánica. In H.L. Morales (Ed.), Actas del I congreso internacional sobre el español de América, 117–​164. San Juan: Academia Puertorriqueña de la Lengua Español. Quilis, A. (1993). Tratado de fonología y fonética españolas. Madrid: Gredos. Rietveld, A.C.M., & Gussenhoven, C. (1985). On the relation between pitch excursion size and prominence. Journal of Phonetics, 13, 299–​308. Robles-​ Puente, S. (2011). Absolute questions do not always have a rising pattern: Evidence from Bilbao Spanish. In S.M. Alvord (ed.), Selected Proceedings of the 5th Conference on Laboratory Approaches to Romance Phonology, 98–​107. Somerville, MA: Cascadilla Proceedings Project. Retrieved from Robles-​Puente, S. (2012). Two languages, two intonations? Statements and yes/​no questions in Spanish and Basque. International Journal of Basque Linguistics and Philology (ASJU), 46, 252–​262. Sosa, J.M. (1999). La entonación del español:  Su estructura fónica, variabilidad, y dialectología. Madrid: Cátedra. Sosa, J.M. (2003). La notación tonal del español en el modelo Sp_​ToBI. In P. Prieto (ed.), Teorías de la entonación, 155–​184. Barcelona: Ariel. Toledo, A.G. (2000). Taxonomía tonal en español. Language Design:  Journal of Theoretical and Experimental Linguistics, 3, 1–​20.

Part III

Acquisition patterns

12  How preschoolers acquire the null-​overt contrast in Mexican Spanish Evidence from production Hannah Forsythe, Dan Greeson and Cristina Schmitt

1.  Introduction One of the first things that anglophone learners of Spanish notice is the fact that subjects can be optionally omitted without causing ungrammaticality, as in (1). (1) Primero ∅ estábamos comiendo. First ∅ were eating. “First we were eating.” While the alternation between null and overt subjects is shared by many languages, the precise distribution and frequency of use of these forms varies across particular language varieties. The null/​overt alternation has long been studied from the perspective of sociolinguistic variation (Bayley & Pease-​ Alvarez, 1997; Flores-​Ferrán, 2007; Lastra & Butragueño, 2015; Michnowicz, 2015; Otheguy et al., 2007, 2010; Otheguy & Zentella, 2011; Shin & Otheguy, 2009), parametric variation (Barbosa, 2009, 2011 and references therein; Toribio, 2000), processing (Carminati, 2002; Filiaci et al., 2013; Jegerski et al., 2011; Keating et al., 2011, 2016), and second-​language acquisition (Montrul, 2004; Pérez-​Leroux & Glass, 1999). Yet only recently has attention turned toward the question of how the variability between null and overt subjects is acquired (Shin, 2012, 2016; Shin & Cairns, 2012). This is an important question because, in order for such a widespread, stable alternation to arise in the first place, it must be acquirable by children learning their first language. The question we address in this study is how children acquire this knowledge about the distribution of null and overt subjects. Acquiring the null/​overt contrast involves two steps. First, the learner must realize that her language licenses both null and overt subjects in tensed clauses, without overextending this alternation to nonfinite clauses or to operator constructions such as relative clauses (2), where the gap receives a different analysis. Language acquisition researchers have long known that this step is accomplished early (Austin et al., 1997; Grinstead, 2004). In fact, this

246  Forsythe, Greeson and Schmitt may be the preferred parameter setting in child language, since even children acquiring non-​pro-​drop languages pass through a stage in which they produce and accept null subjects alongside overt subjects (Hyams, 1986; Orfitelli & Hyams, 2012; Valian, 1990; Yang, 2002, among many others). (2) el muchacho que (*él) saludó a Ana the boy that (*he) greeted Ana Second, in those environments where null and overt subjects alternate, the learner must discover the semantic and pragmatic conditions that govern their use. This second step has received much less attention from the L1 research literature and it is where we focus our efforts. We focus specifically on the conditions governing the choice between an overt subject pronoun versus a null pronoun in the dialect of Spanish spoken in Mexico City, Mexico. How do children acquiring this variety of Spanish learn to correctly produce and interpret this contrast? We begin with a description of the knowledge that must be acquired and an explanation of the challenge that this presents to the learner (Section 2). In Section 3, we present a specific hypothesis about how children solve this problem: namely, that they begin by initially restricting their attention to 1st and 2nd person pronominal subjects. We identify three questions that must be answered in order to support this hypothesis: (i) Is the null/​overt distinction in principle acquirable from the 1st and 2nd person alone?; (ii) Do children show knowledge of this distinction in their own production of 1st and 2nd person pronominal subjects?; and (iii) Do children generalize that knowledge to 3rd person pronominal subjects? In Section 4 we present a brief review of the currently available acquisition results. Section 5 presents a corpus study addressing questions (i) – (iii). Results are consistent with the proposal that children acquire the conditions on subject pronoun realization within the domain of 1st and 2nd person subjects first, before generalizing to the 3rd person, and that this process occurs by age six. The chapter concludes with a summary of our findings, their limitations, and their impact. In particular, this study provides the earliest evidence to date of children’s sensitivity to the null/​overt pronoun contrast.

2.  Defining the learning problem An extensive literature reveals that adult speakers of pro-​drop languages tend to associate the null subject with the subject of the preceding clause but have either the opposite bias or no bias at all when interpreting the overt pronoun (Alonso-​Ovalle et  al., 2002; Carminati, 2002; de la Fuente, 2015; Papadopoulou et  al., 2015). First, we describe how this contrast is played out in the production and comprehension of so-​called “consistent” null subject languages like Spanish (Barbosa, 2009, among others). Next, we illustrate

Preschoolers and the null-overt contrast  247 how acquiring this contrast presents a challenge, since it requires the child to engage in the complicated task of pronoun resolution. In Spanish, listeners presented with grammatically ambiguous examples such as (3) (adapted from de la Fuente, 2015) tend to interpret the null subject (3a) as referring to the preceding subject antecedent Juan, whereas they have no such interpretive bias for the overt subject (3b). The subject-​antecedent reading is commonly referred to as the “same-​reference” reading because the same referent is referred to in both subject positions; all other readings tend to be grouped under the term “switch-​reference” (Otheguy & Zentella, 2011). In addition to judgments of ambiguous sentences like (3), online reading times and offline acceptability judgments of pragmatically disambiguated sentences show that the null subject is easier to process and more acceptable when presented in same-​reference contexts, relative to switch-​reference contexts, and vice versa for the overt pronoun (Carminati, 2002; Filiaci et  al., 2013; Jegerski et al., 2011; Keating et al., 2011, 2016). (3) a. Juan llamó a Pedro cuando ∅ estaba en casa. Juan called Pedro when _​_​ was at home. b. Juan llamó a Pedro cuando él estaba en casa. Juan called Pedro when he was at home. Sociolinguistic studies reveal the same biases in production as in comprehension. Across a host of language varieties, speakers tend to produce fewer overt pronominal subjects in same-​reference contexts relative to switch-​ reference contexts; in fact, this is consistently the strongest factor conditioning subject pronoun expression next to its person/​number features (Otheguy & Zentella, 2011; Carvalho, Orozco & Shin, 2015, and references therein). It is important to underscore that this contrast is probabilistic rather than categorical. Even though the null subject in (3a) favors the same-​reference reading more strongly than does the overt subject in (3b), both readings are available for both pronouns. Moreover, because pronoun interpretation is sensitive to multiple discourse-​pragmatic factors, the reading that adults ultimately prefer can be biased either towards the same-​reference reading or the switch-​reference reading, depending on context. For example, were it made known to the listener that Juan was calling Pedro in order to find out where Pedro was at that moment, then the switch-​reference reading would be favored over the same-​reference reading in both (3a) and (3b), i.e., the listener would likely infer that Juan called Pedro when Pedro (not Juan) was at home, regardless of which pronoun the speaker used. Nevertheless, the relative contrast between (3a) and (3b) would still stand, with the overt variant favoring the switch-​reference reading even more strongly than the null variant. The task for the L1 learner, then, is to associate null subjects with an increase in probability of a same-​reference reading, relative to overt pronominal subjects (rather than acquiring a categorical rule). To succeed at this task,

248  Forsythe, Greeson and Schmitt the child must track the relative rates of pronoun realization in same-​reference contexts versus switch-​reference contexts in her input. The problem that this task presents to the learner is that the correlation between same-​and switch-​reference contexts and null and overt pronouns can only be established (i.e., in the child’s mind) once the pronoun’s referent has already been identified. That is, if a child is to verify that a null subject like (3a) in her input indicates reference to the preceding subject, or that an overt subject like (3b) indicates reference to a non-​subject antecedent, she must know in advance what the intended referent of each pronoun is. However, pronoun resolution is a complex task, depending on a multitude of grammatical and discourse factors. Relevant grammatical cues to pronoun reference include the pronoun’s person, number, and gender features, as well as syntactic constraints like Principle B,1 and discourse cues include information from the context, such as what the current topic of conversation is, which discourse referents are most prominent, how events and states are pragmatically related to one another, and relevant real-​world knowledge about these referents and eventualities, to name just a few. The complexity of determining pronoun reference can be illustrated by any discourse situation that offers more than one sufficiently salient referent with the right grammatical features. For example, in the context of a story about Juan and Pedro hatching a plan to skip school, being discovered by the principal, and being disciplined by their fathers, a speaker could utter (3a) or (3b) using either the null or overt pronoun; or as illustrated in (4) the speaker could use any number of more semantically restricted noun phrases, such as a proper name or a modified noun phrase. Because pronouns are semantically underspecified, they could in principle refer to any of the characters in this story, in contrast to the other more semantically restricted DP subjects, as illustrated in (5). (4) Context: A story about Juan and Pedro skipping school, getting caught by the (male) principal, and being disciplined by both of their fathers. Juan llamó a Pedro cuando {Juan/​el niño travieso /​ø/​él} estaba en casa. Juan called Pedro when Juan/​the naughty boy/​pro/​he was at home. (5) Potential referents of the subject DP in (4) a. Juan: {Juan} b. el niño travieso: {Juan, Pedro} c. null subject: {Juan, Pedro, principal, Juan’s father, Pedro’s father} d. overt subject: {Juan, Pedro, principal, Juan’s father, Pedro’s father} What this example shows is not that sentences with null and overt pronominal subjects are impossible to interpret, but that doing so requires an inferential process that should not be taken for granted—​the child must consider

Preschoolers and the null-overt contrast  249 the set of potential referents and narrow it down to the referent she believes is intended by the speaker. Then—​and only then—​can she identify the sentence as a same-​reference or switch-​reference token and create the necessary association between that interpretation and the use of a null or overt subject pronoun.

3.  Proposed learning path In this section we present a specific proposal for how children might establish the probabilistic association between null and overt subject pronouns, and same-​and switch-​reference, respectively. In the remainder of the chapter we present evidence that this proposal is consistent with both the input and production of children acquiring Spanish. One possible learning strategy might be to delay acquisition of the null/​ overt contrast until the child has acquired enough knowledge of pragmatic cues, world knowledge, discourse relations, and other discourse cues to be able to resolve a sufficiently large portion of the subject pronouns in her input. While we agree that this learning path is possible in principle, we argue against it for a conceptual reason. First of all, this explanation forces us to assume that children initially ignore relevant information present in their input. Second, it merely pushes the learning problem back one step. If acquisition of the null/​overt distinction depends on prior acquisition of cues to pronoun resolution, how did children acquire those cues in the first place? We propose that a much simpler learning path becomes available if children initially restrict their attention to a subset of the subjects in their input: 1st and 2nd person pronouns. These pronouns refer to the speaker and addressee, respectively, and therefore present a much more restricted set of potential referents at the outset. Barring special circumstances,2 1st person pronouns refer directly to the speaker him or herself and are therefore completely unambiguous. Second person pronouns merely require the child to infer who the intended addressee is, which in one-​on-​one speech is the child herself. Furthermore these pronouns do not compete with a wide range of definite descriptions as 3rd person pronouns do (Pedro, he, the new neighbor, the principal, etc.). Because of these restrictions, it should be much simpler to identify the intended referent of 1st and 2nd person pronouns and thence to identify same-​and switch-​reference contexts like (6), compared to similar contexts with 3rd person pronouns like (3). (6) a. Tú llamaste a Pedro cuando ∅ estabas en casa. You called Pedro when _​_​ were at home. b. Pedro te llamó cuando tú estabas en casa. Pedro called you when you were at home. Once the child has formed an association between overt 1st and 2nd person pronominal subjects and an increased probability of switch-​reference, she can

250  Forsythe, Greeson and Schmitt then generalize this to all subject personal pronouns. We assume that this generalization is justified because 1st, 2nd and 3rd person pronouns fall into the same paradigm, differing only in the phi-​features that they carry (Charnavel, 2015). In sum, once the child learns that 1st and 2nd person null pronouns are statistically more likely to occur with preceding subject referents and overt variants with other referents, and that this applies to 3rd person pronouns in their input as well, the child can use the same variations in production and comprehension of all subject pronouns. Our proposal is summarized in (7). (7) Proposed learning path a. Step 1, Association: Track the realization of 1st and 2nd person pronominal subjects in same-​and switch-​reference contexts and associate switch-​reference readings with an increased probability of overt realization. b. Step 2, Generalization: Generalize this association to the production/​comprehension of all subject pronouns—​1st, 2nd and 3rd person. To show that this proposal meets the challenge of empirical adequacy, we must at minimum answer the following three questions: Q1: Is the null/​overt contrast in principle acquirable from the distribution of 1st and 2nd person pronouns in children’s input? Q2: Do children show sensitivity to this contrast in their own production of 1st and 2nd person pronouns? Q3: Do children generalize this contrast to the production/​comprehension of 3rd person pronominal subjects after they have acquired this contrast in the domain of 1st and 2nd person? We leave it to future work to show that the alternative strategy (namely, delaying acquisition of the null/​overt distinction until the child has mastered pronoun resolution) is empirically inadequate, in addition to being conceptually undesirable. Before addressing these questions with data of our own, we will review what is currently known from existing data in languages that display the null/​overt contrast.

4.  Acquisition background What we know so far about the developmental path of acquiring the distribution of null and overt subjects comes from studies of spontaneous production, felicity judgments, and pronoun resolution tasks in a variety of languages. We concentrate here on pro-​drop languages with rich agreement. Children learning these languages must associate same-​reference contexts

Preschoolers and the null-overt contrast  251 with a decreased rate of overt pronoun realization, relative to switch-​reference contexts, although the strength of this association may vary across language varieties (Filiaci et al., 2013; de la Fuente, 2015). Production studies from the acquisition literature find that children acquiring Spanish and Catalán begin to produce overt subjects before age two (Bel, 2003; Grinstead, 2004). However, overt pronominal subjects remain infrequent. Sociolinguistic studies that include child subjects corroborate the low frequency of overt pronominal subjects well into middle and late childhood. In Mexican Spanish, first graders (ages six–​seven) overtly realize pronouns at a rate of 6–​8% (Shin, 2012, 2016), and this rate gradually increases to about 10  percent in 5th grade (ages 10–​11)—​far less than the 18–​22% rate found among Mexican adults (Lastra & Butragueño, 2015; Shin & Erker, 2015; Shin & Otheguy, 2013). Despite being infrequent, however, overt pronouns are not randomly distributed. Shin (2016) finds that overt realization is positively associated with switch-​reference contexts among even the youngest age group (six–​seven years). Felicity judgment tasks show that children are much more accepting than adults of null subjects in switch-​reference contexts as well as of overt subjects in same-​reference contexts. Sorace et  al. (2009) asked bilingual children, monolingual children, and adult speakers of Italian to judge which of two speakers produced a better description of an event, with one speaker producing a null subject (8a) and the other an overt subject (8b). Crucially, the event was manipulated such that either Minnie (a same reference context) or another person (a switch-​reference context) had fallen. Monolingual Italian-​ acquiring children ages six–​seven and eight–​ten were just as likely as adults to prefer the overt pronoun in the switch-​reference condition; however, only the eight–​ten-​year-​olds were just as likely as adults to prefer the null pronoun in the same-​reference condition. Bilinguals were overall less adult-​like than their monolingual peers—​even those whose other language was Spanish, another pro-​drop language. (8) a. Minnie ha detto che ø è caduta. Minnie has said that (she) has fallen. b. Minnie ha detto che lei è caduta. Minnie has said that she has fallen.

(Sorace et al., 2009)

Shin & Cairns (2012) used a similar methodology to probe the preferences of Mexican Spanish-​acquiring children from ages 6 to 15. They found a qualitatively similar but quantitatively slower developmental trajectory. In the switch-​reference condition, eight-to-​ten-​year-​olds showed a significant preference for an overt subject, although the strength of this preference did not match adults until age 14–​15. In the same-​reference condition, even the oldest children failed to show a significant preference for the null subject. Finally, studies among L2 learners have shown an even more pronounced version of

252  Forsythe, Greeson and Schmitt this pattern of non-​target behavior (Belletti et al., 2007; Montrul, 2004, 2011; White, 2011, among others). These results suggest a very protracted developmental path; however, these tasks may underestimate what children actually know about the null/​overt contrast at earlier stages in development. The tasks employed in these studies are cognitively complex, requiring the listener to hold two utterances in short-​term memory while making a metalinguistic judgment about them. To lighten the cognitive load of the task in order to better get at children’s knowledge, Papadopoulou et al. (2015) used a less taxing method to probe Greek-​acquiring children’s sensitivity to the null/​ overt distinction from ages 6 to 11 and found more adult-​like performance. Participants listened to a sentence like (9a) while simultaneously viewing a picture that corresponded to either a same-​reference interpretation (pro = the old man), or one of two switch-​reference interpretations (object interpretation: pro = his grandchild; other interpretation: pro = another person). At the end of each sentence, the task was simply to judge whether the sentence matched the picture. Additionally, the audio was self-​paced:  participants would press the spacebar to hear each phrase of the sentence, and their listening times were measured. A second experiment used the same methodology with overt pronouns, as in (9b). (9) a. O papús millúse δinatá ston egonó tu ótan ø δjávaze ena vivlío. The old-​man spoke-​3SG loudly to his grandchild when (he) read-​3SG a book. b. I jajá xerétise tin kipéla ótan aftí pernúse to δromo. The old-​lady greeted-​3SG the girl when she crossed-​3SG the street. Like adults, children of all ages accepted the same-​reference reading of the null subject nearly all the time, they accepted the object reading less often, and they accepted the “other” reading even less often. In the overt subject experiment, children of all ages were like adults in accepting the object reading most of the time and the “other” reading less often, but unlike adults they over-​ accepted the same-​reference reading of the overt pronoun until ages 10–​11. Nevertheless, listening times showed that even in this condition the younger children still processed the object reading faster than the same-​reference reading. Summing up, in languages with a similar null/​overt contrast to Spanish, children as young as six show sensitivity to the association between subject pronoun expression and pronominal reference in their own production as well as their interpretation of ambiguous pronouns. Unfortunately, not much is known about the trajectory of development before this age, after the onset of overt subjects at around two. We help fill this gap by contributing production and comprehension data for Spanish-​acquiring children ages 3 to 6.

Preschoolers and the null-overt contrast  253

5.  Corpus study: subject realization Our proposal claims that children acquire the conditions governing the null/​ overt contrast from the distribution of 1st and 2nd person pronominal subjects in their input. The first question to address is whether the input even provides the necessary information—​after all, 1st and 2nd person pronouns have a different discourse function and therefore may not show the same patterns. We do this by examining naturalistic child-​directed speech by monolingual speakers of Spanish native to Mexico City. The second question is whether children’s own production demonstrates awareness of this contrast, and if so, whether it is acquired first in the domain of 1st and 2nd person subjects before being generalized to the 3rd person. We used data from the Miller–Schmitt corpus (collected in Mexico City in 2008), comprising approximately 649,000 words of spontaneous speech from 11 low-​SES and 14 high-​SES child-​caretaker dyads (child ages 1;6–​5;11). Caretakers were recorded during sessions playing with their children, as well as one session each chatting with another adult. 5.1  Subject characteristics We examined a subset of 103,249 words taken from the speech of five mother-​ child dyads, whose characteristics are summarized in Table 12.1. 5.2  Hypotheses and predictions The learning path hypothesized in Section 3 makes three predictions. First, if the null/​overt distinction is learnable from 1st and 2nd person pronouns in the input, we expect that mothers in our sample will produce significantly more overt 1st and 2nd person pronominal subjects in switch-​reference contexts than they do in same-​reference contexts.

Table 12.1 Subject characteristics of 5 Mexican Spanish-​speaking dyads from the Schmitt-​Miller  corpus Child

Age

Mean length of utterance

Mother word count

Child word count

YGSZ YBM OMJ KUC JRC

3;9 4;8 5;2 4;1 5;11 Mean: 4;9

3.652 3.993 3.87 4.522 3.735 Mean: 4.02

9,608 11,054 11,934 11,721 13,114 Total: 93,024

10,910 8,373 7,314 9,393 10,548 Total: 69,552

254  Forsythe, Greeson and Schmitt Second, if children become attuned to this statistical pattern at some time before six years of age, we expect that some portion of the children in our sample will show the same pattern in their own production of 1st and 2nd person pronominal subjects. Finally, if children generalize constraints from 1st and 2nd person to the 3rd person (rather than the other way around), then we would predict that a slightly older group of children in our sample will show this pattern in their production of 3rd person pronominal subjects as well. 5.3  Coding methods From our sample, we extracted all animate, pronominal subjects of tensed verbs that were preceded by at least one other clause in the same turn, defined as the longest uninterrupted string of speech by one person. These subjects were then coded for form (null, overt) and reference (same, switch) following the definition in Otheguy and Zentella (2011, appendix), which we illustrate in 5.3.2. 5.3.1  Exclusions We excluded the subjects of tensed verbs in several conditions, as described in (10). Following Otheguy and Zentella (2011), we excluded tensed verbs in conditions (a–b); following Shin (2016) we excluded tensed verbs in conditions (c–d). We excluded all subjects of dative-​experiencer predicates (e), since the surface/​syntactic subject is not necessarily the same as the underlying/​ semantic subject in these cases. We excluded generic subjects (f) because these uses are nearly exclusively null. We also excluded any pre-​authored material, such as passages from books, song lyrics, or other recited material, since it corresponds to another register and/​or speaker. (10) Exclusions a. frozen expressions: quién sabe, ¿sabes?, ¿ves?, ¿ya viste?, and so on. b. Subject relatives: El niño que ∅ toca el piano (“The boy that plays the piano””) c. subjects of imperatives: Dime ∅ cuántos años tienes d. the phrase no sé (“I don’t know”) e. Dative experiencer–​predicate constructions: me gusta eso (“I like that”; literally, “that pleases me”), le encanta la playa (“he loves the beach”; literally, “the beach charms him”), etc. f. Generic references to “them” or “people,” such as dicen que… (“they say that…”) and en la escuela me dieron de comer (“at school (they) gave me something to eat”) g. Passages read from books, song lyrics, nursery rhymes, etc.

Preschoolers and the null-overt contrast  255 5.3.2  Definition of same-​reference and switch-​reference We coded a subject as same-​reference if it referred to the same referent as the subject of the preceding tensed verb within the same speaker turn. In the hypothetical example (11), the null subject of tienes sueño (“you’re sleepy”) refers to the same entity as the subject of estás bostezando (“you are yawning”) and would therefore be coded as a same-​reference token. We coded a subject as switch-​reference if it referred to a non-​subject argument of the preceding tensed verb, to an entirely new referent never mentioned before, or to a referent that had been mentioned before, but in a preceding turn (see also example (13), below, for our operational definition of what counts as an eligible preceding verb). In (11), for example, the null subject of estás bostezando (“you are yawning”) would be coded as a switch-​reference token because it refers to the same referent as the preceding object of veo (“I see”). The subject of vamos (“we’ll leave”) would also be coded as switch-​reference because it does not match in person and number features with the subject of the preceding verb tienes sueño (“you’re sleepy”). Finally, we excluded any subjects that were not preceded by another tensed verb in the same turn. For example, the first subject in (11), yo (“I”) has no preceding subject with which to maintain or switch reference, so it is excluded as a token, though it still serves as a suitable “trigger” for coding the subject of the following verb, estás. (11) Yo te veo que ∅ estás bostezando. Si ∅ tienes sueño, ∅ nos vamos. I see you that (you) are yawning. If (you)’re sleepy, (we)’ll leave. In some cases, a subject pronoun was rendered trivially switch-​reference by the properties of the preceding subject—​usually because the preceding subject was non-​referential.3 In these cases, the preceding subject was skipped and the preceding subject before that was used instead. The following criteria were used to determine whether the preceding subject was an eligible “trigger” for coding same-​and switch-​reference: (12) Eligibility of the preceding subject for coding same-​and switch-​reference a. Subjects of imperatives and inanimate subjects were considered eligible triggers, even though they were not themselves coded as tokens. The exception to this rule was set imperative phrases like mira, oye, etc. b. Subjects that do not refer to individuals were considered ineligible (see footnote 3 for some examples). This includes: subjects of frozen expressions, clausal subjects, and expletive subjects (es que, lo que quiero es, etc.) and weather verb subjects. c. We excluded se-​impersonal/​passives as well as 3rd person plural subject like dicen que (“they say that”).

256  Forsythe, Greeson and Schmitt d. We excluded traces of wh-​operators. e. Contra Otheguy & Zentella (2011) we did not consider the grammatical subjects of experiencer-​predicate constructions (e.g., me gusta el agua, literally “water pleases me”) or the subject of presentational haber (e.g., habían tres gatos, “there were three cats”) to be eligible triggers. 5.3.3  Inter-​rater reliability Two different raters coded the form (null, overt) and reference (same-​reference, switch-​reference) of all subject personal pronouns produced by the children and mothers in our sample. To calculate inter-​rater reliability, both raters independently coded a subset consisting of six transcripts (comprising 18% of the data) and their ratings for each token were compared. Inter-​rater reliability was 96.6% (κ=.84) for pronominal form and 95.7% (κ=.91) for reference. 5.3.4  Examples Data from our corpus analyses illustrate that both null and overt subjects can occur in same-​and switch-​reference contexts in children’s naturalistic input. In example (13), produced by the mother YBM (4;5) we find two same-​ reference tokens that illustrate this. The speaker’s turn begins with a null subject, and the subjects of both the following verbs (es, tiene) maintain reference to the same referent (a dancer on a tightrope). In one case, the mother uses an overt subject (ella) and in the second she uses a null subject. (13) ø estuvo a punto de caerse, “(She) was about to fall, pero no, porque {ella} es una experta bailarina but no, because she is an expert dancer y {ø} tiene todo el equilibrio para poder bailar en una cuerda floja! and (she) has all the balance to be able to dance on a tightrope!” The same is true for switch-​reference tokens, as illustrated in (14)–(15), from the same mother. The speaker begins her turn with an imperative, whose (null) subject refers to the child, then she switches reference to herself using an overt pronoun (yo). Finally, in (15), the mother begins her turn referring to herself (with a null subject) and switches reference to her daughter, this time using a null pronoun. (14) Context: Mother encourages daughter to sing a lullaby to some lions. Cántasela “(you) sing it to them, Para que {yo}me siente un ratito so that I can sit down for a sec”

Preschoolers and the null-overt contrast  257 (15) Context: Mother pretending to be a doctor prescribing “luneta” candies. Okey, entonces ø ya no le doy esta receta “Okay, then (I) won’t give you this prescription y ya {ø} no va a comer lunetas nunca más en la vida and now (you) will never eat lunetas ever again in your life.” These examples confirm that in child-​directed speech, as in adult-​directed speech, there is not a categorical requirement to use null subjects in same-​ reference contexts or to use overt subjects in switch-​reference contexts. Rather, if these associations are attested in child-​directed speech, they will be probabilistic in nature. 5.4  Results: overall rate of overt pronoun realization Mothers and children in our sample produced overt pronouns at very similar rates to each other (12.7% for mothers, 12.6% for children). These rates differed somewhat from subject pronoun expression rates found in other studies (see Table 12.2 for comparison), which could be due to differences in coding decisions as well as differences in the nature of the interactions that were recorded. The child-​directed speech produced by the mothers in our sample contains fewer overt pronouns than adult-​directed speech from Mexico City as reported by Lastra & Butragueño (2015) (12.7% versus 21.7%). This may be because child-​directed speech tends to reference familiar items, potentially increasing the rate of null pronouns. Or it may be due to a difference in the sampling method: because we included only tokens preceded by another clause within the turn, we excluded turn-​initial pronouns, potentially under-​sampling the overt variant. On the other hand, the children in our sample produced overt pronouns more often than the slightly older children from other regions in Mexico reported by Shin (2016) (12.6% versus 8%). This difference may be due to the Table 12.2 Rates of overt pronoun use % overt pronouns Mothers (this study) Adult-​directed speech, Mexico City (Lastra & Butragueño, 2015) Children ages 3–​6 (this study) Children ages 6–​7, Querétaro & Oaxaca (Shin, 2016)

12.7% (540/​4,320) 21.7% (443/​2,040) 12.6% (417/​3,314) 8% (148/​1,845)

258  Forsythe, Greeson and Schmitt nature of the interactions between speakers. Data from Shin (2016) is mainly from narrative contexts, which tend to feature a single referent to whom a narrator refers multiple times. In other words, narratives are dominated by same-​reference segments, thereby encouraging more null subjects. In contrast, the Schmitt-​Miller corpus recorded mostly one-​on-​one interactions between parent and child, which may have led to fewer same-​reference segments, and which may also have encouraged children to assimilate to their mothers’ input, leading to more overt subjects. 5.5  Results: pronoun realization in same/​switch-​reference contexts The learning path proposed in Section 3 depends upon the assumption that children’s input provides the necessary statistical information to acquire the null/​overt distinction from 1st and 2nd person subject pronouns alone. Hence, the first prediction to test is whether 1st and 2nd person subject pronouns in child-​directed speech are indeed overtly realized more often in switch-​ reference compared to same-​reference contexts. Figure  12.1 shows the rate and frequency of overt pronouns in mother’s speech (left) and children’s speech (right) in each of these contexts, across 1st, 2nd and 3rd person domains. To test the first prediction, we employed separate chi-​squared tests of proportion to compare the rate of overt pronoun production across same-​ reference and switch-​reference contexts in mother’s speech, for each of the three personal pronouns. For mothers, the effect of reference was significant for 1st and 2nd but not 3rd person subjects. That is, mothers produced significantly more overt pronominal subjects in switch-​reference contexts relative to same-​reference contexts in the 1st (χ (1) = 7.54, p < 0.007) and 2nd person

Mothers

Children

1 0.8 0.6 0.4

same **

0

(55/337) (21/344)

0.2

1

**

ns

(69/434)

(13/180)

switch

**

***

2

(6/78)

(46/311) (11/188)

(28/138)

(9/117)

(5/109)

3

*

1

2

(7/93) (3/123) 3

Figure 12.1 Rate of overt 1st, 2nd, and 3rd person pronouns appearing in same-​and switch-​reference contexts in the speech of mothers (left) and children (right).

Preschoolers and the null-overt contrast  259 (χ (1) = 16.903, p < 0.001), but not in the 3rd person (χ (1) = 0.15, p = 0.70). Thus, the statistical contrast between null and overt pronominal subjects is not only present in the 1st and 2nd person, it may be stronger there. Given that 1st and 2nd person pronominal subjects seem to provide a robust statistical signal, the second prediction to test is whether children are sensitive to this signal and reflect it in their own production of 1st and 2nd person pronouns. To test this, we ran the same chi-​squared tests of proportion on children’s production of overt pronominal subjects. Children produced significantly more overt pronominal subjects in switch-​reference contexts relative to same-​reference contexts across both 1st and 2nd persons (1st person: χ (1) = 8.40, p < 0.004; 2nd person: χ (1) = 7.71, p < 0.006). Thus, this sample provides evidence that even children under six are sensitive to the null/​overt contrast. The third prediction to test is whether children generalize this distinction to the 3rd person. A chi-​squared test of proportions finds a significant difference in the realization of 3rd person pronouns across same-​and switch-​ reference contexts (χ (1) = 5.48, p = 0.019), indicating that, as a group, these children have generalized the null/​overt distinction to the 3rd person as well. Given that this difference was not significant in mothers’ speech, this would seem to indicate that children have inferred a statistical pattern that is actually absent in the input. However, this conclusion is not warranted. The failure to reach statistical significance does not constitute evidence that no such difference exists, since a larger sample of the input may reveal differences that our sample failed to detect. Rather, we take the absence of a significant difference in mothers’ realization of 3rd person pronominal subjects as an indication that the statistical signal provided by the input may be weaker in the 3rd person than it is in the 1st and 2nd persons. In order to establish that children generalize knowledge acquired in the 1st and 2nd person to the domain of the 3rd person, rather than the other way around, we need evidence that the null/​overt distinction is acquired earlier in the 1st and 2nd person. Figure 12.2 provides a more detailed view of the developmental trajectory of the null/​overt contrast across individual children, separating 1st and 2nd person from 3rd person. All children in our sample produced overt pronouns more often in switch-​reference contexts compared to same-​reference contexts, but the difference reached significance at a younger age in the 1st and 2nd person. One-​sided chi-​squared tests of proportion for each individual child revealed significantly more overt 1st and 2nd person pronouns in switch-​versus same-​reference contexts for YBM (4;8, χ (1) = 6.93, p = 0.004) and OMJ (5;1, χ (1) = 3.77, p = 0.026) and a marginal difference for the oldest child, JRC (5;11, χ (1) = 1.97, p = 0.08). For 3rd person pronominal subjects, however, this difference was significant only in the speech of the oldest child (JRC, 5;11, χ (1)  =  2.87, p  =  0.045). This suggests that children first acquire the null/​overt contrast as it applies to 1st and 2nd person pronominal subjects before generalizing to the 3rd person.

260  Forsythe, Greeson and Schmitt 1st and 2nd person subjects

1

Animate 3rd person subjects

same switch

0.8 0.6 0.4

ns

ns

**

*

KUC (4;5)

YBM (4;8)

OMJ (5;1)

(.)

ns

JRC (5;11)

YGSZ (3;9)

ns

ns

ns

*

0.2 0 YGSZ (3;9)

KUC (4;5)

YBM (4;8)

OMJ (5;1)

JRC (5;11)

Figure 12.2 Rate of overt pronominal subjects produced by individual children, in same-​reference and switch-​reference contexts. Left:  1st and 2nd person pronominal subjects. Right: animate 3rd person pronominal subjects.

If this sample of five children is representative of acquisition more generally, then it suggests that typically developing children begin to acquire the null/​overt contrast around age four and a half. However, a larger sample, preferably longitudinal in nature, would be required to make this claim more confidently. 5.5  General discussion and study limitations Our proposal generated three predictions that are supported by the production data in our sample of naturalistic speech. First, analysis of the input reveals that the null/​overt distinction is in principle acquirable from 1st and 2nd pronouns alone: mothers produce significantly more overt 1st and 2nd pronominal subjects in switch-​reference contexts relative to same-​reference contexts. In fact, this statistical signal may even be stronger in the 1st and 2nd person than it is in the 3rd person, where the same contrast fails to reach significance. Assuming that the input produced by these mothers can be generalized to caretaker speech at similar developmental stages, then this would suggest that children not only can learn the null/​overt distinction from 1st and 2nd person subjects, it may actually be easier to do so. Second, children’s own production reveals that they have picked up on this statistical signal early in life:  they too produce significantly more overt 1st and 2nd person pronouns in switch-​reference compared to same-​reference contexts. Third, children’s production of 3rd person subject pronouns shows the same statistical pattern, indicating that they have generalized the null/​ overt contrast to apply to all persons. Furthermore, the individual-​level analysis reveals earlier acquisition of this distinction in the 1st and 2nd persons (between 4;5 and 4;8), relative to the 3rd person (by 5;11), indicating that the

Preschoolers and the null-overt contrast  261 generalization is made from 1st and 2nd persons to the 3rd person, rather than running in the other direction. Our proposal makes one prediction that we have not tested here. If children generalize knowledge of the null/​overt contrast from 1st and 2nd person to the 3rd person, then this knowledge should be available to them in the comprehension of 3rd person pronominal subjects as well—​not just in production. If children acquire the null/​overt contrast around age four and a half, as our individual-​level analysis suggests, then we would expect children of that age to comprehend the contrast between null and overt 3rd person subjects at that same age. Interestingly, comprehension data from Forsythe (2017) shows that children age 4;7–​6;4, but not younger, use the null/​overt contrast to help resolve grammatically ambiguous 3rd person subject pronouns (see also Forsythe, Greeson & Schmitt, 2019). There are two main limitations to this study: its small sample size and its limited comparability with the sociolinguistic literature. The overall number of tokens on which this study is based is comparable to that reported in other studies on the spontaneous production of null and overt pronouns (e.g., Lastra & Butragueño 2015, six-to-seven-​year-​olds in Shin 2016). Nevertheless, the number of individuals involved is smaller, especially considering the wide age range of the children in our sample (3;9–5;11). Currently, we are working to increase our sample size. A second limitation is that our sampling method prevents us from making a direct comparison with standard sociolinguistic studies on this topic. Breaking with the standard of Otheguy & Zentella (2011), we made the decision to limit our analysis to pronoun reference within a single turn, in order to focus our attention on pronouns with locally available antecedents. While we ultimately replicated the effect of same-​reference versus switch-​reference contexts found by all standard sociolinguistic studies, we did find a slightly different overall rate of pronoun realization than other studies on this variety of Spanish, and our results should therefore be viewed with appropriate caution.

6.  Conclusion In this study, we asked how children acquiring Spanish as their first language can learn the contrast between null and overt subjects given the complexity of the task. Specifically, we sought a way around the problem of having to identify pronominal antecedents in order to acquire the association between null subjects and same-​reference, on the one hand, and overt pronominal subjects and switch-​reference, on the other. We proposed that children solve this problem by first tracking the realization of 1st and 2nd person pronominal subjects, whose reference is easier to infer, and we provided evidence consistent with that proposal from the Schmitt-​Miller corpus of Mexico City parent-​child speech (Miller & Schmitt, 2012). Finally, in addition to providing a novel proposal for how children acquire the conditions governing the use of null and overt pronominal subjects, this

262  Forsythe, Greeson and Schmitt study provides the first evidence we know of that children under age six are sensitive to the conditions governing the use of null and overt subjects, filling a crucial gap in the literature on this otherwise well-​studied alternation.

Notes 1 Principle B refers to the constraint that prohibits pronoun coreference with antecedents in certain structural positions. Roughly, the principle states that pronouns must be free in their domain, where “domain” roughly corresponds to a clause. This explains the contrast between (i), where the pronoun him cannot refer to the subject of its clause Bill, and (ii), where the pronoun him can refer to Bill when Bill is the subject of a higher clause. (i) Bill saw him. him ≠ Bill (ii) Bill said that John saw him. him = Bill 2 Kratzer (2009) and Heim (2008) identify cases of “fake indexicals” where 1st and 2nd person pronouns do not refer directly to the speaker or hearer, but instead receive a bound variable reading. For example, the sentence Only I eat what I cook, implies that there is no other person x such that x eats what x cooks. To produce this meaning, the embedded pronoun must be referentially dependent on the matrix pronoun. 3 We use the term “non-​ referential subjects” to refer to expletive subjects (i), impersonals (ii), and wh-traces (iii). In these examples, the subject pronoun ella (“she”) fails to refer to the same entity as the preceding subject, not because these are genuine cases of switch-​reference, but because the preceding subject simply does not refer to an entity in the first place. (i) Cuando ø llueve, ¿va a llevar ella su sombrero? When (it) rains, is she going to take her hat? (ii) ø se dice que ella va a llevar su sombrero. (They) say that she is going to take her hat. (iii) ¿Quién dijo que ella va a llevar su sombrero? Who says that she is going to take her hat?

References Alonso-​Ovalle, L., Fernández-​Solera, S., Frazier, L., & Clifton, C. (2002). Null vs. overt pronouns and the topic-​focus articulation in Spanish. Italian Journal of Linguistics, 14, 151–​170. Austin, J., Blume, M., Parkinson, D., del Prado, Z.N., & Lust, B. (1997). The status of pro-​drop in the initial state: Results from new analyses of Spanish. Contemporary Perspectives on the Acquisition of Spanish, 1, 37–​54. Barbosa, P.P. (2009). Two kinds of subject pro. Studia Linguistica, 63(1), 2–​58. Barbosa, P.P. (2011). Pro-​ drop and theories of pro in the minimalist program Part  1:  Consistent null subject languages and the pronominal-​ agr hypothesis. Language and Linguistics Compass, 5(8), 551–​570.

Preschoolers and the null-overt contrast  263 Bayley, R., & Pease-​Alvarez, L. (1997). Null pronoun variation in Mexican-​descent children’s narrative discourse. Language Variation and Change, 9(3), 349–​371. Bel, A. (2003). The syntax of subjects in the acquisition of Spanish and Catalan. Probus, 15(1), 1–​26. doi:10.1515/​prbs.2003.003 Belletti, A., Bennati, E., & Sorace, A. (2007). Theoretical and developmental issues in the syntax of subjects: Evidence from near-​native Italian. Natural Language & Linguistic Theory, 25(4), 657. Carminati, M.N. (2002). The processing of Italian subject pronouns (Doctoral dissertation). Retrieved from ProQuest Information & Learning (AAI3039345). Carvalho, A.M., Orozco, R., & Shin, N.L. (2015). Subject Pronoun Expression in Spanish: A Cross-​Dialectal Perspective. Washington, DC: Georgetown University Press. Charnavel, I. (2015). Let you be bound to me (and me to you). 33rd WCCFL Poster Session Proceedings. SFU’s Working Papers in Linguistics, 18–​27. Retrieved from de la Fuente, I. (2015). Putting pronoun resolution in context:  The role of syntax, semantics, and pragmatics in pronoun interpretation (Doctoral dissertation). Université Paris Diderot. Filiaci, F., Sorace, A., & Carreiras, M. (2013). Anaphoric biases of null and overt subjects in Italian and Spanish: A cross-​linguistic comparison. Language, Cognition and Neuroscience, 29(7), 825–​843. DOI: 10.1080/​01690965.2013.801502 Flores-​Ferrán, N. (2007). A bend in the road: Subject personal pronoun expression in Spanish after 30  years of sociolinguistic research. Language and Linguistics Compass, 1(6), 624–​652. Forsythe, H. (2017). Discourse and grammatical cues in the acquisition of Spanish pronouns (Doctoral dissertation). Michigan State University. Forsythe, H., Greeson, D., & Schmitt, C. (2019). Learnability in Romance:  How indirect input helps children acquire the contrast between null and overt subjects. In M.M. Brown & B. Dailey (eds.), Proceedings of the 43rd Boston University Conference on Language Development, 231–​ 244. Somerville, MA: Cascadilla Press. Grinstead, J. (2004). Subjects and interface delay in child Spanish and Catalan. Language, 80(1), 40–​72. Heim, I. (2008). Features on bound pronouns. In D. Harbour, D. Adger & S. Bejar (eds.), Phi Theory:  Phi Features across Modules and Interfaces, 35–​56. Oxford University Press. Hyams, N. (1986). Language Acquisition and the Theory of Parameters. Dordrecht: Reidel Publishing Company. Jegerski, J., VanPatten, B., & Keating, G.D. (2011). Cross-​linguistic variation and the acquisition of pronominal reference in L2 Spanish. Second Language Research, 27(4), 481–​507. Keating, G.D., VanPatten, B., & Jegerski, J. (2011). Who was walking on the beach?: Anaphora resolution in Spanish heritage speakers and adult second language learners. Studies in Second Language Acquisition, 33(2), 193–​221. Keating, G.D., Jegerski, J., & VanPatten, B. (2016). Online processing of subject pronouns in monolingual and heritage bilingual speakers of Mexican Spanish. Bilingualism: Language and Cognition, 19(1), 36–​49.

264  Forsythe, Greeson and Schmitt Kratzer, A. (2009). Making a pronoun: Fake indexicals as windows into the properties of pronouns. Linguistic Inquiry, 40(2), 187–​237. Lastra, Y., & Butragueño, P.M. (2015). Subject pronoun expression in oral Mexican Spanish. In A. Carvalho, R. Orozco & N. Shin (eds.), Subject Pronoun Expression in Spanish:  A Cross-​Dialectal Perspective, 39–​57. Washington, DC:  Georgetown University Press. Michnowicz, J. (2015). Subject pronoun expression in contact with Maya in Yucatan Spanish. In A. Carvalho, R. Orozco & N. Shin (eds.), Subject Pronoun Expression in Spanish: A Cross-​Dialectal Perspective, 103–​122. Washington, DC: Georgetown University Press. Miller, Karen L, & Schmitt, Cristina (2012). Variable input and the acquisition of plural morphology. Language Acquisition, 19(3), 223–261. Montrul, S. (2004). Subject and object expression in Spanish heritage speakers:  A case of morphosyntactic convergence. Bilingualism: Language and Cognition, 7(2), 125–​142. Montrul, S. (2011). Multiple interfaces and incomplete acquisition. Lingua, 121(4), 591–​604. Orfitelli, R., & Hyams, N. (2012). Children’s grammar of null subjects: Evidence from comprehension. Linguistic Inquiry, 43(4), 563–​590. Otheguy, R., & Zentella, A. C. (2011). Spanish in New  York:  Language Contact, Dialectal Leveling, and Structural Continuity. Oxford: Oxford University Press. Otheguy, R., Zentella, A.C., & Livert, D. (2007). Language and dialect contact in Spanish in New York: Toward the formation of a speech community. Language, 83(4), 770–​802. Otheguy, R., Zentella, A.C., & Livert, D. (2010). Generational differences in pronominal usage In Spanish reflecting language and dialect contact in a bilingual setting. In M. Norde, B. de Jonge & C. Hasselblatt (eds.), Language Contact: New Perspectives, 45–​62. Amsterdam: John Benjamins. Papadopoulou, D., Peristeri, E., Plemenou, E., Marinis, T., & Tsimpli, I. (2015). Pronoun ambiguity resolution in Greek:  Evidence from monolingual adults and children. Lingua,155, 98–​120. doi:10.1016/​j.lingua.2014.09.006 Pérez-​Leroux, A.T., & Glass, W.R. (1999). Null anaphora in Spanish second language acquisition: probabilistic versus generative approaches. Second Language Research, 15(2), 220–​249. Shin, N.L. (2012). Variable use of Spanish subject pronouns by monolingual children in Mexico. In K. Geeslin & M. Díaz-​Campos (Eds.), Selected Proceedings of the 14th Hispanic Linguistics Symposium, 130–​141. Somerville, MA: Cascadilla Proceedings Project. Shin, N.L. (2016). Acquiring constraints on morphosyntactic variation:  children’s Spanish subject pronoun expression. Journal of Child Language, 43(4), 914–​947. Shin, N.L., & Cairns, H.S. (2012). The development of NP selection in school-​age children: Reference and Spanish subject pronouns. Language Acquisition, 19(1), 3–​38. doi:10.1080/​10489223.2012.633846 Shin, N.L., & Erker, D. (2015). The emergence of structured variability in morphosyntax:  Childhood acquisition of Spanish subject pronouns. In A. Carvalho, R. Orozco & N. Shin (eds.), Subject Pronoun Expression in Spanish: A Cross-​Dialectal Perspective, 169–​190. Washington, DC: Georgetown University Press. Shin, N.L. & Otheguy, R. (2009). Shifting sensitivity to continuity of reference: Subject pronoun use in Spanish in New York City. In Manel Lacorte & Jennifer Leeman

Preschoolers and the null-overt contrast  265 (eds.), Espanõ l en Estados Unidos y en otros contextos: Cuestiones sociolingüísticas, políticas y pedagógicas, 111–​136. Madrid: Iberoamericana. Shin, N.L., & Otheguy, R. (2013). Social class and gender impacting change in bilingual settings:  Spanish subject pronoun use in New  York. Language in Society, 42(4), 429–​452. Sorace, A., Serratrice, L., Filiaci, F., & Baldo, M. (2009). Discourse conditions on subject pronoun realization: Testing the linguistic intuitions of older bilingual children. Lingua, 119(3), 460–​477. doi:10.1016/​j.lingua.2008.09.008 Toribio, A.J. (2000). Setting parametric limits on dialectal variation in Spanish. Lingua, 110(5), 315–​341. doi:10.1016/​s0024-​3841(99)00044-​3 Valian, V. (1990). Null subjects: A problem for parameter-​setting models of language acquisition. Cognition, 35(2), 105–​122. White, L. (2011). Second language acquisition at the interfaces. Lingua, 121(4), 577–​590. Yang, Charles D. (2002). Knowledge and Learning in Natural Language. Oxford: Oxford University Press.

13  Acquisition of null subjects by heritage children and child L2 learners Michele Goldin

1.  Introduction Comparison of child first language (L1), child second language (L2), heritage language and adult L2 acquisition demonstrates that these processes follow different developmental paths. As such, it is important to our understanding of language acquisition as a whole to investigate the variables that shape bilingual developmental trajectories. This study is concerned with the bilingual grammars of children living in the United States who are either heritage speakers of Spanish or who are learning Spanish as an L2 in an immersion school. Specifically, we address bilingual children’s judgments of null subjects, an area of grammar that falls at the syntax-​pragmatics interface, in both English and Spanish. Our aim is to explore whether these two types of bilinguals show similar or different patterns of acquisition of null subjects in both of their languages. A growing body of literature is dedicated to child L2 acquisition, particularly regarding its similarities and differences in various grammatical areas with respect to child L1 and adult L2 (Schwartz, 2004; Lakshmanan, 1995; Meisel, 2007; 2008; Pladevall Ballester, 2010; Unsworth, 2007, 2016; Unsworth et al., 2014). Teasing these apart is valuable for our general understanding of language acquisition. Based on evidence from a variety of studies, Schwartz (2003, 2004) concludes that in the domain of inflectional morphology, child L2 acquisition is more like child L1 acquisition, but in the domain of syntax, child L2 acquisition is more like adult L2 acquisition because there is L1 influence. In the long term, L2 children typically outperform L2 adults. Thus, a great amount of L2 research has addressed age of acquisition (AoA) effects in order to identify when the ability to acquire an L2 to nativelike levels starts to decline (see Unsworth, 2016, for a recent review). Simultaneous bilinguals (2L1) acquire both languages before the age of three, while L2 children begin exposure to the second language after the grammatical and lexical foundations of the first language are in place (between ages three and four) (Guasti, 2002; Meisel, 2004; Schwartz, 2004; Unsworth, 2007). Hence, this has been pinpointed as the developmental moment at which qualitative differences in morphosyntactic production seem to arise (Meisel, 2008; Unsworth, 2016).

Null subjects by heritage and L2 learners  267

2.  Null subjects 2.1  Pronominal subject realization English is an overt subject language where null subjects are extremely rare in declarative clauses and never allowed in embedded clauses (Chomsky, 1981; Haegeman, 2009; Barbosa, 2011; Torres Cacoullos & Travis, 2019), as in sentences (1a) and (1b): (1) a. Sandra told us that she went to the store. b. *Sandra told us that went to the store. Spanish, on the other hand, is a null subject language in which a subject’s number and person can be identified in the verb’s inflection without the need for a subject pronoun (for a discussion, see Camacho, 2013), and in which null and overt pronoun distribution is guided by a series of pragmatic constraints. While researchers have identified a range of more detailed constraints (for a review, see Carvalho, Orozco & Shin, 2015), in this chapter we focus on topic shift and topic continuation (Frascarelli, 2007). Topic shift is the introduction of a new topic in discourse and topic continuation is the reference to a previously mentioned topic as in sentences (2a) and (2b): (2) a. Sandra fue al supermercado. (pro) Compró manzanas. b. Sandra fue al supermercado. Ella compró manzanas. “Sandra went to the supermarket. She bought apples.” Generally, Spanish speakers favor null subjects overall, but topic shift contexts increase the likelihood of overt subject expression (Rothman, 2009; Camacho, 2013; Carvalho et  al., 2015). In Spanish, overt subjects serve to remove referential ambiguity when new subjects are added to the discourse. The null subject parameter is derived from the Principles and Parameters model of UG which includes the Extended Projection Principle (EPP), a guiding condition in which all sentences must have a syntactic subject (Chomsky, 1981). Languages with null subjects, like Spanish, satisfy the EPP through verbal agreement morphology that allows for the licensing (a condition on the syntactic representation) and identification (a condition on the interpretation) of a null pronoun as a subject (Rizzi, 1986; for a review see Barbosa, 2011). In languages like English with a null setting of the parameter, this principle is satisfied by overt pronouns because the language’s verbal agreement morphology lacks pronominal features to satisfy the EPP requirement (Camacho, 2013). Thus, Spanish identifies null subjects by recovering the person and number values encoded in its verbal morphology, while English resorts to a full pronoun instead. While there are many more analyses of the null subject parameter, we will take these to be the main

268  Michele Goldin differences between Spanish and English grammars regarding the licensing and identification of subjects. 2.2  Syntax-​pragmatics interface The differences between null and overt subjects across languages like English and Spanish have proven to be fertile ground for testing hypotheses about the syntax-​pragmatics interface, a proposal first put forth in acquisition literature by Hulk and Müller (2000) and Müller and Hulk (2001). They suggest that cross-​linguistic influence is likely to occur when discourse-​pragmatic context influences the possibilities of syntactic structure and that bilingual children, presented with competing evidence from two languages, may need more time to acquire that structure than monolingual children. This is referred to as an interface delay in which bilinguals’ variability is due to properties in the narrow syntax possibly being acquired before those conditioned by the syntax-​pragmatics interface (Rothman, 2009). For L2 learners, delays at the interface became coined the ‘Interface Hypothesis’ (IH) (Sorace & Filiaci, 2006). The IH predicts that both syntactic and pragmatic conditions are acquirable, but that even very advanced L2 speakers will continue to exhibit residual optionality in their pragmatic distribution of null and overt subjects (Sorace, 2011). Thus, difficulties in acquisition at the syntax-​pragmatics interface are evident not only in 2L1 (Argyri & Sorace, 2007; Sorace, Serratrice, Filiaci & Baldo, 2009; Goldin, under review), but also in L2. A review follows of subject acquisition in both monolingual and bilingual children as well as L2 speakers.

3.  Previous studies 3.1  Monolingual children Monolingual children acquiring overt subject languages pass through a missing subject stage that converges with adult-​like grammar at around age 3;6 (Orfitelli & Hyams, 2008; 2012; Bloom, 1990; Valian, 1991). The same stage also exists in null subject languages, but children begin to produce null and overt pronouns in pragmatically felicitous ways very early on (Grinstead, 2000; Montrul, 2004b; Serratrice, 2005; Austin, Sánchez & Pérez-​Cortes, 2017).1 Crucially, this temporary period coincides with the gradual acquisition of the inflectional system (Grinstead, 2000) and also co-​occurs with root infinitives (Grinstead 2004; Orfitelli & Hyams, 2008; Grinstead, De la Mora, Vega-​Mendoza & Flores, 2009). During this stage, children produce main clause declaratives with inflected verb forms that are grammatical in the adult language as well as a large percentage of infinitival verb forms that are not (Wexler, 1994). Orfitelli & Hyams (2008) provided the first experimental evidence to link the null subject stage with the root infinitive stage, showing that monolingual English children exit both stages around age 3;5. There

Null subjects by heritage and L2 learners  269 is debate as to whether subject omission in monolingual child language is attributable to a difference between child and adult grammars (Hyams, 1986; Jaeggli & Hyams, 1988; Hyams & Wexler, 1993; Rizzi, 2005) or if it is the result of immature processing capabilities (Bloom, 1990; Valian, 1991), but we will focus on the grammatical account as it pertains to the data presented later in this study. Hyams (1986), Jaeggli & Hyams (1988) and Hyams & Wexler (1993) all consider children of overt subject languages to have an incomplete grammar of their adult versions. Hyams (1986, 2011) proposes that this phenomenon reflects an initial setting of the null subject parameter, as found in the Principles and Parameters model of UG. Infants start off with preset parameters at a universal value that is correct for some languages but not for others. As children grow older, these parameters are set and reset at the appropriate values for their target language. Similarly, Müller & Hulk (2001) point to the notion of a Minimal Default Grammar (as originally posited by Roeper,1999) as the initial setting in UG. They propose that certain properties are part of the universal, initial representation in the grammar of a child and that children adapt to the syntactic rules of the specific language they receive in their input. However, as previously mentioned, bilingual children may persist longer at this universal stage due to conflicting input from two partially overlapping languages. 3.2  Bilingual children These developmental patterns have led researchers to investigate pronoun development in bilingual children acquiring an overt subject language together with a null subject language (Paradis & Navarro, 2003; Argyri & Sorace, 2007; Sorace et al., 2009; Liceras, Fernández Fuertes & Alba de la Fuente, 2012; Silva-​Corvalán, 2014; Montrul & Sánchez-​Walker, 2015; Austin et al., 2017; Montrul, 2018; Goldin, under review), all showing evidence for cross-​linguistic transfer. Sorace et  al. (2009) propose that the language with the most economical syntax–​pragmatics interface system (like English) may influence the language with a more complex interface system (like Spanish) where null and overt subjects are both viable and regulated by subtle discourse-​pragmatic constraints. Indeed, most studies have found this to be the case, evidenced by an overuse of overt pronouns in the null subject language and the infelicitous application of the discourse-​pragmatic principles guiding pronoun use (e.g., Silva-​Corvalán, 2014; Montrul & Sánchez-​Walker, 2015). Data from comprehension and grammaticality judgment experiments also shows that bilingual children tend to accept overt pronouns in the null subject language in contexts in which monolinguals prefer a null subject, especially when they are dominant in the overt subject language (Argyri & Sorace, 2007; Sorace et al., 2009; Goldin, under review). According to Liceras et  al. (2012) who investigate the directionality of cross-​linguistic influence, the transfer of null subjects from a null subject

270  Michele Goldin language to an overt one is highly unlikely because overt subjects are the key overlapping structure. However, certain external factors may allow for cross-​linguistic influence from a null subject language to an overt one such as sociolinguistic conditions and language dominance (Austin et al., 2017), or learning environment (Goldin, under review). 3.3  L2 children and L2 adults A handful of studies have explored L2 acquisition of null subject properties in English/​ Spanish bilingual children using spontaneous speech data (Hilles, 1991; Lakshmanan, 1994) and experimental production data (Pladevall Ballester, 2010, 2012, 2016). For L2 English children, obligatory overt subjects emerge quickly in the acquisition process and are highly correlated with the development of verbal morphology (Hilles, 1991; Lakshmanan, 1994; Pladevall Ballester, 2012). For L2 Spanish children, the syntactic properties of null subjects are acquired early on, but the discourse-​pragmatic properties remain challenging even for older and more advanced children (Pladevall Ballester, 2010). The same has been found for adult L2 learners of Spanish since properties at the syntax–​pragmatics interface have shown learnability problems due to their increased complexity (Sorace, 2011). L2 learners with L1 English acquire the syntax of null subjects but show delays in acquiring the discourse distribution of null and overt subject pronouns in both interpretation and production (Pérez-​ Leroux & Glass, 1997, 1999; Rothman, 2009; Clements & Domínguez, 2017, among others).

4.  The present study This study aims to explore the developmental patterns of 2L1 heritage children and child L2 acquirers of Spanish in similar social and learning environments. 4.1  Research questions, hypotheses and predictions This study is guided by three research questions, the first of which is: RQ1: Does AoA affect acceptability of null subjects in Spanish? AoA should have an effect on bilingual children’s acceptability of null and overt subjects. Studies in child and adult L2 acquisition have shown that the pragmatic constraints of null subjects present challenges (IH, Sorace & Filiaci, 2006). While these also exist in 2L1 acquisition, we hypothesize that L2 children, who have had less exposure to Spanish, will show less sensitivity to topic shift and topic continuation contexts. Thus, we expect that heritage children will perform more accurately than L2 children.

Null subjects by heritage and L2 learners  271 The second research question is: RQ2a: Does morphological proficiency modulate null subject acceptance in Spanish? RQ2b: Does morphological proficiency modulate null subject acceptance in English? Grinstead (2000) observed that for monolingual Spanish children, the pragmatically felicitous use of null and overt subjects co-​occurs with the acquisition of verbal morphology. If this is the case for monolingual children, it is reasonable to assume that the same order of acquisition exists for bilingual children. Therefore, the prediction is that as bilingual children’s verbal morphology in Spanish increases, so will their understanding of the pragmatic constraints governing the use of null and overt subjects. In English, Orfitelli & Hyams (2008) showed that the null subject stage coincides with a period of root infinitives or uninflected verbs. This period tapers off between the ages of three and four, just as the null subject stage does. If this is true for monolingual children, then acquisition of verbal morphology in English dominant bilingual children should modulate null subject acceptance in English. We predict that as verbal morphology increases, overt subject acceptance will increase, and null subject acceptance will decrease. Our final research question is: RQ3: Are there cross-​linguistic effects in subject acceptance from Spanish to English? Knowing that properties at the syntax–pragmatics interface are more vulnerable to cross-​linguistic influence (Hulk & Müller, 2000), it is possible that both heritage bilinguals and child L2 learners experience cross-​linguistic effects in subject acceptance, even in their dominant language. However, if, as Liceras et al. (2012) and Sorace et al. (2009) suggested, transfer of null subjects from a null subject language to an overt one is highly unlikely, then the prediction in this study is that there will be no evidence of transfer from Spanish to English. 4.2  Participants Three groups of children ages 4;0 to 7;0 participated in this research. The first was a group of heritage speakers (n=20, mean age=5;5) recruited at a Spanish-​immersion school in New Jersey who had attended the school for at least one academic year. The second was a group of L2 learners of Spanish (n=17; mean age=5;2) who attended the same immersion school for at least one academic year. None of the children had attended the school for more than two years. The third was a group of monolingual English speakers (n=17; mean age=5;8) attending English monolingual schools in New Jersey, recruited through personal contacts. The heritage speakers had

272  Michele Goldin either one or two parents who spoke Spanish in the home and they were born in the United States or arrived before the age of two. Therefore, they were exposed to both languages before the age of three and are considered simultaneous bilinguals. In the L2 group, parents were monolingual English speakers with limited knowledge of Spanish and the children began to learn Spanish in the immersion school after age 2;6 or 3. All the English monolingual children had little to no exposure to Spanish.2 Parents of heritage children were asked country of origin on the language background questionnaire so that children acquiring a Caribbean variety of Spanish were not included.3 Furthermore, the bilingual children’s teachers also did not speak Caribbean Spanish. 4.3  Procedure Children’s morphological proficiency in each language was assessed using a modified version of the BESA (Peña et  al., 2014) (see Table  13.1 for mean scores by group in both languages). Specifically, the sections on verbal morphology were used. Parents were asked to complete a language background questionnaire for each child which included information such as age of first exposure and current language use. Following Sorace et  al. (2009), an acceptability judgment task was chosen for this study because such tasks have proven successful in assessing even very young children’s understanding of grammatical features (e.g., Orfitelli & Hyams, 2012). Two versions of the task were created, one in Spanish and one in English; both were delivered via PowerPoint presentation presented on a laptop. Of the bilingual children, half received the English task first and half the Spanish task first. The experimenter spoke to the participants in the language in which the experiment was being carried out. English monolinguals were administered the task in English only. 4.4  Materials The acceptability judgment task, adapted from Sorace et al. (2009), consisted of a series of video clips showing four character puppets (Mickey Mouse, Minnie Mouse, Donald Duck and Goofy) with eight experimental items (four Table 13.1 BESA proficiency scores by group Group

Spanish mean

Spanish SD

English mean

English SD

Heritage L2 Monolingual

68.18 56.67 N/​A

19.42 18.35 N/​A

86.50 87.65 97.06

17.71 12.63 5.72

Null subjects by heritage and L2 learners  273 +TS and four -​TS items) and four distractors. In the experimental items, either Mickey or Goofy performed an action which was commented upon either by himself ([-​topic shift] (-​TS) condition) or by the other character that had watched the action but was not involved in it ([+topic shift] (+TS) condition). Then, the participants saw Minnie and Donald each say a subordinate clause one after the other. One character would begin the subordinate clause with a null subject and the other with an overt subject (he/​él). The order in which the grammatical and the ungrammatical sentences were presented by Minnie and Donald was counterbalanced throughout. The following examples illustrate the types of sentences used in the Spanish task 3(a) and 3(b) and in the English task 3(c) and 3(d): 3 (a) -​topic shift condition (-​TS) Mickey cries Mickey: “¡He llorado!”4      “I have cried!” Minnie: Mickey dijo que ha llorado.     “Mickey said that (he) has cried.” Donald: Mickey dijo que él ha llorado.     “Mickey said that he has cried.” (b) +topic shift condition (+TS) Mickey eats Goofy: “¡Mickey ha comido!”     “Mickey has eaten!” Minnie: Goofy dijo que ha comido.     “Goofy said that (he) has eaten.” Donald: Goofy dijo que él ha comido.     “Goofy said that he has eaten.” (c) -​topic shift condition (-​TS) Mickey jumps Mickey: “I jumped!” Minnie: Mickey said that jumped. Donald: Mickey said that he jumped. (d) +topic shift condition (+TS) Goofy coughs. Mickey: “Goofy coughed!” Minnie: Mickey said that he coughed. Donald: Mickey said that coughed. After listening to the comments of all the characters, the children (who had been told that the characters were learning English or Spanish) were

274  Michele Goldin asked to decide whether Minnie Mouse or Donald spoke “better” English or Spanish. Their responses were audio recorded and coded for null and overt subject acceptability in –​topic shift (-​TS) conditions and +topic shift (+TS) conditions. The distractor items had a similar structure to the experimental ones, but the characters in the second video uttered true or false statements rather than ungrammatical ones in order to check that the participants had understood the task and were focusing on the sentences presented to them. By necessity, the tasks were slightly different in each language. In English, children were asked to choose between grammatical sentences with overt subjects and ungrammatical sentences with null subjects, whereas in Spanish, children had to choose between pragmatically felicitous or infelicitous pronouns. To succeed in the English task, participants had to rely purely on their syntactic knowledge of English as an overt subject language where null subjects are never allowed in subordinate clauses. However, in Spanish, children needed to integrate their syntactic knowledge of Spanish as a null subject language with their knowledge of the pragmatic constraints that guide the distribution of null and overt subject pronouns in topic shift or topic continuation circumstances (Sorace et al., 2009).

5.  Results The data, a set of 432 data points, were analyzed in R version 3.3.2 (R Development Core Team, 2012) using a generalized linear mixed effects model (GLMM) for each language to examine subject acceptance (null, overt) as a function of group (monolingual, heritage, L2), topic (+/​-​topic shift), and morphological proficiency.5 Given the categorical nature of the participants’ responses, the data were modeled using GLMMs with a binomial linking function. The predictors “group” and topic were dummy coded with monolingual participants and + topic shift set as the reference levels. Participant and item were the random effects with random slopes for age and proficiency. Main effects and higher order interactions were tested using nested model comparisons. Mean morphological proficiency in each language (English: 90.18 /​Spanish: 62.43) was centered at 0. The analyses for the English and the Spanish data are reported separately. Table 13.2 Total of mean expected responses by group Group

Language

Pronoun type

Mean response

Heritage

English Spanish Spanish English Spanish Spanish English

overt null overt overt null overt overt

.63 .49 .53 .63 .44 .49 .79

L2 Monolingual

Null subjects by heritage and L2 learners  275

% Proportion overt preferences

1.00

0.75

0.50

0.25

0.00 Non-immersion

Group

Topic continuity

Immersion Mean +/– 95% CI Topic shift

Figure 13.1 Proportion of overt pronoun acceptance in Spanish in topic shift and topic continuity conditions

5.1  Pronominal preferences in Spanish Figure 13.3 reports the mean of overt and null pronoun choices in Spanish. To recall, in topic shift conditions, expected responses were overt subjects and in topic continuity conditions, expected responses were null subjects. Both the heritage and L2 groups performed similarly and there was no effect of group (χ2(1) = 0.26, p = 0.61), topic (χ2(1) = 0.002, p = 0.96), age (χ2(1) = 0.08, p  =  0.78), or proficiency (χ2(1)  =  0.002, p  =  0.96). Again, responses to distractor items were checked and both groups performed similarly. The heritage group had a mean score of .83 and SD=.38, and the L2 group had a mean score of .79, SD = .40. 5.2  Pronominal preferences in English In English, the analysis yielded a main effect of group (χ2(2) = 6.80, p < 0.04). The monolingual group responded with an overt pronoun significantly more often than the L2 group (β = 0.84; SE = 0.28; z = −​3.04; p < .002). The heritage group performed similarly to the L2 group and there were no significant differences between the two (β = −​1.59; SE = 2.41; z = 0.00; p = 1.00) (See Figure 13.3). There was no effect of topic in English (χ2(1) = 1.06, p = .30), but there was a main effect of proficiency (χ2(1) = 3.69, p = 0.05). There were no higher order interactions.

276  Michele Goldin

% Proportion overt preferences

1.00

0.75

0.50

0.25

0.00 L2

Heritage

Monolingual

Group

mean +/–95% CI

Figure 13.2 Proportion of overt subject acceptance by group in English

HL

L2

MO

Proportion overt preferences

1.00

0.75

0.50

0.25

–40 –20

0

–40 –20

0

–40 –20

0

prof (centered) L2

Heritage

Monolingual

Figure 13.3 Null and overt acceptance in English by group as a function of English morphological proficiency

Null subjects by heritage and L2 learners  277 The responses to distractor items were checked to ensure that children were indeed paying attention to the task. In English, all three groups performed similarly. The heritage group had a mean score of .88 and SD=.33, and the L2 group had a mean score of .86, SD = .36. The monolingual group had a mean of .90, SD=.31. Figure  13.3 shows the proportion of overt acceptance as a function of morphological proficiency in English and group. For the monolingual and heritage groups, following our hypothesis, as verbal morphology increased so did acceptance of overt subjects. However, for the L2 children, the effect of morphological proficiency was much smaller, at this stage in acquisition. Possible explanations for this will be discussed further in section 6.

6.  Discussion In this study, three groups of children (English monolinguals, simultaneous heritage bilinguals, and L2 learners) completed an acceptability judgment task to investigate their understanding of null subjects in both English and Spanish. The findings revealed that both groups of bilingual children accepted significantly more ungrammatical null subjects in English than the monolingual children, but that there were no differences in Spanish between the bilingual groups regardless of the discourse context. Additionally, a relationship between English verbal morphology and null subject acceptance was found in the monolingual and heritage children, but for the L2 learners this correlation was much smaller. These results establish differences among monolinguals, simultaneous bilinguals and child L2 learners in their acquisition patterns of null subjects and have potential implications for theories of AoA which will be discussed further in this section. The first research question addressed whether AoA had an effect on null subject acceptability and the results do not support our hypothesis that heritage speakers would outperform L2 children due to having had more exposure to Spanish. Rather, both the simultaneous bilinguals who began acquiring English and Spanish before the age of three and the child L2 learners who acquired English before age three and Spanish after age three, patterned similarly in their responses to the acceptability judgment task in Spanish. There was no statistical difference in their responses to either (-​TS) or (+TS) conditions, suggesting they had an understanding that both are grammatical and possible in Spanish, but an uncertainty as to the pragmatic conditions under which to use them. These results align with the notion of delays at the syntax –pragmatics interface as predicted by Hulk & Müller (2000) and Müller & Hulk (2001) for 2L1 acquisition as well as the Interface Hypothesis for L2 learners (Sorace & Filiaci, 2006).6 However, by contrast, the pattern of results in English suggests that both groups of bilingual children accept more null subjects (37%) than their monolingual peers (21%). Interestingly, in English, there was also no preference for choosing a null subject in either the (-​TS) or (+TS) condition, suggesting that the children were not relying on

278  Michele Goldin pragmatic guidance for null and overt pronoun distribution. There was, therefore, no effect of AoA as both bilingual groups performed similarly to each other in both languages. The second research question addressed the relationship between morphological proficiency and null subject acceptance in both languages. In Spanish, contrary to the predictions that would have supported our hypothesis, there was no correlation. Knowing the delay that exists in acquiring the pragmatic constraints guiding null and overt pronoun use for both simultaneous bilinguals (i.e., Argyri & Sorace, 2007; Sorace et  al., 2009) and L2 learners (Rothman, 2009; Pladevall Ballester, 2010, 2016), it is not surprising that acquisition of verbal morphology does not coincide with target-​like pragmatic use of pronouns in Spanish. Additionally, the results of studies such as Argyri & Sorace (2007) and Sorace et al. (2009) provide evidence that as simultaneous bilingual children get older, they continue to exhibit residual optionality in their pragmatic distribution of null and overt subjects. In English, however, an interesting pattern emerged especially because English was the dominant language for all three groups of children. As in previous studies (Hyams, 1986; Orfitelli & Hyams, 2008, 2012), the monolingual children provided evidence in support of our hypothesis that there is a relationship between acquisition of verbal morphology and pronoun acceptance. As the children’s proficiency increased, fewer null subjects were accepted in favor of mostly overt ones, demonstrating near achievement of an adult-​like grammar. The heritage simultaneous bilinguals presented a similar correlation, even though, as seen in Figure  13.3, they accepted significantly more null subjects than the monolingual children. Given that these children were English-​dominant simultaneous bilinguals, it is unclear whether this correlation is similar to their monolingual peers because they are English dominant or because the syntactic and pragmatic properties of subject use in Spanish are more complex to acquire than the simpler grammatical rules of English. Finally, and surprisingly, the L2 learners presented a different pattern of acquisition in which the effect of morphological proficiency was much smaller. This could be explained as an interruption of the monolingual path of development due to the addition of a new language at age three or four, one that provides competing evidence for null subjects. The L2 children in this study provide a unique canvas for analyzing the acquisition of null subjects in English because they were exposed to the L2, Spanish, at age three or four, just when monolingual English speakers begin to emerge from the null subject stage. Rather than continuing along the same path as their monolingual counterparts and rejecting increasingly more null subjects, they show evidence of remaining in the null subject stage for a longer period of time. This seems to occur in spite of continuing to acquire verbal morphology at the same rate as their English monolingual peers (mean L2 morphological proficiency scores were significantly higher in English—​87.65—​than in Spanish—​56.67). Therefore, the smaller effect of

Null subjects by heritage and L2 learners  279 morphological proficiency in the L2 children (which is indeed found in the monolinguals and simultaneous bilinguals) suggests that the introduction of a new language at age three or four can affect the age of acquisition of overt subjects in English without impacting acquisition of verbal morphology. This brings us to the final research question of this study which addressed whether there would be evidence of cross-​linguistic influence from Spanish to English. In contrast to previous studies (i.e., Paradis & Navarro, 2003; Silva-​ Corvalán, 2014; Sorace et al., 2009), the data obtained here do not support our prediction that English would not be influenced by Spanish. The children accepted a higher amount of non-​target-​like null pronouns in English despite being English dominant. These findings complement those of Austin et  al. (2017) in which Spanish dominant bilingual children overproduced non-​target-​like null pronouns in English. While the results of the previous study can be explained as evidence of cross-​linguistic transfer from the dominant language, the current study reveals a subset of English dominant bilingual children, even Spanish L2 learners, who behave in a similar way in their acceptability of null subjects in English. For the simultaneous bilinguals, this finding can be explained by early bilingual input and for the L2 learners, it may be cross-​linguistic influence from the introduction of Spanish at age three or four. In both cases, the result appears to be a prolonging of the null subject stage due to competing input, as posited by Müller and Hulk (2001). The data support the theories put forth by Hyams (1986) (as well as Jaeggli & Hyams, 1988; Hyams & Wexler, 1993) that children of overt subject languages, like English, have an incomplete grammar of their adult versions.7 Various studies have found that the directionality of influence goes from the overt subject language to the null subject language, regardless of language dominance (for a review, see Sorace & Serratrice, 2009). Furthermore, Liceras et al. (2012) proposed it would be unlikely for a null subject language to influence an overt subject language because overt subjects are the key overlapping structure. Thus, it is possible that cross-​linguistic effects from English are not yet visible at the ages tested in this study, four to seven. Rather, in bilingual children who are confronted with input from two partially overlapping languages like English, in which only overt subjects are allowed, and Spanish, where both are viable, this early universal stage appears to persist longer.

7.  Conclusion The findings of this study provide new insight into the comprehension of pronominal subjects in the bilingual acquisition of English and Spanish. This study sought to investigate similarities and differences in the development of null subject comprehension of heritage speakers of Spanish and child L2 learners of Spanish educated in an immersion school in the United States. The results of the bilingual children’s English can be explained as a prolonging of the null subject stage due to bilingual input from birth for the

280  Michele Goldin heritage speakers and cross-​linguistic effects from Spanish for the L2 learners. Based on evidence from previous studies in which bilingual children between the ages of six and ten perform at ceiling in English, regardless of language of dominance (Argyri & Sorace, 2007; Sorace et al., 2009), one can presume that later in childhood both groups of English dominant bilinguals will converge with English monolinguals in their acceptability of null subjects in English. However, this age range is outside the scope of this study and should be investigated in future research. This knowledge contributes to the acquisition literature because while much attention has been paid to how the L1 influences a child’s L2 (Lakshmanan, 1995; Schwartz, 2003, 2004; Unsworth, 2007), less is known about how the introduction of a second language in childhood affects the L1 when the L1 continues to be the dominant language, which has implications for linguistic theory and highlights the importance of research in child bilingualism. The results of this study add to our understanding of AoA, the relationship between the acquisition of verbal morphology and null subject acceptability in English, and the directionality of cross linguistic influence in bilingual children.

Notes 1 On the other hand, in comprehension, Shin and Cairns (2012) found that monolingual Spanish children in Mexico did not converge with adult-​like pragmatic distribution until after age nine. 2 Some children received 30 minutes per week of Spanish instruction at school, but had no functional competence in the language and, therefore, it was determined that this was not enough exposure to make any significant difference. 3 Studies have found that these Spanish speakers tend to produce more overt pronouns than speakers of other dialects (Camacho, 2008). 4 We recognize that the present perfect tense was chosen for the Spanish task, though it is not the most widely used among Latin American and heritage speakers. However, this did not seem to have any effect on the children’s understanding of the phrases and experimental items. 5 Age was not included as a predictor in this model because age and proficiency were found to be correlated (r  =  .52). Therefore, only morphological proficiency was included. 6 The results also align with previous studies of judgments from Spanish-​speaking monolingual children (Shin & Cairns, 2012). This is despite the task items having been presented in perfect tense, a less common form in Latin American varieties of Spanish, which may have played a role in degraded acceptability. 7 A reviewer points out that these proposals address children under age three while the children in this study were ages four–​six. I cautiously suggest that the results obtained demonstrate an extension of the null subject stage precisely because they appear to be similar to what Hyams (1986) noted as occurring for younger ages.

Null subjects by heritage and L2 learners  281

References Argyri, E., & Sorace, A. (2007). Crosslinguistic influence and language dominance in older bilingual children. Bilingualism: Language & Cognition, 10(1), 79. Austin, J., Sánchez, L., & Pérez-​Cortes, S. (2017). Null subjects in the early acquisition of English by child heritage speakers of Spanish. In S. Perpiñán, D. Heap, I. Moreno-​Villamar & A. Soto-​Corominas (eds.), Proceedings of the 44th Linguistic Symposium on Romance Languages, 209–​227. Amsterdam: John Benjamins. Barbosa, P.P. (2011). Pro-​ drop and theories of pro in the minimalist program part  1:  consistent null subject languages and the pronominal-​ agr hypothesis. Language & Linguistics Compass, 5(8), 551–​570. Bloom, P. (1990). Subjectless sentences in child language. Linguistic Inquiry, 21, 491–​504. Camacho, J. (2008). Syntactic variation: The case of Spanish and Portuguese subjects. Studies in Hispanic and Lusophone Linguistics, 1(2), 415–​433. Camacho, J. (2013). Null Subjects. Cambridge: Cambridge University Press. Carvalho, A, Orozco, R., & Shin, N.L. (eds.). (2015). Subject Pronoun Expression in Spanish:  A Cross-​ Dialectal Perspective. Washington, DC:  Georgetown University Press. Chomsky, N. (1981). Lectures on Government and Binding. Dordrecht: Foris. Clements, M., & Domínguez, L. (2017). Reexamining the acquisition of null subject pronouns in a second language:  Focus on referential and pragmatic constraints. Linguistic Approaches to Bilingualism, 7(1), 33–​62. Frascarelli, M. (2007). Subjects, topics and the interpretation of referential pro: An interface approach to the linking of (null) pronouns. Natural Language & Linguistic Theory, 25(4), 691–​734. Goldin, M. (under review). An exploratory study of the effect of Spanish immersion education on the acquisition of null subjects in child heritage speakers. Grinstead, J. (2000). Case, inflection and subject licensing in child Catalan and Spanish. Journal of Child Language, 27, 119–​155. Grinstead, J. (2004). Subjects and interface delay in child Spanish and Catalan. Language, 80(1), 40–​72. Grinstead, J., de la Mora, J., Vega-​Mendoza, M., & Flores, B. (2009). An Elicited Production Test of the Optional Infinitive Stage in Child Spanish. In J. Crawford, K. Otaki & M. Takahashi (eds.), Proceedings of the 3rd Conference on Generative Approaches to Language Acquisition North America (GALANA 2008), 36–​45. Somerville, MA: Cascadilla Proceedings Project. Haegeman, L. (2009). Understood subjects in English diaries. On the relevance of theoretical syntax for the study of register variation. Multilingua – Journal of Cross-​ Cultural and Interlanguage Communication, 9(2), 157–​200. Hilles, S.  (1991). Access to universal grammar in second language acquisition. In Eubank, L. (ed.), Point Counterpoint: Universal Grammar in the Second Language, 305–338. Amsterdam: John Benjamins. Hulk, A., & Müller, N. (2000). Bilingual first language acquisition at the interface between syntax and pragmatics. Bilingualism:  Language and Cognition, 3(3), 227–​244. Hyams, N. (1986). Language Acquisition and the Theory of Parameters. Dordrecht: Reidel.

282  Michele Goldin Hyams, N. (2011). Missing subjects in early child language. In J. de Villiers & T. Roeper (eds.), Handbook of Generative Approaches to Language Acquisition, 13–​ 52. Dordrecht: Springer. Hyams, N., & Wexler, K. (1993). The grammatical basis of null subjects in child language. Linguistic Inquiry, 24, 421–​459. Jaeggli, O., & Hyams, N. (1988). Morphological uniformity and the setting of the null subject parameter. In Proceedings of the North Eastern Linguistic Society (NELS), 18, 238–​253. Amherst. MA: GLSA, University of Massachusetts. Lakshmanan, U. (1994). Universal Grammar in Child L2 Acquisition: Morphological Uniformity and Null Subjects. Amsterdam: John Benjamins. Lakshmanan, U. (1995). Child second language acquisition of syntax. Studies in Second Language Acquisition, 17, 301–​329. Liceras, J., Fernández Fuertes, R., & Alba de la Fuente, A. (2012). Overt subjects and copula omission in the Spanish and the English grammar of English–​Spanish bilinguals:  On the locus and directionality of interlinguistic influence. First Language, 32(1–​2), 88–​115. Meisel, J.M. (2004). Early differentiation of languages in bilingual children. In K. Hyltenstam & L. Obler (Eds.), Bilingualism across the Life Span:  Aspects of Acquisition, Maturity, and Loss, 13–​ 40. Cambridge:  Cambridge University Press. Meisel, J. (2007). Age of onset in successive acquisition of bilingualism:  Effects on grammatical development. Working Papers in Multilingualism, 80, 3–​32. Hamburg: University of Hamburg. Meisel, J. (2008). Child second language acquisition or successive first language acquisition? In B. Haznedar & H. Gavruseva (eds.), Current Trends in Child Second Language Acquisition: A Generative Perspective, 55–​80. Amsterdam: John Benjamins. Montrul, S. (2004a). Subject and object expression in Spanish heritage speakers:  A case of morphosyntactic convergence. Bilingualism: Language and Cognition, 7(2), 125–​142. Montrul, S. (2004b). The Acquisition of Spanish:  Morphosyntactic Development in Monolingual and Bilingual L1 Acquisition and Adult L2 Acquisition. Amsterdam: John Benjamins. Montrul, S. (2018). Heritage language development: Connecting the dots. International Journal of Bilingualism, 22(5), 530–​546. Montrul, S., & Sánchez-​Walker, N. (2015). Subject expression in bilingual school-​age children in the United States. In A. Carvalho, R. Orozco & N. Shin (eds.), Spanish Subject Pronoun Expression in Spanish: A Cross-​Dialectical Perspective, 231–​247. Washington, DC: Georgetown University Press. Müller, N. and Hulk, A. (2001). Crosslinguistic influence in bilingual language acquisition:  Italian and French as recipient languages. Bilingualism:  Language and Cognition, 4(1), 1–21. Orfitelli, R., & Hyams, N. (2008). An experimental study of children’s comprehension of null subjects:  Implications for grammatical/​ performance accounts. In H.  Chan,  H. Jacob, & E. Kapia (eds.), BUCLD 32 Proceedings, 335–​346. Somerville, MA: Cascadilla Press. Orfitelli, R., & Hyams, N. (2012). Children’s grammar of null subjects: Evidence from comprehension. Linguistic Inquiry, 43(4), 563–​590.

Null subjects by heritage and L2 learners  283 Paradis, J., & Navarro, S. (2003). Subject realization and crosslinguistic interference in the bilingual acquisition of Spanish and English: What is the role of input? Journal of Child Language, 30, 371–​393. Peña, E., Gutierrez-​Clellen, V., Iglesias, A., Goldstein, B., & Bedore, L. (2014). Bilingual English Spanish Assessment (BESA). San Rafael, CA: AR Clinical Publications. Pérez-​Leroux, A.T., & Glass, W.R. (1997). OPC effects on the L2 acquisition of Spanish. In A.T. Pérez-​Leroux & W.R. Glass (eds.), Contemporary Perspectives on the Acquisition of Spanish, Vol. 1, 149–​165. Somerville, MA: Cascadilla Press. Pérez-​Leroux, A.T., & Glass, W.R. (1999). Null anaphora in Spanish second language acquisition: Probabilistic versus generative approaches. Second Language Research, 15, 220–​249. Pladevall Ballester, E. (2010). Child L2 development of syntactic and discourse properties of Spanish subjects. Bilingualism: Language and Cognition, 13(2), 185–​216. Pladevall Ballester, E. (2012). Child L2 English acquisition of subject properties in an immersion bilingual context. Second Language Research, 28(2), 217–​241. Pladevall Ballester, E. (2016). Developmental asynchrony in the acquisition of subject properties in Child L2 English and Spanish. Language Acquisition, 23(4), 387–​406. R Development Core Team. (2012). R:  A language and environment for statistical computing. Retrieved from Rizzi, L. (1986). Null objects in Italian and the theory of pro. Linguistic Inquiry, 17, 501–​558. Rizzi, L. (2005). On the grammatical basis of language development:  A case study. In G. Cinque & R. Kayne (eds.), The Oxford Handbook of Comparative Syntax, 70–​109. New York: Oxford University Press. Roeper, T. (1999). Universal bilingualism. Bilingualism, Language and Cognition, 2, 169–​186. Rothman, J. (2009) Pragmatic deficits with syntactic consequences?: L2 pronominal subjects and the syntax–​pragmatics interface. Journal of Pragmatics, 41, 951–​973. Schwartz, B. (2003). Child L2 acquisition: Paving the way. In B. Beachley, A. Brown & F. Conlin (Eds.), Proceedings of the 27th Annual Boston University Conference of Language Development [BUCLD], 26–​50. Somerville, MA: Cascadilla Press. Schwartz, B. (2004). Why child L2 acquisition? In J. van Kampen & S. Baauw (eds.), Proceedings of GALA 2003, Vol. 1, LOT Occasional Series 3, 47–​66. Utrecht: LOT Publications. Serratrice, L. (2005). The role of discourse pragmatics in the acquisition of subjects in Italian. Applied Psycholinguistics, 26, 437–​462. Shin, N., & Cairns, H. (2012). The development of NP selection in school-​age children: Reference and Spanish subject pronouns. Language Acquisition, 19(1), 3–​38. Silva-​Corvalán, C. (2014). Bilingual Language Acquisition: Spanish and English in the First Six Years. Cambridge: Cambridge University Press. Sorace, A. (2011). Pinning down the concept of “interface” in bilingualism. Linguistic Approaches to Bilingualism, 1(1), 1–​33. Sorace, A., & Filiaci, F. (2006). Anaphora resolution in near-​native speakers of Italian. Second Language Research, 22, 339–​368. Sorace, A., & Serratrice, L. (2009). Internal and external interfaces in bilingual language development:  beyond structural overlap. International Journal of Bilingualism, 13(2), 195–​210.

284  Michele Goldin Sorace, A., Serratrice, L., Filiaci, F., & Baldo, M. (2009). Discourse conditions on subject pronoun realization: Testing the linguistic intuitions of older bilingual children. Lingua, 119, 460–​477. Torres Cacoullos, R., & Travis, C.E. (2019). Variationist typology: Shared probabilistic constraints across (non-​)null subject languages. Linguistics, 57(3). DOI: https://​doi. org/​10.1515/​ling-​2019-​0011. Unsworth, S. (2007). Child L2, Adult L2, Child L1:  Differences and similarities. A  study on the acquisition of direct object scrambling in Dutch. Language Acquisition, 14(2), 215–​217. Unsworth, S. (2016). Early child L2 acquisition: Age or input effects? Neither, or both? Journal of Child Language, 43, 603–​634. Unsworth, S., Argyri, F., Cornips, L., Hulk, A., Sorace, A., & Tsimpli, I. (2014). On the role of age of onset and input in early child bilingualism in Greek and Dutch. Applied Psycholinguistics, 35, 765–​805. Valian, V. (1991). Syntactic subjects in the early speech of American and Italian children. Cognition, 40, 21–​81. Wexler, K. (1994). Optional infinitives, head movement and the economy of derivations in child grammar. In D. Lightfoot & N. Hornstein (eds.), Verb Movement, 305–​350. Cambridge: Cambridge University Press.

14  Demonstratives in Spanish Children’s developing conceptualization of interactive space Naomi L. Shin and Jill P. Morford

1. Introduction1 Although demonstratives are among the first words that children learn (Diessel, 2006), children’s comprehension and production of demonstratives do not necessarily correspond to adults’ complex mapping of demonstrative form and meaning (Clark & Sengul, 1978). For example, Turkish adults’ demonstrative selection depends on both spatial distance and joint attention, yet Turkish four-​ year-​ olds appear to be sensitive to spatial distance but not joint attention (Küntay & Özyürek, 2006). In this chapter we examine Spanish-​speaking adults’ and children’s demonstrative use to investigate the developmental trajectory of demonstrative use in Spanish. Languages generally have at least two demonstratives reflecting a distance scale:  a proximal demonstrative used for referents near the deictic center and a distal demonstrative used for referents some distance from the deictic center (Diessel, 1999, see also Anderson & Keenan, 1985; Cooperrider, 2016; Coventry, Valdés, Castillo, & Guijarro-​Fuentes, 2008). Demonstrative selection has also been shown to involve factors related to intersubjectivity, such as the negotiation of joint attention of the referent in the discourse (e.g., Peeters & Özyürek, 2016; Piwek, Beun & Cremers, 2008; Rybarczyk, 2015). The current pilot study examines Spanish-​speaking adults’ (N=10) and children’s (N=8) demonstrative selection within the speaker-​addressee shared space. More specifically, we ask whether spatial distance and intersubjectivity influence adults’ and children’s production of esta and esa during a puzzle completion task.

2.  Spanish demonstratives Spanish has three demonstratives, which inflect for gender2 and number, as shown in Table 14.1. In this chapter we focus on speakers’ selection of demonstrative type, that is, whether they choose from Set 1, Set 2 or Set 3, setting aside differences in gender and number. As such, we use the label esta to refer to any demonstrative in Set 1, esa for any demonstrative in Set 2, and aquella for any demonstrative in Set 3.

286  Naomi L. Shin and Jill P. Morford Table 14.1 Spanish demonstratives

Set 1 Set 2 Set 3

Masculine singular

Masculine plural

Feminine singular

Feminine plural

este ese aquel

estos esos aquellos

esta esa aquella

estas esas aquellas

Various proposals in the literature have focused on spatial distance to account for demonstrative selection among Spanish-​ speaking adults. According to the distance from speaker account, esta is used for referents near the speaker, esa for referents a bit further away from the speaker, and aquella for referents much further away (Coventry et al., 2008; Hottenroth, 1982; Kemmerer, 1999). The speaker/​addressee account is similar, but takes into account distance from the addressee, too: esta is used for referents near the speaker, esa for referents near the addressee, and aquella for referents at a distance from both the speaker and the addressee (Alonso, 1968; Bello, 1847). A third proposal, the dyad-​oriented account, is that speakers use esta for objects in the interlocutors’ shared space (Jungbluth, 2003). Experimental research has provided empirical evidence that lends some support to all three of the spatial distance accounts described above. In Coventry et al.’s (2008) experiment, participants (48 speakers of English and 52 speakers of Spanish) sat at a long table that was divided into quadrants. Each quadrant contained three colored dots. All dots in a quadrant were spaced equidistantly from the participant. Colored geometrical shapes were placed on the dots, and participants were instructed to point at the shapes and to use a demonstrative, a color, and a shape to describe the them (e.g., “this red triangle”). Half the participants used a 70 cm stick to point rather than their finger. Furthermore, the experimenter sat next to the participant in half the experimental sessions, and across from the participant at the other end of the table in the other half of the sessions. Results showed that the closer the object, the more often participants produced proximal demonstratives (English this and Spanish este). Spanish speakers preferred este for the first quadrant (closest to the participant), ese for the second quadrant, ese or aquel for the third quadrant, and aquel for the fourth quadrant. The use of the 70 cm stick significantly increased participants’ use of proximal this and este for objects in quadrants further from the participant, demonstrating that extended reach influences speakers’ sense of proximity. In addition, Spanish-​ speaking participants used este rather than ese to refer to objects in quadrant 2 more often when the experimenter sat across from the participant, which supports Jungbluth’s (2003) proposal that este is used for objects in the interlocutors’ shared space. Although concrete spatial distance plays a role in demonstrative selection, there is also evidence that speakers construe space subjectively. Kemmerer

Demonstratives in Spanish  287 (1999) argues that proximal and distal referents can be used to refer to the same object, regardless of actual distance. Enfield (2003) finds that the distal demonstrative in Lao can sometimes be used for a referent that is physically very close to the speaker. As Langacker (2008) puts it, the distinction between this/​these and that/​those “need not be spatial. If someone says I really like this pen, the proximity coded by this might be spatial (the speaker is holding the pen), temporal (the speaker is holding it now), functional (the speaker is using it), attitudinal (the speaker likes it), or any combination of these” (p. 283). Indeed, Gutiérrez-​Rexach’s (2002) Spanish example ese idiota de Juan illustrates the use of ese to achieve metaphorical distancing that connotes disdain. Another metaphorical extension of spatial distance takes into account the intersubjective nature of referent identification. Consider Piwek et  al.’s (2008) helpful analogy:  If you knock on a door and nobody answers, you knock harder in order to get someone’s attention. The analogy in discourse is the failure to coordinate mental reference (Langacker, 2008). If an addressee fails to identify a referent, the speaker must bring the addressee’s mental state closer to her own. Proximal demonstratives are the linguistic analogue to knocking harder at a door: the less accessible a referent, the more speakers will rely on proximal rather than distal demonstratives (Kirsner & Van Heuven, 1988; Leonard, 1995; Piwek et al., 2008).

3.  Acquisition of demonstratives Children’s acquisition of demonstratives begins at an early age (Clark & Sengul, 1978). In English and Spanish, children’s first demonstratives are produced around 21 months of age (Fenson et al., 1994; Rodrigo, González, de Vega, Muñetón-​Ayala & Rodríguez, 2004), and are typically used to draw the addressee’s attention to a referent. Clark and Sengul (1978) identify three dimensions of demonstrative usage in English that children must master, all of which apply to Spanish demonstratives as well. First, demonstratives are deictic—​they point to referents, a function closely tied to coordinating attention. Second, demonstratives shift in their meaning on the basis of who is speaking. Third, demonstratives have a locative dimension, specifying more proximal or distal points in space relative to the speaker. These dimensions of demonstratives are foundational, but most studies to date have not specified that demonstratives also have an intersubjective dimension, requiring the speaker to evaluate whether the speaker and addressee have a shared conception of space. Rodrigo and colleagues investigated the transition from gestural to verbal deixis in Spanish using longitudinal naturalistic observation of eight children:  four children were observed from 12  months to 24  months, and four children were observed from 24 months to 36 months. They found that children initially used pointing gestures to coordinate shared attention with their caretakers, but gradually introduced demonstratives and nominals in combination with points. Forms such as este/​esto/​esta were produced at four times the

288  Naomi L. Shin and Jill P. Morford rate of forms such as ese/​eso/​esa at this age range. Further, the investigators found a much higher rate of reference to proximal locations than to distal locations in interactions with the caretaker, although they did not report whether children used a proximal or non-​proximal demonstrative when referencing proximal locations. Studies of demonstrative comprehension in young children learning English provide additional insight into the mastery of the spatial dimension of demonstratives. Tanz (1980) tested children’s comprehension of the English this/​that contrast using a “Find the penny” experimental task. Children ages 2;6–​5;3 were seated across from the experimenter, who hid a penny under one of two plates while the child wasn’t looking. The experimenter introduced two puppets who sat next to the respective plates, and gave the children hints about where the penny was hidden, saying “This plate has a penny under it,” or “That plate has a penny under it.” Tanz found that the youngest children could reliably select the plate that was proximal to a puppet in response to “This plate,” but their selection of the distal plate in response to “That plate” was at chance until they were over four years of age. When she asked the children to do the hiding and to provide the cues to where the penny was hidden, the youngest children understood the deictic function of demonstratives, but did not yet distinguish demonstrative selection according to distance. Instead, they would use the same demonstrative for both referents, such as “This one” vs. “This one” or “That one does but that one doesn’t,” while relying on the co-​ occurring pointing gesture to distinguish the location. Only the five-​year-​olds were able to use demonstratives to distinguish proximal vs. distal referents. Although these results indicate a gradual increase in children’s comprehension of demonstratives, Tanz emphasizes that there was individual variability in the performance of children at every age. Clark and Sengul (1978) also tested comprehension experimentally, but included a condition in which the experimenter sat next to the child, and one in which the experimenter sat across from the child. This manipulation allowed them to determine whether the children were using themselves as the deictic center, or whether the speaker was interpreted as the deictic center. In other words, if the experimenter says, “This one,” does the child select a referent close to the child or close to the experimenter? Results showed that it was twice as common for children to select themselves as the deictic center instead of the experimenter. Yet, some of the youngest children showed a preference for interpreting the experimenter as the deictic center. Thus, not all children follow the same path to mastery of demonstratives. At the same time, and like Rodrigo et al. (2004) and Tanz (1980), Clark and Sengul found that children mastered proximal uses of demonstratives earlier than non-​proximal uses. Two studies that consider the role of intersubjective factors on children’s demonstrative selection yield potentially contradictory conclusions. Skarabela, Allen, & Scott-​Phillips (2013) analyzed spontaneous conversational data from four Inuktitut-​speaking children, ages 2;0 to 3;6, who were recorded speaking to family and friends. Demonstratives in Inuktitut can occur as either clitics

Demonstratives in Spanish  289 or independent forms. The authors found that while the use of demonstrative clitics was rare at these ages, they were almost always used in the presence of joint attention. They argue that this reflects children’s sensitivity to the impact of joint attention on form selection: when joint attention is present, the children could introduce abbreviated forms, because the need to guide the addressee’s attention to the referent was less critical. The authors conclude that very young children display pragmatic awareness, taking the addressee’s focus of attention into account in their choice of demonstrative. In contrast, Küntay & Özyürek’s (2006) study of Turkish suggests that children do not master intersubjective dimensions of demonstrative use until they are about six years old. Turkish encodes a three-​way demonstrative contrast with two demonstratives distinguishing proximal and distal referents within the addressee’s visual attention, and a third demonstrative used for referents (proximal or distal) that are not within the addressee’s visual attention (Özyürek, 1998; Özyürek & Kita, 2000). Due to the obligatory marking of the attentional state of the addressee, the study of Turkish demonstratives provides an important contrast to other languages. Küntay & Özyürek asked 12 children and six adults to complete a collaborative building task and evaluated their spontaneously produced demonstratives. They found that four-​year-​olds are sensitive only to the distance contrast in Turkish demonstratives, while six-​year-​olds show emerging sensitivity to the intersubjective dimension. Specifically, the six-​year-​olds used the demonstrative required for referents outside the addressee’s visual attention, but only when the referent was spatially proximal, indicating that they had discovered how it contrasts with the proximal demonstrative, but not with the distal demonstrative. This pattern of results indicates that even though children’s earliest demonstratives are used to capture the addressee’s attention, intersubjective monitoring and manipulation of the addressee’s attentional state during an interaction is more challenging than tracking the distance of referents relative to the self. On first reflection, the results from Turkish may seem to contradict the findings from Inuktitut. However, both studies provide evidence that young children are sensitive to the addressee’s attentional state. They differ in the degree to which the demonstrative is used to shift the addressee’s attention. Two-​to three-​year-​old speakers of Inuktitut can select a less salient form to talk about a referent in the addressee’s focus of attention. By contrast, even six-​year-​old speakers of Turkish have not yet learned that a demonstrative that can get the addressee’s attention for a proximal referent could also shift that addressee’s attention to a distal referent.3 In sum, young children produce demonstratives among their earliest words, and recognize their deictic function for directing attention. They develop sensitivity to the shifting deictic center and to the encoding of distance between three and five years of age, with mastery of proximal demonstratives preceding mastery of distal demonstratives. By six years of age, children show evidence of using demonstratives to shift the addressee’s attentional state in

290  Naomi L. Shin and Jill P. Morford languages that obligatorily encode both spatial distance and attentional state, but do not yet exhibit fully adult-​like use of demonstratives. What is missing from studies of demonstrative acquisition to date is an evaluation of children’s developing sensitivity to both spatial and intersubjective dimensions of demonstrative selection in languages that do not mark intersubjective contrasts obligatorily. Further, while the Rodrigo et al. study provides a detailed account of demonstrative usage from 12 to 36  months, the acquisition of Spanish demonstratives in the pre-​school to early elementary years has yet to be investigated, despite the fact that children achieve important social and cognitive milestones during these years. The current study evaluates the influence of spatial and intersubjective factors on demonstrative selection in children acquiring Spanish between three and eight years of age. We hypothesize that sensitivity to the spatial dimension of demonstratives will emerge earlier than the intersubjective dimension of demonstratives, as children acquiring English and Turkish demonstrate early mastery of the proximal-​distal contrast between four and six years of age, but children acquiring Turkish do not yet fully master the use of demonstratives to shift the addressee’s attentional state by six years of age. A second reason to predict earlier emergence of the spatial contrast in children acquiring Spanish is that spatial and intersubjective dimensions are conflated with respect to form contrasts. Hence, the multifunctionality of demonstrative forms in Spanish is predicted to protract the rate of mastery. However, since Turkish six-​year-​olds are already beginning to acquire the third demonstrative form, it is possible that similarly aged children acquiring Spanish will be sensitive to the addressee’s focus of attention, and modify their choice of demonstrative to take the addressee into account. For this reason, we have designed a task in which the location of referents changes across trials, as well as whether or not the participant and the experimenter’s focus of attention is shared. Given the lack of prior studies that probe intersubjective modulations to demonstrative selection in Spanish, we first evaluate demonstrative selection in a group of adult native speakers to ascertain that the task will elicit changes in demonstrative selection when there is a disruption to the shared conception of space by the speaker and the addressee. We then present the results of our study of children’s demonstrative selection using the same task.

4.  Methods 4.1  Participants Our study included 18 Spanish speakers, 10 adults and 8 children, all of whom resided in New Mexico. The adults’ ages ranged from 19 to 35. All were bilingual in Spanish and English, and considered Spanish to be their first language. Nine were born in a country in which Spanish is the dominant

Demonstratives in Spanish  291 language (Colombia, Costa Rica, Mexico, or Spain), and one participant was born in California. All lived in the US at the time of their participation in the experiment. Self-​evaluation for Spanish fluency on a scale of 0 to 6 was on average 5.8 for speaking and 6.0 for understanding.4 The eight child participants included four children between three and five years old (average age 4;3) and four between six and eight years old (average 7;6). All eight of these children are being raised by caregivers who were born and raised in Spanish-​speaking countries. The caregivers also reported that more Spanish than English was spoken in the home. In other words, these children were exposed to an abundant amount of Spanish. 4.2  Materials and procedure The experimenter sat opposite the participant at a table and placed a 25-​piece bordered puzzle on the table in front of the participant, with a band dividing the participant and the puzzle board from the experimenter’s side of the table. Thirteen puzzle pieces were placed on the participant’s side of the barrier (the proximal region) and 12 were placed on the experimenter’s side (the distal region) (Figure 14.1). The experimenter explained that the task was to put the puzzle together following two rules: The participant may not touch the pieces and the participant may not reach across the barrier created by the band. The experimenter asked the participant a sequence of questions to elicit the selection of puzzle pieces for placement in the puzzle. Intersubjectivity was manipulated by asking two types of questions. For find it questions (N=25), the experimenter described what the piece looked like, e.g., “¿Cuál pieza tiene el ojo del dinosaurio verde?” “Which piece has the green dinosaur’s eye?” The second question type was misunderstanding questions (N=10): after the participant identified a piece as a response to a find it question, the experimenter explicitly ignored the participant’s choice and selected a different piece, while saying ¿ Esta? “This one?”5 These misunderstanding questions were predicted to elicit responses in which the participant corrected the experimenter and

Figure 14.1 Puzzle task

292  Naomi L. Shin and Jill P. Morford clarified which puzzle piece was the intended referent. The entire session was videotaped. 4.3  Coding Coders first transcribed all participant responses to the questions, as well as gestures that accompanied demonstratives. For the current study, only responses that included demonstratives esta/​este or esa/​ese were analyzed (N=628).6 These responses were categorized according to: 1 Demonstrative type:  esta and esa. Recall that we do not explore any variations in gender in this study; therefore, the labels esta and esa are notations for este/​esta/​estos/​estas and ese/​esa/​esos/​esas, respectively. The feminine forms were more common in our data set, and there were no instances of neuter forms. 2 The target location of the participant’s demonstrative:  proximal (for puzzle pieces on the participant’s side of the barrier) vs. distal (for puzzle pieces on the experimenter’s side of the barrier). 3 Question type: find it and misunderstandings

5.  Results 5.1  Adults’ demonstrative selection Results from a mixed effects binary logistic regression show that both spatial distance and question type impact adults’ demonstrative selection. The model reported in Table  14.2 includes Demonstrative type (esta vs. esa) as the dependent variable, with esta set as the application value, and two independent variables:  Barrier side (participant’s side, experimenter’s side) and Question type (find it, misunderstanding).7 Participant is included as a random variable.

Table 14.2 Mixed effects binary logistic regression, Demonstrative selection, 10 Spanish-​speaking adults, 346 tokens N tokens (Intercept)

Estimate .17

Std. Error .37

Barrier side (Reference Level: Experimenter’s side, N =146) Participant’s side 200 2.15 .32 Question type (Reference Level: Find it, N =279) Misunderstanding 67 1.14 .44

z value .48

p-value .63

6.71

< .001

2.61