Functional Approaches to Spanish Syntax: Lexical Semantics, Discourse and Transitivity

Functional Approaches to Spanish Syntax Lexical Semantics, Discourse and Transitivity

Edited by J. Clancy Clements and Jiyoung Yoon

Functional Approaches to Spanish Syntax

This page intentionally left blank

Functional Approaches to Spanish Syntax Lexical Semantics, Discourse and Transitivity Edited by

J. Clancy Clements Indiana University

and Jiyoung Yoon University of North Texas

Editorial matter and selection and Chapter 1 © J. Clancy Clements and Jiyoung Yoon 2006 Individual chapters © contributors 2006 All rights reserved. No reproduction, copy or transmission of this publication may be made without written permission. No paragraph of this publication may be reproduced, copied or transmitted save with written permission or in accordance with the provisions of the Copyright, Designs and Patents Act 1988, or under the terms of any licence permitting limited copying issued by the Copyright Licensing Agency, 90 Tottenham Court Road, London W1T 4LP. Any person who does any unauthorized act in relation to this publication may be liable to criminal prosecution and civil claims for damages. The authors have asserted their rights to be identified as the authors of this work in accordance with the Copyright, Designs and Patents Act 1988. First published 2006 by PALGRAVE MACMILLAN Houndmills, Basingstoke, Hampshire RG21 6XS and 175 Fifth Avenue, New York, N.Y. 10010 Companies and representatives throughout the world. PALGRAVE MACMILLAN is the global academic imprint of the Palgrave Macmillan division of St. Martin’s Press, LLC and of Palgrave Macmillan Ltd. Macmillan® is a registered trademark in the United States, United Kingdom and other countries. Palgrave is a registered trademark in the European Union and other countries. ISBN-13: 978–1–4039–9406–6 ISBN-10: 1–4039–9406–4 This book is printed on paper suitable for recycling and made from fully managed and sustained forest sources. A catalogue record for this book is available from the British Library. Library of Congress Cataloging-in-Publication Data Clements, J. Clancy. Functional approaches to Spanish syntax : lexical semantics, discourse and transitivity / edited by J. Clancy Clements and Jiyoung Yoon. p. cm. Includes bibliographical references and index. ISBN 1–4039–9406–4 1. Spanish language – Syntax. I. Yoon, Jiyoung. II. Title. PC4361.C54 2005 465—dc22 2005051277 10 9 8 7 6 5 4 3 2 1 15 14 13 12 11 10 09 08 07 06 Printed and bound in Great Britain by Antony Rowe Ltd, Chippenham and Eastbourne

This book is dedicated to the memory of Clements’ teacher Eugenio Coseriu

This page intentionally left blank

Contents List of Tables and Figures

ix

Acknowledgements

xii

Notes on the Contributors

xiii

1

Introduction J. Clancy Clements and Jiyoung Yoon

2

Semantic and Discourse-Pragmatic Factors in Spanish Word Order Belén López Meirama

7

Continuity and Episodic Structure in Spanish Subject Reference Llorenç Comajoan

53

3

1

4

Gustar-Type Verbs Victoria Vázquez Rozas

80

5

Primary and Secondary Object Marking in Spanish J. Clancy Clements

115

6

Null Direct Objects in Spanish J. Clancy Clements

134

7

Transitivity and the Syntax of Inalienable Possession in Spanish Richard Winters

8

Ser-estar in the Predicate Adjective Construction J. Clancy Clements

9

Spanish Adjective Position: Differences between Written and Spoken Discourse Richard J. File-Muriel

151 161

203

10 Adjective Placement and Noun Semantics in Spanish Mariche García-Bayonas

219

11 Transitivity and Spanish Non-Anaphoric se J. Clancy Clements

236

vii

viii Contents

12 ‘Juan salió contento’: Semantic Constraints on Small Clauses in Adjunct Position Jiyoung Yoon

265

13 Causative hacer and dejar Carmen Ruiz-Sánchez

278

Index

301

List of Tables and Figures

Tables 1.1 1.2 2.1

2.2 2.3 2.4 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 3.10 3.11 3.12 4.1 4.2 4.3 4.4 5.1 5.2

Component parts of Transitivity Component parts of individuation Comparison of semantic differences and syntactic patterns of prototypically transitive clauses sentences v. clauses sentences with a gustar-type verb construction (In)definiteness, (non-)referentiality and topichood The theme–rheme dichotomy The overlay between thematic and syntactic structure Percentage rates of third-person pronouns in different Spanish varieties Percentage rates of zero, pronoun, and full NPs in different Spanish varieties Nominal devices for topic continuity Subject nominal devices used in the narratives Nominal devices to introduce new characters Nominal devices to introduce same subjects from previous clauses Nominal devices to introduce different subjects from previous clauses Look-back distance for definite NPs, pronoun, and zero when they refer to a different subject from the previous clause Overall hits and misses for the episodic model Hits and misses for new referents, old referents and different subjects Nominal devices in episodes 5 and 6 Nominal devices to introduce continuous subjects in Spanish, English and Japanese Degree of topicality in indirect objects Frequencies and percentages of animate v. inanimate subjects in two-participant clauses Frequencies and percentages of clauses functioning as subject Frequencies and percentages of preposition and postposition of subject and indirect object for gustar Nominative-accusative v. ergative-absolutive marking Direct–indirect v. primary–secondary object marking ix

3 3

18 21 26 29 54 55 56 60 61 61 62 63 63 64 69 69 84 88 89 97 117 122

x

List of Tables and Figures

5.3 5.4

The etymological Spanish pronominal system Frequency of le as a function of the referents’ animacy in (1) writings of St Teresa; (2) speech of present-day Castilian professional women; (3) speech of present-day Castilian rural speakers 5.5 A probable result of leísmo in a variety of the Castilian pronominal system 5.6 General pronominal system found in the Moratín play El sí de las niñas 5.7 Probable object-marking system based on pronoun use in Moratin’s El sí de la niñas 6.1 Lazard’s definiteness scale 8.1 Characterization of verb classes 8.2 Examples of Spanish adjective types in terms of features 8.3 Comparison of commonly occurring verb forms of ser and estar 8.4 Form-by-form frequency comparison of different forms of ser and estar 8.5 Copula-adjective combination according to adjective type and reading for both animate and inanimate subject referents 9.1 Spoken and written style 9.2 Position of As to Ns for the all the data according to their respective weights 9.3 Position of As to Ns in the spoken data according to their respective weights 9.4 Position of As to Ns in the written data according to their respective weights 9.5 Preposed adjectives in spoken discourse 9.6 Adjective position 9.7 One v. two adjectives modifying same noun 9.8 Distribution of semantic categories 9.9 Chi-square test for Table 9.8 data (semantic class) 9.10 Diachronic distribution of A position in texts 9.11 Mean word length of As by semantic class 10.1 Interpretations of alto funcionario and funcionario alto in adults and children 10.2 Interpretations for azafata alta and alto piloto in adults and children 10.3 Semantic interpretations for N-pequeño and pequeño-N in adults and children 10.4 ‘Miserable’ v. ‘poor’ interpretation for preposed v. postposed pobre

124

125 125 126 127 143 169 173 182 183

198 210 210 210 210 211 212 212 212 213 214 214 225 225 226 227

xi List of Tables and Figures

10.5 10.6 10.7 10.8

11.1 12.1 13.1 13.2

‘Mere’ v. ‘simple-minded’ interpretation for preposed v. postposed simple Chi-square results for the adults Chi-square results for the children The significance of specific–general and general–specific ordering in the instrument, calculated with tokens from the specific semantic interpretation in adult and child groups Characterization of Vendlerian Aktionsart categories Licensing factors of adjunct predicates in relation to Transitivity Characterization of causative dejar Characterization of causative hacer

227 228 228

229 251 272 288 289

Figures 5.1 8.1 8.2 8.3

8.4 13.1 13.2 13.3 13.4 13.5

Possibilities for a simple lexical split of case marking: two two-way sub-systems, ‘accusative’ v. ‘ergative’ The distribution of ILPs and SLPs over states, activities and events Prototypical coding of the world in language based on time stability Time stability of adjectives with no underlying process/event v. adjectives with an underlying process v. adjectives with an underlying event Spanish adjective classification Reference frame of hacer Reference frame for causative hacer with activities Reference frame for main verb and causative dejar Implied reference frame of causative dejar in certain cases Reference to the result of an event in small clauses with an implied event or state

116 166 169

170 173 295 296 296 297 297

Acknowledgements This volume has benefited greatly from discussions with students over the years. We are grateful to those students for all their input and also to an anonymous reviewer whose comments shaped the argumentation of many of the contributions. Thanks go also to Jill Lake for presenting this project to Palgrave Macmillan, to Manolo Triano López for translating Chapters 2 and 4 and to Richa Clements for proofreading some of the typescript. J. CLANCY CLEMENTS JIYOUNG YOON The editors, contributors and publishers are grateful to Georgetown University Press for permission to use Chapter 6, Mouton de Gruyter for permission to use Chapter 5, and Elsevier Publishers for permission to use Table 6.1. Every effort has been made to contact all copyright holders, but if any have been inadvertently omitted the publishers will be pleased to make the necessary arrangement at the earliest opportunity.

xii

Notes on the Contributors J. Clancy Clements is Associate Professor of Linguistics and Spanish and Portuguese at Indiana University, Bloomington, USA. He received his MA (1979) in Spanish from the Universität Tübingen, Germany, and his PhD (1985) in Romance Linguistics from the University of Washington, Seattle. His main areas of interest are contact linguistics and functional syntax, with a focus on varieties of Iberian Romance languages. His writings include The Genesis of a Language: The Formation and Development of Korlai Portuguese (1996) and The Linguistic Legacy of Spanish and Portuguese: Colonial Expansion and Language Change (forthcoming), four co-edited volumes, as well as over 30 articles on language-contact phenomena and functional linguistics. Llorenç Comajoan is Assistant Professor of Spanish at Middlebury College, USA. His main areas of interest are second-language acquisition, language variation and language policy. Richard File-Muriel is a doctoral student in the Department of Spanish and Portuguese at Indiana University, Bloomington, USA. His main interests are phonetics, phonology and contact linguistics. Mariche García-Bayonas is Assistant Professor in the Department of Romance Languages at the University of North Carolina at Greensboro, USA. Her main areas of interest are second-language acquisition, phonetics, phonology and sociolinguistics. Belén López Meirama is Professor of Spanish in the Department of Spanish at the Universidad de Santiago de Compostela, Spain. Her main areas of interest are functional syntax and syntax of Spanish, with a focus on the syntax of the sentence. Her writings include La posición del sujeto en la cláusula monoactancial en español (1997), and various articles on Spanish syntax and pedagogically oriented writings. Victoria Vázquez Rozas is Professor in the Department of Spanish in the Universidad de Santiago de Compostela, Spain. She was a visiting professor at the Romansk Instituto of the University of Copenhagen (1989) and in the Spanish and Portuguese Department at the University of California at Santa Barbara (2001). She has published numerous studies on syntax and discourse from a functionalist perspective, including El complemento indirecto en español (1995). She is a contributor to the development of the Syntactic Data Base (http://www.bds.usc.es). Carmen Ruiz Sánchez is a doctoral student in the Department of Spanish and Portuguese at Indiana University, Bloomington, USA. Her three main areas of xiii

xiv Notes on the Contributors

specialization are Hispanic sociolinguistics, second-language acquisition and functional syntax. She received her BA (1999) in English philology from the University of Seville, Spain, and her MA from Indiana University in TESOL/Applied Linguistics (2001) and Hispanic Linguistics (2003). Richard Winters is Assistant Professor in the Department of Modern Languages at the University of Louisiana at Lafayette, USA. His main research interests are functional and formal syntax, with a focus on the Romance languages. Jiyoung Yoon is Assistant Professor of Spanish and Spanish Linguistics and a coordinator of the first- and second-year Spanish programme at the University of North Texas (Denton). She received her MA at the Universidad de Guadalajara in Mexico (1994) and her PhD at Indiana University (2002). She specializes in functional syntax and semantics as well as foreign language instruction and pedagogy. Her work includes a forthcoming monograph on Spanish small clause constructions and articles on semantic and syntactic phenomena in Spanish from both functional and construction-grammar approaches. She is currently working on a cross-linguistic study of subject and object coding.

1 Introduction J. Clancy Clements and Jiyoung Yoon

Since about 1970 the study of syntax has increasingly benefited from semantics-based approaches. In generative grammar, work by Bresnan (1982, 2001), Levin and Rappaport Hovav (1995) and Pollard and Sag (1987, 1994), among others, have acknowledged and explored the intricate links between lexical semantics and syntactic structure. In the functional/cognitive-oriented literature, work by Bybee (1985, 2005), Bybee, Perkins and Pagliuca (1994), Dean (1992), Dik (1997), Croft (1991, 2001), Foley and Van Valin (1984), Givón (1995, 2001), Goldberg (1995), Gutierrez-Ordóñez (1997), Halliday (1985), Hopper and Thompson (1980, 1984), Langacker (1987, 1991), MacWhinney (1999) Thompson and Hopper (2001), Tomasello (1998, 2003), Van Valin and LaPolla (1997), Vazquez Rozas (1997), Verhaar (1990) and many others have significantly furthered the study of the multifarious connections between syntax on the one hand and lexical semantics, discourse and cognition on the other. Levin and Rappaport Hovav (1995: 1) note that the hypothesis that the syntactic properties of lexical items (specifically verbs) are determined by their meaning is traceable as far back as P¯an.ini in the 6th century BCE. They suggest that there is a causal relation between the regularities linking arguments bearing certain semantic roles and particular syntactic expressions: [t]o the extent that the semantic role of an argument is determined by the meaning of the verb selecting it, the existence of linking regularities supports the idea that verb meaning is a factor in determining the syntactic structure of sentences. The striking similarities in the linking regularities across languages strongly suggest that they are part of the architecture of language. From the perspective of language acquisition, Bates and Goodman (1999) offer substantial empirical evidence in first-language (L1) acquisition to support this causal relation between the lexicon and grammar (comprising syntactic structure and also grammatical relation marking), arguing for a 1

2

Introduction

unified lexicalist approach to the processes involved in grammar acquisition. Numerous studies by them and their colleagues reveal the close relation between the development of the L1 lexicon on the one hand, and the development of L1 grammar on the other. That is, ‘[t]he dependence of early grammar on vocabulary size is so strong and the nonlinear shape of this function is so regular that it approaches the status of a psychological law …’ (1999: 51). The syntax of Spanish is intricately connected to the verb semantics and discourse structure. In Chapter 2, for example, it is shown for intransitive verbs that the more agentive and definite a subject is, the more likely it is to appear preverbally; and the less agentive and definite a subject is, the more likely it is to appear postverbally. Spanish syntax is also sensitive to the argument structure of verbs. In Chapter 5, Spanish is argued to be a (SV/VS-VO) subject–verb/verb–subject–verb–object type language (Dryer, 1997), with a slight predominance of verb–subject order in intransitive clauses but an overwhelming predominance of subject–verb order in transitive clauses. In Chapter 4, it is shown that two-argument verbs of the gustar type have overwhelmingly experiencer subjects, and as these are low in agency they appear overwhelmingly in postverbal positions. The syntax of Spanish is also sensitive to the semantics of the noun phrase (NP). Animacy of object arguments is shown to underlie the ergative pattern discovered in the Spanish pronoun system, discussed in Chapter 5. Overt v. null pronominalization of direct objects, examined in Chapter 6, is found to depend on the semantic nature of the direct object. Thus, the semantics of the verb and its argument NPs in Spanish has direct repercussions for its syntax. One seminal work that has informed much of the above-mentioned research, and one which addresses the intimate connection between lexical semantics, morphosyntax and discourse, is Hopper and Thompson’s (1980) Transitivity Hypothesis. Traditionally, transitive clauses are those whose verbs take a direct object (DO). Hopper and Thompson, however, cast the notion more broadly, viewing Transitivity as a set of component parts that takes into consideration all aspects of the carrying-over of an action (or the transferring of energy) from one participant to another. They argue that Transitivity is scalar (2001: 28) and that it ‘can be broken down into its component parts, each focusing on a different facet of the carrying-over in a different part of the clause’ (1980: 253). By breaking Transitivity down into its component parts, it is possible to characterize clauses, not as transitive or intransitive, but rather as more or less transitive. On this view, Transitivity becomes a question of degree, and syntactic structure and marking can be accounted for by positing degrees (high or low, more or less) of Transitivity. The factors making up the notion of Transitivity, adapted from Hopper and Thompson (1980) and shown in Tables 1.1 and 1.2, are lexical semantic properties of verbs/predicates and nouns and discourse properties of noun

J. Clancy Clements and Jiyoung Yoon 3

phrases (for example definiteness, referentiality). There are two observations to be made regarding the tables. First, the reference to objects in Table 1.1j, 1.1k and Table 1.2 refers to both direct as well as indirect objects. Hopper and Thompson (1980: 259–60) note that indirect objects should be considered Transitive objects, just as direct objects are. Secondly, the item included in Table 1.1i (in small caps and bold) is an addition to Hopper and Thompson’s original proposal, an extension originally proposed in a similar framework by Verhaar (1990: 94–7), who distinguishes between what he calls intransitive verbs of actorhood, high on the Transitivity scale such as the equivalents of ‘swim’, ‘talk’ and ‘run’, and intransitive verbs of undergoing, low on the Transitivity scale such as equivalents of ‘fall’, ‘melt’ and ‘faint’. By adopting this extension, it is possible to include intransitive clauses on the Transitivity scale as well. Thus, with the help of the values ‘high/low’ in Table 1.1 and ‘individuated/non-individuated’ in Table 1.2 one can measure the degree of Transitivity of any given clause, and use this as a basis for predicting the covariance of semantic and syntactic features of different elements that co-occur in a clause. Table 1.1 Component parts of Transitivity

a. PARTICIPANTS b. KINESIS c. ASPECT d. PUNCTUALITY e. VOLITIONALITY f. AFFIRMATION g. MODE h. AGENCY i. INDIVIDUATION of S(ubject) j. AFFECTEDNESS of O k. INDIVIDUATION of O

High

Low

Two or more participants, A(gent) and O(bject) action telic punctual volitional affirmative realis A high in potency

One participant non-action atelic nonpunctual nonvolitional negative irrealis A low in potency

S HIGHLY INDIVIDUATED O totally affected O highly individuated

S NON-INDIVIDUATED O not affected O nonindividuated

Table 1.2 Component parts of individuation (Table 1.1i and 1.1k) Individuated

Non-individuated

proper human, animate concrete singular count referential, definite

common inanimate abstract plural mass nonreferential, indefinite

4

Introduction

The Transitivity Hypothesis states that in any given clause, elements of the clause should co-vary in terms of their high- or low-Transitivity features. Two examples from Spanish will illustrate the point. In Chapter 6, overt v. null direct object pronominalization is examined. It turns out that overt pronominalization is not necessary if the direct object is equivalent to a mass noun or a bare plural. Whereas the pronominalization of el libro in (1.1a) with lo is obligatory in Spanish syntax (see 1.1b v. 1.1c), the pronominalization of a mass noun can be with a null, as shown in (1.2): (1.1)

a. ¿Compraste el libro? ‘Did you buy the book?’ b. Sí, lo compré. ‘Yes, I bought it.’ c. *Sí, compré. ‘Yes, I bought.’

(1.2)

a. ¿Compraste cafe? ‘Did you buy coffee?’ b. Sí, compré Ø;. ‘Yes, I bought (some).’

In terms of the Transitivity Hypothesis, this can be analysed as follows. Overt pronominalization co-varies with higher Transitivity, while null pronominalization co-varies with lower Transitivity. This is calculated using the component parts in Individuation listed in Table 1.2. The direct object NP el libro ‘the book’ in (1.1a) is count, referential and definite, whereas the direct object NP café ‘coffee’ in (1.2a) is a noncount (mass), nonreferential and indefinite. The NP with a higher number of high-Transitivity (Individuation) traits is overtly pronominalized in Spanish. This is discussed in more detail in Chapter 6. The other example involves the verb pair dormir ‘sleep’ and dormirse ‘fall asleep’. In Chapter 11, we argue that the presence of se co-varies with higher Transitivity. In this example, as well as in many others, it is true. In terms of the component parts of Transitivity in Table 1.1, the predicate dormir ‘sleep’ is atelic. That is, it has duration and does not contain an inherent endpoint. By contrast, dormirse ‘fall asleep’ has no duration (it is punctual) and is thus telic. In other words, the point at which the event occurs is its endpoint. Thus, sentence (1.3a) is lower in Transitivity than sentence (1.3b): (1.3)

a. Marta durmió en casa de su amiga. ‘Marta slept at her friend’s house.’ b. Marta se durmió en casa de su amiga. ‘Marta fell asleep at her friend’s house.’

J. Clancy Clements and Jiyoung Yoon 5

The present volume adopts the Transitivity Hypothesis, as well as other models in the area of lexical semantics, syntax and discourse, as the basis for examining various morphosyntactic, syntactic and discourse phenomena in Spanish. The topics include word order (Chapters 2 and 4), null arguments (Chapters 3 and 6), ergative patterns in object marking (Chapter 5), inalienable possession (Chapter 7), ser-estar (Chapter 8), non-anaphoric se (Chapter 11), small-clause phenomena (Chapter 12) and causative constructions (Chapter 13). Adjective placement is analysed in a number of ways. Chapter 9 (‘Syllable weight and adjective placement’) demonstrates statistically that processing considerations involving syllable weight of nouns and adjectives play an important role in adjective placement. In Chapter 10 (‘Spanish adjective placement and noun semantics’), it is suggested that the semantic interpretation of a noun (specific v. general) is linked to adjective pre- v. postnominal placement respectively. The application of probabilitistic methodology is shown effective in identifying the most significant variables affecting adjective placement and the rates of grammaticalization apparent in this phenomenon (Chapters 9 and 10). Moreover, the use of data from large corpora (Chapters 2, 4, 5, 6, 8, 9, 11, 12, 13) or statistical analysis of elicited/collected data (Chapters 3, 7) lends strong empirical support to the findings on overt v. null arguments, word order, case-marking systems and adjective placement. We hope that this volume will contribute to filling the void we perceive in the literature on Spanish functional syntax and in general usage-based analysis of natural language. To date, this is the first comprehensive study of Spanish syntax from the general functional perspective. We hope it will encourage other scholars to develop and extend this line of research. References Bates, E. and J. Goodman (1999) On the Emergence of Grammar from the Lexicon. The Emergence of Language, ed. B. MacWhinney. Malwah, NJ: Lawrence Erlbaum, pp. 29–79. Bresnan, J. (2001) Lexical-Functional Syntax. Malden, MA: Blackwell. —— (ed.) (1982) The Mental Representation of Grammatical Relations. Cambridge, MA: MIT Press. Bybee, J. (2005) ‘The Impact of Use on Representation: Grammar is Usage and Usage is Grammar’, Presidential address at the Linguistics Society of America annual meeting. Oakland, CA, 6–9 January. —— (1985) Morphology. Amsterdam: Benjamins. Bybee, J., R. Perkins and W. Pagliuca (1994) The Evolution of Grammar. Chicago: University of Chicago Press. Croft, W. (2001) Radical Construction Grammar. Oxford: Oxford University Press. —— (1991) Syntactive Categories and Grammatical Relations. Chicago: University of Chicago Press. Dean, P. (1992) Grammar in Mind and Brain. Exploration in Cognitive Syntax. Berlin: Mouton de Gruyter.

6

Introduction

Dik, S. (1997) The Theory of Functional Grammar. 2 vols, revd edn. Berlin: Mouton de Gruyter. Dryer, M. (1997) ‘On the Six-way Word Order Typology’, Studies in Language, vol. 21(1), pp. 69–103. Foley, W. and R. Van Valin (1984) Functional syntax and universal grammar. Cambridge: Cambridge University Press. García, E. (1975) The role of theory in linguistic analysis: the Spanish pronoun system. Amsterdam: North Holland. Givón, T. (2001) Syntax: an introduction. 2 vols. Amsterdam: Benjamins. —— (1995) Functionalism and grammar. Amsterdam: Benjamins. Goldberg, A. (1995) Constructions. A Construction Grammar Approach to Argument Structure. Chicago: University of Chicago Press. Gutiérrez Ordoñez, S. (1997) Principios de sintaxis funcional. Madrid: Arco Libros. Hopper, P. J. and S. A. Thompson (1984) ‘The Discourse Basis for Lexical Categories in Universal Grammar’, Language, vol. 60, pp. 703–52. —— (1980) ‘Transitivity in Grammar and Discourse’, Language, vol. 56, pp. 251–99. Langacker, R. (1991) Foundations of Cognitive Grammar II. Stanford: Stanford University Press. —— (1987) Foundations of Cognitive Grammar I. Theoretical Prerequisites. Stanford: Stanford University Press. Levin, B. and M. R. Hovav (1995) Unnacusativity: At the Syntax-lexical Semantics Interface. Cambridge, MA: MIT Press. MacWhinney, B. (ed.) (1999) The Emergence of Language. Malwah, NJ: Lawrence Erlbaum. Pollard, C. and I. Sag (1994) Head-driven Phrase Structure Grammar. Chicago: Chicago University Press. —— (1987) Information-Based Syntax and Semantics. Stanford, CA: Center for the Study of Language and Information. Thompson, S. A. and P. J. Hopper (2001). ‘Transitivity, Clause Structure, and Argument Structure: Evidence from Conversation’. in J. Bybee and P. J. Hopper (eds), Frequency and the Emergence of Linguistic Structure. Amsterdam: Benjamins, 27–60. Tomasello, M. (2003) The New Psychology of Language. Cognitive and Functional Approaches to Language Structure. Volume 2. Mahwah, NJ: Lawrence Erlbaum. —— (1998) The New Psychology of Language. Cognitive and Functional Approaches to Language Structure. Volume 1. Mahwah, NJ: Lawrence Erlbaum. Van Valin, R. and R. LaPolla (1997) Syntax: Structure, Meaning and Function. Cambridge: Cambridge University Press. Vázquez Rozas, V. (1995) El complemento Indirecto en español. Santiago de Compostela: Universidade de Santiago de Compostela. Vendler, Z. (1967) Linguistics in Philosophy. Ithaca, NY: Cornell University Press. Verhaar, J. (1990) ‘How Transitive is Intransitive?’, Studies in Language, vol. 14, pp. 93–168.

2 Semantic and Discourse-Pragmatic Factors in Spanish Word Order1 Belén López Meirama

Introduction Spanish is usually characterized as a free-word-order language that typologically belongs to the group of subject–verb–object (SVO) languages. In this respect, we will address two things. First, a typological description of Spanish word order is too vague given that Spanish has many intransitive clauses in which subjects are postverbal (VS order). In this chapter, we will argue that Spanish is a subject-initial, that is, an SVO/SV, language.2 Secondly, the expression ‘free order’ cannot be construed as ‘arbitrary order’ since the variations in the order of constituents are never gratuitous. Following Thompson (1978: 20), we claim that Spanish is a language that uses constituent order for both grammatical and discourse-pragmatic purposes: From a typological point of view, then, we can show that some languages utilize predicate-argument order primarily for pragmatic purposes, and some primarily for grammatical purposes. There are also languages which use predicate-argument order for both purposes without giving priority to either; Spanish may be such a language. It is not uncommon for word order to be considered a way of identifying grammatical relations between the constituents of the sentence, generally related to case marking and agreement. In typological approaches to language, it is a common practice to establish possible relationships between different markings (see, for instance, Forster and Hofling, 1987). In this sense, word order can play an important role in morphologically poorer languages such as English or French. However, in morphologically richer languages such as Spanish, in which word order does not play a crucial role in identifying grammatical relations in the sentence, word order variations are used at a discourse-pragmatic level ‘to establish a social or affective stance, to create text-level structures from sentence-level resources, and to communicate in a 7

8

Spanish Word Order

way which is optimally matched to the text receiver’s cognitive capacities’ (Downing 1995: 9). In Spanish, there are many factors that influence the order of constituents, some of which are strictly formal. For example, it is claimed that Spanish follows the quasi-universal rule of placing the heaviest constituents towards the end of the sentence (see Hernanz and Brucart, 1987, Fernández Soriano, 1993, or Gawelko, 1995). It is also argued that in certain sentence types, subjects can appear postverbally, as in the case of relative sentences, in which the relative pronoun is the element that initiates the sentence (for example, el reloj que me regaló mi padre [lit. the watch that me-gifted my father] ‘the watch my father gave me as a present’). Subjects also appear obligatorily postverbally in absolute (that is nonfinite) constructions, as in al llegar el médico … [lit. upon arriving the doctor] ‘when the doctor arrived …’, llegado el médico … [lit. arrived the doctor] ‘once the doctor had arrived …’. Moreover, some descriptive grammars of Spanish posit rules such as those in Butt and Benjamin (1988: 519) given below: ●

●

● ●

Spanish does not separate prepositions from the noun or pronoun that they modify … In general, only no should separate a preposition from its infinitive. Set phrases, particularly set verbal phrases like tener que ‘to have to’ should not be broken up by the insertion of other words. As a rule, words should not be inserted between haber and a participle. Unstressed object pronouns (me, te, se, la, lo, le, nos, os, los, las, les) are never separated from their verb.

Formal factors such as those just mentioned show minimal or no variation and thus will not be examined here. Instead, we will focus our analysis on the most important semantic and discourse-pragmatic factors that affect the word order in Spanish. Since the topic is rather complex and related to many different issues, we will focus mainly on two issues: the so-called basic sentence,3 that is, the mono-transitive sentence and the intransitive monoargumental structure, and the most relevant instances of basic word order alteration.4 In the next section we discuss the most relevant semantic factors affecting word order in the sentence, that is, agentivity and definiteness of the subject,5 and we deal in the third section with the most important discoursepragmatic factors affecting word order in Spanish. Starting with the most generalizad notion of topic (that which is spoken about, that which a sentence is about), we point out the link between the topic and other discoursepragmatic notions, such as cohesion, emphasis and the distinction between old and new information. We hasten to note, however, that the link between these notions and topic does not assume identification between the notion of topic and the others. Moreover, we shall attempt to illustrate the clearest

Belén López Meirama 9

types of a marked topic in Spanish, focusing above all on the fronting of the direct object (DO). We see our contribution to the ongoing discussion of word order in Spanish to be twofold. We present arguments in favour of considering the distinction between unergative and unaccusative to refer to construction types and not to verbs. That is, we argue that it is more reasonable to speak of verbs that prefer an unergative or unaccusative construction rather than to speak of unergative or unaccusative verbs. Second, we argue that it is possible to define the notions of theme, rheme, topic, and focus in terms of tendencies and patterns instead of hardfast rules, and illustrate our arguments with data gleaned from Arthus. Abbreviation that we have used for original texts cited are listed at the end of the chapter.

Semantic factors Linguists have approached the subject of word order by analysing certain semantic factors affecting the distribution of the elements in a sentence; factors such as polarity, modality, aspect, and so on. For example, Knauer (1989) discusses how polarity can affect word order in a sentence, arguing the Spanish double negation (NEG) construction favours a postverbal subject (for example todavía no ha venido nadie [lit. yet NEG has come no one] ‘nobody has come yet’; Hernanz and Brucart (1987) argue that subject inversion is practically obligatory in a marked modality such as interrogative and imperative types. In this chapter, we will focus on two semantic factors which have generated most interest to date in the bibliography about word order in Spanish: the agentivity and the definiteness of the subject. In the generative framework, the semantic verb type plays an important role in characterizing word order. For example, intransitive verbs, that is, monadic verbs, have been divided into two semantic classes: the unergative class, which are constructed with a preverbal agentive subject; and the unaccusative class, constructed with a postverbal, non-agentive subject. Although the Unaccusative Hypothesis is not without its drawbacks (see below), it has pointed out an indisputable fact about Spanish grammar: the relevance of the semantic role of the subject with regard to its position within a clause. In general, a semantically less prototypical subject of a transitive clause tends to be in postverbal position. In this regard, we will examine a one-argument construction with VS order, and two two-argument constructions: indirect object-verb-subject (IVS) and adverbial adjunct-verbsubject (AVS).6 In addition to agentivity and animacy, linguists have also pointed out that definiteness is a very important factor affecting the likelihood of an element occupying a topical position. As we will see later in this section, definiteness transcends the purely semantic sphere, and thus can be best characterized as semantic-pragmatic (we choose, however, not to discuss definiteness because

10 Spanish Word Order

it would break the internal coherence of the section, which deals with the most representative cases of marked order in Spanish). Agency in a subject Unaccusative constructions Some generative studies on word order in Spanish claim that unaccusativity is a factor that triggers a postverbal subject in one-argument clauses. In other words, the subjects of the so-called ‘unaccusative verbs’, when receiving the -role ‘patient’, are generated in D-Structure in the position that corresponds to objects (see de Miguel Aparicio, 1989, and Fernández Soriano, 1993). Therefore, this hypothesis postulates that unergative intransitive verbs are generated with a preverbal subject, and that the unaccusative intransitive verbs are generated with a postverbal subject, as shown in the following examples from Arthus:7 (2.1)

(2.2)

a. Mil quinientos hombres avanzan con el agua hasta la cintura, su fusil recargado en los antebrazos. ‘One thousand five hundred men advanced waist-deep in water, their rifles resting on the forearms.’ (DIEGO: 91, 18) b. Los dos equipos jugarán el sábado, a partir de las diez, en Vigo. ‘Both teams will play Saturday at ten in Vigo.’ (1VOZ: 47) a. Cuando se habla del carácter de los norteamericanos, casi siempre aparece la palabra ingenuidad. ‘When people talk about the nature of North Americans, the word ‘naivety’ surfaces almost always.’ (TIEMPO: 49, 38) b. ¡Es estúpido! Cada día, todos los días del año, muere gente. ‘That’s stupid! Every day, all year round, people die.’ (MIRADA: 119, 19) c. Pero últimamente las cosas se han torcido para ti. Ha surgido otro candidato, un rival. ‘But as of late things have been going bad for you. Another candidate has emerged, a rival.’ (HOMBRE: 34, 30)

The verbs avanzar ‘to advance’ and jugar ‘to play’ in (2.1) are unergative. That is, they are accompanied by an agent subject that thus appears in preverbal position, whereas the verbs aparecer ‘surface’, morir ‘die’, and surgir in (2.2) are considered unaccusative because they possess a patient subject that appears in postverbal position. Two additional subgroups within the group of unaccusative verbs have revealed themselves to be important: verbs of movement/direction (e.g.: arribar ‘arrive’, ascender ‘ascend’, bajar ‘go down, caer ‘fall’, descender ‘descend’, entrar ‘go in’, llegar ‘arrive’, partir ‘depart’, salir ‘leave’, subir ‘go up’, venir ‘come’, and so on) and existential verbs (e.g., abundar ‘be abundant,

Belén López Meirama 11

teem’, acaecer ‘happen’, durar ‘last’, escasear ‘be scarce’, figurar ‘appear’, menudear ‘happen frequently, ocurrir ‘take place’, quedar ‘remain’, subsistir ‘subsist’, and so forth).8 However, Spanish still lacks a complete list of unergative verbs organized in semantic classes such as the one provided for English by Levin and Rappaport Hovav (1995). Nevertheless, research done on word order includes a sufficient number of verbs as to allow the Unaccusativity Hypothesis to be tested on a copious corpus of present-day Spanish. While such a study is beyond the scope of this chapter, a cursory examination of data provided by Arthus still provides us with highly illustrative examples. For instance, comparing the uses of the verbs avanzar ‘advance’, jugar ‘play’, aparecer ‘appear’ and morir ‘die’ in (2.3) and (2.4) with the ones discussed in (2.1) and (2.2), we note not only that clauses containing avanzar and jugar can be constructed with a postverbal subject (cf. 2.3), but that clauses with aparecer y morir are also found with preverbal subjects, as in (2.4). (2.3)

(2.4)

a. Cuando las piernas me llevaron a su vera se abrió el semáforo y avanzó el tráfico rodado. ‘When my legs led me to her, the trafficlight turned green, and the traffic moved on.’ (LABERINTO: 38, 23) b. Pero cuando después hace la interpretación de esos datos, entra a jugar su sagacidad. ‘But when he interprets those data, his cleverness begins to play a part (in the situation).’ (BAIRES: 53, 4) a. Todas las mujeres de la casa aparecieron de repente. ‘All the women of the house suddenly showed up.’ (SUR: 15, 9) b. El viudo de Xius murió dos años después. ‘Xius’s widower died two years later.’ (CRÓNICA: 41, 28)

Even more revealing than these examples are those instances regarding verbs of movement provided by the corpus, with which one can verify that the majority of such verbs can be equally constructed with an agentive subject (animate entity that volitionally controls an activity definable as movement), as in the (a) examples in (2.5)–(2.8), or with a nonagentive subject (a usually inanimate entity that is introduced in discourse by means of a presentative verb), as in the corresponding (b) examples.9 (2.5)

a. El viejo asoma con gesto contrariado que se torna jubiloso al ver al joven. ‘The old man leans out with an expression that turns jubilant when he sees the youth.’ (SONRISA: 253, 9) b. Vuelve atrás: estaban hablando, hablan, no importa qué, hablan. Asoma la sonrisa horrible.

12 Spanish Word Order

(2.6)

a.

b.

(2.7)

a.

b.

(2.8)

a.

b.

‘Come back: they were talking, they talk, it doesn’t matter what they talk about, they talk. That horrible smile turns up.’ (MIRADA: 33, 17) El hijo sale y vuelve pronto con dos hermosas peras y un cuchillo, sobre un plato que deja en la mesilla. ‘The son leaves and comes back with two gorgeous pears and a knife on a plate that he sets on the table.’ (SONRISA: 26, 13) Abrí los grifos y salió el agua tibia y parduzca. ‘I opened the faucets and the water flowed out tepid and brownish.’ (LABERINTO: 92, 23) Tía Delia vino enseguida, y entre ella y mamá me obligaron a seguir viviendo. ‘Aunt Delia came at once, and between her and mom they forced me to continue to live.’ (SUR: 39, 13) … tenía yo treintaiún años. Entonces, pues, empezaron a venir los hijos. Tengo cinco hijos. ‘I was thirty-one. Then the children started coming. I have five children.’ (SEVILLA: 110, 3) Sin saber cómo, el hijo de David había vuelto; estaba otra vez sentado frente a él. ‘Without knowing how, David’s son had come back; he was again sitting in front of him.’ ( JÓVENES: 177, 12) El lago, la lluvia, el ángel … Volvían las imágenes que despertaban sus recuerdos. ‘The lake, the rain, the angel … The images that awakened his/her memories were coming back.’ (CARTA: 137, 14)

Moreover, it is worth noting that the majority of verbs are polysemic, having commonly several meanings, with one or more involving movement and others involving existential, presentative, or stative readings. Consider the following examples, which involve different meanings of the same verb commonly found in monolingual Spanish dictionaries. (2.9)

(2.10)

a. ¡David bailaba delante del Tabernáculo! (meaning ‘dance’) ‘David was dancing in front of the Tabernacle!’ (OCHENTA: 88, 24) b. Por las páginas del libro bailaba la luz del sur. (meaning ‘oscillate’) ‘The sunlight was dancing across the pages of the book.’ ( JÓVENES: 57, 34) a. Los chuchos corren de un extremo a otro. (meaning: ‘move fast’) ‘The mutts run from one place to another.’ (PAISAJES: 57, 12) b. Corre el año 1850 en París y está anocheciendo. (meaning ‘go by’) ‘It was the year 1850 in Paris, and it was getting dark.’ (ZORRA: 7, 6)

Belén López Meirama 13

(2.11)

a. Tu vecina pasa. No sé si me habrá visto. (meaning: ‘go by/ through a place’) ‘Your neighbor is passing by. I don’t know if she saw me.’ (DIEGO: 87, 19) b. Por eso tienes los nervios así, destrozados. Pero bueno, ya pasó todo. (meaning ‘cease, end’) ‘That’s why your nerves are on edge. But everything is over now.’ (SUR: 88, 24)

However, the fact that many verbs may display both constructional alternatives is not the only objection to the approach we are examining. More serious is the observation that the Unaccusative Hypothesis has been applied to the study of word order in excessively reductionist terms. That is, the Unaccusative Hypothesis leads us to believe that an argument accompanying an intransitive verb can only be semantically equivalent to either the subject of the transitive clause (i.e., the verb is unergative) or the direct object (i.e., the verb is unaccusative). Although a prototypical transitive clause is a case in which agent and patient are most clearly defined and differentiated, an intransitive clause offers a whole gamut of possibilities with regard to semantic functions served by its only argument. The possibilities range from the most prototypical patient (found in the traditionally called ‘reflexive passives’) to the most prototypical agent found in active sentences. The following examples, which only represent a sampling, are organized in two series: pronominal sentences, shown in (2.12), and active sentences (2.13), in which the (b) example presents a subject whose degree of agentivity lies between that of (a) and (c) examples. (2.12)

(2.13)

a. Arruti se despidió y se fue. ‘Arruti said goodbye and left.’ (HISTORIAS: 89, 28) b. Al fondo del zaguán se abre la puerta del ascensor y … ‘At the end of the hallway, the door of the elevator opens and …’ (SONRISA: 151, 13) c. Se editarán mil copias del dibujo realizado por la niña de seis años. ‘A thousand copies of the six-year-old girl’s drawing will be published.’ (3VOZ: 32) a. La enfermera sonrió buscando mi complicidad. ‘The nurse smiled, looking for my complicity.’ (DIEGO: 145, 10) b. Pero gusta más el carácter nuestro, parece que estamos más en familia. ‘But our character is more pleasing, it seems that we are among friends.’ (SEVILLA: 291, 10) c. Comieron, y cuando entraron en la plaza ya había mesas ocupadas en las terrazas. Abundaban los hombres.

14 Spanish Word Order

‘They ate, and when they entered the square there were already full tables on the terraces. There were many more men.’ (JÓVENES: 174, 34) This variability is related to an issue addressed in Hopper and Thompson’s (1980) seminal work, namely, that the Transitivity of a sentence involves its semantic structure, which in turn is determined by sets of different factors. It turns out that all of the factors are properties relevant to the study of word order in Spanish. In this regard, it is essential to underscore that such properties are associated with tendencies, that is, they do not set rules. Based on the discussion and the observations so far, the so-called Unaccusative Hypothesis can be refined as follows: ●

●

●

With respect to word order, it is more appropriate to talk about unaccusative and unergative constructions, rather than unaccusative and unergative verbs. In addition to agentivity, there are other semantic features that influence the pre- or postverbal placement of the arguments: a study on word order in Spanish should also include at least definiteness, which we will discuss later in the chapter. Semantic features never set categorical rules of distribution of the arguments. They do, however, mark tendencies in a perceptible and measurable way.

Two-argument constructions with postverbal subject Several researchers have proposed to extend the Unaccusative Hypothesis to some two-argument sentences, such as those with a subject-indirect object configuration with an experiencer subject (for example, gustar ‘like’), and those assuming a location whose nucleus is normally a verb of movement (for example, entrar ‘go in’). In both cases, we have a so-called internal subject in postverbal position. Examples are given in (2.14): (2.14)

a. A Miguel le gusta el cebiche. (Indirect object–Verb–Subject) ‘Miguel likes ceviche.’ b. Por esa ventana entra el viento. (Adverbial adjunct–verb–subject) ‘Through that window the wind comes in.’

With regard to Spanish grammar in particular, Hatcher’s (1956) work is noteworthy is that it has been cited as an example of semantic classification of unaccusative verbs in Spanish. In it we find the classes that allow, and even trigger, the presence of a locative element,10 as shown in the examples in (2.15), taken from Hatcher (1956). The example in (2.15a) corresponds to

Belén López Meirama 15

a verb that belongs to the come-class. The verb in sentence (2.15b) belongs to the existence-presence-class, while the verb in (2.15c) is from the production-class. (2.15)

a. A mis ojos acudían las lágrimas como temerosas. Tears would well up in my eyes, as if they were tremulous.’ b. Muy cerca de aquí está el palacio. ‘The palace is very near here.’ c. Y en sus ojos se cristalizaba un recuerdo inefable. ‘And in his/her eyes an ineffable memory was crystallized.’

Secondly, the discussions by Morales de Walters (1982) and Delbecque (1991) address the question of verb semantics and subject position, and include mention of psych verbs. The following examples have been taken from Morales de Walters and correspond to the class labelled by this author as ‘verbs of psychological reaction’. (2.16)

a. Me fascina la cocina francesa. me fascinates the cuisine French ‘I love French cuisine.’ b. Nos gusta ver los logros de nuestro trabajo. us likes to see the fruits of our work ‘We like to see the fruits of our work.’

Examples (2.14–2.16) reveal the need to review both types of constructionlocative and psych verbs, albeit briefly, in order to account for an unmarked order that does not coincide with the basic word order in Spanish. With respect to locative verbs, the structure containing a locative verb illustrated in (2.14b), termed ‘locative inversion’, is considered by many linguists as a diagnostic of unaccusativity. They posit a series of properties in Spanish with the purpose of demonstrating the equivalence between the subject of this structure and the direct object of the transitive sentence. Such properties, however, fail a detailed analysis, and thus cannot be adduced as evidence (in this respect, see López Meirama, 1997a, and Cifuentes Honrubia, 1999). Against the position that the locative inversion construction is an example of syntactic unaccusativity, Levin and Rappaport Hovav (1995) provide an alternative account for the properties of the unaccusative type that this construction exhibits (that is, the postverbal subject and the preverbal locative or directional element). The key point, in their view, is the discursive function that the construction serves. In other words, the construction is used as a presentational type of focus used to introduce the referent of the postverbal NP. Levin and Rappaport Hovav, as well as Birner (1992, 1994), argue that the general function of all inversions, including the locative one, is that of using

16 Spanish Word Order

sentence initial-position to link relatively unfamiliar information to already familiar information in the discourse. The information represented by postverbal NPs must always be less familiar – but not necessarily new – than that represented by preverbal NPs. That is, the locative inversion introduces less familiar information within the context of more familiar information. The construction, therefore, is presentational, but in a more general sense than the one we attribute to this term. The verb in this construction must be informationally light and, if an appropriate context is provided, it may belong, at least in principle, to any semantic type of verb (including the unergative types), although the canonical types are represented by those referring to existence (for example existir ‘exist’) and appearance (for example aparecer ‘appear [in the punctual sense]’) because these do not add information to that conveyed by the preverbal NP. The following examples illustrate the difference brought about by the positional change at the discourse level between a subject and a locative element. This difference is clear with existence and appearance verbs as in examples (2.17, 2.18) and movement verbs as in examples (2.19, 2.20): (2.17)

(2.18)

(2.19)

a. Allí donde mirase descubría señales de su existencia. Él habitaba ya de alguna manera entre nosotros. ‘Wherever I looked, I would find signs of his existence. Somehow, he was already living among us. (SUR: 100, 4) b. La residencia de Lalín no sólo acoge a los ancianos de la comarca del Deza … Últimamente en la residencia habitan también tres niños. ‘The nursing home in Lalín not only admits the aged from the Deza area. As of late, three children have lived in the nursing home, too.’ (3VOZ: 32, 2) a. Los recién casados aparecieron poco después en el automóvil descubierto, abriéndose paso a duras penas en el tumulto. ‘The newly-weds showed up afterwards in the convertible.’ (CRÓNICA: 48, 24) b. Al llegar a este punto, detrás de nosotros aparecen unas personas que permanecen de pie. Son nuestros ‘ángeles’ y van a hablar en nuestro lugar. ‘At this point, behind us, some people show up who remain standing. They are our “angels”, and are going to talk on our behalf.’ (JÓVENES: 45, 17) a. … el hecho es que Pablo Vicario entró solo en la pocilga a buscar los otros cuchillos. ‘… the fact of the matter is that Pablo Vicario walked into the pig sty by himself to look for the other knives.’ (CRÓNICA: 64, 8)

Belén López Meirama 17

(2.20)

b. La luz no regresaba y por las ventanas mal cerradas entró el silbido del viento. ‘Light was not being restored, and through the badly-shut windows the wind blew in.’ ( JÓVENES: 51, 23) a. Pasa un rato y el niño se despierta. El viejo acude a la alcobita y consigue volverle a dormir. ‘A while goes by, and the child wakes up. The old man goes to the bedroom and manages to put him back to sleep.’ (SONRISA: 158, 8) b. Los niños celíacos de Galicia celebran el domingo su fiesta anual en Santiago . . . A la fiesta acudirán pequeños de las cuatro provincias. ‘The celiac children from Galicia are celebrating their annual party in Santiago on Sunday. Children from the four provinces will go to the party.’ (3VOZ: 28, 3)

It is not difficult to see that in the (b) examples, the locative elements convey to the interlocutor old information or, at least, more familiar or accessible information than the elements appearing as subjects. Thus, the locative elements in these examples occupy a topic position (AVS order), whereas in the (a) examples it is the subject that carries more familiar information and occurs in topic position. As expected, the subject and topic coincide (SVA order) in these examples. Regarding psychological verbs, the structure containing a psychological verb exemplified in (2.14a) has been accounted for in the literature by appealing to the notion of the unaccusative construction. Fernández Soriano (1993: 130), for instance, claims the following: Another type of verbs, closely linked to the unaccusative verbs seems to display properties similar to those of the ‘psychological’ verbs, discussed in Belletti and Rizzi (1987). Succinctly, the hypothesis of these authors is that verbs such as gustar ‘like’, asustar ‘scare’, preocupar ‘worry’, and so on, also have ‘internal’ subjects whose basic postverbal position is normally reserved for the direct objects of transitive verbs accompanied by an agentive subject. The excessive reductionism of the Unaccusative Hypothesis also manifests itself in this type of psychological verb construction. Therefore, the conclusions so for advanced are equally applicable in this case. That is, it is possible to find examples of this construction involving a preverbal subject. Although such instances do not invalidate the Unaccusative Hypothesis entirely, they nevertheless strongly suggest that there are several semantic factors at play regarding constituent order in Spanish. The Arthus corpus provides instances

18 Spanish Word Order

of the most representative verbs of this group: (2.21)

a. ¿Tienes preferencia por algún pintor concreto? Sí, Sorolla me gusta muchísimo. ‘Do you have any preference for any painter in particular? Yes, I like Sorolla a lot.’ (SEVILLA: 279, 30) b. ¿Qué me pasa? Lo de siempre. Mi estómago vuelve a molestarme. ‘What’s the matter? The same old story. My stomach is bothering me again.’ (CARTA: 87, 20) c. Cuando ríes mucho, algo te pasa. ‘When you laugh a lot, something happens to you.’ (CAIMÁN: 21, 24)

A more promising approach, in our view, is that of Vázquez Rozas (1995, and Chapter 4 of this volume), according to which the two-argument construction (subject–indirect object) is considered a non-prototypical two-argument semantic configuration in that it reveals a low degree of Transitivity based on the parameters proposed in Hopper and Thompson (1980) (cf. Chapter 1, this volume). The semantic differences between syntactic patterns of prototypically transitive clauses and clauses containing a gustar-type-verb construction can be outlined as in Table 2.1 (adapted from García-Miguel 1995: 94). Table 2.1 Comparison of semantic differences and syntactic patterns of prototypically transitive clauses sentences v. clauses sentences with a gustar-type verb construction Subject-Predicate-Direct Obj

Subject-Predicate-Indirect Obj

Dynamic Non-involuntary subject Affected (modified) 2nd argument  Animate 2nd argument  Animate 1st argument

Stative Involuntary subject 2nd argument not affected physically Animate 2nd argument  Animate 1st argument

In fact, some verbs have different features depending on the configuration in which they are used. As we can see in the following examples from Arthus, the intransitive meaning, which corresponds to the pattern with indirect object, shows a departure from a prototypical transitive sentence: (2.22)

a. El ridículo personaje de Action Française admira ahora su propia obra con orgullo casi paterno. (meaning: ‘to esteem, to value’) ‘The ridiculous character in Action Française now admires his own work with almost paternal pride.’ (PAISAJES: 78, 29)

Belén López Meirama 19

(2.23)

(2.24)

b. Era Julián el que hablaba, y a David le admiró su valentía para contradecir al muchacho. (meaning: ‘surprise’) ‘It was Julián who was speaking, and David was amazed at Julian’s courage in proving the boy wrong.’ (JÓVENES: 140, 32) a. El conquistador invariablemente aprovecha las rivalidades entre los estados y las divisiones internas. (meaning: ‘take advantage of’) ‘The conqueror invariable takes advantage of the rivalries among states and internal divisions.’ (TIEMPO: 88, 34) b. ¿A quién aprovechaba aquel endemoniado desbarajuste? (meaning: ‘be advantageous’) [lit. to whom was advantageous that darn mess] ‘Who benefited from that darn mess?’ (PAISAJES: 15, 4) a. Aunque parezca extraño, en el norte pasamos un calor terrible. (meaning: ‘suffer’) ‘Although it may seem strange, we suffer high temperatures in the north.’ (MADRID: 116, 7) b. Siempre me pasan cosas malas. (meaning: ‘happen, occur’) ‘Bad things always happen to me.’ (SUR: 62, 3)

These examples display differences in the animacy of the participants of the sentences. In the transitive sentences (the [a] examples), the subject is an animate entity and the object is inanimate. The the intransitive sentences (the [b] examples), it is the opposite. Moreover, in the examples (2.22a) and (2.23a) the subject of the transitive acts volitionally, in (2.22b) and (2.23b) the respective subjects do not exhibit any volition at all. Definiteness of the subject Among the semantic properties associated with the constituents of the sentence that favour the appearance of one as a topic, the feature of definiteness stands out and it has attracted a good deal of attention among many researchers. An indication of this attention is the coining of the notion ‘Definiteness Effect’. Essentially, definiteness is said to impose on the NP subject of a sentence the restriction that it must be definite (Belletti 1987: 173). With regard to the order of constituents in a sentence, the Definiteness Effect means that definite phrases (that is, proper nouns, personal pronouns, phrases headed by definite determiners or possessives and some quantifiers) cannot form part of a presentative structure (in English with expletive there), nor can they be the only argument of a sentence whose verb is unaccusative. In other words, in Belletti’s view, unaccusative verbs are constructed with an indefinite subject generated in object (postverbal) position at D-structure, that is, at deep structure. Applying this analysis to present-day Spanish, it is easy to find counterexamples to this principle in the relatively extensive database used for this study.

20 Spanish Word Order

The following small sample shows that so-called ‘unaccusative verbs’ also appear with definite subjects, preverbally (2.25) as well as postverbally (2.26): (2.25)

(2.26)

a. Le ofrecí mi vida a cambio de tu salvación. Yo moriría antes de cumplir los diez años. ‘I offered him/her my life in exchange for your salvation. I would die before turning ten.’ (SUR: 22, 25) b. La historia ocurrió antes de 1930. ‘The story happened before 1930.’ (HISTORIAS: 101, 10) c. Es el atardecer. El día va a decaer. Tu vecina pasa. ‘It is getting dark. The day is going to fade. Your neighbour walks by.’ (DIEGO: 87, 19) a. ANA. – Estábamos contentos y teníamos miedo. ‘We were happy and scared.’ TEO. – Claro, había muerto Franco. ‘Of course. Franco had died.’ (AYER: 50, 14) b. [Fue] una imposición militar extranjera la que estableció el comunismo en Polonia. Ocurrió lo mismo, con pequeñas variantes, en Checoslovaquia. ‘It was a foreign military imposition which established communism in Poland. The same happened, with small variations, in Czechoslovakia.’ (TIEMPO: 195, 31) c. Ha pasado la medianoche y el tiempo es tormentoso. ‘Midnight has passed, and the weather is stormy.’ (COARTADA: 23, 3)

On the other hand, it is questionable to justify the postverbal position of an indefinite subject on the grounds of verb class, since it would be difficult to account for the instances of indefinite subject NPs with verbs that are considered unergative, such as the examples in (2.27): (2.27)

a. A esa hora asoman una tras otra las estrellas y canta lejos alguien que vuelve del campo. ‘At that time, the stars come out one after the other, and, in the distance, somebody coming back from the countryside sings.’ (SONRISA: 322, 23) b. Ahora estudia mucha gente. ‘Now many people study.’ (MADRID: 151, 19) c. Ladra un perro; ladra agresivamente. ‘A dog is barking. It is barking aggressively.’ (DIEGO: 88, 22)

Belén López Meirama 21

These examples, however, are not direct counterexamples to the Definiteness Effect. Instead, they reveal two fundamental drawbacks that weaken it. The first is that it is presented as a rule but in fact reflects a tendency: in Spanish, definite phrases do not obligatorily precede the verb nor do non-definite ones follow it, although oftentimes this is the case. The second drawback is that the restriction is linked exclusively to verb class, when actually we are dealing with a semantic feature linked to the subject argument. This is why the tendency towards postposing non-definite phrases is detected in both unaccusative and unergative structures, as evidenced in the examples we have seen so far. In fact, at the time when Milsark (1974, 1977) established the notion of the ‘Definiteness Principle’, the so-called ‘thematic hierarchies’ or ‘topicality hierarchies’ began to proliferate.11 These hierarchies make it evident that the tendency for an argument to function as Topic (that is, to occupy a sentenceinitial position) is directly related to certain semantic features of the arguments, such as animacy and definiteness. Such features cannot be reduced to a dichotomy. In other words, the influence of definiteness on the tendencies found in argument pre-or postplacement cannot be accounted by means of a purely formal interpretation based on a definite–indefinite binary opposition. In our view, it is clearly advantageous to adopt a logical tradition to grammatical studies, in which definiteness is interpreted as that which is identifiable. We can, for example, appeal to theoretical usefulness in explaining definiteness within the more general framework of reference. Thus, we can easily justify why a definite referential subject is more likely to occupy a topical position and a non-referential subject more likely to occupy a non-topical position. Givón (1984: 434) states that the reason for this is that human communication fundamentally (or prototypically) deals with real events and individual referential entities. Likewise, humans typically talk about animate and agentive entities that control some action, hence the propensity to organize sentences primarily around real and concrete entities identifiable in discourse by the interlocutor. Moreover, it is specific individuals (rather than types or classes) that carry out actions. This tendency can be visualized through the hierarchy in Table 2.2. Therefore, proper nouns occupy the highest position in the hierarchy since they are inherently definite, which accounts for the high frequency with which they appear in preverbal position. In addition to proper nouns, Table 2.2 (In)definiteness, (non-)referentiality and topichood Definite

Indefinite Topic

Non-Referential

22 Spanish Word Order

personal and demonstrative pronouns and phrases with demonstrative determiners share the highest position in the hierarchy. As the examples in (2.28) illustrate, the NPs occupying the highest position in the hierarchy appear in preverbal position: (2.28)

a. Toniolo murió enseguida, la malaria. ‘Toniolo died very quickly, malaria.’ (SONRISA: 163, 31) b. Inf. A. – Mm. Nosotros hablamos a la mañana. ¿Te acordás? ‘Uhm. We talked in the morning, remember?’ (BAIRES: 42, 1) c. ¿Puedo lavarme las manos? Las llevo manchadas de pintura y esto delata más que nada. ‘Can I wash my hands? I have paint on them, which gives one away more than anything else.’ (HOMBRE: 21, 11) d. Estas heridas sanarán dentro de unos días. ‘These wounds will heal in a few days.’ (ZORRA: 44, 13)

At the other end of the hierarchy, we find determiner-less NPs, or NPs without determiners or modifiers (bare nouns), usually employed to convey the generic sense. These occupy a non-topic position. The Arthus data provide numerous instances of bare NPs of one-argument sentences in postverbal position and some are listed in (2.29) (see also 2.2b above): (2.29)

a. Pasaban con altoparlantes y después empezó a desfilar gente con las banderas. ‘They were passing with bullhorns, and then people carrying flags started to parade.’ (BAIRES: 74, 10) b. Apenas quedaba ya luz en la habitación. ‘There was hardly any light left in the room anymore.’ (CAIMÁN: 74, 13) c. Empieza a soplar viento. Una nueva tormenta parece acercarse. ‘The wind is beginning to blow. A new storm seems to be approaching.’ (2INFAN: 90, 15)

In López Meirama (1997a), we supply and comment in detail on statistical data with reference to one-argument sentences. Summarizing those findings, we found roughly 85 per cent preplacement of subjects that were relatively high on the hierarchy (those with a proper noun, a personal pronoun or a deictic determiner), and around 90 per cent postplacement of subjects relatively low on the hierarchy (noun phrases without determiner). Such findings underscore the argument we have proposed that the placement of the subject relative to the verb is not only a matter of verbal semantics. It is for this reason that we propose speaking of unergative and unaccusative construction types rather than of unergative and unaccusative verbs.

Belén López Meirama 23

This approach also allows the consideration of other significant factors that affect subject order, a key one of which is variable definiteness of the subject.

Discourse-pragmatic factors Introduction Unlike the wider range of interlinguistic variability in the ordering of syntactic elements – typology, for example, has established three fundamental types of basic word orders for transitive sentences: SOV, SVO and VSO – ordering on the discourse-pragmatic level is more homogeneous. Thus, the sequence theme–rheme is the most natural order, that is the least marked order, in the majority of world languages. Some linguists have formalized this fact. For instance, Bossong’s (1980: 51) ‘principe de linéarisation thématorhématique’, or Tomlin’s (1986) ‘Theme First Principle’ would govern basic word order, along with other functional principles. It is doubtless the case that Spanish follows this principle, and that as Reyes (1985: 577–8) points out, the theme–rheme pattern tends to prevail even at the expense of the subject–predicate syntactic pattern. This means that although theme and subject frequently coincide, there are many examples in which the theme of the sentence is a functional element other than a subject. In the previous section we discussed that certain semantic features contribute to word-order flexibility. In this section we examine the extent to which word-order variation in Spanish depends on discourse-pragmatic factors. From a discourse-pragmatic viewpoint, the sentence-initial slot is unquestionably a ‘privileged’ position. We shall propose that this position is used for different discourse-pragmatic purposes, mainly for preserving discursive cohesion and for accommodating given (or presupposed) information. Moreover, the sentence-initial position is used for emphasis, since it is an intonationally prominent position. At the same time, it is also true that the discourse-pragmatic functions linked to sentence-initial position can also be rendered through other means. That is, given that new or given information relates to the structure of the text and to other factors such as ellipsis and anaphora, its status is not only expressed by its relative position within the sentence. Furthermore, intonation plays a significant role in emphasizing expression, thus implying that, theoretically, an element can be emphasized in any position in a sentence. Later in this section we provide an analysis of the most relevant cases of marked positioning of subjects due to emphasis. A good number of studies on word order analyse the distribution of elements as a manifestation of specific discourse-pragmatic functions in such a way that a given position is associated with a particular function at the discourse-pragmatic level. As it relates to sentence-initial position, such an approach is not without its problems. At its best, it renders partial analyses that leave many cases unaccounted for. In our view, it is desirable to develop

24 Spanish Word Order

a theory that allows for the interweaving of all the relevant factors for subject order relative to the verb without blurring them. With particular regard to the notion of theme (and the theme–rheme structure), however, many definitions have been advanced and various semantic and pragmatic properties have been associated with them. To illustrate, we highlight those that we believe are the most common: ‘that which is spoken about, that which the sentence is about’, ‘known or given information’, ‘presupposed information’, ‘element around which the interest of the speaker is centred’, ‘determined or generic element’, ‘element that begins the sentence / preposed to the verb / dislocated to the right or left’. Of all these, the undoubtedly most extended definition is the first one. Hidalgo (2003: 39) notes in this respect that the meaning of aboutness ‘appears in practically all the definitions of the notion [of theme] that are found in the writings and the models that have treated the question’. Hidalgo (2003: 39) finds this important because it ‘reflects the relationship of the term to the pretheoretical and intuitive notion of the term as used in common usage, as “theme or matter which something is about”, such as a book or a movie, that is, a specific fragment of discourse’.12 This will be the concept of theme we will work with in this chapter. In adopting such a general definition as ‘what is spoken about’ or ‘what we speak about’ allows us to associate it with different discourse-pragmatic functions. In general, ‘we’ (in the generic sense) speak of that which is given, presupposed, or inferable information. Discourse will be coherent if what is spoken about allows us to link what will be communicated with what has already been communicated. However, the theme can be sometimes an element that a speaker wants to highlight, and does not have to be necessarily given or inferable information. Later in the chapter we analyse the most relevant cases of marked theme in Spanish. The theme and other discourse-pragmatic functions The presence of different discourse-pragmatic functions associated with syntactic elements has been a constant point of debate in the grammatical tradition surrounding the Spanish language, although it has generally been discussed from an impressionistic viewpoint. In particular, some manuals of grammar provide a definition of the sentence that is not strictly syntactic, whereby the subject–predicate distinction is comparable to the theme–rheme dichotomy (which further implies that subject and theme coincide in unmarked cases). That is, grammarians have long believed that speakers mold the information into fully meaningful units that in turn are divided into two segments. Functional Grammar has overcome the limitations of traditional grammar by separating the syntactic level from the discoursepragmatic (or informational) one (see, for instance, Dik, 1978, and Givón, 1984). With respect to the latter level, we espouse Halliday’s concept of thematic structure (as opposed to syntactic structure), whereby a sentence is divided into theme and rheme.

Belén López Meirama 25

However, the distinction between the syntactic and discourse-pragmatic levels cannot per se clearly establish the concept of theme since, as we have pointed out, the element the sentence is about (which in Spanish occupies the preverbal position) is generally the initial position and associated with different informational factors.13 It would, therefore, be convenient to clearly differentiate these factors from one another. In other words, the thematic or preverbal position is discourse-pragmatically relevant, and speakers use it for several communicative purposes. Old/given information The sentence-initial position is often occupied by given (or old) information, which has led to its being identified with the theme, a view that attained great prestige especially in the 1970s, thanks to the work by the Prague School, and the so-called ‘functional sentence perspective’. Among the authors that have applied this perspective to the word order in Spanish are Contreras (1976), Silva-Corvalán (1977), Delbecque (1979), Morales de Walters (1982) and de Miguel (1989). Identifying given or old information with the theme has a serious drawback: the distinction between old and new information does not constitute a dichotomy but rather a continuum, whereas the theme–rheme structure is binary, since the placing of an element in a sentence can only be preverbal or postverbal. Silva-Corvalán (1984: 4), for instance, claims that ‘even though it is possible that there is no old/new information continuum, these values are surely relative’. That is, the information can be more or less old or new, and word order ‘responds to the relative values of newness of the information’ (1984: 4). Along similar lines, Mendieta and Medina (1997: 459) analyse the notion of ‘pragmatic function’ among the variables used in their analysis of preverbal objects. The authors argue for the existence of four different functions, which are defined in terms of their degree of newness, and which are ordered along a four-point continuum: (a) the relation function; (b) lexical cohesion by repetition; (c) lexical cohesion by association; and (d) rhematic anteposition. In the first three cases, a preverbal object introduces given information (although in [c] the information is old for been deduced from the text or the situation), whereas in the fourth case (d) (rhematic anteposition) an object introduces new information. In fact, there are many examples in which the identification between theme and given information is untenable as shown in the following examples (the first one appears in Mendieta and Medina [1997: 461]) (for this section, we have underlined the item that occupies a thematic position): (2.30)

a. … pero no les aconsejo que vayan allí, porque las chinches las sacan a peso. ‘… but I advise you not to go there, because bedbugs they take out by the truckload.’ (México, p. 100)

26 Spanish Word Order

b. ¡Madonna, déjame vivir un mes más que el Cantanotte, por favour! ¡Un cirio te llevaré, el más gordo que encuentre! ‘Madonna, let me live one month more in the Cantanotte, please! A candle I will bring you, the fattest one I can find!’ (SONRISA: 98, 4) c. NÉSTOR – Espera. Yo querría saber, Rufina, si Charito puede seguir yendo a los ensayos. ‘Wait. I wanted to know, Rufina, if Carlos may continue going to rehearsal.’ RUFINA – Que haga lo que quiera. Remedio ya no tiene. ‘He can do whatever he wants. It’s hopeless now.’ (CAIMÁN: 78, 5) Examples (2.30b) and (2.30c) support the hypothesis put forward by Mendieta and Medina (1997: 461) that preverbal objects introducing new information tend to be associated with the feature [-definite]. This hypothesis is by no means original. The association between the feature [definite] of an argument and its capacity to express old/new information has already been addressed in studies on word order since thematic hierarchies began to spring up. In these hierarchies, however, the relationship is extended to the theme–rheme dichotomy, and can be represented as in Table 2.3. Table 2.3 The theme–rheme dichotomy Theme

Rheme

Old information [definite]

New information [definite]

Nevertheless, the possibility that a non-definite element may function as theme in Spanish has led some researchers to reject the conceptualization of theme as ‘given information’ and replace it with others. For instance, Reyes (1985) argues that the notion of given information is rather imprecise and that it would be clearer to use the notion of presupposed information,14 which, in her view, encompasses all the degrees of mutual knowledge between speaker and hearer, and allows one to integrate the representation of unknown entities or individuals into the concept of theme. Zubizarreta (1999: 4,219), in turn, defines theme as ‘what the sentence is about’, and claims the following: It should be noted that an indefinite phrase … can also function as theme … We know that the indefinite nominal phrase (unlike the definite one) introduces new information. This proves that the dichotomy of new vs. old information is not appropriate to define the theme of a sentence.

Belén López Meirama 27

Moreover, the new/old dichotomy is not reflected only in word order, since it is related to other grammatical aspects of languages, such as ellipsis and anaphora. In light of these considerations, we conclude that, even though the element spoken about (the theme of the sentence) constitutes frequently known information, one should not make the exclusive identification between theme given/known information. As Hidalgo (2003: 60) notes, ‘the informational status of discourse referents [their nature as given or new] is a discourse notion, clearly differentiated from the function of theme, understood in the sense of “aboutness”, because of which one cannot identify one aspect with the other in the organization of discourse’. Discourse cohesion Closely related to the discussion so far is the notion of theme as a cohesive element that is necessary for the text (understood as an informational unit) to be constructed coherently, as pointed out by Silva-Corvalán’: The progression from the known to the unknown information creates cohesion in discourse. Therefore, when C (the complement) represents the known information in a sentence, whether because its referent has been evoked or inferred, it is placed in initial position, where it serves a textual-linking function (translation by the editors). (1984: 6) The claim that sentence-initial position is where cohesive elements are usually placed is bolstered by the fact that connectors, whether they are syntactic or discursive, normally occupy this position, at least in Spanish, adding a sense of connection, continuity, and, therefore, coherence to the text, as we can see in the following examples: (2.31)

a. No sé si ayer estabas en la reunión. ‘I don’t know if yesterday you were in the meeting.’ b. Escuché muchas veces aquella melodía que tanto habíamos amado. ‘Many times I listened to that melody that we had loved so much.’

With respect to the notion of theme and thematic structure, there is no identification between these concepts and that of the term ‘cohesive element’, although the initial position is frequently used for this function. This can be demonstrated by the positional tendencies of adjuncts. Consider, for instance, the sentences in (2.32), in which the theme repeats what has been mentioned immediately before: (2.32)

a. Había dos asignaturas que eran una rémora en la carrera, era el Latín, por una parte, y el Arte. El Arte no había más remedio que aprobarlo.

28 Spanish Word Order

‘There were two courses that were an obstacle in our degree: Latin, on one hand, and Art, on the other hand. Art, you just had to pass it.’ (SEVILLA: 202, 39) b. Mi padre no era de aquí, pero vivió casi toda su vida en Sevilla. ‘My father was not from here, but lived almost all his life in Seville.’ ¿Y tu madre? ‘And your mother.’ Mi madre creo que nació aquí. ‘My mother, I think she was born here.’ (SEVILLA: 275, 13) We also find cases, given in (2.33), in which the theme is syntactically disconnected with the sentence it begins, but where the cohesiveness of the sentence containing the elements is unquestionable: (2.33)

Inf B. – ¡Yo tengo unos líos!, que ya no sé ni medir los versos, fíjate. ‘I am so confused that I don’t even know anymore how to measure the verses, imagine!’ Inf. A. – Eso te coges cualquier libro, cualquier manual. ‘[For] that, you get yourself any book, any manual.’ (MADRID: 410, 5)

Therefore, there is an undeniably close relationship between sentence-initial position and discursive coherence, but we repeat that such a link cannot lead us to an absolute link between theme and a cohesive element. Cohesion can also be achieved through means other than word order, such as intonation or the use of deictic elements, particularly anaphoric elements. Emphasis The discursive importance of the sentence-initial position also surfaces when a speaker employs it to emphasize a particular element. Jiménez Juliá (2000) claims that the theme has a potentially contrastive function, since it implies a choice of one over another, and thus it is ‘very likely to carry the intonational focus at a given moment, according to M.A.K. Halliday’s description’ (2000: 156, translated from the original). Halliday defines focus as the ‘point of prominence within the message’ (1967: 203); that is, the element that receives the most intonational emphasis as a result of heightened informational interest on the part of the speaker. Halliday claims that a speaker presents the focus as a non-recoverable element from the preceding discourse, that is, it is considered new information. In any case, it is indisputable that theme and focus are not synonymous. We merely point out that an emphasized element very frequently occupies

Belén López Meirama 29

the rheme (that is, postverbal) position, as illustrated in the following examples (the focus appears in small capitals): (2.34)

a. Y aún pienso que podré resistir mucho más, ir más allá de donde pueda ir cualquier mujer, acercarme a las fieras, a su orgullosa autosuficiencia. Quiero abastecerme YO MISMA de libertad y de aventura. ‘And I still think that I will not be able to hang on much longer, go beyond where any woman can go, approach the beasts, their prideful self-sufficiency. I want to stock up on freedom and adventure MYSELF.’ (ZORRA: 30, 1) b. Y la acompañé, o me acompañó ELLA a mí, en la noche sin tiempo. ‘And I accompanied her, or SHE accompanied me into the timeless night.’ (CAIMÁN: 71, 15)

However, we must reiterate that the functions of theme and focus are not mutually exclusive. Since two features that allow us to identify the focus are on one hand, a marked (that is, unexpected) position within a sentence, and on the other hand, intonational emphasis, it seems clear that a marked theme can become the focus of the predication. Later in this section we will discuss specifically the marked theme. For now, we focus on the most relevant cases of a marked theme in Spanish. The thematic structure of Spanish As we have already pointed out, we take the thematic structure to be a binary mechanism of distributing information through which a theme is presented and something about it, that is, the rheme, is predicated. This informational structure is superimposed, as it were, onto the semantic and syntactic structure in the sense that the elements in the informational structure also make up these semantic and syntactic structures. However, these three structures (syntactic, semantic and informational) do not exist independently of one another, but rather are interconnected in all languages. Apart from the semantic tier, a sentence is thus the result of the superimposition of two structures: the semantic and the informational. In a nominative-accusative language such as Spanish, it is not uncommon that both theme and subject coincide, as illustrated in Table 2.4.

Table 2.4 The overlay between thematic and syntactic structure Thematic structure Syntactic structure

Theme SUBJ

PRED

Rheme DO

IO

30 Spanish Word Order

Due to this, researchers usually regard a prototypical sentence as one in which the subject is the element spoken about. Thus, in grammars (for example in that of the Academia, 1973), we find examples such as El niño duerme en la cuna, ‘The child sleeps in the crib’, or Juan dio una limosna a nuestro vecino, ‘Juan gave some money to our neighbour’, in which El niño and Juan are both subject and theme. However, it is debatable how respresentative such sentences are in Spanish, given that it is more frequent to find other structures, especially in oral texts, such as aquí duermo ‘here I sleep’, le pegó una paliza ‘s/he beat him/her up’, or Por asuntos comerciales viajaba mucho mi hermano mayor entre España y América ‘For commercial reasons, my older brother would travel a lot between Spain and America’ (in Academia, 1973). In this regard, we make the following two observations. First, not all sentences in Spanish are thematic, nor do they respond to the theme–rheme structure. The reason for this is that the morphological system of Spanish is rich enough to be able to do without obligatory full NPs or overt subject pronouns once they have been introduced into discourse. This is frequently observed in the case of the subject, which is the first candidate to occupy a thematic position, as pointed out earlier. In fact, the majority of subjects in Spanish are ‘morphological’.15 Consider the following fragment from Crónica de una muerte anunciada [Chronicle of a Death Foretold ], by Gabriel García Márquez: Bayardo San Román, el hombre que devolvió a la esposa, había venido por primera vez en agosto del año anterior: seis meses antes de la boda. Llegó en el buque semanal con unas alforjas guarnecidas de plata que hacían juego con las hebillas de la correa y las argollas de los botines. Andaba por los treinta años, pero muy bien escondidos, pues tenía una cintura angosta de novillero, los ojos dorados, y la piel cocinada a fuego por el salitre. Llegó con una chaqueta corta y un pantalón muy estrecho, ambos de becerro natural, y unos guantes de cabritilla del mismo color. Magdalena Oliver había venido con él en el buque y no pudo quitarle la vista de encima durante el viaje. ‘Bayardo San Román, the man who had given the wife back, had come for the first time in August of the previous year: six months before the wedding. He came on the weekly boat with silver-adorned saddlebags that matched the buckles on his belt and the rings on his ankle boots. He was around 30 years old, but concealed his age very well, because he had the narrow waist of a bullfighter, golden eyes, and skin broiled by saltpeter. He came wearing a short jacket and tight-fitting pants, both made of authentic calfskin, and a pair of kidskin gloves of the same color. Magdalena Oliver had come with him on the boat, and had not been able to take her eyes off him during the voyage.’ In the text, there are seven predicates, marked in boldface, which function as nuclei of their respective sentences. Among these sentences, only the first

Belén López Meirama 31

and the penultimate ones satisfy the characterization according to which the subject and the theme are the same argument as unmarked cases. The other five sentences illustrate the abundance of athematic structures in Spanish: note that once a topic has been introduced into discourse, it is not normally expressed again by an overt NP or pronoun for several sentences (cf. Comajoan, Chapter 3 of this volume, for details). Second, Table 2.4 represents the unmarked theme, which as expected coincides with the subject. However, as just noted, the morphological richness of Spanish may be what underlies the separation of Spanish syntactic and thematic structure: the position of an NP within the sentence is not an identifying feature of the subject, except in extreme cases, such as modernización significa abolición del pasado / abolición del pasado significa modernización ‘modernization means abolition of the past / abolition of the past means modernization’. Thus, the theme is not necessarily the subject and often may be the DO or IOs, as in (2.35): (2.35)

lo he visto en el supermercado. a. A Juan OBJ Juan him.ACC have.1s seen in the supermarket. ‘Juan I saw in the supermarket.’ b. A mi gata le pica la herida. OBJ my cat her.DAT itches the wound ‘My cat’s wound itches.’

Having made these preliminary remarks, let us now discuss the issue of themes coinciding with elements of a sentence other than the subject. Given that the distributional possibilities of constituents in Spanish are many, we will limit ourselves to discussing the most representative cases. More specifically, we will focus on what might be considered a prototypical sentence (that is, independent, active, declarative, affirmative) and examine cases of thematization of the DO. In general, when the literature refers to thematized DOs, it provides examples such as (2.35a). In fact, an examination of a large corpus of presentday Spanish reveals that this example represents the most common type of DO thematization, as illustrated in examples in (2.36), extracted from Arthus: (2.36)

lo puso Arturo, naturalmente. a. El dinero the money it.ACC put Arturo of course ‘The money Arturo put it in, of course.’ (AYER: 67, 18) b. Este muelle lo construyó mi padre antes de que yo naciera. this dock it.ACC constructed my father before that I was.born ‘This dock my father built it before I was born.’ (CINTA: 32, 12) c. Al parque lo habían invadido las grandes hierbas locas. the park it.ACC had invaded the big grasses crazy ‘The park, the big crazy grasses had invaded it.’ (DIEGO: 146, 15)

32 Spanish Word Order

This type of construction exhibits two basic features that are normally underscored by researchers who have analysed it: first, the existence of pronominal reduplication (Suñer, 1988, and García-Miguel, 1991, among others, speak of an agreement marker); second, a postverbal subject. We might thus claim that theoretically a monotransitive sentence presents two possible distributions of subject and object: SVO (unmarked), and OVS (marked). Nevertheless, the following observations are in order. First, note that most instances of preverbal object are accompanied by non-overt subject. That is, the subject is exclusively expressed by morphological means, since it is usually considered known information. The examples (2.37) corroborate this point: (2.37)

a. ¿Te gustan los toros? A mí no. No, a mí me cansan. ‘Do you like bullfights? Not me. No, they tire me.’ (SEVILLA: 30, 30) b. Esa parte de la historia la conozco mejor. ‘That part of the story I know it better.’ (PASAJERO: 63, 2) c. El «Ya» lo leíamos en casa los domingos. ‘«Ya» we used to read it at home on Sundays.’ (MADRID: 82, 17)

Hatcher (1956) employs the abbreviation ‘O lo V’ to identify this construction. Given that overt subject deletion also occurs when the DO occupies a rhematic position, we should refine our nomenclature for accuracy: SVO/VO is an unmarked order in a transitive sentence, whereas OVS/OV (or, in Hatcher’s terms, O–lo–VS / O–lo–V) is a marked sequence, resulting from DO thematization. Secondly, with a lexically realized subject, DO thematization often takes place alongside a postverbal subject, but not in all cases, which means that the thematic position may be occupied by both arguments. Thus, in Spanish there are other possibilities of constituent distribution apart from the ones discussed earlier. In the examples in (2.38), the two elements that make up what we call the multiple theme are numbered with a subscript:16 (2.38)

(2.39)

OSV a. El artículo de Fillmore1 yo2 lo podría haber elaborado mucho mejor. ‘Fillmore1’s article2 I could have developed it better.’ (BAIRES: 36, 38) b. Pero a a Emilia1 esta idea2 no acababa de hacerla feliz. ‘But Emilia1 this idea2 didn’t make her totally happy.’ (LABERINTO: 148, 21) SOV a. ¿El árabe? Yo1 el árabe2 lo aprobé con éste, con P.

Belén López Meirama 33

‘Arabic? Well, Arabic1 I2 passed it with this guy, with P.’ (MADRID: 418, 3) b. Una buena alfombra1, como una buena capa, todo2 lo tapa. ‘A good carpet1, like a good cloak, covers everything2.’ (HOTEL: 31, 34) There are even cases in which a speaker changes the relative order of both thematic elements, as shown in the example in (2.40), where the speaker wavers when constructing the subject: (2.40)

Pues ha cambiado todo, porque a mí estos … , esta construcción de ladrillos1 a mí2 me enferma. ‘Well, everything has changed, because me, this … , this brick construction1 makes me2 sick. (MADRID: 269, 31)

On the other hand, some studies on themes in Spanish argue that the theme can be in external position of the sentence, which has led researchers to establish different types of themes, and even to distinguish theme from another discourse-pragmatic function. The first approach is found in Zubizarreta (1999: 4,218–24), who, following Cinque (1983), claims that the ‘theme of the sentence’ becomes grammaticalized through two constructions: the Hanging Topic construction and Left-Dislocation. The second approach is epitomized by López García (1996), who distinguishes theme from topic and claims that ‘the theme merely represents a modification of the gradual order of informational succession of the elements in the sentence’, whereas ‘the topic entails a privative view based on configurationality’ (1996: 486 translated from the original). The analysis of a large corpus of real texts in modern Spanish reveals that the dissociation of the theme from a syntactic structure belongs almost exclusively to the realm of colloquial Spanish (since it is practically inexistent in narratives), and it appears to varying degrees. What follows is a synopsis of the most relevant types of dissociation found in our corpus. The theme can be separated from a sentence by means of a pause,17 although our data reveal that, unless the theme is used in a spoken language (or in texts that imitate it, such as dialogues in novels and plays), a pause is primarily used for prosodic reasons when the theme is long, as shown in (2.41): (2.41)

a. A un chico a quien se le escapara un chiste atrevido o un taco delante de una señorita, se le catalogaba inmediatamente como un grosero. ‘Any boy who would let an off-color joke or a swear word slip out before a young lady one immediately would brand him a rude person.’ (USOS: 194, 10)

34 Spanish Word Order

b. Lo que hizo la Reforma protestante en la esfera de las creencias y los sentimientos religiosos, lo ha hecho la Unión Americana en la esfera secular. ‘What the Protestant Reform did in the realm of religious beliefs and feelings, the American Union has achieved in the secular realm.’ (TIEMPO: 37, 9) There is sometimes a discrepancy between the element in thematic position and the coreferential clitic pronoun, as in (2.42): (2.42)

a. Encuestador: ¿Pero no crees que para esto de la medicina hay que tener mucha sangre fría también? Researcher: ‘But don’t you think that in this medicine business one has to be a coolheaded person, too?’ Inf. – Sí; pero eso la adquieres a lo largo del tiempo. ‘Yes, but that you acquire over time.’ (MADRID: 133, 11) b. Creo que consigue muchas cosas por su inteligencia y otras muchas cosas lo consigue por su dinero. ‘I think that s/he gets many things because of his/her intelligence and many other things s/he gets it through his/her money.’ (MADRID: 97, 18) c. Los estudios superiores me dices que lo hiciste aquí en la Facultad. ‘Your studies in higher education you’re telling me that you did it here in the school.’ (SEVILLA: 41, 4)

That is, in all these examples, there is a lack of grammatical gender and/or number agreement between the lexical DO and the corresponding doubled pronoun: eso – la ‘that-NEUT’ – it-FEM’, otras muchas cosas – lo ‘many other things-FEM – it-MASC’, and los estudios superiores – lo ‘higher education studies-MASC – it-MASC’. We also find a certain degree of dissociation when the preverbal element is a DO without the expected DO marker a, that is, an animate DO without the prescriptively obligatory personal a. (2.43)

a. Mis amigos de la Universidad los adoro todavía. ‘My friends from college I still adore them.’ (SEVILLA: 255, 29) b. De todas formas, muchos de los actores de ahora los conocemos a través de la televisión. ‘Anyway, many of today’s actors we know them through TV.’ (MADRID: 276, 4)

Belén López Meirama 35

c. Porque, no sé, los demás los encuentro muy distanciados. ‘Because, I don’t know, the others I find them rather distant.’ (SEVILLA: 58, 13) The dissociation is even clearer when the thematized element is not the DO of a transitive sentence, as in the examples in (2.44): (2.44)

a. Y ¿la enseñanza media? La enseñanza media fui a una academia. ‘And secondary education? [For my] secondary education I went to an academy.’ (SEVILLA: 273, 9) b. La gente universitaria está mucho más unida, de forma que Granada, por ejemplo, vas en tiempo de curso o en tiempo de vacación y notas una diferencia enorme. ‘College people are much closer to one another, so Granada, for example, you go [there] during the period of classes, or over a break, and you notice an enormous difference.’ (SEVILLA: 160, 2)

Examples in (2.43) and (2.44) suggest that in extreme cases the theme may become an element outside of a sentence, an element that is similar to conjunctive adjuncts and modals in Halliday’s (1985) sense. In our view, however, the significant difference between these types of elements must be taken into consideration: the discourse-pragmatic value (more specifically, the informational value) of adjuncts is exclusively cohesive. Although these thematized elements qualify as being ‘spoken about’ within the sentence, they cannot be identified with any of the sentence’s arguments because they are not marked as such. Nevertheless, we should bear in mind that these examples almost exclusively belong to spoken language, which is characterized by constructional immediacy. This sense of urgency sometimes generates what is called Anacoluthon. That is, since the thematic position in unmarked cases is occupied by the subject argument, and given that the subject in Spanish lacks formal marking (for example, prepositions), it follows that a speaker may associate non-marking of an element with the grammatical relation of subject. As a result, a speaker sometimes begins his/her utterance with a noun phrase, although he/she then constructs a sequence in which that phrase should be introduced by a linking particle (en/durante la enseñanza media ‘in/during secondary education’; a Granada ‘to Granada’).18 In order to support this account, we offer examples in which an element that functions as theme is introduced by a marking that does not correspond to the meaning of the construction. For example, as the speaker in (2.45) goes along, s/he changes the syntactic structure with which s/he expresses the content s/he wants to communicate. In (2.45), the speaker begins with con Jacqueline ‘with Jacqueline’ as the theme and then marks this adjunct

36 Spanish Word Order

clause as an indirect object in se le ha creado demasiado mito ‘one has created to her to much myth’: (2.45)

Es que con Jacqueline creo que se le ha creado demasiado mito y no se la considera ya como una mujer igual que las demás. ‘It’s that with Jacqueline I think one has created around her a lot of myth, and she is not considered a woman like the others.’ (MADRID: 97, 25)

Another typical feature of spoken language is that the theme can be introduced by specific marking. In this regard, López García (1996: 485), following Metzeltin (1990), distinguishes what he calls a topicalizer, such as en lo referente a ‘with regard to’, en lo tocante a ‘with respect to’, ‘as for’, en relación con ‘in relation to’, and so on, from the lexical expression of its topicalized nature, which is realized through expressions like hablando de ‘speaking of’, la cuestión es que ‘the thing is that’, and so on. Similarly, Zubizarreta (1999: 4,220–1) points out that what she calls hanging topics can be optionally preceded by the expressions en cuanto a ‘as for’ and con respecto a ‘with respect to.’ It seems clear, then, that this way of marking an element implies a disassociation between the theme and the internal syntactic structure of the sentence. Some examples from Arthus are given in (2.46): (2.46)

a. En cuanto al grupo cuarto, cabe destacar que lo encabezan el Ignacio Gago y el San Mamed. ‘As for the fourth group, we should mention that it is led by Ignacio Gago and San Mamed.’ (1VOZ: 51, 6) b. La carrera en cuanto dificultad de, de superarla, de sacarla creo que no tiene problema mayor. ‘The degree, as far as difficulty of, of bettering it and completing it is concerned, I don’t think it’s a big problem.’ (MADRID: 100, 13) c. En cuanto a la Virgen, todos, creo, que estamos de acuerdo. La más representativa de Sevilla es La Macarena. ‘As for the Virgin [Mary], all of us, I think, agree. The most representative [Virgin] of Seville is La Macarena.’ (SEVILLA: 40, 25)

Another property that Zubizarreta considers typical of the hanging topic, and one which differentiates it from left-dislocation, is that ‘the relationship between the theme and a certain sentence-internal position is that of coreferentiality; there cannot be a relationship of grammatical dependence’ (1999: 4,224, translated from the original). This entails that within a sentence we can find not only a clitic (example 2.47a), but also a stressed pronoun as

Belén López Meirama 37

in (2.47b), an epithet as in (2.47c), or a phrase with which the theme establishes an inalienable relationship as in (2.47d) (examples from Zubizarreta 1999): (2.47)

a. En cuanto al hermano, parece que los padres lo contemplan mucho. ‘As for the brother, it seems that his parents have him on their mind a lot.’ b. En cuanto al hermano, parece que los padres hablan de él todo el tiempo. ‘As for the brother, it seems that his parents talk about him all the time.’ c. En cuanto al hermano, parece que el desgraciado se lleva bien con todo el mundo, inclusive con los padres. ‘As for the brother, it seems that the poor soul gets along with everybody, even with his parents.’ d. [Context: an argument about Juan’s cars]: En cuanto al BMW, parece que los frenos le fallan constantemente. ‘As for the BMW, it seems the brakes are constantly failing.’

In Arthus, we find examples of coreference with an element other than the clitic, even in the absence of thematization marking (that is, even without expressions such as en cuanto a ‘as for’). Some are listed in (2.48): (2.48)

a. Y luego, claro, coches, pues se veían muy pocos. ‘And then, of course, cars, you would see very few [of them].’ (MADRID: 159, 7) b. No sé, yo, vos sabés que los tanos no tengo nada en contra de ellos, pero … ‘I don’t know, but you know that Italians I have nothing against them, but …’ (BAIRES: 413, 27) c. Bueno, pues, ese animalito, trabajo, trabajo, lo que la gente llama trabajo puede que no tenga mucho. (MADRID: 219, 22) ‘Well, that little animal, work, work, what people call work, he probably doesn’t have much [to do]’ (cf. Clements Chapter 6 this volume, who considers this type a case of null object (pronominalisation)).

Extraposition is also possible in Spanish: an element that belongs to a sentence integrated within a complex syntactic structure is placed at the beginning of the structure, that is, outside of the sentence of which it is a constituent. Rivero (1980) considers such constructions to be left-dislocations,

38 Spanish Word Order

in which the subject has been fronted: (2.49)

a. A la jovencita no convenía apearla de su pedestal de sueños. ‘The young lady, it wasn’t a good idea to knock her off her pedestal of dreams.’ (USOS: 157, 36) b. Pero en fin, el arte me gusta apreciarlo, me gusta, me encanta, sí. ‘But anyway, art I like to appreciate it, I like it, I love it, yes.’ (SEVILLA: 92, 35) c. Éste de la Gran Vía ya no me acuerdo ahora cómo se llama. ‘This one on Gran Vía, I don’t remember what is called anymore.’ (MADRID: 249, 7)

Arthus, moreover, reveals that although it is possible to find examples in written texts as in (2.49a), extraposition is more common in spoken language, such as (2.49b) and (2.49c). In addition, many examples are found in reported speech (with matrix-sentence predicates such as creer ‘believe’, saber ‘know, learn’, and suponer ‘suppose’, generally in the first person), suggesting that this structure usually stems from the speaker’s involvement in the text, shown by the example in (2.50): (2.50)

La mentalidad creo que no me atrevo a decir que sea distinta. ‘The mentality, I think I don’t dare to say that it’s different.’ (SEVILLA: 269, 18)

As a corollary of what we have discussed so far, we believe that thematic structure in Spanish is highly productive and displays a higher degree of independence relative to syntactic structure than in languages such as French or English. The productive use of thematic structure in Spanish is, in our view, precisely what makes Spanish a so-called free-word-order language. The distributional combinations of elements are manifold, especially in the oral register. However, the possibilities are not arbitrary in that the placing of elements in a particular position depends on different communicative strategies that a speaker may adopt while producing his/her utterance. In the following section, we will show that emphasis also generates a particular word order. Focus and subject position In the previous section we claimed that the discourse-pragmatic structure is a complex network of different factors. This complexity has caused a good deal of confusion and heterogeneity regarding the notion of ‘theme’ and the traits that characterize it (cf. Casielles-Suárez, 2004). One of these factors is contrast, which can be defined as the use of various means by a speaker to draw attention to one of the elements in an utterance. In other words,

Belén López Meirama 39

contrast is given as a result of a speaker’s choice among a limited number of candidates, that is the elements in an utterance, with the element chosen being the so-called ‘contrastive focus’ (Chafe, 1976). One of the most representative cases of contrastive focus is the equational structure (or cleft and pseudo-cleft sentence), as illustrated in the examples in (2.51) in which different distributional possibilities are shown. In Spanish, this structure has some variants with the connectives si ‘if’ and porque ‘because’, examples of which are given in (2.52): (2.51)

(2.52)

a. Lo que no acepta mi soberbia es el error, mi visión deformada de ti. ‘What my pride cannot accept is the mistake, my deformed view of you.’ ( JÓVENES: 69, 8) b. Pero horchata y zarzaparrilla era lo que se bebía … la CocaCola es zarzaparrilla. ‘But horchata [a sweet milky drink made of tiger nuts] and sarsaparilla was what one would drink. Coca Cola is sarsaparilla.’ (MADRID: 259, 28) c. Son nuestros amigos mexicanos los que me han animado a pensar que puedo ganarme la vida en México, dando lecciones. ‘It was our Mexican friends who encouraged me to think that I can earn a living in Mexico teaching.’ (DIEGO: 40, 19) a. Si algo acertado he hecho en la vida, Emi, te lo aseguro, ha sido compartirla contigo. ‘If something right I’ve done right in life, Emi, I assure you, it’s been sharing it with you.’ (CINTA: 118, 27) b. Diría que lo que sucede hoy aquí es porque no nos aguantamos las unas a las otras ni un cochino día más. ‘I would say that what is happening here today is because we can’t stand one another even one more friggin’ day.’ (HOTEL: 48, 24)

Nevertheless, a discourse-pragmatic factor whose presence is practically constant in human communication (speaking, like living, is the product of successive choices) cannot be expressed only through a few specific structures. In any sentence, one of the constituents can be highlighted as a constrastive focus. In marked modalities, a possibility turns into a rule, and thus, the sentence-initial element in exclamatory sentences and in indirect, partial questions becomes the focus of the utterance:19 (2.53)

a. ¡Ya, ya, BUENAS LABORES debía de hacerle a ella! ‘OK, OK, GOOD WORK [I/she/he] should do for her.’ (SONRISA: 163, 29)

40 Spanish Word Order

b. ¿CUÁNTAS VECES te habrías adormecido escuchándolo desde tu habitación? ‘HOW MANY TIMES would you have dozed off listening to him from your room?’ (SUR: 40, 21) It is often said that in Spanish prosodic prominence is the mechanism that enables us to identify the focus. The pairs of examples in (2.54) and (2.55) (from Hernaz and Brucart, 1987) illustrate this point: (2.54)

(2.55)

a. Pedro se casará con María. ‘Pedro will marry María.’ b. PEDRO se casará con María (y no Luis). ‘PEDRO (not Luis) will marry María.’ a. María detesta las acelgas. ‘María hates chard.’ b. María detesta LAS ACELGAS (y no los bombones). ‘María hates CHARD (not chocolates).

Unfortunately, such examples are provided without appropriate contexts (even though the authors regard the focus as a discourse-pragmatic value that warrants analysis at the discourse level, instead of the sentence level). They are indeed invented sentences, not examples extracted from recordings of actual speech that are accompanied by the pertinent prosodic study. Arthus is not the most appropriate corpus to find examples containing this discourse-pragmatic value, since it is largely made up of written texts.20 Nevertheless, the information that we have been able to glean from it suggests that speakers usually combine intonational prominence with a modification of the word orders. On some occasions, speakers even make use of specific focalization markings (like the ones for the theme in the previous section) that we might label as focalizers: (2.56)

a. (repetition of [2.34a]) Quiero abastecerme YO MISMA de libertad y de aventura. ‘I want to stock up on freedom and adventure MYSELF.’ (ZORRA: 30, 1) b. Contra las esperanzas de Mercedes, el abuelo no ofrecía indicios de mejoría. Durante los últimos días, había telefoneado con frecuencia a médicos y hospitales o acudido ELLA MISMA a visitarlos. ‘Contrary to Mercedes’ hopes, her grandfather showed no signs of recovery. On the last days, she had often called doctors and hospitals, or SHE HERSELF had gone to visit them.’ (TERNURA: 125, 25)

Belén López Meirama 41

(2.57)

c. – Anda, anda, a tu siestecita …. ‘There, there, take your little nap …’ – Bueno, pero te acuestas TÚ TAMBIÉN. ‘OK, but YOU go to sleep, too.’ (SONRISA: 332, 36) a. HASTA LA MUERTE la consideras como una prerrogativa exclusivamente tuya. ‘EVEN DEATH you consider a prerogative exclusively your own.’ (CINTA: 22, 20) b. Pero HASTA ESO lo hizo con tanto sigilo que … ‘But EVEN THAT he did [it] in so much secrecy that …’ (CRÓNICA: 50, 24) c. Hace rato que entró en los llanos, ya NI GUARDACRUCEROS hay. ‘It’s been a while that he got into the plains. There are absolutely no GUARDACRUCEROS left’ [i.e., people who wave a flag in order to stop traffic and thus prevent accidents]. (DIEGO: 67, 23)

Based on an examination of these examples, we deduce that the focus is placed in a postverbal position if the subject of the sentence is a focalized element and in preverbal position if the focalized element is a complement. We believe word order in focalization can be more accurately characterized as follows: the focalized element is placed in a marked position, that is non-canonical position, in relation to its syntactic and/or semantic features (cf. Gili Gaya, 1961: 81–97). With regard to subjects, we can claim that, in general, the postverbal position (in particular, the absolute final position; Fant, 1984), is emphatic, but so is the preverbal position if, for instance, the subject is a determiner-less noun. The examples in (2.58) and (2.59) for subject as focus and (2.60) for object as focus (none of which include the focalizing element) serve to illustrate this point: a. ERNESTO – [Hablando de cuatro hermanos] Seguimos igual. Cuando uno está en dificultades, allí acudimos TODOS, ¿verdad? Ernesto: (speaking of four brothers) ‘The situation is the same. Each of us is having problems. EVERYBODY goes there, right?’ (CINTA: 89, 14) b. Miguel no podía dejar de mirar con asombro a un lado y a otro, sin advertir siquiera sus propias toses. Sí las advirtió en cambio LA ABUELA. ‘Miguel could not stop looking around in astonishment, without even noticing his own coughing. THE GRANDMOTHER, however, did notice it.’ (TERNURA: 6, 11) (2.59)21 a. … consultaban en vano el mapa con la nueva nomenclatura de las calles: no entendían ni pío. AMBULANCIAS Y COCHES PATRULLA aullaban inútiles.

(2.58)

42 Spanish Word Order

(2.60)

‘… they were vainly going over the map with the new street terminology. They did not understand a thing. AMBULANCES AND (PAISAJES: 15, 14) POLICE CARS howled pointlessly.’ b. Se ha combatido contra el extranjero, contra la antipatria: MORAL Y HONOR pueden dormir tranquilos. ‘One has combated foreigners [and] anti-homeland element. MORAL AND HONOR can sleep in peace.’ (HISTORIAS: 57, 14) a. ESTA IMPRESIÓN, si no otra peor, debimos de causarle a la recepcionista. ‘THAT IMPRESSION, if not a worse one, we probably gave the receptionist. (LABERINTO: 181, 24) b. MONTSERRAT – Yo no me mancho las manos con comida. Sólo tomaré una copa de champagne y una pata de pollo con el hueso envuelto en papel de plata. ‘I am not going to get my hands dirty with food. I will only have a glass of champagne and a chicken leg with the bone wrapped in aluminum foil.’ ROCÍO – Pues yo PAPEL DE PLATA puedo darte. De lo otro, nada. ‘Well, I can give you the ALUMINUM FOIL, but none of the other stuff.’ (HOTEL: 66, 37)

From our discussion so far, we can deduce that a focused element in initial position is also a thematic element, apart from being the topic. However, this runs against a widespread stance, especially among generativists, whereby there have to be two different types of fronting, illustrated in the following examples in (2.61) taken from Hernanz and Brucart (1987): (2.61)

a. A Sansón lo traicionó Dalila. ‘Samson, Delilah betrayed [him].’ b. A SANSÓN traicionó Dalila. ‘SAMSON Delilah betrayed.’

From this standpoint, only (2.61b) can be paraphrased with a clefted structure: Fue a Sansón (y no a otro) a quien traicionó Dalila ‘It was Samson (not another man) that Delilah betrayed’. Zubizarreta (1999: 4,240) explains these differences as follows: A preverbal-focus construction is distinguished from left-dislocation not only from a prosodic and interpretative viewpoint, but also from a syntactic one. A focalized direct object does not allow accusative clitic doubling. In contrast, a dislocated direct object obligatorily requires the presence of the accusative clitic. (Translated from the original)

Belén López Meirama 43

Given this stance, one could conclude that thematic structure and informational structure of contrastive focus are incompatible, since they cannot cooccur in the same sentence. However, this is not always the case. The speaker can combine both structures when s/he thematizes an element belonging to a cleft structure, as shown in the examples in (2.62), taken from Arthus: (2.62)

a. El campo lo que encuentro es que es un poco aburrido. ‘The countryside what I find is that it is a bit boring.’ (SEVILLA: 136, 34) b. Ahora, ¿Esto lo que gasta es gas o petróleo? ‘Now, that, what it runs on is it natural gas or normal gas?’ (BAIRES: 103, 34) c. Y allí los coches lo que no pueden es circular porque nada más es para aparcarlo. ‘And there, cars what they cannot do is drive around, because it is for parking only.’ (SEVILLA: 166, 31)

On the other hand, the concepts of ‘theme’ and ‘focus’ as used here cannot lead us to Zubizarreta’s (1999: 4,227) conclusion that the same phrase cannot function as theme and focus at the same time. In fact, there is no reason why an element which the sentence is about (that is, the theme) cannot be at the same time the result of a choice on the speaker’s part (that is, the focus). It is clear that (2.61a) and (2.61b) are different sequences from a discoursepragmatic standpoint; the presence or absence of the coreferential clitic is not, of course, arbitrary. Although it may be that the Arthus corpus is not the most appropriate source for studying the discourse-pragmatic value of an element of a sentence that manifests itself through intonational emphasis, it is indeed a good source for real oral and written language. The instances provided by the corpus indicate that in a large portion of the cases the coreferential clitic appears in sequences in which the postverbal subject is also the contrastive focus or, to be more precise, in sequences in which there is a postverbal element that is also a focus of contrast. Examine the examples in (2.63) and (2.64), which have been contextualized in order to highlight the contrastive value of the focused elements: (2.63)

Inf. A – Mira, ya al fin y al cabo, estamos lanzados … ‘Look, finally, we have thrown ourselves into it.’ Inf. C – EL SEGUNDO AÑO hacéis, ¿no? ‘You are doing THE SECOND YEAR, right?’ Inf. B – ¡No!, ¡primero! ‘No! First!’ Inf.C – ¿Y son tres?

44 Spanish Word Order

(2.64)

‘And are they three?’ Inf. B – ¡Cinco! ‘Five!’ (MADRID: 440, 24) Hasta cuarto, lo hice en el mismo colegio que hice la primaria, y ya, después QUINTO lo hice EN EL INSTITUTO, sexto libre y, ahora, C.O.U. otra vez en el instituto. ‘Up until fourth grade, I did at the same school that I did my elementary education, and then THE FIFTH YEAR I did IN THE INSTITUTE, sixth I tested out of, and now the Pre-University Course again in the institute.’ (SEVILLA: 55, 21)

Based on the discussion so far, we believe that the examples in (2.61) can have two interpretations (with the corresponding intonational change shown in small capitals). The sentence (2.61a) can be interpreted (with stress marked) as A SANSÓN lo traicionó DALILA, that is, fue a Sansón (y no a otro) a quien traicionó Dalila (y no otra) ‘it was Samson (not another man) that Delilah (not another woman) betrayed’. It can also be interpretated as A SANSÓN lo traicionó DALILA, that is, fue Dalila (y no otra) quien traicionó a Sansón ‘It was Delilah (not another woman) that betrayed Samson’. In this respect, we have already mentioned that sentences generally lack an overt subject in the majority of cases containing direct object thematization. Therefore, it would be erroneous to think that the postverbal occurrence of this subject is due only to the thematized object. In our view, such an occurrence is also discourse-pragmatically relevant, which may explain why (2.61b), in which the subject is not a contrastive element, is not representative of Spanish, as evidenced in the Arthus corpus where out of a total of roughly 160,000 sentences we found not one example of the type illustrated by (2.61b). A good source of examples (recurrent throughout Arthus) that seems to corroborate this hypothesis can be found in a neuter demonstrative eso used as the thematized element. For instance, the example in (2.65) can be said with an additional coreferential clitic, illustrated in (2.66); such an addition would indicate the focalization of the element in final position, shown in small capitals in (2.66): (2.65)

(2.66)

a. ESO comentábamos con Lita, ¿no? ‘THAT we were discussing with Lolita, weren’t we?’ (BAIRES: 44, 22) b. ESO hizo por mí la Salvinia. ‘THAT, Salivinia did for me.’ (SONRISA: 261, 26) a. ESO lo comentábamos CON LITA, ¿no? (posible respuesta: ‘no, con Laura’) ‘THAT we were discussing WITH LITA, weren’t we?’ (possible answer: ‘no, with Laura’)

Belén López Meirama 45

b. ESO lo hizo por mí LA SALVINIA. (posible continuación: ‘aunque tú creas que fue Eva’). ‘THAT SALVINIA did for me (possible follow-up:’even though you think it was Eva’). In fact, there are numerous examples featuring postverbal subjects, as in (2.67), in which the context clearly shows that they carry contrastive focus: (2.67)

VÍCTOR – Pero para dejar a salvo el idealismo de tu juventud no dudarás en acusarla de delatora, aún a sabiendas de que es mentira. VÍCTOR JOVEN – ESO lo hiciste TÚ. A mí, ni siquiera me han detenido todavía. VÍCTOR – ‘But in order to salvage the idealism of your youth, you will not hesitate to accuse her of being an informer, even though you know that it’s a lie.’ VÍCTOR JOVEN – ‘THAT YOU did. I haven’t even been arrested yet.’ (HOMBRE: 35, 29)

Our interpretation of this type of structures is a tentative one that could be either confirmed or refuted in a thorough study on contrastive focus in Spanish based on a large corpus of oral texts. To our knowledge, such a study has yet to be carried out, although we do have some general works on the discourse-pragmatic context in Spanish, among which Fant (1984) stands out. The author likens LOS PLATOS los lava FERNANDO ‘The dishes Fernando washes them’ to a clefted structure such as Lo que vende son CALCULADORAS’ What he sells are calculators’. Fant (1984: 76) claims that both sequences are automatically assigned a contrastive value, which allows them to be combined, as illustrated by the example (2.68) taken from Fant (1984: 131). (2.68)

… a Carlos le voy a dar dos tortas si no pone más cuidado al lavar los platos. ‘Carlos I’m going to give [him] a couple of slaps if he doesn’t pay more attention when doing the dishes.’ Espérate, papá, Carlos es quien quita la mesa; los platos los lava Fernando. ‘Wait, Dad, Carlos is the one who clears the table, but it’s Fernando that does the dishes.’

In any case, it is indisputable that contrastive emphasis in Spanish makes use of word order as a common means of expression, and thus becomes another factor that affects the variability in the order of constituents. For this reason, contrastive emphasis should receive more attention in studies

46 Spanish Word Order

on constituent order in Spanish, to be examined alongside thematization and topicalization.

Summary In this chapter, we have not offered an exhaustive treatment of constituent order in Spanish, but rather we have tried to point out the most relevant semantic and discourse-pragmatic aspects that affect it, using real-text examples from modern Spanish. With respect to the semantic aspects affecting Spanish constituent order, we have examined two: agentivity and the definiteness of the subject noun phrase. Here, we noted the tendency displayed by Spanish to postpose non-agentive and non-definite subjects and we suggested the possibility of recasting two hypotheses discussed in the generative grammar literature (the Unaccusative Hypothesis and the Definiteness Effect) as two parts of a complex array of factors that affect Spanish word order. Although the usefulness of these notions to account for the constituent order phenomena is evident, we consider it risky to establish taxonomic laws (for example verb is an unaccusative verb) with concrete reference to constituent order in Spanish, a language about which it is preferable to speak in terms of wordorder tendencies. That is, we prefer to speak in terms of unergative and unaccusative constructions, not unergative or unaccusative verbs. On the other hand, we have found that different features overlap considerably. That is, in the linear organization of elements in each utterance by a given speaker, various factors interact in each case, semantic as well as discourse-pragmatic factors. The reliance on unaccusativity and/or definiteness alone can become, thus, a reductionist enterprise. As regards discourse-pragmatic factors, we have centred our analysis on thematic structure in discourse, attempting to offer a basic notion of a topic that is still in need of highly detailed studies. Thus, we consider the work presented in this chapter as preliminary in various ways.22 First, it must be noted that the notion ‘theme’ has been defined at times as ‘known/given/ old information’, at others as ‘a cohesive element’, and even as a ‘potentially contrastive element’ (cf. Casielles-Suárez, 2004). In this chapter, where we have preferred the definition of a theme as ‘that which is spoken about in the sentence’, we observed that the theme can frequently be associated with the features just mentioned, but in no way can it be defined by these features. The element in sentence-initial position, about which something is predicated in the sentence, is customarily known/old/given information, but it can also be new information. Such an element is customarily a cohesive element in discourse, but cohesion can also be achieved using other means. At times, a theme may be defined as contrastive, but contrastive focus, apart from the possible trait of emphatic intonation, can be independently characterized as appearing in a non-default position, that is, a marked position. That is, no one trait defines one and only one notion.

Belén López Meirama 47

We also briefly examined Spanish thematic structure, pointing out its characteristic features. On the one hand, we observed the frequency with which we find athematic clauses in Spanish, something rarely discussed in the literature. On the other hand, we discussed the most characteristic cases of a ‘marked theme’ in transitive clauses, whereby we noted that Spanish uses a wide variety of mechanisms to mark the theme, above all in spoken language. Again, these facts suggest that in studying Spanish constituent order it is more accurate to speak of tendencies or patterns than hard and fast rules. Notes 1

2

3

4 5

6 7 8

9

For this paper we have used data from the Base de datos sintácticos, compiled at the University of Santiago Compostela from a large corpus on present-day Spanish (Arthus). The compilation was made possible through successive research projects under the supervision of Prof. D. Guillermo Rojo. For further information, the reader is directed to http://www.bds.usc.es. Some authors, such as Delbecque (1991: 261), claim that Spanish is an SVO/VS language, whereas others, De Miguel Aparicio (1989: 758) among them, argue that it is an SVO/SV-VS language. This last characterization coincides with Dryer’s (1997) proposed typology. As Givón (1995: 32) points out, ‘the main, declarative, affirmative, active clause has been tacitly assumed, in grammatical description ever since the Greeks, to be the privileged, unmarked clause type (bold is my emphasis).’ This assumption is currently shared by different frameworks. In this chapter, we will not discuss cleft constructions or clauses containing a copula. In López Meirama (1997a and 1997b), we offer a detailed account of the reasons why Spanish is to be considered a subject-initial language. Here we remind the reader that Spanish is considered a nominative-accusative language because both the intransitive- and transitive-sentence subjects (S and A respectively) are marked in the same way, and distinctly from the object (O). As is well-known, in the languages of the world we find two opposing tendencies with regard to topicalization: the more common one consists of the preference for the topicalization of agentive items; the less common tendency, however, reveals a preference for the topicalization of patients. Bearing in mind the syntactic generalizations we are referring to, we can establish two patterns of topic selection: the nominative-accusative {S,A} and the ergative-absolutive {S,O} patterns. Spanish is generally considered a nominative-accusative language, since the subject is usually the most agentive argument of the predication (which, obviously, does not mean that all subjects are agents). This implies that the pattern of topic selection is {S,A}, and, consequently, that the basic order is SVO/SV (however, see Clements, Chapter 5 of this volume, for a discussion of ergative patterns in Spanish). ‘I’ represents ‘indirect object’, and ‘A’ stands for ‘adverbial adjunct’. In the examples, taken from Arthus, the verbal nucleus of each example has been italicized. In relation to this class some researchers also allude to the verb classes in Hatcher (1956), who subdivides them in ‘verbs of existence-presence’, ‘verbs of continuingremaining’, ‘verbs of occurrence’, and so on. See López Meirama (1997a: 158ff) for further discussion.

48 Spanish Word Order 10

Contreras (1978: 74) explicitly states that in addition to presentative verbs accompanied by only one patient, there are others that also include a locative or an expression of origin. 11 Some examples are: Hawkinson and Hyman (1974), Givón (1976), Silverstein (1976), Dixon (1979) and Lazard (1984). 12 The translations from Spanish are ours ( JCC, JY). 13 There are cases in which sentence-initial position is occupied by a cohesive element, or by an adjunct, as in the following examples from our corpus. (i)

Francamente la televisión a mí me aburre. ‘Frankly, television bores me.’ (SEVILLA: 94, 8) (ii) De repente, a Renato le alarma en el viejo cierta expresión doliente. ‘Suddenly, a certain suffering expression of the old man alarms Renato.’ (SONRISA: 142, 33) 14 15

16

17

18

19 20

21

22

That is, the speaker’s supposition about what the hearer can accept as true in each stage of the conversation (Reyes, 1985: 576). A tally of the Arthus data yielded the following figures: out of a total of 149,882 sentences with subject, 53,669 (35 per cent) contain an overt subject and 96, 213 sentences (65 per cent) have a null subject. Jiménez Juliá (2000: 164) proposes that more than one element in preverbal position can be considered having two different structures, made up of either two linked themes, Yo//, de eso// (no tengo) ni idea ‘I, about that, (I don’t have) any idea’, or a presentative theme linked to a focalized unit (in small caps), as in Eso A MÍ no me gusta nada [lit. that to me NEG me pleases nothing] ‘THAT, I don’t like at all’, or A mí ESO no me gusta nada [lit. to me that NEG me pleases nothing] ‘I don’t like THAT at all’. Following Metzeltin (1990), López García (1996: 485) considers the pause as a topicalization strategy in Spanish, in cases as the following: Dulces, no quiero ‘Candies, I don’t want’, or A estos jóvenes recelosos, un exceso de solicitud y de comprensión les parece aún más sospechoso que la rigidez ‘To these distrustful youths, an excess of solicitude and of understanding seems to them even more suspicious than rigidity.’ This could also account for instances of spurious agreement, such as ¿Qué cosas te interesan más conocer? ‘What things are you most interested in knowing?’ (SEVILLA: 75, 24), in which there is agreement between main-clause interesan [interest. PRESENT.3P.] and qué cosas, the DO of the embedded-clause verb conocer. Given that this type of focus is typically accompanied by prosody, the focus element in these examples is highlighted in small capitals. In fact, all cases of focalization are interpretations of this author based on a careful reading of the texts. Thus, the use of Arthus, especially for the examples in this chapter, should be considered a way of illustrating the most prominent tendencies in Spanish constituent order. According to Suñer (1982: 224), the coordination of determiner-less nouns is one of the strategies that Spanish uses to present a situation as contrastive: ‘They do not obey the [Naked Noun Constraint] because they are contrastive; that is, each element of the list receives heavy stress because the focus of the sentence has been attracted to this preverbal subject position.’ Here, the recent work by Hidalgo Downing (2003) could serve well as a point of departure.

Belén López Meirama 49

References Academia, Real … Española (1973) Esbozo de una nueva Gramática de la Lengua Española. Madrid: Espasa-Calpe. Belletti, A. (1987) Los verbos inacusativos como asignadores de caso. Sintaxis de las lenguas románicas, in V. Demonte and M. Fernández Lagunilla (eds). Madrid: El Arquero, pp. 67–230. Belletti, A. and L. Rizzi (1987) Los verbos psicológicos y la teoría temática. Sintaxis de las lenguas románicas, in V. Demonte and M. Fernández Lagunilla (eds). Madrid: El Arquero, pp. 60–122. Birner, B.J. (1992) The Discourse Function of Inversion in English. doctoral dissertation. Northwestern University, Evanston, IL. —— (1994) ‘Information Status and Word Order: An Analysis of English Inversion’, Language, vol. 70, pp. 233–59. Bossong, G. (1980) ‘Variabilité positionnelle et universaux pragmatiques’, Bulletin de la Société de Linguistique de Paris, vol. 75, pp. 39–67. Butt, J. and C. Benjamin (2000) A New Reference Grammar in Modern Spanish, 3rd ed. London: Edward Arnold. Casielles-Suárez, E. (2004) The Syntax-Information Structure Interface: Evidence from Spanish and English. London: Routledge. Chafe, W.L. (1976) ‘Giveness, Contrastiveness, Definiteness, Subjects and Topics’, in C.N. Li (ed.), Subject and Topic. New York: Academic Press, pp. 25–55. Cifuentes H., José Luis (1999) ‘Inacusatividad y movimiento,’ Revista Española de Lingüística, vol. 29, pp. 35–61. Cinque, G. (1983) ‘Topic constructions in some European languages and “Connectedness”’, Connectedness in Sentence, Discourse and Text in H. van Riemsdijk and K. Ehlich (eds). Tiburg: Katholieke Hogeschool. Contreras, H. (1976) A Theory of Word Order with Special Reference to Spanish. Amsterdam: North Holland (Spanish version: El orden de palabras en español. Madrid, Cátedra, 1978). Delbecque, N. (1979) Etude quantitative d’un phénomène de variation syntaxique: la position du sujet en espagnol. Doctoral Dissertation. Universidad Católica de Lovaina. —— (1991) El orden de los sintagmas. La posición del regente. Gramática española: enseñanza e investigación, vol. II, Gramática. Ediciones Universidad de Salamanca. Dik, S. (1978) Functional Grammar. Amsterdam: North Holland. De Miguel Aparicio, E. (1989) ‘Sujetos invertidos en las construcciones ergativas del castellano. Los conceptos de tema y rema’. Lenguajes naturales y lenguajes formales, IV/2, in Carlos Martín Vide (ed.) Universidad de Barcelona, pp. 753–66. De Miguel Aparicio, E. and S. Olga Fernández (1988) ‘Proceso-acción y ergatividad: las construcciones impersonales en castellano’. Lenguajes naturales y lenguajes formales, III/2, in Carlos Martín Vide (ed.) Universidad de Barcelona, pp. 643–51. Dixon, Robert M.W. (1979) ‘Ergativity’. Language, vol. 55, pp. 59–138. —— (1987) ‘Introduction’, in R.M.W. Dixon (ed.), Studies in Ergativity. Amsterdam: Elsevier Science Publishers B.V., pp. 1–16. Downing, P. (1995) ‘Word order in discourse’, in Pamela Downing and Michael Noonan (eds), Word Order in Discourse. Amsterdam: John Benjamins, pp. 1–27. Fant, L. (1984) Estructura informativa en español. Estudio sintáctico y entonativo. Acta Univ. Upsaliensis 34. Fernández Soriano, O. (1993) ‘Sobre el orden de palabras en español’, Dicenda (Cuadernos de Filología Hispánica) vol. 11, pp. 113–52.

50 Spanish Word Order Forster, J.F. and C.A. Hofling (1987) ‘Word order case and agreement’, Linguistics, vol. 25, pp. 475–99. García-Miguel, José María (1991) ‘La duplicación de complemento directo e indirecto como concordancia’, Verba, vol. 18, pp. 375–410. —— (1995) Las relaciones gramaticales entre predicado y participantes. Lalia, Series Maior, no. 2. Santiago de Compostela: Universidad de Santiago de Compostela. Gawelko, M. (1995) ‘Sobre el orden de palabras básico en español’. Anuario de Lingüística Hispánica (Universidad de Valladolid) vol. 11, pp. 183–92. Gili Gaya, S. (1961) Curso superior de sintaxis española. Vox: Barcelona. Givón, T. (1976) ‘Topic, pronoun and grammatical agreement’, in Charles N. Li (ed.), Subject and Topic, New York, Academic Press. Givón, T. (1984) Syntax: A Functional-Typological Introduction, vol. I. Amsterdam: John Benjamins. —— (1995) Functionalism and Grammar. Amsterdam: John Benjamins. Halliday, M.A.K. (1967) ‘Notes on Transitivity and theme in English, Part 2’. Journal of Linguistics, vol. 3, pp. 199–244. —— (1985/1994)2 An Introduction to Functional Grammar. London: Edward Arnold. Hatcher, A. Granville (1956) Theme and underlying question. Two studies of Spanish word order. Word 12, supplement 3. Hawkinson, A.K. and L.M. Hyman (1974) ‘Hierarchies of natural topic in Shona’, Studies in African Linguistics, vol. 5, pp. 147–70. Hernanz, M.L. and J.M. Brucart (1987) La sintaxis. Barcelona: Ed. Crítica. Hidalgo D., Raquel (2003) La tematización en el español hablado. Madrid: Gredos. Hopper, P. J. and S.T. (1980) ‘Transitivity in grammar and discourse’, Language, vol. 56, pp. 251–99. Jiménez Juliá, T. (2000) ‘Tema en español y en inglés: dos conceptos enfrentados’, Bulletin of Hispanic Studies vol. 77, pp. 153–76. Knauer, G. (1989) ‘Verbo y orden de palabras en la estructura oracional del español actual’ Linguistische Arbeits-Beritche, vol. 68, pp. 43–52. Lazard, G. (1984) ‘Actance variation and categories of the object’, in Frans Plank (ed.), Objects. Towards a Theory of Grammatical relations. London: Academic Press in pp. 269–92. Levin, B. and M.R. Hovav (1995) Unaccusativity: at the Syntax-Lexical Semantics Interface. Linguistic Inquiry Monographs 26. Cambridge, Massachusetts: The MIT Press. López García, Á. (1996) Gramática del español II. La oración simple. Madrid: Arco Libros. López Meirama, B. (1997a) La posición del sujeto en la cláusula monoactancial en español. Lalia, Series Maior, no. 7. Santiago de Compostela: Universidad de Santiago de Compostela. López Meirama, B. (1997b) ‘Aportaciones de la tipología lingüística a una gramática particular: el concepto orden básico y su aplicación al castellano’. Verba, vol. 24, pp. 45–82. Mendieta, E. and I. Medina (1997) ‘Anteposición de objeto en el habla culta de México y Madrid’. Revista Española de Lingüística vol. 27, pp. 447–77. Metzeltin, M. (1990) Semántica, pragmática y sintaxis del español. Heidelberg: Gottfried Egert Verlag. Milsark, G. (1974) Existential Sentences in English. PhD dissertation. Massachussets Institute of Technology. Cambridge MA. —— (1977) ‘Toward an explanation of certain peculiarities in the existential construction in English’, Linguistic Analysis, vol. 3, pp. 1–30.

Belén López Meirama 51 Morales de W.A. (1982). ‘La posición del sujeto en el español de Puerto Rico a la luz de la clase semántica verbal, la oposición tema-rema y el tópico oracional’, Lingüística Española Actual, vol. 4, pp. 23–38. Reyes, G. (1985) ‘Orden de palabras y valor informativo en español’, in Julio Fernández Sevilla et al. (ed.), Philologica Hispaniensa in Honorem Manuel Alvar II. Madrid: Gredos pp. 567–88. Rivero, M. L. (1980) ‘On left-dislocation and topicalization in Spanish’, Linguistic Inquiry, vol. 11, pp. 363–93. Silva-Corvalán, C. (1977) ‘A discourse study of some aspects of word order in the Spanish spoken by Mexican-Americans in West Los Angeles’. Unpublished M.A. Thesis. —— (1984) ‘Topicalización y pragmática en español’, Revista Española de Lingüística vol. 14, pp. 1–19. Silverstein, M. (1976) ‘Hierarchy of features and ergativity’, in Robert M. W. Dixon (ed.), Grammatical Categories of Australian Languages Camberra: Australian Institute of Aboriginal Studies pp. 112–71. Suñer, M. (1988) ‘The role of agreement in clitic doubled constructions’, Natural Language and Linguistic Theory, vol. 6, pp. 391–434. Thompson, S. (1978) ‘Modern English from a typological point of view: some implications of the function of word order’, Linguistische Berichte, vol. 54, pp. 19–35. Tomlin, R.S. (1986) Basic Word Order: Functional Principles. Kent: Croom Helm. Vázquez, R.V. (1995) El complemento indirecto en español, col. Lalia, Series Maior, no. 1. Universidad de Santiago de Compostela. Zubizarreta, M.L. (1999) ‘Las funciones informativas: tema y foco’, in Ignacio Bosque and Violeta Demonte (eds), Gramática descriptiva de la lengua española (Chapter 64), (eds) 215–44. Madrid: Espasa Calpe.

Abbreviations used for the original texts cited Essays Bunge, M., Lingüística y filosofía, Ariel, Barcelona, 1983. Martín Gaite, C., Usos amorosos de la postguerra española, Anagrama, Barcelona, 19888. TIEMPO Paz, O., Tiempo nublado, Seix Barral, Barcelona, 1983. HOMILÍA Sánchez Ferlosio, R., La homilía del ratón, El País, Madrid, 1986. LING

USOS

Narrative Aldecoa, J.R., Porque éramos jóvenes, Seix Barral, Barcelona, 1986. Bioy Casares, A., Historias desaforadas, Alianza, Madrid, 1986. CARTA Colinas, A., Larga carta a Francesca, Seix Barral, Barcelona, 1986. GLENDA Cortázar, J., Queremos tanto a Glenda, Alfaguara, Madrid, 19814. CRÓNICA García Márquez, G., Crónica de una muerte anunciada, Mondadori, Madrid, 1987. SUR García Morales, A., El sur (seguido de Bene), Anagrama, Barcelona, 1985. PAISAJES Goytisolo, J., Paisajes después de la batalla, Montesinos, Barcelona, 1982. MIRADA Guelbenzu, J.M., La mirada, Alianza, Madrid, 1987. TERNURA Martínez de Pisón, I., La ternura del dragón, Anagrama, Barcelona, 19883. LABERINTO Mendoza, E., El laberinto de las aceitunas, Seix Barral, Barcelona, 1982. DIEGO Poniatowska, E., Querido Diego, te abraza Quiela y otros cuentos, Alianza/Era, Madrid, 1987. JÓVENES

HISTORIAS

52 Spanish Word Order SONRISA

Sampedro, J.L., La sonrisa etrusca, Alfaguara, Madrid, 1985.

Theatre Alonso de Santos, J.L., Bajarse al moro, Antonio Machado, Madrid, 19872. Buero Vallejo, A., Caimán, Espasa-Calpe, Madrid, 1981. AYER Díaz, J., Ayer, sin ir más lejos, Antonio Machado, Madrid, 1988. OCHENTA Diosdado, A., Los ochenta son nuestros, Antonio Machado, Madrid, 1990. COARTADA Fernán Gómez, F., La coartada, Antonio Machado, Madrid, 1987. HOTEL Gala, A., El hotelito, Antonio Machado, Madrid, 1988. ZORRA Nieva, F., Te quiero, zorra, Antonio Machado, Madrid, 1989. 1INFAN Olmo, L. y P. Enciso, Teatro infantil I, Antonio Machado, Madrid, 1987. 2INFAN Olmo, L. y P. Enciso, Teatro infantil II, Antonio Machado, Madrid, 1987. PASAJERO Reina, M.M., El pasajero de la noche, Antonio Machado, Madrid, 1988. CINTA Reina, M.M., La cinta dorada, Antonio Machado, Madrid, 1989. HOMBRE Salom, J., Un hombre en la puerta, Martín Pancorbo, 1984. MORO

CAIMÁN

Oral speech Barrenechea, A.M., El habla culta de la ciudad de Buenos Aires. Materiales para su estudio, tomo II, Instituto de Fil. y Lit. Hispánicas Dr Amado Alonso, Buenos Aires, 1987. MADRID Esgueva, M. y M. Cantarero (eds.), El habla de la ciudad de Madrid. Materiales para su estudio, CSIC (Miguel de Cervantes), Madrid, 1981. SEVILLA Pineda, M.A. de (ed.), Sociolingüística andaluza 2. Materiales de encuestas para el estudio del habla urbana culta de Sevilla, Universidad de Sevilla, 1983. BAIRES

Newspapers 1VOZ La Voz de Galicia (30 October 1991). 2VOZ La Voz de Galicia (22 November 1991). 3VOZ La Voz de Galicia (23 November 1991).

3 Continuity and Episodic Structure in Spanish Subject Reference Llorenç Comajoan

Subject reference in Spanish In Spanish, as in other Romance languages, human subject reference is encoded by three linguistic devices: lexical noun phrases (NPs), personal pronouns and null subjects (zero).1 For instance, Participant 3 in this study produced the three types of devices to refer to a human referent (el niño, ‘the boy’) in the following narrative segment: (3.1) El niño recoge una de las banastas de de peras y Ø se las lleva con él en la bicicleta. Ah mientras Ø se lleva las peras Ø se cruza con otra niña que viene en bicicleta y, al cruzarse, en un camino que es más bien estrecho, él pierde el sombrero, Ø se despista mirando hacia atrás y Ø se cae. ‘The boy takes one of the baskets of pears and Ø takes them with him on the bike. Ah while Ø is taking the pears Ø passes another girl who is coming on a bike and when passing, in a road that is narrow, he loses his hat, Ø distracts himself looking back, and Ø falls.’ Research in Spanish nominal reference has been conducted within different linguistic theories and methodologies (see reviews in Luján, 1999; SilvaCorvalán, 2001). The major research questions in Spanish functional linguistics have been related to two main topics: the variable use of pronouns in different Spanish varieties and the linguistic explanations for the variability of use. The variability in Spanish nominal devices can be examined by studying the use of devices in different varieties (for example, comparing Puerto Rican and Madrid Spanish) or the use of different devices within one variety (for example, rates of NP, pronoun, and zero in Puerto Rican Spanish). 53

54 Continuity and Episodic Structure in Spanish Subject Reference Table 3.1 Percentage rates of third-person pronouns in different Spanish varieties Pronoun 3rd sing. % 8 10 12 14 26 39 43 48

Pronoun 3rd pl. % 8 9 6 14 16 25 23 22

Variety

Study

Madrid Puente Genil (Spain) Spain Madrid Mexican (Los Angeles) Puerto Rican (San Juan) Puerto Rican (Boston) Puerto Rican (New York City)

Cameron (1992) Ranson (1991) Rosengren (1974) Enríquez (1984) Silva-Corvalán (1994) Cameron (1992) Hochberg (1986) Flores-Ferrán (2002)

Most Spanish speakers intuitively notice that Spanish subject pronouns are more frequent in certain varieties (such as Puerto Rico Spanish) than in others (for example, Madrid Spanish), and research results have shown that different varieties of Spanish indeed display different rates of pronoun/NP use. Table 3.1 displays results from studies in different varieties that have focused on third person nominal reference and shows that the percentage use of pronouns ranges between 8% and 48%. However, the data on the variability of use of pronouns need to be taken cautiously, because the types of discourse, the total number of nominal devices under study, and the syntactic positions are not always the same. Table 3.2 includes a summary of studies that have specifically studied nominal reference. Some of the discourse types for these studies include written literary texts (Rosengren, 1974), semi-guided or spontaneous personal oral conversations (Givón, 1983; Bentivoglio, 1983; Cameron, 1992; Flores-Ferrán, 2002; Ranson, 1991), and elicited spoken narratives (for example, using The Pear Film, Blackwell, 1994; Clancy, 1980). Some studies (for example, Bentivoglio, 1983, 1992; and Blackwell, 1994) analyse three linguistic devices, but others (such as Cameron, 1992; Flores-Ferrán, 2004) have included only zero and pronouns. The role of different syntactic contexts in the analysis of nominal reference is evident in the data from Blackwell (1994). When the use of nominal devices in subject and object position were analysed, full NPs and zero were the most common devices (38% and 36%, respectively), followed by pronouns (26%). However, when only anaphoric instances in subject position were examined, the proportions differed (90% were zero, 5% were pronouns, and 5% were NPs). Table 3.2 shows that the use of pronouns is variable (ranging between 20% and 45%) and that zero and NPs are used in higher proportions than pronouns.

Llorenç Comajoan 55 Table 3.2 Percentage rates of zero, pronoun, and full NPs in different Spanish varieties Zero 40 56 36 90 55 79 55

Pronoun 23 24 26 5 45 21 45 40 39 24 20 22

Full NPs 37 20 38 5

Variety

Study

Caracas Caracas Aragon (Spain) Aragon (Spain) Puerto Rican (San Juan) Madrid Puerto Rican (New York City) Puerto Rican (Boston) Puerto Rican Puente Genil (Spain) Madrid Mexican (Los Angeles)

Bentivoglio (1983) Bentivoglio (1992) Blackwell (1994) (S and O) Blackwell (1994) Cameron (1992) Cameron (1992) Flores-Ferrán (2004) Hochberg (1986) Morales (1986) Ranson (1991) Enríquez (1984) Silva-Corvalán (1994)

Functional approaches in the study of noun referentiality In order to explain the variable use of NPs, pronouns, and zero in Spanish, researchers in functional linguistic approaches have isolated a series of internal and external factors (Flores-Ferrán, 2002; Morales, 1982; Silva-Corvalán, 2001). Internal factors include switch reference, person/number of the verbal inflection, tense-mood-aspect of the verb, verbal semantics (for example, perception v. cognition verbs), ambiguity, functional compensation, and specificity. External factors include narrative style, contact with another language, social class, and the speakers’ gender and age. The most studied internal factor is switch reference, which describes two types of relationships between NPs: ‘When these two NPs have different referents, they are “switch” in reference. When these two NPs share the same referent, they are “same” in reference’ (Cameron, 1995: 4). Studies following a variationist approach have shown that switch reference is a strong factor in the use of Spanish NPs. For instance, Silva-Corvalán (1982) examined conversational data of 24 speakers in the Spanish-speaking community of West Los Angeles and studied what variables could account for the (non)expression of subjects. The results showed that the factors that favoured absence of subject were old information, unambiguous verb, and same reference; whereas the factors that least favoured absence of subject were ambiguous verb form, new information, and switch reference. In a comparison of San Juan and Madrid Spanish, Cameron (1994) found that 66% of pronouns were used to switch reference in San Juan Spanish (compared to 35% to refer to the same reference). In Madrid Spanish, the percentages were

56 Continuity and Episodic Structure in Spanish Subject Reference

38% and 14%, respectively.2 Further research has emphasized the importance of going beyond switch reference and has shown that degree of discourse connectedness, reference chains, and perseveration need to be taken into consideration as well (Bayley and Pease-Alvarez, 1997; Cameron, 1995; Cameron and Flores-Ferrán, 2004). This study takes previous research into consideration and expands on it by examining two approaches to the study of nominal reference in Spanish: the distance and episodic approaches. The distance approach Early studies of noun referentiality derived from Givón’s (1983a) theory of nominal reference and topic continuity, which argued that the function of nominal reference within discourse was to assist the speaker and the hearer in keeping track of the participants in discourse. Givón’s (1983a, b, c) methodology provided quantitative data in three measures of topic continuity: (a) referential distance (look-back) assessed the gap between a referent in a clause and its previous mention; (b) persistence or decay mirrors the previous measure and assessed for how long a referent was mentioned in discourse after first mention; and (c) potential interference or ambiguity examined how several referents interact in discourse (that is, how reference is established when more than one referent are present in the context). Results using this methodology showed that different linguistic devices could be distributed along a continuum of topic accessibility in which zero coded the most continuous/accessible topics, and referential indefinite NPs coded the most discontinuous topics (Givón 1983a: 17, see also Givón, 2001b: 463) (Table 3.3).3 Bentivoglio (1983) applied Givón’s (1983a, c) methodology to analyse Latin American Spanish (Mexico City, Caracas, and Santiago de Chile).

Table 3.3 Nominal devices for topic continuity Continuity

Nominal device

Most continuous/accessible topic zero anaphora unstressed/bound pronouns or grammatical agreement stressed/independent pronouns right-dislocated definite NPs neutral-ordered definite NPs left-dislocated definite NPs contrastive topicalization cleft/focus constructions referential indefinite NPs Most discontinuous/inaccessible topics

Llorenç Comajoan 57

She found that verbal agreement (that is, zero) and stressed pronouns were the most continuous devices, whereas indefinite existentials (‘there is X’) and definite NPs modified by a relative clause were the most discontinuous. As far as the distribution of the different nominal devices, the data from Bentivoglio (1983) showed that zero was the most common device (40%), followed by NPs (37%) and pronouns (23%). In a further study, Bentivoglio (1993) studied Caracas Spanish and found that NPs had different rates of use according to their function. The most common grammatical roles for NPs were locative, object, and subject of an intransitive verb; and the least common was subject of a transitive verb. In addition, most NPs coded inanimate referents, except for subjects of transitive verbs, which were animate in 91% of the cases. Most NPs were used to track referents already introduced into the discourse, 23% were used to code truly new referents, and only 25.5% were used to code human participants. The data from Bentivoglio (1993) included very few human subjects coded by NPs (5% of NPs). Whenever they were used and they coded new referents, they were always (100%) introduced by NPs. These data showed that full NPs can fulfill a variety of roles and that coding for new information was not their major role, except for the case of subjects in transitive verbs, which tended to be human. The episodic approach In a critique of the recency or distance approach used in the studies by Givón and Bentivoglio, Tomlin (1987) argued that this approach had two strengths – generability and simplicity – but that it could not account for two types of counterexamples: NPs that refer to an antecedent found in the previous clause, and pronouns that persist for several clauses (Tomlin, 1987; Fox, 1987). As an alternative to the recency approach, Tomlin (1987: 456) argued for an episodic/paragraph model that ‘considers the alternation between noun and pronoun to be a function of the limited capacity of working memory, which is manifested in the text artifact primarily through its paragraph, or episodic, organization.’ In discussing the strengths and weaknesses of this approach, Tomlin (1987) explained that it was superior to the distance approach, because it could account for noun and pronoun use in different types of text and created a link between language and psychology. Nevertheless, the episodic approach was challenging from a methodological perspective, because it is difficult to define constructs such as paragraph, episode, and focus in texts. In order to test the episodic model, Tomlin (1987) used a set of 21 slide pictures with a story and asked three groups to produce a story in English. Group 1 saw the slides one at a time, group 2 saw the slides in odd pairs, and group 3 saw the slides in even pairs.4 The rationale behind the experiment was that the differences in the pairing of the slides would produce boundaries of episodes that would trigger differences in the use of nouns and pronouns. Thus, if participants used nouns after a boundary (a new pair of

58 Continuity and Episodic Structure in Spanish Subject Reference

slides), regardless of the distance from the previous mention, it would be considered evidence for the episodic approach; but if participants used nouns and pronouns to refer to previous mention, regardless of the boundaries, it would be considered evidence for the recency approach. The results indicated that all three groups produced similar measures of discourse density (proportion of propositions to episodes) and episode boundary: NPs were used after an episode boundary and pronouns within an episode boundary. In a second experiment, 10 participants watched a videotaped cartoon and produced an online oral description. The data showed that the participants produced more propositions per clause than in the slide experiment, but the proportion of NPs after an episode boundary and pronouns within an episode was the same as for the slide experiment. In sum, Tomlin (1987) provided evidence for the episodic approach, but this study still needs to be replicated in other languages and types of narratives. The current integrates research in switch reference and topic accessibility in order to investigate (a) the use and function of Spanish NPs, pronouns, and zero in Peninsular Spanish, and (b) to what extent the data provide evidence for the distance or episodic approaches.

The study Materials Spanish narratives were elicited using The Pear Film from Chafe (1980), a seven-minute video-film used for studies in cognitive, cultural, and linguistic aspects of narrative production (see Appendix A for a summary of the plot). The data in this study come from the analysis of approximately half of the story narrated in the film: from the beginning of the story until the end of the fourth paragraph in the summary of Appendix A (when the boy riding his bike falls and the pears spill out of the basket onto the ground). Participants The 16 participants (11 female, 5 male; average age 29.5) in this study were native speakers of Peninsular Spanish varieties that do not neutralize verbal inflection (Barcelona, Valencia, Mallorca, Madrid, Bilbao, Burgos, Salamanca, Guipúzcoa, Ciudad Real, Palencia, and Valladolid). All participants were students at a United States university at the time of data collection. Their average length of stay in the United States was three years. Procedure The participants were informed that they would participate in a study about the different ways people told stories. They watched the complete film and were asked to retell the story as if they were telling it to somebody who had not seen it. It was stressed that they should tell it in as much detail as

Llorenç Comajoan 59

possible. For the recording of the data, the researcher provided the participants with a tape recorder and asked them to retell the story in a separate room by themselves. Coding categories Each narrative was divided into clauses that contained one verbal form. Only clauses with human subject referents in which one of the variables (NP, pronoun, or zero) could be inserted were considered (for example, direct object relative clauses were analysed, but subject relative clauses and clauses with se le were coded in a separate category). Infinitives in modal constructions were analysed as part of the matrix clause. Constructions with the verb parecer ‘to seem’ were not considered, because it was not clear whether their subject was a human referent or an expletive subject. Verb phrases containing comments from the participants were not considered. Each clause was coded into the following categories: referent, introduction of referents, linguistic form, referential distance, and episode boundary. The human referents of the film were the following: man picking up pears, man with the goat, boy, and girl. Each subject was coded as ‘new’ if it was the first mention of the referent. If it was not the first mention, it was coded as ‘different subject’ or ‘same subject’ depending on whether the previous clause had the same or a different subject. For the coding of the linguistic form of subjects, the following categories were used: definite NP, indefinite NP, pronoun, and zero. The category ‘Definite NPs’ included definite NPs, demonstrative adjectives, and otro/a ‘another’. The category ‘Other’ included a few other devices (for example, gerunds, pronominal demonstratives, and los dos ‘both’). Relative clauses were coded as direct object relative clauses (with an NP, pronoun, or zero subject) or relative clauses (with a definite or indefinite NP subject antecedent). Finally, clauses with hay ‘there is/are’ that referred to human subjects were coded as hay-clauses, and clauses with ergative constructions (with se [le]) (for example, se le cae el sombrero ‘the hat falls [off his head]’) were coded as se (le) clauses. For the analysis of the distance approach, the data were coded for referential distance. Referential distance (or look-back) measured how many clauses intervened between one referent and its last mention. Following Givón (1983a, c), new referents were given a measure of 20, and old referents were given a number ranging from 1 (mentioned in the previous clause) up to 20 (the highest number given in this study was 13). Thus, referential distance measured switch/same reference when the look-back value was 1 and a pronoun, an NP, or zero was used. The look-back measure referred to mention of the referent regardless of whether the previous mention was in subject position or not. The procedure for coding episodic boundary and events closely followed Tomlin (1984, 1987). Tomlin defined an event as ‘the total action occurring

60 Continuity and Episodic Structure in Spanish Subject Reference

between event boundaries’ (1984: 121) and an event boundary as ‘the loci of abrupt visual change within the video’ (1984: 121). Two main types of abrupt visual change can occur in the video segment used for this study: a camera cut and the loss or gain of characters. The analysis identified 28 events (camera cuts) and six episodes (see Appendix B). The relationship between episode boundary and linguistic form (NP, pronoun, or zero) was coded using a system of hits and misses (following Tomlin, 1987). A hit was defined as an NP at the beginning of an episode or the use of a pronoun/zero inside an episode. A miss was defined as an NP inside an episode or a pronoun/zero at the beginning of an episode. Clauses that could not be readily classified into one of the events and episodes of the film were excluded from the percentage analysis of hits and misses and were coded as ‘out’ cases. The excluded clauses were instances of extra information that was not in the event/episode analysis of the film (Appendix B), events that were not retold in the order in which they occurred in the film (for example, descriptions of characters after first mention), clauses with se le, and subject relative clauses.

Results Nominal devices for new and old referents The most common device to code human subject referents in the narratives was zero (68.5%), followed by NPs (29%) and pronouns (2.5%) (Table 3.4).

Table 3.4 Subject nominal devices used in the narratives Participant 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Total

NP 4 7 3 9 6 9 13 5 10 7 4 8 6 12 8 6 117 (29%)

Pronoun 0 0 2 0 1 0 0 0 0 4 1 0 0 1 1 0 10 (2.5%)

Zero 15 8 11 14 16 32 23 9 18 18 19 15 23 24 14 16 275 (68.5%)

Llorenç Comajoan 61 Table 3.5 Nominal devices to introduce new characters Participant 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Total

NP indef.

NP def.

Zero

Pronoun

0 0 0 1 0 0 0 0 0 0 0 0 0 1 0 0

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

2 (3.5%)

0 (0%)

0 (0%)

3 1 2 1 4 1 2 2 2 1 3 4 1 2 3 2 34 (56.5%)

RC NP indef. 0 1 1 1 0 0 0 2 1 1 0 0 2 1 0 0 10 (16.5%)

RC NP def.

hay NP indef.

Other

0 0 1 0 0 0 1 0 1 1 1 0 0 0 0 1

0 1 0 1 0 2 0 0 0 1 0 0 0 0 0 1

1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0

6 (10%)

6 (10%)

2 (3.5%)

Table 3.6 Nominal devices to introduce same subjects from previous clauses Participant

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Total

NP indef.

NP def.

0 0 0 0 0 0 0 0 1 0 0 0 1 0 0 0

0 1 1 1 0 3 4 0 2 0 0 1 2 5 3 3

2 (1%)

26 (10%)

Zero

12 4 9 12 12 24 14 6 12 15 10 9 19 20 12 13 203 (78.5%)

Pronoun

0 0 1 0 0 0 0 0 0 3 0 0 0 0 1 0 5 (2%)

RC NP indef.

RC NP def.

(se) le

0 0 0 0 1 0 2 0 1 3 0 1 1 0 1 1

1 0 0 0 2 0 1 0 0 0 0 0 0 0 0 0

0 0 0 0 0 2 1 1 0 1 1 0 0 0 1 1

11 (4%)

4 (1.5%)

8 (3%)

For the introduction of new characters, indefinite NPs were the most common device (56.5%), followed by relative clauses (RC) with indefinite NPs (16.5%), and hay-clauses (‘there is/are’-clauses) with indefinite NPs (10%). New referents were never coded by zero or pronouns (Table 3.5). The most frequent device to code subject continuity (same subject as previous clause) was zero (78.5%). Pronouns (2%) and indefinite NPs (1%) were barely used (Table 3.6).

62 Continuity and Episodic Structure in Spanish Subject Reference Table 3.7 Nominal devices to introduce different subjects from previous clauses Participant 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Total

NP def.

NP indef.

Zero

Pronoun

RC NP indef.

RC NP def.

(se) le

Other

1 4 0 6 2 6 7 3 4 6 1 3 2 4 2 1

0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0

3 4 2 2 4 8 9 3 6 3 9 6 4 4 2 3

0 0 1 0 1 0 0 0 0 1 1 0 0 1 0 0

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

1 0 0 0 1 0 0 0 1 0 0 0 0 1 0 0

1 0 0 0 0 0 1 1 2 0 0 1 0 0 0 0

0 0 0 0 0 1 0 0 1 0 0 0 0 0 0 0

52 (36.5%)

1 (1%)

72 (51%)

5 (3.5%)

0 (0%)

4 (3%)

6 (4%)

2 (1.5%)

Subject discontinuity (reference to a different subject from the previous clause) was marked with zero (51%) and definite NPs (36.5%). The percentage rate of pronouns for discontinuity was very low (3.5%) (Table 3.7). These results confirm the different functions fulfilled by nominal devices in Spanish narratives: Indefinite NPs mostly code new subjects, zero tends to code subject continuity, and definite NPs and zero code subject discontinuity. Subject pronouns were infrequent in the Spanish narratives in this study, and their function is discussed in the following section. The recency approach versus the episodic approach The coding of discontinuity in this study was along the following continuum: zero (51%)  definite NP (36.5%)  pronoun (3.5%) (Table 3.6). The high percentage of zero to refer to a different subject from the previous clause is contrary to what in the distance approach predicts, because zero is the most continuous device. However, the percentage rates of zero to code continuity (78.5%) and discontinuity (51%) confirm that – although zero can be used to refer to different subjects from the previous clauses – its main function is to code continuity. An analysis of the look-back measure shows the difference between the use of zero for same and different subjects from the previous clause: the average look-back measure of zero when it referred to the same subject (78.5%) was 1, whereas it was 2.5 when it referred to different subjects (4.5 for NP; 3 for pronoun) (Table 3.8). The analysis of hits and misses relating episode boundaries and the use of NP, pronoun, and zero showed that 81.5% of the data were hits and could be accounted by an episodic approach (NPs at the beginning of an episode, pronoun/zero inside an episode) (Table 3.9).

63 Table 3.8 Look-back distance for definite NPs, pronoun, and zero when they refer to a different subject from the previous clause Participant

Def. NP

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

1 2, 5, 7, 1

Average

4.5

Pronoun

3 2, 2, 3, 4, 4, 5 3, 2 2, 2, 2, 11, 7 1, 2, 5, 12, 3, 4, 5 3, 4, 4 2, 4, 13, 8 2, 2, 10, 11, 3, 33 2 4, 3, 2 4, 11 2, 3, 6, 7 6, 6 5

5

3

2

3

Zero 2, 4, 4 2, 3, 3, 3 2, 3 2, 2 1, 2, 2, 3 1, 2, 2, 2, 3, 4, 7, 9, 3, 1 2, 2, 2, 2, 2, 2, 2, 3, 3 1, 2, 3 1, 2, 2, 2, 2, 3 1, 1, 2 2, 2, 2, 3, 3, 4, 10, 2, 2 2, 2, 3, 3, 3, 8 2, 2, 2, 2 1, 1, 3, 3 2, 4 1, 2, 3 2.5

Note: 1 indicates that the referent was mentioned in the previous clause, 2 indicates that it was mentioned two clauses back, and so on. The look-back measure refers to mention of the referent regardless of whether the previous mention was in subject position or not.

Table 3.9 Overall hits and misses for the episodic model Participant 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Total

Hit1

Hit2

18 11 11 19 18 32 26 11 18 19 19 14 25 28 17 16

15 10 9 17 14 32 24 9 16 18 16 10 24 25 14 14

Miss 1 3 5 4 4 9 10 0 6 3 1 2 3 7 5 5

302 (81.5%)

–

68 (18.5%) Hit 1

–

267 (80%)

68 (20%) Hit 2

64 Continuity and Episodic Structure in Spanish Subject Reference Table 3.10 Hits and misses for new referents, old referents, and different subjects Hit

Miss

Out

New referents NP indef. Zero

33 0

(100%) (0%)

0 0

(0%) (0%)

1 0

Old referents Same subject NP def. Zero

3 179

(1.5%) (82%)

22 14

(10%) (6.5%)

1 10

19 57

(19%) (57%)

19 6

(19%) (6%)

14 9

Different subject NP def. Zero

New referents were always introduced by indefinite NPs at the beginning of episodes (33 tokens, 100%), In order to avoid a possible bias in favour of the episodic approach, two analyses of hits and misses were calculated. Hit 1 included all the subject referents, whereas Hit 2 excluded new referents. Overall, the difference in the misses between the two counts was less than 2 percentage points (18.5% based on Hit 1, 20% based on Hit 2). Table 3.10 provides an analysis of hits and misses for new and old referents. Overall, the data show that the episodic model can account for most uses of NPs and zero in the Spanish narratives. New referents were always coded by indefinite NPs and were all at the beginning of episodes. Regarding old referents, zero was mostly used inside episodes. Specifically, the use of zero inside episodes represented 82% of the data for subject continuity and 57% for subject discontinuity. The largest percentage of misses in the episodic model occurred in the coding of discontinuity and old referents when they were coded by definite NPs (19%).

Discussion This section discusses the data from this study in light of previous research and focuses on three aspects: the use of NP, pronouns, and zero; the evidence for the recency and episodic approaches; and counterexamples to the episodic model. In the narrative data from this study, human referents were coded by zero or NPs (68.5% and 29%, respectively), and only in a few instances by pronouns (2.5%). Overall, the participants produced more NPs and fewer pronouns than previously reported (Tables 3.1 and 3.2). This difference may be due to the fact that the data in previous studies were conversational and included more than one participant, whereas in the current study participants produced monologic retellings of The Pear Film. The presence of an interlocutor in the retellings is very likely to influence the production of

Llorenç Comajoan 65

NPs, pronouns and zero. For instance, Tannen (1980) compared Greek and English narratives of The Pear Film and found differences in the use of nominal devices in the languages in part because the Greek participants tended to assume that the researcher collecting the data had already heard the narrative several times. However, despite differences in the methodology (but using the same film), Blackwell’s (1994) Spanish data on pronoun use in subject position (5-6%) were similar to those of the current study (3.5%). In this study, the nominal devices that introduced new characters were indefinite (indefinite NPs, relative clauses with indefinite NPs, and hayclauses with indefinite NPs). These results confirm Bentivoglio’s (1983) claim that the most discontinuous devices for human subjects were indefinite existentials and presentatives. Bentivoglio’s analysis did not classify relative clauses with indefinite NPs as a separate category, but her data showed that the most continuous device after indefinite existentials and presentatives were definite NPs modified by a relative clause. The strong relationship between indefiniteness and introduction of new characters also supports Givón’s (1983a) continuity hierarchy and Levinson’s (1987) extragrammatical pragmatic principles (Levinson, 1987), whereby The speaker takes into account the hearer’s knowledge and provides statements that are informationally appropriate to the situation. The limited number of subject pronouns (10 tokens) produced by the participants in this study allows for a close examination of their use. The context in which the subject pronouns were used had two characteristics in common: the pronouns referred to the boy in the narrative (8 of the 10 instances), and they occurred in episodes 5 and 6 (see Appendix B) (8 of the 10 instances). The boy was the main character in the story, which can account for the use of pronouns to focus on this character. At first, it could be argued that the subject pronouns in episodes 5 and 6 were used to disambiguate; because in those episodes the boy and the girl are riding their bicycles, and, when they pass each other from opposite directions, the boy loses his hat and falls. In the retelling of this episode, the use of a pronoun may specify who the referent for each verb is. For instance, in (3.2a and b) the subject pronoun él ‘he’ disambiguates in two occasions the referent who loses the hat and falls: (3.2)

Participant 3 (2 tokens)5 el niño recoge una de las banastas de de peras the boy takes one of the baskets of pears y Ø se las lleva con él en la bicicleta and Ø takes them with him on the bike ah mientras Ø se lleva las peras ah while Ø is taking the pears Ø se cruza con otra niña Ø passes another girl que viene en bicicleta y al cruzarse, en un camino que es más bien estrecho,

66 Continuity and Episodic Structure in Spanish Subject Reference

who is coming on a bike and when passing, in a road that is narrow, (a) ➔él pierde el sombrero he loses his hat Ø se despista mirando hacia atrás Ø distracts himself looking back y Ø se cae and Ø falls (b) ➔y y como consecuencia cae él de la bici, las peras and and as a consequence he falls from the bike, the pears y su sombrero se pierde and his hat is lost y ah además Ø se hace daño en una pierna. and ah in addition Ø hurts himself in a leg. In (3.3), the subject pronoun disambiguates the referent of the person who is distracted: (3.3)

Participant 5 (1 token) el niño este va por ahí con la bicicleta con el cesto el canasto este de de peras en su bicicleta de manera que es bastante dificultosa ¿no? porque pesan y todo esto this boy goes around there on the bike with the basket this basket of pears on his bike in a way that is pretty difficult, right? because they weigh and all this pasa una niña en una en otra bicicleta a girl passes by on another bike Ø se cruza con él Ø passes him ➔él se distrae he distracts himself Ø choca contra una piedra la bicicleta Ø hits the bike against a stone y Ø se y se cae. and Ø falls.

Participant 10 produced 4 subject pronouns, but only one of the tokens had a disambiguation function. The pronoun in (3.4d) makes it clear that the boy was the referent who got off the bike, not the girl: (3.4)

Participant 10 (4 tokens) hay un señor que está subido a una escalera de mano frente a un árbol

Llorenç Comajoan 67

there is a man who is at the top of a ladder in front of a tree (a) ➔él está cogiendo fruta verde, unas peras he is picking up green fruit, pears y Ø lo hace de una forma muy artesanal muy calmado así como una a una and Ø does it in a very careful way very calmly like one by one eh nuestro protagonista vuelve a subir al árbol our protagonist climbs up the tree again y Ø sigue haciendo su tarea and Ø keeps doing his task Ø recoge una a una las piezas las frutas Ø picks up the pieces of fruit one by one (b) ➔mientras él está subido while he is up hay una hay una secuencia que there is a scene that eh un niño vestido así como un poco tipo boy scout pasa con una bici ah a boy dressed like a boy scout goes by on a bike y el supuesto boy scout, por llamar de alguna forma, no sé, tendrá también unos doce años and the supposed-to-be boy scout, to give him some name, I don’t know, must also be twelve years old. (c) ➔ah el al cruzarse él se queda mirándola ah when passing he looks at her y el sombrero bueno Ø se chocan bueno no se chocan and the hat well Ø crash into each other well they do not crash into each other eh Ø gira la cabeza ah Ø turns his head around y se le escapa el sombrero and the hat flies from him y nada Ø se tropieza con una piedra and Ø hits a stone y Ø se cae and Ø falls la chica yo creo que ni se da cuenta the girl I think does not realize (d) ➔pero él tiene que bajarse y demás. but he has to get off and all that.

68 Continuity and Episodic Structure in Spanish Subject Reference

Finally, in (3.5), the pronoun in él mismo ‘he himself’ disambiguates the reference regarding which participant falls (the boy): (3.5)

Participant 15 (1 token) entonces en esto Ø se cruza con una chica then Ø passes a girl y al mirarla se le cae el sombrero and when looking at her his hat falls ➔y él mismo va detrás de las peras. and he goes behind the pears (falls).

However, not all pronouns have a disambiguating function. For instance, the pronouns in (3.6) and (3.7) do not disambiguate, because other contextual information makes it clear who the referent is. In (3.6), the feminine direct object pronoun la ‘her’ unambiguously refers to the girl; and in (3.7), the girl has not been introduced yet: (3.6)

Participant 11 (1 token) y en esto que en el camino ah en pues en la otra dirección viene una chica and in all this on the road ah then in the other direction a girl is coming ➔él se la queda mirando he stands there looking at her y la chica pues le da le quita al chico al niño el sombrero que Ø llevaba. and the girl then takes the hat Ø was wearing.

(3.7)

Participant 14 (1 token) el niño también lleva un sombrero porque hace mucho sol the boy is also wearing a hat because it is very hot ➔en el camino cuando él va con la bicicleta on the road when he is on his bike se cruza con una niña de largas trenzas passes a girl with long braids que también va en bicicleta. who is also on a bike.

The subject pronouns in (3.6) and (3.7) show that the participants used subject pronouns in contexts that are referentially ambiguous (two referents performing similar actions in similar situations) even though the contexts are not linguistically ambiguous. Thus, a subject pronoun may anticipate ambiguity and provide extra information to track reference of the participants.

Llorenç Comajoan 69

Finally, in three instances (3.4a, 3.4b, and 3.4c), subject pronouns were used in nonambiguous contexts and did not refer to the boy in the narrative. All three tokens were in subordinate clauses and were produced by the same speaker. Only 6 of the 16 participants produced subject pronouns in their narratives to refer to the boy in episodes 5 and 6. In the same episodes, the remainder of the participants used NPs, non-subject pronouns, or zero (Table 3.11). In sum, these results show that in the Spanish data for this study the expression of nonnew human subject referents was expressed through devices along the following continuum: zero  NP  pronoun. The ordering of the three devices does not follow Givón’s (1983) topic-accessibility hierarchy (Table 3.3). This may be due to differences in the use of pronouns in different languages. For instance, Clancy (1980) compared English and Japanese data using The Pear Film and found that the Japanese participants produced more NPs and zero than the English participants (Table 3.12). The proportion of use of zero in Japanese and Spanish was similar (73.2% and 78.5%, respectively). Clancy (1980) argued that the difference in the use

Table 3.11 Nominal devices in episodes 5 and 6 Participant

Nominal device

1 2 4 6 7 8 9 12 13 16

se le (unstressed dative pronoun) NP NP NP NP le (unstressed dative pronoun) se le (unstressed dative pronoun) NP NP zero

Table 3.12 Nominal devices to introduce continuous subjects in Spanish, English, and Japanese English

Japanese

Spanish (this study)

NP

15.7%

26.8%

Pronoun Zero

63.8% 20.5%

– 73.2%

10% (def. NP) 1% (indef. NP) 2% 78.5%

Note: The Japanese and English data are from Clancy (1980).

70 Continuity and Episodic Structure in Spanish Subject Reference

of NPs between English and Japanese was probably due to the fact that Japanese only has two devices for continuity, whereas English has three. Spanish has as many devices as English, but the Spanish data on the use of NPs and zero were closer to the Japanese than the English data. These results indicate that similar devices may fulfill distinct functions in different languages. Thus, pronouns in English can be used anaphorically (in low referential distance) or to mark contrast (in higher referential distance) (Givón, 1983c); whereas in Spanish the anaphoric function is carried out by zero and to a lesser extent by pronouns. Further research needs to compare data elicited through similar tasks in order to replicate these results and study the specific functions of nominal devices within narratives. Further research needs to compare data using the same elicitation task in order to replicate these results and investigate how the same devices in different languages fulfill different functions. The data from this study can shed light on the recency and episodic approaches to noun referentiality. Research following the recency approach found that zero and pronouns were the most continuous devices, followed by NPs of different kinds (Bentivoglio, 1983, table 3.3). The general results for the current study do not provide support for the recency approach, because the data showed that NPs were more continuous than pronouns, and zero was more discontinuous than NPs. Specifically, old referents were mostly marked by zero (78.5%), followed by definite NPs (10%) and pronouns (2%) (Table 3.6); and zero marked both continuity (78.5%) and discontinuity (51%). Table 3.8 showed that the average look-back distance was 2.5 clauses for zero, 3 clauses for pronouns, and 4.5 for definite NPs. This pattern supports the results of the distance approach except that a shorter distance might be expected for zero. The fact that zero persisted for an average of 2.5 clauses suggests that distance alone cannot explain the use of nominal devices, because zero continued to be used even when discontinuity (produced by a new referent, a different human referent, a different nonhuman referent, or a comment from the speaker) occurred. Cameron (1995) referred to reference chains for cases in which same-reference devices (for example, zero) persisted over intervening switch-reference subjects, and he found a gradation in the use of subject pronouns in chains that extended two referents back (higher possibility of pronoun use when the preceding referents were also switch reference in regard to each other and the target). Cameron’s (1995) reference chains capture the limitations of the distance approach but do not explain what triggers the use of a specific nominal device at the end of the chain. A possible explanation is that the nominal devices persist or decay depending on the episodic structure of the text. The analysis of hits and misses in Table 3.10 provided evidence for an episodic model in which pronouns and zero are used inside episodes, and NPs are used to begin new episodes. The percentage values of the episodic model in the current study (81.5% hits, 18.5% misses) were similar to the values in Tomlin (1987) (84% hits and 16% misses). The remaining 16% of the

Llorenç Comajoan 71

data in Tomlin (1987) were counterexamples to the episode model and were intraepisode uses of NPs or uses of pronouns at the beginning of new episodes. The intraepisode uses of NPs were found in evaluations, in the performance of one individual speaker, and in cases of ambiguity resolution that were supposed to be filtered out but were not. In this study, no exceptions to the introduction of new referents were found, because they were always introduced with indefinite NPs, relative clauses, or hay-clauses and never by pronouns or zero. Regarding the introduction of same subjects, the intraepisode uses of definite NPs (10%) can be grouped under three categories: 1 Most intraepisode uses of definite NPs were descriptions of the characters, as in (3.8): (3.8)

Participant 15 en la historia aparece un hombre in the story a man appears que está recogiendo peras de un árbol who is picking up pears ➔el señor tiene bigote the man has a moustache Ø tiene un pañuelo en el cuello rojo y un delantal Ø has a red handkerchief around the neck and an apron entonces el hombre se afana en coger muchas peras Then the man labors at picking up many pears que Ø las va bajando del árbol por una escalera. that Ø is bringing them down the tree with a ladder.

2 Three instances of intraepisode definite NPs were provided by two participants, who used them with a demonstrative (este, ‘this’), as in 3.9: (3.9)

Participant 7 ➔y ah mientras este campesino sigue recogiendo la fruta and ah while this peasant keeps picking up fruit viene un niño con un sombrero en la cabeza a boy comes with a hat on his head ➔viene este niño en bicicleta this boy comes on a bike Ø es un niño Ø is a boy que es más pequeño que la bicicleta who is smaller than the bike prácticamente Ø no puede ni ir en su bicicleta ¿no? practically Ø cannot ride on his bike, right?

72 Continuity and Episodic Structure in Spanish Subject Reference

pero de lo pequeño que Ø es comparado con la bici but Ø is so small compared to the bike ➔ah entonces este este niño que debe tener ¿qué sé yo? unos ah siete u ocho años rubito muy delgadito ah se se da cuenta del enorme cesto de peras. ah then this this boy who must be what do I know? about seven or eight years old blond very skinny ah notices the big basket of pears. 3 Another instance of a miss is when one of the characters is part of a collective subject, as in (3.10): (3.10)

Participant 7 entonces el niño las peras y la bicicleta se van todos al suelo. then the boy the pears and the bike go to the ground (fall).

Finally, in a few instances, intraepisode uses of definite NPs did not have a clear-cut explanation, as in (3.11): (3.11)

Participant 15 entonces el hombre se afana en coger muchas peras then the man labors at picking up many pears que Ø las va bajando del árbol por una escalera that Ø brings them down using a ladder Ø las va dejando abajo Ø is bringing them down ➔entonces cuando el hombre recoge un capazo then when the man takes a basket Ø vuelve a subir arriba Ø goes back on the tree y en esto Ø sigue recogiendo peras ¿no? and Ø keeps picking up pears, right?

Only 6% of the uses of zero that referred to the same subject as the previous clause occurred at the beginning of an episode (Table 3.10). A few of the counterexamples occurred in subordinate clauses that referred to a new episode, whose main clause was part of the previous episode (example 3.12). The others did not have a clear explanation (example 3.13): (3.12)

Participant 3 el niño recoge una de las banastas de de peras the boy picks up one of the baskets of pears y Ø se las lleva con él en la bicicleta and Ø takes them with him on the bike ah mientras Ø se lleva las peras

Llorenç Comajoan 73

ah while Ø is taking the pears ➔ Ø se cruza con otra niña Ø passes another girl que viene en bicicleta who is coming on a bike. (3.13)

Participant 11 y la chica pues le da le quita al chico al niño el sombrero and the girl then hits takes from the boy the hat que Ø llevaba that Ø was wearing y entonces en esto de que quitar estar mirándole a ella pues Ø no se da cuenta bien de la carretera and then in all this looking at her then Ø does not realize about the road ➔y en una y ah Ø tropieza eh con una piedra. and Ø hits a stone.

The misses for definite NPs were instances of definite NPs inside an episode (19%). These counterexamples could be classified into two categories: 1 A definite NP to refer to a different subject within an episode usually occurred after a comment from the speaker or a description of a referent: (3.14)

Participant 3 después aparece un niño en una bicicleta también con sombrero por cierto then a boy appears on a bike also with a hat by the way y Ø lleva un pañuelo puesto ah como un cowboy a la manera de cowboy and Ø is wearing a handkerchief like a cowboy, cowboy style ➔el niño recoge una de las banastas de de peras the boy picks up one of the baskets y Ø se las lleva con él en la bicicleta and Ø takes the with him on his bike.

2 Definite NPs to refer to a different subject within an episode were also used to disambiguate (as previously discussed): (3.15)

Participant 4 cuando el niño iba en la bicicleta when the boy was on his bike Ø encontró por el camino otra niña, una niña con dos trenzas, muy guapa,

74 Continuity and Episodic Structure in Spanish Subject Reference

Ø found on the road another girl, a girl with two braids, very beautiful, que venía en la contrar en la dirección contraria who was coming on the opposite direction ➔y el niño llevaba un sombrero de paja en su cabeza para protegerse del sol and the boy was wearing a straw hat on his head to protect himself from the sun y la niña cuando pasó delante suyo and the girl when passed in front of him Ø se lo quitó Ø took it from him Ø se lo tiró Ø pulled it from him ➔el niño, sorprendido, miró hacia atrás the boy, surprised, looked towards his back y de repente Ø se cayó de la bicicleta con las peras en el suelo and suddenly Ø fell from his bike with the pears on the ground. The misses (6%) in the introduction of different subjects by zero were due to the methodology of the study. Once the episodes and events were identified in the film, the same division of events and episodes was applied to all narratives. Most participants provided more than one clause for each episode, but in a few cases only one clause was provided. In order to be consistant in the coding of the data, these cases were coded as individual episodes, even though from the perspective of the speaker they were likely to be part of the previous episode. To summarize, the analysis of hits and misses supports the episodic model that argues for the use of NP at the beginning of an episode and zero/pronouns inside the episode. The model cannot account for all the data, but most counterexamples can be grouped under meaningful categories (for example, NPs within an episode to describe a character).

Conclusion This study has shown that zero and NPs were the two most common nominal devices to refer to subjects in Spanish narratives. Subject pronouns were used infrequently and mostly in ambiguous contexts. Reference to same and different subjects from the previous clause was expressed by zero. The high use of zero for reference to same and different subjects can be considered

Llorenç Comajoan 75

counterevidence to the distance approach (Givón, 1983 a, b, c; Bentivoglio, 1983). The results indicated that a high percentage of zero and definite NPs (80%) can be explained by a model that has the segmentation of narratives into episodes and events as the main variable (Tomlin, 1987). The remaining 20% of the data was explained by some exceptions and other more specialized variables that interacted with episodes and events, such as descriptions of characters, comments from the speaker, subordination, and disambiguation. These findings add to the growing body of research in nominal reference in Spanish. Specifically, they provide evidence for the role of episodes from a cognitive and linguistic perspective. This episodic model overcomes the main weaknesses of the recency approach, namely the use of NPs for antecedents found in the previous clause and the use of zero/pronouns that persist for several clauses. However, further research needs to investigate the episodic model in order to replicate the results found in this study and refine the model. First, future studies need to apply the episodic model in a variety of types of discourse to explore how conversations with several interlocutors differ from retellings produced as monologues. Second, further studies also need to explore the processing mechanisms and narrative strategies that make some speakers use pronouns and others use different nominal devices (only six of the sixteen participants used subject pronouns). Third, the results showed that new characters were introduced by indefinite NPs (isolated or in relative clauses) and hay-clauses, which is evidence for the role of clause structure and syntax in the introduction of new characters. Thus, further research needs to go beyond the study of NP, zero, and pronouns to examine how syntax interacts with nominal reference. Finally, a complete episodic model must incorporate other factors that have been found to correlate with the use of NPs, pronouns, and zero (such as verbal inflection and regional/social variation) in order to ascertain whether episodes and scenes can subsume other factors.

Appendix A: Summary of The Pear Film (from Chafe, 1980: xiii–xiv) The film begins with a man picking up pears on a ladder in a tree. He descends the ladder, kneels, and dumps the pears from the pocket of an apron he is wearing into one of three baskets below the tree. He removes a bandana from around his neck and wipes off one of the pears. Then he returns to the ladder and climbs back into the tree. Toward the end of this sequence we hear the sound of a goat, and when the picker is back in the tree a man approaches with a goat on a leash. As they pass by the baskets of pears, the goat strains toward them, but is pulled past by the man and the two of them disappear in the distance. We see another closeup of the picker at his work, and then we see a boy approaching on a bicycle. He coasts in toward the baskets, stops, gets off his bike, looks up at the

76 Continuity and Episodic Structure in Spanish Subject Reference picker, puts down his bike, walks toward the baskets, again looking at the picker, and lifts up a basket full of pears. He puts the basket down near his bike, lifts up the bike and straddles it, picks up the basket and places it on the rack in front of his handlebars, and rides off. We again see the man continuing to pick pears. The boy is now riding down the road, and we see a pear fall from the basket on his bike. Then we see a girl on a bicycle approaching from the other direction. As they pass, the boy turns to look at the girl, his hat flies off, and the front wheel of his bike hits a rock. The bike falls over, the basket falls off, and the pears spill out onto the ground. The boy extricates himself from under the bike, and brushes off his leg. In the meantime we hear what turns out to be the sound of a paddleball, and then we see three boys standing there, looking at the bike boy on the ground. The three pick up the scattered pears and put them back in the basket. The bike boy sets his bike upright, and two of the other boys lift the basket of pears back onto it. The bike boy begins walking his bike in the direction he was going, while the other three boys begin walking off in the other direction. As they walk by the bike boy’s hat on the road, the boy with the paddleball sees it, picks it up, turns around, and we hear a loud whistle as he signals to the bike boy. The bike boy stops, takes three pears out of the basket, and holds them out as the other boy approaches with the hat. They exchange the pears and the hat, and the bike boy keeps going while the boy with the paddleball runs back to his two companions, to each of whom he hands a pear. They continue on, eating their pears. The scene now changes back to the tree, where we see the picker again descending the ladder. He looks at the two baskets, where earlier there were three, points them, backs up against the ladder, shakes his head, and tips up his hat. The three boys are now seen approaching, eating their pears. The picker watches them pass by, and they walk off into the distance.

Appendix B: Division of the Pear Story into events and episodes

Ev. Ep. Description

Character

1

1

scene-setting, countryside

scene-setting, intro. of man picking pears

2 3 4

1 1 1

man picking up pears in the tree one pear falls to the ground man goes down the ladder, empties pears from the pocket of an apron, and cleans one pear

5

2

6

2

one man comes from the left side, the other man climbs up the ladder, the man has a goat the man and the goat go by the tree and leave

Sec. Sec./Ep. 3

50

5 2 40

intro. of man 25 with goat 6

31

Llorenç Comajoan 77

7 1

man picking up pears in the tree

man picking more pears

3

8 1 9 1 10 1

closeup of a hand picking a pear man picking up pears in the tree face of man picking pears in the tree

1 1 3

11 3

a boy riding a bike can be seen from a boy distance man picking up pears in the tree the boy riding a bike is getting closer the boy bikes closer to the tree and gets off his bike the boy touches a basket full of pears and looks at the man up in the tree man picking up pears in the tree the boy looks up and picks a basket, gets on his bike and puts the basket in front of the bike, and leaves man picking up pears in the tree boy riding his bike with difficulty closeup of pears in the basket in front of the bike boy riding his bike

2

12 3 13 3 14 3 15 3 16 3 17 3

18 19 20 21

3 3 3 3

22 4 23 4 24 5 25 5

26 6 27 6 28 6

a girl riding a bike can be seen from a distance boy riding his bike

girl

boy and girl riding bikes from opposite directions, they are going to pass each other closeup of boy, girl passes him, hat flies, and boy notices his hat is off

boy and girl

closeup of a stone, the boy’s bike hits a stone pears fall to the ground boy on the ground, he feels his injured leg

accident, boy

8

69

2 15 8 3 2 25

3 4 2 3 5

7

2 3

6

3

2

14

4 8 185

185

Note: Ev.  event, Ep.  episode, Sec.  seconds of each event, Sec./Ep.  seconds of each episode.

Notes 1 2

NP refers to full lexical NPs and zero refers to null subjects. Despite the fact that the relative percentage rates seemed to differ in San Juan and Madrid Spanish, VARBRUL weights were not different. Similar rates were found in New York City Puerto Rican Spanish (Flores-Ferrán, 2004, Cameron and FloresFerrán, 2004). Other studies (Cameron, 1992, 1994; Flores-Ferrán, 2004) have confirmed the central role of switch reference in explaining pronoun use in Spanish. 3 It needs to be noted that, in further research, Givón (1992, 1995) moved beyond discourse, relating his previous studies to cognition (attention and episodic memory) and arguing for the discrete rather than scalar properties of topicality (Givón, 1992: 19).

78 Continuity and Episodic Structure in Spanish Subject Reference 4

Group 1 saw each slide separately, whereas groups 2 and 3 saw slides in pairs. That is, group 1 (singles condition) saw slide 1, followed by slide 2, followed by slide 3, and so on. Group 2 (odd condition) saw slide 1, followed by the pair of slides 2 and 3, followed by slides 4 and 5, and so on. Group 3 (even condition) saw slides 1 and 2 together, followed by slides 3 and 4, and so on. 5 Each line of data represents one clause. The nominal devices under discussion are underlined. The symbol ➔ refers to the line under discussion.

References Bayley, R. and L. Pease-Alvarez (1997) ‘Null Pronoun Variation in Mexican-descent Children’s Narrative Discourse’, Language Variation and Change, vol. 9, pp. 349–71. Bentivoglio, P. (1983) ‘Topic Continuity and Discontinuity in Discourse: A Study of Spoken Latin-American Spanish’, in T. Givón (ed.), Topic Continuity in Discourse: A Quantitative Cross-Language Study. Philadelphia: Benjamins, pp. 259–311. —— (1992) ‘Linguistic Correlations between Subjects of One-Argument Verbs and Subjects of More-Than-One-Argument Verbs in Spoken Spanish’, in P. Hirschbühler and K. Koerner (eds), Romance Languages and Modern Linguistic Theory. Philadelphia: Benjamins, pp. 11–24. —— (1993) ‘Full NPs in Spoken Spanish: A Discourse Profile’, in W. Ashby, M. Mithun, G. Perissinotto and E. Raposo (eds), Linguistic Perspectives on the Romance Languages. Philadelphia: Benjamins, pp. 211–24. Blackwell, S. (1994) ‘A Neo-Gricean Pragmatic Approach to Spanish NP Anaphora’, unpublished PhD dissertation, University of Pittsburgh, Pennsylvania. Brown, C. (1983) ‘Topic Continuity in Written English Narrative’, in T. Givón (ed.), Topic Continuity in Discourse: A Quantitative Cross-Language Study. Philadelphia: Benjamins, pp. 317–41. Cameron, R. (1992) ‘Pronominal and Null Subject Variation in Spanish: Constraints, Dialects, and Functional Compensation’, unpublished PhD dissertation, University of Pennsylvania. —— (1994) ‘Switch Reference, Verb Class, and Priming in a Variable Syntax’, Papers from the 30th regional meeting of the Chicago Linguistic Society: Vol. 2: The Parasession on Variation in Linguistic Theory, ed. by K. Beals et al., pp. 27–45. Chicago: Chicago Linguistic Society. —— (1995) ‘The Scope and Limits of Switch Reference as a Constraint on Pronominal Subject Expression’, Hispanic Linguistics, vol. 6/7, pp. 1–27. —— (1996) ‘A Community-Based Test of a Linguistic Hypothesis’, Language in Society, vol. 25, pp. 61–111. Cameron, R. and Flores-Ferrán, N. (2004) ‘Perseveration of Subject Expression across Regional Dialects of Spanish’, Spanish in Context, vol. 1, pp. 41–65. Chafe, W. (1980) The Pear Stories. Norwood, NJ: Ablex. Clancy, P. (1980) ‘Referential Choice in English and Japanese Narrative Discourse’, in W. Chafe (ed.), The Pear Stories. Norwood, NJ: Ablex, pp. 127–202. Enríquez, E. (1984) El pronombre personal sujeto en la lengua española hablada en Madrid. Madrid, Spain: Consejo Superior de Investigaciones Científicas. Flores-Ferrán, N. (2002) Subject Personal Pronouns in Spanish Narratives of Puerto Ricans in New York City: A Sociolinguistic Perspective. Munich, Germany: Lincom Europa. —— (2004) ‘Spanish Subject Personal Pronoun Use in New York City Puerto Ricans: Can We Rest the Case of English Contact?’ Language Variation and Change, vol. 16, pp. 49–73.

Llorenç Comajoan 79 Fox, B. (1987) Discourse Structure and Anaphora. New York: Cambridge University Press. Givón, T. (ed.) (1983a) Introduction. Topic Continuity in Discourse: A Quantitative CrossLanguage study. Philadelphia: Benjamins, pp. 5–41. —— (1983b) ‘Topic Continuity in Discourse: The Functional Domain of Switch Reference’, in J. Haiman and P. Munro (eds), Switch-reference and Universal Grammar. Philadelphia: Benjamins, pp. 51–82. —— (1983c) ‘Topic Continuity in Spoken English’, in T. Givón (ed.), Topic Continuity in Discourse: A Quantitative Cross-Language Study. Philadelphia: Benjamins, pp. 347–63. —— (1992) ‘The Grammar of Referential Coherence as Mental Processing Instructions’, Linguistics, vol. 30, pp. 5–55. —— (2001) Syntax. Vol. I. Philadelphia: Benjamins. Hochberg, J. (1986) ‘Functional Compensation for /s/ Deletion in Puerto Rican Spanish’, Language, vol. 62, pp. 609–621. Levinson, S. (1987) ‘Pragmatics and the Grammar of Anaphora: A Partial Pragmatic Reduction of Binding and Control Phenomena’, Journal of Linguistics, vol. 23, pp. 379–434. Luján, M. (1999) ‘Expresión y omisión del pronombre personal’, in I. Bosque and V. Demonte (eds), Gramática descriptiva de la lengua española: Volume 1. Madrid, Spain: Espasa, pp. 1277–315. Morales, A. (1982) ‘La posición de sujeto en el español de Puerto Rico a la luz de la clase semántica verbal, la oposición tema-rema y el tópico oracional’, Lingüística Española Actual, vol. 4, pp. 23–37. —— (1986) ‘La expresión de sujeto pronominal en el español de Puerto Rico’, in Gramáticas en contacto: Análisis sintácticos sobre el español de Puerto Rico. San Juan, PR: Playor, pp. 89–100. Ranson, D. (1991) ‘Person Marking in the Wake of /s/ Deletion in Andalusian Spanish’, Language Variation and Change, vol. 3, pp. 133–152. Rosengren, P. (1974) Presencia y ausencia de los pronombres personales sujeto en español moderno. Stockholm: Acta Universitatis Gothoburgensis. Silva-Corvalán, C. (1982) ‘Subject Expression and Placement in Mexican-American Spanish’, in J. Amastae and L. Elías-Olivares (eds), Spanish in the United States: Sociolinguistic Aspects. New York: Cambridge University Press, pp. 93–119. —— (1993) ‘On the Permeability of Grammars: Evidence from Spanish and English Contact’, in W. Ashby, M. Mithun, G. Perissinotto and E. Raposo (eds), Linguistic Perspectives on the Romance Languages. Philadelphia: Benjamins, pp. 19–43. —— (1994) Language contact and change: Spanish in Los Angeles. Oxford: Clarendon Press. —— (2001) Sociolingüística y pragmática del español. Washington, DC: Georgetown University Press. Tannen, D. (1980) ‘A Comparative Analysis of Oral Narrative Strategies: Athenian Greek and American English’, in W. Chafe (ed.), The Pear Stories. Norwood, NJ: Ablex, pp. 51–87. Tomlin, R. (1984) ‘The Treatment of Foreground-Background Information in the On-line Descriptive Discourse of Second Language Learners’, Studies in Second Language Acquisition, vol. 6, pp. 115–142. —— (ed.) (1987) ‘Linguistic Reflections of Cognitive Events’, in Coherence and Grounding in Discourse. Philadelphia: Benjamins, pp. 455–79.

4 Gustar-Type Verbs Victoria Vázquez Rozas

Introduction When comparing the verbal structures of English and Spanish, grammatical accounts underscore the differences between English like and Spanish gustar. Despite their closeness in meaning, these predicates exhibit a divergent syntactic behaviour: whereas like codes as subject the entity that experiences a certain feeling, and as object the stimulus responsible for that feeling, gustar expresses the experiencer through an indirect object (or dative) and the stimulus through the subject, illustrated in the examples (4.1) and (4.2). (4.1) (4.2)

I liked the book. Me gustó el libro.

Other verbs that share structural features with gustar appear in the following examples: (4.3) (4.4) (4.5) (4.6)

A Miguel ya no le apetecía jugar al parchís. (TERNURA: 50, 4) ‘Miguel did not feel like playing Parcheesi anymore.’ El dulce les repugnará. (GLENDA: 57, 11) ‘The cake will disgust them.’ Le pareció reconocer la música que sonaba. (CARTA: 185, 2) ‘S/he thought s/he recognized the music that was playing.’ Nunca una mujer le importó como Viviana. (HISTORIAS: 60, 2) ‘Never before had a woman mattered to him as Viviana did.’

In view of these examples, it could be argued that in Spanish there is a particular selection of a syntactic pattern typical of verbs of feeling or emotion. The situation, however, is more complex since the range of the so-called verba sentiendi (that is, psychological verbs) also comprises verbs 80

Victoria Vázquez Rozas 81

that adhere to a transitive syntactic pattern, with a coding similar to that for English like: (4.7) (4.8) (4.9) (4.10)

La había amado en secreto varios años. (CRÓNICA: 11, 1) ‘He had loved her in secret for several years.’ La Mujer Pirata odiaba al abuelo. (TERNURA: 99, 13) ‘The Woman Pirate hated the grandfather.’ Yo detestaba a los hombres altaneros. (CRÓNICA: 35, 6) ‘I detested haughty men.’ Tía Delia creía haber visto a abuela. (SUR: 49, 10) ‘Aunt Delia thought that she had seen Grandma.’

Therefore, the same semantic class of predicates is represented by, at least, two classes of syntactic constructions.1 This chapter presents a study of gustar-type verbs (GTVs) with the Transitivity Hypothesis model advanced by Hopper and Thompson (1980). In the following section we adduce arguments in favour of considering indirect objects (IOs) as part of the core argument structure of GTVs. We will show that constructions containing GTVs exhibit low Transitivity. We then move on to examine the syntax and semantics of the GTVs as compared to transitive constructions with emotion verbs. We argue that there is a semantic distinction between the two structures and offer typological data in support of this claim that suggest the existence of a cognitive basis for the constructional contrast. We then examine the properties of those verbs that fluctuate between the DO and IO constructions. It is found that the alternation between the two structures follows from the predictions made in the Transitivity Hypothesis. We argue, however, that it is difficult to determine at which point a change from one construction to another will take place, given that we are not dealing with categorical rules but rather with statistically significant tendencies.

The Transitivity Hypothesis In Spanish, the most frequent syntactic configuration of biactant predications corresponds to the transitive construction, in which the arguments take on the grammatical relations of subject and direct object. We have taken our examples from the Arthus corpus (Archivo de Textos Hispánicos de la Universidad de Santiago) and the quantitative data provided here have been obtained from the Base de Datos Sintácticos (BDS), based on an analysis of the corpus under Professor Guillermo Rojo’s supervision.2 According to the data from the BDS, of a total of 91,465 clauses with two participants, 68,011 (74.36%) display the subject–direct object pattern. From a semantic viewpoint, the transitive construction prototypically encodes actions initiated by

82

Gustar-Type Verbs

an agent and transferred to a patient. Thus, despite their syntactic configuration, examples (4.7–4.10) do not conform to the transitive protoytpe. With regard to GTVs, their subject–indirect object pattern deviates from the transitive prototype in that they exhibit a configuration that can be best characterized as experiencer-stimulus, which is clearly distinct from the transitive agent–patient configuration. The hypothesis defended here is that both levels – syntactic and semantic – are intimately related, and that the manifestation of a particular nontransitive configuration (that is, one with an indirect object) constitutes the syntactic manifestation of the semantically marked character (that is, marked with respect to the transitive prototype) of the clause in question. Following Hopper and Thompson’s (1980) framework, it will become evident that GTVs determine the lower Transitivity of the clauses that they configure, in opposition to the higher Transitivity of the prototypically transitive clauses. The present study will analyse the semantic characteristics of the clauses with GTVs, with special emphasis on some of the Transitivity parameters proposed by Hopper and Thompson (1980: 252). Participants of the verb The first parameter mentioned by Hopper and Thompson (1980) refers to the number of participants or arguments contained in a clause. It is understood that a highly transitive clause will have two or more participants, whereas a clause lower in Transitivity will typically exhibit one participant. Although Hopper and Thompson (1980: 252, and endnote 1) do not explicitly define the concept of participant, they cite Dixon’s (1979) A (for Agent) and O (for Object), which in a restrictive interpretation coincide with the subject and object of a prototypically transitive clause. It follows that the English clause ‘Jerry likes beer’ (1980: 254) contains two participants, whereas the Spanish clause Me gusta la cerveza ‘I like beer’ features only one, since ‘the experiencer … appears in an oblique case’ (ibid.). However, by denying an O(bject) status to the dative experiencer in Me gusta la cerveza, Hopper and Thompson (1980) weaken their assertion that ‘[w]e make no claims about the grammatical relations that the NP arguments referring to these participants [A & O] might bear to the verb’ (1980: 252, footnote 1). In order to properly tackle the study of the GTVs, it is necessary to overcome the limitations inherent in the traditional analysis of clauses, which is based on a surface interpretation of the Transitivity/Intransitivity dichotomy. Such an approach merely reduces the distinction to the presence or absence of a certain type of clausal constituent, the direct object, considered ‘the object’, with other possible complements being demoted to an ancillary or peripheral status. Valency (or dependency) grammar, which takes Tesnière’s (1959) work as its starting point, distinguishes between actants and circumstants according to their governed nature. Another, more formal distinction is superimposed

Victoria Vázquez Rozas 83

on this one; it identifies with the circumstants the presence of prepositional marking or of certain case affixes. In contrast to the circumstants, the actants lack prepositional marking and correspond to less-marked cases, such as nominative and accusative. Nevertheless, the difference between actants and circumstants (or core v. oblique) cannot be reduced to the selection of certain concrete markings, but rather is felt by many linguists to exist at a more general level, with case-marking being just one of its manifestations (Thompson, 1997: 60). Although Thompson (1997: 61) points out that ‘[l]anguages differ in the extent to which they make a morphological distinction between noun phrases serving core and oblique roles’, core participants are regarded as those that take on the functions of subject and direct object, which are precisely the arguments that lack prepositional marking in English. On the other hand, from a typological viewpoint, core functions are represented by the only argument of an intransitive verb, and the agent and patient of a transitive verb (S, A and O in Dixon’s, 1979, terminology). In this sense, we arrive at a rather restrictive interpretation of the concept of core arguments, since it mirrors the traditional distinction between transitive and intransitive. With regard to GTVs, there are reasons to argue that, apart from the subject, the indirect object (or dative) is also a core function in the clause. From a semantic viewpoint, it displays enough heterogeneity to prevent a simple association between syntactic function and semantic role. An argument with the function of indirect object can be an experiencer (Me gusta la cerveza ‘I like beer’), a recipient (Le mentí a Juan, ‘I lied to Juan’), a goal (Le entregué los documentos a tu hermano ‘I handed the documents to your brother’), a benefactive (Le preparé la cena ‘I prepared dinner for him’), or a possessor (Le lavé la cara al niño ‘I washed the child’s face’), and so on. In considering the syntactic features of core functions, we might expect that the process of grammaticalization of expressive devices would lead to less marking in morphological encoding. However, this does not mean that core functions are universally characterized by syntactic features such as a lack of prepositional marking, as in English. In Spanish, for example, the presence of the preposition a with animate direct objects, as in He visto a María ‘I have seen Mary’, does not deny the core character of the direct object, nor does its presence justify the alignment of the indirect object with oblique participants. There is a significant index of expression shared by subject, direct object and indirect object – the three core functions of the Spanish clause – which is the agreement with the verb. This agreement is marked by inflectional categories for the subject function (person and number suffixes on the verbal stem), and by unstressed pronouns for the direct and indirect object functions, (proclitics or enclitics to the verbal form). These agreement markers are illustrated in the example in (4.11), where se marks the indirect object and is coreferential with Juan, la marks the feminine singular direct object

84

Gustar-Type Verbs

and is coreferential with la bicicleta, and the suffix –mos marks 1pl and is coreferential with nosotros. (4.11)

La bicicleta se la regalaremos nosotros a Juan. The bicycle IO.3 DO.3fsg give-FUT1pl we to John ‘We will give the bicycle to John.’

The presence of the clitic totals nearly 100 per cent in the GTV examples. That is, based on the count of the tokens of gustar with the subject–predicate– indirect object structure, 1,218 of the 1,221 examples in the database (99.75%) contained the clitic. With regard to the semantico-pragmatic implications, the central functions encode the most prominent participants in the process described by the predication, which is reflected in the degree of topicality they display when compared to the degree of topicality of noncentral or peripheral participants. The indirect object undoubtedly presents a high degree of topicality, as shown in the distribution of its categorical realizations in our corpus. Within the framework of the Topic Continuity Theory (Givón, 1983), clitics would display the highest topicality; the clitic conjunction and full form would imply a lower degree of topicality; and, finally, with the full form without clitic being the least topical. The BDS data regarding of indirect objects provide the results in Table 4.1. Thus, there are semantic and syntactic reasons supporting the idea that the indirect object in Spanish is a core function, and not an oblique one. In contrast to Hopper and Thompson (1980), we claim that the clauses in (4.2–4.6) have two participants, thus matching the number of participants present in prototypically transitive clauses. However, there are obvious syntactic differences between a direct and an indirect object in Spanish. Apart from the differences in case between the accusative (direct object) and dative (indirect object) clitics (lo, los, la, las v. le, les respectively), the full NP indirect object is obligatorily marked by the preposition a, whereas the direct object is marked with a only when it possesses certain features involving animacy and/or definiteness.

Table 4.1 Degree of topicality in indirect objects Clitic only Clitic  full form Full form Total

9,727 1,654 1,085 12,466

78.66% 13.38% 8.77% 100%

Clitics highly topical, clitic  full form less topical, full form only least topical.

Victoria Vázquez Rozas 85

There are also differences in the possibility of passivization, which is blocked for indirect objects, as illustrated by the examples in (4.12) and (4.13): (4.12)

(4.13)

a. La nueva maestra les gusta a los niños. ‘The children like the new teacher.’ b. *Los niños son gustados por la nueva maestra. ‘The children are liked by the new teacher.’ a. Los niños quieren a la nueva maestra. ‘The children love the new teacher.’ b. La nueva maestra es querida por los niños. ‘The new teacher is loved by the children.’

An explanation for the impossibility of passivization in the case of (4.12b) involves the selection of semantic roles of GTVs. One must bear in mind that the function of passivization is to make into a theme (in the theme–rheme sense) a constituent other than the subject, that is, the direct object in Spanish. However, in GTVs the indirect object is treated as an unmarked theme, and as such there is no need to passivize it (cf. Table 4.4). Another syntactic difference between the Spanish direct and indirect objects is relativization. Whereas direct objects allow relative que without preposition, indirect objects require that the preposition a precede the relative:3 (4.14) (4.15)

La maestra que quieren los niños. ‘The teacher that the children love.’ Los niños a los que les gusta la nueva maestra. ‘The children that like the new teacher.’

Nevertheless, the analysis of a copious and diverse corpus reveals that the differences between the direct and indirect object are not as clear-cut as they seem in light of the above examples, since there are numerous verbs that show vacillation between both constructions. The uncertainty is compounded by those cases that feature a first- or second-person object clitic, whose form does not allow us to discern between accusative and dative: (4.16) (4.17)

Las leyes ya no me afectan. ‘The laws do not affect me any more.’ Te fastidia recordar el pasado. ‘It bothers you to remember the past.’

(PASAJERO: 34, 12) (AYER: 35, 4)

Even when a third-person clitic is involved, the vacillation between accusative and dative turns out to be somewhat random: (4.18) Lo que realmente lo preocupaba era una ceremonia, de fecha próxima, en la que se presentaría ante la reina (HISTORIAS: 131, 18)

86

Gustar-Type Verbs

‘What really worried him was a ceremony, coming up soon, at which he would introduce himself to the queen.’ (4.19) Dentro de cincuenta años, esos bienes que tanto le preocupan no le servirán de nada (HISTORIAS: 70, 10) ‘In fifty years from now, those possessions that worry him so much will be useless.’ Such difficulty in distinguishing direct from indirect objects represents another argument in favour of considering indirect objects as a participant and not as an oblique in Spanish. Kinesis, aspect and punctuality Three of the Transitivity parameters (kinesis, aspect and punctuality) proposed by Hopper and Thompson (1980) deal with ‘aspectuality’. The authors contend that highly transitive clauses are action clauses, that is, dynamic clauses involving telicity and punctuality. On the other hand, low Transitivity corresponds to static (that is, non-dynamic) clauses, which are atelic and non-punctual. If our approach to GTVs is correct, we would expect the predicates of GTVs take stative, atelic and nonpunctual clauses, and indeed the behaviour of GTVs seems to point in this direction. Some GTVs exhibit incompatibility with the perfective aspect; and although we find this with a reduced number of verbs (atañer ‘concern’, competer ‘be incumbent upon’, concernir ‘concern’, convenir ‘suit, be good for’, incumbir ‘be up to, concern’, which are inherently imperfective), this shows the correlation between perfectivity and high Transitivity on one hand, and imperfectivity and low Transitivity on the other.4 (4.20)

(4.21)

a. Os conviene refrescaros. ‘It is good for you guys to freshen up.’ (CAIMÁN: 37, 9) b. *? Os convino refrescaros. ‘Freshening up was good for you guys.’ a. A ti no te incumbía hablar de ellos realmente. ‘It was really not up to you to talk about them.’ (SEVILLA: 192, 17) b. *? A ti no te incumbió hablar de ellos realmente. ‘It was really not up to you to talk about them.’

A feature that generally characterizes stative clauses as opposed to dynamic ones is that the former do not carry a habitual or frequentative interpretation when used in the present (4.22), contrary to what happens with dynamic clauses, as shown by the sentence in (4.23): (4.22) A Ana le gusta Juan (*todas las semanas/*cada tarde/*con frecuencia). ‘Ana likes Juan (every weekend /every evening /frequently).’

Victoria Vázquez Rozas 87

(4.23) Ana visita a su abuela (todas las semanas / cada tarde / con frecuencia). ‘Ana visits her grandmother (every weekend / every evening / frequently).’ The incompatibility between stativity and habituality not only affects GTVs but also stative predicates with transitive argument structure, such as those in (4.7–4.10) above. In the same vein, the progressive-based test has been used as a key argument in the distinction between dynamic and stative. It is generally contended that stative predicates are anomalous in the progressive (cf. Marín Gálvez, 2000: 71). Indeed, as we have confirmed through the BDS, the verb type that combines with the estar  gerund construction (analogous to English to be  -ing) is the one which denotes acts and operations particularly evident to the faculties of sensation, (Fernández Ramírez, 1960: 534), such as those in (4.24) and (4.25): (4.24) (4.25)

El viejo ya está abriendo a Simonetta. (SONRISA: 147, 13) ‘The old man is already opening [the door] to Simonetta.’ Mira: yo la estoy acariciando. (TERNURA: 31, 18) ‘Look: I am caressing her.’

Nevertheless, it turns out that practically any verb, supported by the appropriate context, can be used in the estar  gerund construction, although not all verbs convey the same semantic implications in this construction. Following Halliday’s (1985: 109) stance regarding English, it can be similarly claimed that, in Spanish, the estar  gerund construction is the marked option with nondynamic clauses, and that its use in such instances – undoubtedly infrequent – does not add a semantic feature that can labelled as ‘progressive’, but rather other features, such as ‘ingressive’ or ‘intensive’: (4.26) (4.27)

Me está apeteciendo un café. ‘I am feeling like having a coffee.’ No nos está gustando nada tu comportamiento ‘We are not liking your behaviour at all.’

Another test that has been adduced to prove the difference between stative and dynamic situations is the possibility (or lack thereof) of combining parar de ‘stop’(de Miguel, 1999). It seems that parar de is only compatible with clearly dynamic situations: (4.28) (4.29)

*Esa película no para de gustarme ‘That movie does not stop pleasing me.’ *La propuesta no paró de interesarles. ‘The proposal did not stop interesting them.’

88

Gustar-Type Verbs

In sum, there are reasons to believe that GTVs have stative (and thus, low Transitive) clauses. Therefore, GTVs contrast with prototypically transitive predicates, which have active (and thus, highly Transitive) clauses. Nevertheless, among the clauses exhibiting the transitive syntactic pattern (subject–direct object), we also find abundant examples of stativity (with verbs such as those cited in 4.7–4.10), thus making it impossible to establish a direct relationship between syntactic pattern and the stative or dynamic nature of the situation denoted by the verb. Volition and agency Among the Transitivity features proposed by Hopper and Thompson (1980), there are two that refer to the semantic characteristics of the participant A: volition and agency. Higher Transitivity is correlated with the volitional and agentive aspect of A, whereas the absence of volition and agency in A is related to a lower Transitivity in the clause. Volition and agency refer to the degree of involvement of participant A in the situation described by a given clause. Volition depends on the voluntary and conscious nature of A’s participation, while agency is given, according to Hopper and Thompson (1980), by Silverstein’s Animacy hierarchy, since it largely coincides with the degree of topicality of the entity in question.5 Both volition and agency are interrelated, as the agentive potentiality, and in particular a high degree of animacy, is a necessary but not sufficient condition for volition. A number of researchers working on typologically unrelated languages have demonstrated that the possibilities of using a noncanonical transitive coding increase when the animacy of the subject is lowered in the Animacy Hierarchy (Dixon, 1979: 85ff; Comrie, 1981: 121; and Lazard 1994: 200–4). This tendency is also confirmed by Spanish data. The subject–direct object clauses and the subject–indirect object clauses display a highly different picture regarding the animate v. inanimate character of their subjects (Table 4.2). The low degree of animacy that characterizes the subject in the subject– indirect object pattern is also corroborated by the frequency of clausal subjects in this pattern when it is contrasted with the rather exceptional presence of a subject clause in the transitive pattern (Table 4.3). Table 4.2 Frequencies and percentages of animate v. inanimate subjects in two-participant clauses Animate Direct–object clauses Indirect–object clauses

58,318 (85.75%) 1,879 (28.35%)

Inanimate 9,693 (14.25%) 4,748 (71.65%)

Note: These data include all the subject–indirect object clauses in our database, not only the data pertaining to GTVs.

Victoria Vázquez Rozas 89 Table 4.3 Frequencies and percentages of clauses functioning as subject

Direct-object clauses Indirect-object clauses

Clausal subject

Others

589 (0.86%) 1,666 (25.14%)

67,422 (99.14%) 4,748 (74.86%)

Examples of GTVs with clausal subjects are the following: (4.30) (4.31)

Me agrada que penséis lo mismo (COARTADA: 50,33) ‘It pleases me that you guys think the same.’ Al viejo le gusta que ella no bisbisee oraciones (SONRISA: 182, 14) ‘The old man likes that she doesn’t whisper prayers.’

The data from Table 4.3 show that the clauses containing the subject–indirect object pattern clearly deviate from the transitive prototype, which includes among its components a highly agentive and, therefore, a highly animate subject, while a clausal subject such as that in (4.30–4.31) occupies the lowest level on the Animacy Hierarchy, and consequently exhibits the lowest potentiality of agency. Thus, those verbs that are combined alternatively with a direct object (accusative clitic) or an indirect object (dative clitic) reject the transitive pattern when they are combined with a clausal subject. Let us compare (4.32) and (4.33): (4.32)

(4.33)

a. Le sorprendió mucho el comienzo de la Guerra. ‘The beginning of the War surprised her/him a lot.’ b. Le sorprendió mucho que comenzase la Guerra. ‘It surprised her/him greatly that the War began.’ a. El comienzo de la guerra la sorprendió en París. ‘The beginning of the war surprised her in Paris.’ b. *Que comenzase la guerra la sorprendió en París.6 ‘That the war began surprised her in Paris.’

The tendency shown by certain verbs to adhere to the transitive pattern if the subject is animate, and to the indirect object pattern if the subject is inanimate, had already been observed by grammarians such as Cuervo (1874: fn. 121) and Fernández Ramírez (1951: 192), and, more recently, by García (1975: 307–16). Copious examples from our corpus corroborate that tendency, illustrated by the following pairs: (4.34) Una chica de Nueva York toma el barco a una isla del Caribe, donde la [DO] espera el novio para casarse. Parece una chica muy

90

Gustar-Type Verbs

buena, y llena de ilusiones, que le cuenta todo al capitán del barco, que es buen mocísimo, y él mira al agua negra del mar, porque es de noche, y después la mira a ella como diciendo ‘esta no sabe lo que le [IO] espera.’ (BMA, 163–4) ‘A girl from New York takes the ship to a Caribbean island, where her fiancé is waiting to get married. She seems like a very good girl, full of hope, who tells everything to the ship’s captain, who is staring into the black water, because it is nighttime, and then he starts looking at her as if to say “she doesn’t know what she’s in for”.’ (4.35) a. Sabía que el sólo decírselo a él iba a aliviarla, pero no lo hizo para no preocuparlo [DO]. (ATC, 484) ‘She knew that her just telling him was going to relieve her, but she did not tell him in order not to worry him.’ b. Lo que más le [IO] preocupaba de la muerte al doctor Urbino era la vida solitaria de Fermina Daza sin él. (ATC, 75) ‘What worried Dr. Urbino most about death was Fermina Daza’s lonely life without him.’ With respect to the feature of volition, which according to Hopper and Thompson’s (1980) view characterizes the participant A in canonical transitive clauses, it is absent in clauses with GTVs. Since inanimate subjects, which typically appear with GTVs (cf. Table 4.2), are inherently nonvolitional, the criterion to determine the presence of volition in participant A would be those GTV constructions containing an animate subject, as exemplified below: (4.36) Hoy precisamente, doy una comida y quiero presentarte a mis amigos, les vas a encantar. (DIEGO: 138, 11) ‘Precisely today, I am hosting a meal, and I want to introduce you to my friends. They are going to like you.’ (4.37) JAVIER – Ganas dinero y follas con quien te apetece. ADELA – Y, a veces, con quien me conviene. (CINTA: 102, 23) ‘JAVIER – You make some money and ball whoever you please.’ ‘ADELA – And, sometimes, who (also) suits me.’ (4.38) Que no seas pelmazo, que ya no me importa José. (OCHENTA: 67, 29) ‘Don’t be a bore, I am no longer interested in Jose.’ (4.39) ¿Quién de los dos le gustaba al francés? ( JÓVENES: 154, 4) ‘Which of the two did the Frenchman like?’ Among the signs that are adduced to show the volitional or controlled character of a situation, are the compatibility with the imperative, the possibility

Victoria Vázquez Rozas 91

of functioning as complement of a verb of command/obligation, and the combinability with manner adverbials (see Dik, 1989: 96–7; Di Tullio, 1996: 225). None of these tests yield positive results with clauses containing GTVs: (4.40)

*Encántales a mis amigos. ‘Make my friends like you.’ (4.41) *Te sugiero / ordeno que le convengas a Juan. ‘I suggest / order that you be right for Juan.’ (4.42) *La convencí / persuadí de que le gustase al francés. ‘I convinced / persuaded her that the Frenchman like her.’ (4.43) *José le importa deliberadamente / cuidadosamente / a propósito a Pedro. ‘Pedro matters to Jose deliberately / carefully / on purpose.’

Similarly, the impossibility of constructing an impersonal passive depends on the non-agentive character of the predication (see Levy, 1994: 357): (4.44)

*Se les encanta a los amigos. ‘One is enjoyed by the friends.’ (Rough translation)

Therefore, the semantic implications of the animate subjects accompanying GTVs include neither agentivity, nor volition, nor control of the situation described by the clause, and are interpreted in a way similar to inanimate subjects, as a cause or stimulus of the state in which the the experiencer finds her/himself. Affectedness and individuation of the O(bject) With regard to the features of the O argument, Hopper and Thompson (1980) claim that the degree of Transitivity is correlated with the affectedness and individuation of this participant. A clause will become more transitive to the extent to which O becomes affected by the process described in the clause, as is the case, for instance, with created or altered Os. With respect to the O appearing with GTVs, coded as the indirect object, it would be difficult to assert that it is affected in Hopper and Thompson’s (1980: 252–3) sense, that is, ‘[t]he degree to which an action is transferred to a patient is a function of how completely that patient is affected’, since the clauses containing these predicates do not refer to an action, nor can it be claimed that the participant O becomes affected. Rather, O is the basis of a stative situation. The fact that the affectedness feature does not apply to the O of GTVs is conditioned by the fact that such a participant is not the patient in an action clause, but rather, the experiencer of a state. For its part, the ‘individuation’ feature is interpreted as a cluster concept with the following properties: proper noun, human-animate, concrete,

92

Gustar-Type Verbs

singular, countable and referential-definite. Such properties refer to the distinctiveness of O, whose Transitivity is lowered to the extent that O is characterized as common, inanimate, abstract, plural, uncountable and nonreferential. In extreme cases, we find clauses that do not really have two participants, but rather one participant (A) with O-incorporation. In this view, the Os in sentences such as el niño tiene hambre (lit. the boy has hunger) ‘the boy is hungry’, la peonza da vueltas (lit. the top gives spins) ‘the top is spinning’, los escolares hacen novillos (lit. the student make young bulls) ‘the students play hooky’, esa cara da miedo (lit. that face gives fright) ‘that face is scary’, are not considered arguments, but rather they are part of complex predicates, with the resulting clauses being interpreted as intransitive.7 Hitherto we have seen that clauses containing GTVs are characterized as having low Transitivity. However, upon examining features of the O participant of the GTV clauses in our corpus, we find that the degree of individuation of the entity would correspond in Hopper and Thompson’s (1980) hypothesis, to clauses of high Transitivity. Indeed, almost all examples with GTVs contain a human, concrete, countable, and referential-definite indirect object. There are, however, some observations by Hopper and Thompson that seem to contradict considering a highly individuated O as an indicator of high Transitivity. That is, the authors claim that the Transitivity of the clause can be reduced ‘when there is an anomalous A–O relationship, viz. when the O is higher than the A in the [Agency] hierarchy’ (ibid.: 273). Indeed, GTVs favour a type of construction in which the O participant surpasses A on the Transitivity hierarchy, such that the high animacy of O would reduce the degree of Transitivity of the clause. In order to minimize the effect of such a contradiction, Hopper and Thompson (1980: 273) state that it ‘is the reduction of the “Agency” of the A which accounts for the anomaly in the A–O relationship, not the fact that the O is high on the hierarchy.’ This statement reestablishes the authors’ initial idea that the human-animate character of O is a sign of high Transitivity. Nevertheless, we offer the following data, which counter the above idea. First, if we assume that the canonical or deviant character of a certain syntactic configuration is correlated with its frequency, those features that characterize the object most frequently will be the ones that will allow us to qualify the object as prototypical. With regard to the animacy of the O participant in transitive constructions, the data from Spanish are convincing: of a total of 68,010 subject–direct object clauses only 12,834 (18.9%) display animate Os, from which we can conclude that the prototypical O will be inanimate. Second, according to the criterion of syntactic marking, a prototypical O will exhibit a morphologically less marked construction.8 Therefore, we must conclude that those Os marked with the preposition a (typically animate and definite) cannot be considered prototypical objects. Third, the inanimate character of O is also given by the semantic definition of Transitivity. It is assumed that a prototypical transitive clause refers to a

Victoria Vázquez Rozas 93

transfer from an agent to a patient, roles which are defined by Langacker (1991: 210) as follows: The archetypal ‘agent’ role is that of a person who volitionally carries out physical activity which results in contact with some external object and the transmission of energy to that object. The polar opposite of an agent is an inanimate ‘patient’, which absorbs the energy transmitted by externally initiated physical contact and thereby undergoes some change of state.9 Finally, the correlation between the different parameters that make up the Transitivity Hypothesis leads us to consider the inanimate O as a feature of high Transitivity. If Hopper and Thompson’s approach were correct, we would expect an animate O to be correlated with the parameters of kinesis, volition, agency, and so on that characterize highly transitive clauses. However, there are copious verbs in Spanish that show the opposite situation, that is, they present more features of high Transitivity when they have an inanimate O (coded as a direct object) than when they have an animate O (coded as an indirect object) (discussed later in this chapter).

‘Dative subjects’ In light of the relationship between clauses such as (4.45) and (4.46), the existence of a relation between the indirect object in (4.45) and the subject in (4.46) was proposed in early formalist accounts: (4.45) (4.46)

A María le gusta la música. ‘María likes music.’ María ama la música. ‘María loves music.’

The relationship was analysed transformationally as a rule of subject–object exchange called FLIP by Lakoff (1970: 126), and Psych-movement by Postal (1971).10 For its part, Case Grammar attributed the same interpretation, in terms of case frames, to both constructions, based on its supposed synonymy (see Fillmore, 1968: 30). Later, Relational Grammar interpreted constructions illustrated by (4.45) as involving inversion, which consisted of demoting the underlying subject to a surface indirect object, and promoting the underlying object to a surface subject, known as unaccusative advancement. Among the specific arguments for the existence of ‘dative-subject’ verbs in Spanish is that dealing with nominalizations. As noted by Fernández-Soriano (1999: 125), the nominalization of a three-argument verb such as entregar ‘hand in’ retains the preposition a for the indirect object, as in (4.47), whereas the indirect object of a verb like faltar ‘to lack’ – which is generally

94

Gustar-Type Verbs

thought of as a type of GTV – is obligatorily preceded by de, the preposition that appears with subjects: (4.47)

La entrega del premio a Juan ‘The presentation to Juan of the award.’ (4.48) La falta de valor de / *a Juan (nominalization of A Juan le falta valor.) ‘Juan’s lack of courage.’ (‘Juan lacks courage’.) In Fernández-Soriano’s view, the obligatory selection of de in (4.48) suggests that Juan may be an underlying (that is, a deep) subject, a so-called ‘dative subject’. As for the nature of the subject in GTVs, such as gustar ‘like’, its status as an underlying object has often been pointed out, appealing to the fact that it does not retain the preposition de, typical of subjects in nominalizations, but rather it is accompanied by the preposition por, as shown in (4.49): (4.49)

a. A Juan le gustan las cerezas. ‘Juan likes cherries.’ b. El gusto de Juan por las cerezas. ‘Juan’s liking for cherries.’

Campos (1999:1,560) mentions the coreferentiality test with temporal infinitival constructions, according to which the indirect object position of GTVs (for example gustar in 4.50) controls the subject position of the infinitive. Thus, in terms of its control properties, the indirect object in (4.50) corresponds to the subject position of verbs denoting dynamic situations (for example escribir ‘write’), where the subject controls the infinitive subject position, as indicated in (4.51): (4.50) (4.51)

A Lucii le gustaba Ronnyj antes de ei,*j conocer a Otto. ‘Lucyi liked Ronnyj before ei,*j meeting Otto.’ A Ronnyj le escribía Lucyi antes de ei,*j conocer a Otto. ‘Lucyj used to write to Ronnyi before ei,*j meeting Otto.’

However, other tests employed by generative grammarians to identify the subject do not support the presence of ‘dative subjects’ with GTVs. For example, Fernández-Soriano (1999: 123–4) attempts to show that the compatibility with small clauses is possible in constructions such as those in (4.52)–(4.56): (4.52)

Me falta café. ‘I need coffee.’

Victoria Vázquez Rozas 95

(4.53) (4.54) (4.55) (4.56)

Me pasa algo. ‘Something is wrong with me.’ Me consta que eres el mejor. ‘I know for a fact that you are the best.’ Nos dieron las dos. ‘It was two o’clock (and we were still engaged in some activity)’. A Marta le fue bien en Buenos Aires. ‘It went well for Marta in Buenos Aires.’

Nevertheless, of the examples in (4.52–4.56), only (4.55) and (4.56) are compatible with a small clause co-indexed with the indirect object, as shown in (4.57) and (4.58), respectively: (4.57)

Nosi dieron las dos borrachosi ‘We were drunk at two o’clock.’ (4.58) Mal mei fue ausentei pero peor presentei. ‘It went badly for me when I was absent, but much worse when I was present.’

However, the constructions featured in (4.52), (4.53) and (4.54) are incompatible with indirect object small clauses.11 This incompatibility is found with to all GTVs. However, GTVs do admit subject small clauses, as exemplified in (4.59): (4.59)

A Maríai Juanj le desagrada borrachoj/ *borrachai. ‘Maríai dislikes Juanj drunkj/*i.’

Both Relational Grammar, as well as Generative Transformational Grammar give precedence to the behavioural properties over the coding properties, minimizing the importance of the coding that identifies subjects (agreement with the verb, absence of preposition) from indirect complements (dative clitic showing agreement, the preposition a). For our part, we interpret such coding as crucial in the syntactic characterization of arguments. Moreover, we do not consider appropriate the distinction of several (more or less deep) levels of syntactic function. However, we acknowledge the similarities between the sentences such as (4.45) (A María le gusta la música ‘María likes music’) and (4.46) (María ama la música ‘María loves music’). Unlike the generativist formulations described above, we believe that the relationship between constructions such as (4.45) and (4.46) is to be interpreted in terms of different semantic and discourse values. Although both sequences can refer to the same reality, the different syntactic coding of the participants is not random at all, but rather it fits a different semantic and discourse configuration. With respect to the corresponding English

96

Gustar-Type Verbs

sentences ( John likes music and Music pleases John), Danes (1968: 61) proposes two different semantic patterns: the first would consist of ‘bearer of attitude – attitude – object of attitude’, whereas the second would be made up of ‘source (cause) – effecting – recipient of effection’. Before studying the features that distinguish the clauses in (4.45) and (4.46), we will discuss the similarities between both structures, starting with a diachronic reference to the issue at hand. In the earlier stages of their respective histories, both English like as well as Spanish gustar had a different argument structure than what they have now. In Old English, the element that currently takes on the form of subject was coded as the object, as in him like oysters v. he likes oysters.12 With regard to gustar, it was used as a transitive verb in Old Spanish. In the sixteenth century this transitive pattern coexists with a pattern with prepositional object, and by the eighteenth century documentation emerges showing gustar with ‘dative subjects’.13 However, the structural changes are not restricted to the verbs like and gustar. As Whitley (1998) points out, many of the so-called psych verbs have presented different syntactic patterns over time, a variability that is not observed in prototypically transitive verbs, which are more syntactically stable. Focusing on the evolution of gustar, we observe that the transition from the transitive to the ‘dative subject’ argument structure has been accompanied by a semantic change that adheres to Hopper and Thompson’s (1980) Transitivity Hypothesis. As Whitley indicates (1998: 138), ‘[i]n the semantic change of gustar from a tasting agent to a satisfied experiencer, there was a decrease in “kinesis”, “volitionality” and “agency”.’ However, regardless of the direction of the structural variation, it is unquestionable that the change reveals a close link between the subject of the transitive pattern and the indirect object of the intransitive one. From a typological perspective, this relationship is supported by data from several languages. For example, in the Indo-European context it is shown in the emergence of the verb of possession ‘have’ from a dative construction containing the copula ‘be’, as exemplified by the Latin examples in (4.60)–(4.61): (4.60)

(4.61)

Mihi est liber to me (dat) is book (nom) ‘I’ve got the book.’ Habeo librum I have book (acus) ‘I’ve got the book.’

The dative, or indirect object, represents a certain degree of activity on the part of the participant, which is linked to its prototypically animate character. This degree of activity sets the participant apart from the direct object

Victoria Vázquez Rozas 97

(characterized as non-active), while bringing it closer to the subject, which in turn corresponds to a prototypically active participant. The relationship between indirect object and subject also manifests itself in the fact that the dative can code the agent complement of the passive construction in languages such as Old Persian and Pre-Classical Latin (see Herslund, 1988: 292–3), Mongolian (Comrie, 1976: 276), or Japanese (Marantz, 1984: 140).14 Another characteristic that draws the indirect object toward the subject and moves it away from the direct object is the feature of independent existence that we attribute to the entity represented by the indirect object. This is one of the properties that Keenan (1976) assigns to the subject.15 Indeed, the independent existence characterizes the indirect object that accompanies GTVs. Finally, the pragmatic-informative properties also attest to the close relationship between the indirect object in a clause like (4.45) and the subject in clauses such as (4.46). In both instances, it is the relatively more active participant that occupies the thematic position, independently of the syntactic function it performs. The data about the position with respect to the verb that have been mentioned for gustar clearly indicate the thematization of the indirect object vis-à-viss the subject. Contrary to the most common association between subject and discourse theme, in clauses with GTVs the unmarked discourse theme is not the subject, but rather the indirect object, and as such it occupies the first position in the clause, whereas the subject follows the verb (Table 4.4).16 Table 4.4 Frequencies and percentages of preposition and postposition of subject and indirect object for gustar Subject17 Preposed Postposed Implicit, clitic, or fixed order Total

Indirect object

125 (9.98%) 678 (54.15%) 449 (35.86%)

223 (18.26%) 29 (2.37%) 969 (79.36%)

1,252

1,221

Direct construction v. inverse construction We begin this section with a terminological note. Following the terminology employed to refer to the patterns under consideration, we will consider a ‘direct construction’ the pattern found in clauses such as María ama la música ‘María loves music’ (in 4.46), and ‘inverse construction’ the pattern identifiable in a María le gusta la música ‘María likes music’ (in 4.45).18 As we have seen in the previous section, some authors claim that the differences

98

Gustar-Type Verbs

between these constructions are limited to the most superficial syntactic level, and that underlyingly we would be dealing with an identical configuration. An argument against this view is the existence of sentence pairs containing the same lexical items that exhibit divergent interpretations: (4.62)

(4.63)

(4.64)

a. María admira la rapidez con que dibujas. ‘María admires the quickness with which you draw.’ b. A María le admira la rapidez con que dibujas. ‘María is astonished by how fast you draw.’ a. Juan apetece una casa propia. ‘Juan fancies having his own house.’ b. A Juan le apetece una casa propia. ‘Juan fancies having his own house.’ a. Antonio repugna el dulce. ‘Antonio finds the sweet disgusting.’ b. A Antonio le repugna el dulce. ‘The sweet disgusts Antonio.’

If we adopt a functional approach, we need to account for the existence of two different constructions for the expression of the same content. That is, we must justify the functional output of the two constructions. Whitley (1998: 130) suggests that there is difference in terms of control between the constructions: Verbs that stayed in or moved to type 1 [the direct construction] suggest a common denominator of choice or control. One can choose to love, hate, hope, use, even take responsibility for it, while pleasure (gustar, placer), pain (doler), and sorrow (pesar) in type 3 [the inverse construction] are feelings that simply happen to the experiencer. Nevertheless, the attribution of control or responsibility to the subject participant of the direct construction is debatable. It is not clear to what extent one can love, hate or admire something or somebody as a result of a conscious, volitional effort. If we try strictly linguistic tests, the result is not any clearer. It is true that some verbs like amar ‘love’ or pensar ‘think’ are compatible with the imperative: (4.65) (4.66)

‘Amaos los unos a los otros’, dijo Jesucristo. ‘ “Love one another”, said Jesus Christ.’ Piénsalo con calma. ‘Think it over calmly.’

However, it does not seem possible to use the imperative with verbs such as apetecer ‘fancy, feel like’, repugnar ‘disgust’. Nevertheless, certain contrasts

Victoria Vázquez Rozas 99

between direct and inverse constructions lend credence to the idea that the direct construction implies a certain degree of agentivity on the part of the subject that is absent from the inverse construction. Thus, in saying he tenido una idea ‘I have had an idea’ or he pensado que podríamos ir a la fiesta ‘I thought that we could go to the party’, we assume responsibility for the mental activity represented, whereas if we say se me ha ocurrido una idea ‘an idea has occurred to me’ or se me ha ocurrido que podríamos ir a la fiesta ‘it has occurred to me that we could go to the party’, we do not admit our participation in the making of the thoughts that are reproduced. Thus, they would escape the conscious control of our mind. Proof regarding how the different degree of agentivity affects the selection of the construction is given by the verbs recordar ‘remember’ and olvidar ‘forget’. Although both refer to similar cognitive processes, the language attributes more agentivity to the action of remembering than to that of forgetting. Therefore, only the latter verb allows the inverse construction:19 (4.67)

(4.68)

a. María recordó el cumpleaños de Juan.’ ‘María remembered Juan’s birthday.’ b. *A María se le recordó el cumpleaños de Juan.20 ‘María remembered Juan’s birthday.’ a. María olvidó el cumpleaños de Juan. ‘María forgot Juan’s birthday.’ b. A María se le olvidó el cumpleaños de Juan. ‘María forgot Juan’s birthday.’

On the other hand, the differences in control between direct and inverse constructions are also reflected in the (im)possibility of functioning as complement of a verb that implies volition, such as intentar ‘try’. Thus, we can say intentó olvidar a María ‘S/he tried to forget María’, but not *intentó que se le olvidara María. From a typological point of view, there are reasons to think this. Like Spanish, other languages are equipped with more than one way of conceptualizing psychic processes (see Wierzbicka 1999: 58–9). This is found in Russian, in which on can express the fact of ‘being sad’ with three different constructions: (4.69)

a. On byl grusten. he-NOM was-MASC sad-MASC b. Emu bylo grustno he-DAT (it)was-NEUT sad(ADV)NEUT c. On grustil he-NOM sad(VERB)-PAST.MASC.

(Wierzbicka, 1999: 59)

100

Gustar-Type Verbs

As Wierzbicka (1999: 60) points out: All these sentences can be roughly glossed as ‘he was sad’, but in fact they differ in meaning. In particular sentence (b) implies that the sadness was involuntary and was, so to speak, ‘happening to the experiencer’, whereas (c) implies active involvement by the experiencer, and suggests that he is bringing about his own sadness by thinking certain thoughts. The semantic and constructional parallelism between (4.69b) and (4.69c) above with the inverse and direct construction, respectively, is evident. A similar situation is found in Polish, which presents direct constructions as well as inverse ones in order to represent the same type of processes: (4.70)

a. Ania podziwiala Piotra. Ania:NOM admired Piotr:ACC ‘Ania admired Piotr.’ b. Ani imponuje Piotr. Ania:DAT impressed Piotr:NOM ‘Piotr impressed Ania. / Ania was impressed by Piotr.’ (Dhbrowska, 1997: 70)

Thus, the experiencer is coded, according to the construction, as nominative or dative. Dhbrowska asks herself about the basis for this distinction and concludes that it is based on the existence of two ways of conceptualizing mental experiences, which revolve around two different theories. On the one hand, we have the ‘craftsman model’, according to which the mental experience is conceived as an action undertaken by the experiencer who manipulates mental objects (ideas, images, experiences). This model corresponds to the coding of experiencer as nominative. On the other hand, we have the ‘mental arena model’, which conceives the mind as a container of ideas that executes the activities of thinking, feeling, and so on, as spontaneous processes that take place in the experiencer’s personal sphere. According to this model, the sensations, feelings and beliefs are conceived by attributing to them an independent existence outside the experiencer.21 This model corresponds to the dative experiencer (see Dhbrowska, 1997: 77). The existence of the two types of constructions, direct and inverse, would corroborate the cognitive validity of both theories. According to Dhbrowska for Polish, the coding of the experiencer through the nominative is interpreted as the default option, with the dative construction requiring a special motivation (ibid.: 79). We can also explain the existence of two different codings for psychological processes through the prototypical configuration of the biactant predications, which can be summed up as the transference of an action from an

Victoria Vázquez Rozas 101

agent to a patient. The verbs that adjust more naturally to this semantic configuration have been called ‘primary transitive verbs’,22 and pose no problems with regard to the syntactic construction that they select: the agent is coded as subject and the patient as direct object. Situating ourselves within Dowty’s (1991) framework, which deals with a set of semantic implications for the Proto-Roles of Agent and Patient, we can say that for ‘primary transitive verbs’ the subject has all the Proto-Agent features, whereas the object displays all the properties of the Proto-Patient.23 Nevertheless, in the case of psychological processes, the selection of subject and object is less obvious, since the participants are not so evidently specialized as Proto-Agents or Proto-Patients. Rather, we need to acknowledge that both the experiencer and the stimulus have Proto-Agent properties: the experiencer has property b (‘sentence and/or perception’) of the Agent Protorole, whereas the stimulus has property c (‘causing an event or change of state in another participant’) of the Agent Proto-role. Thus, both the experiencer and the stimulus can be candidates to be coded as subject, as is the case in Spanish.

Indirect object v. direct object The list of verbs that combine exclusively with indirect object and show the semantic configuration stimulus-experiencer is relatively limited: (4.71) agradar ‘please’, alcanzar ‘affect’, be sufficient for’, apetecer ‘fancy’, atañer ‘concern’, competer, concernir ‘concern’, convenir ‘be good for’, desagradar ‘displease’, doler ‘hurt’, extrañar ‘find strange’, gustar ‘like’, importar ‘matter’, incumbir ‘concern’, parecer ‘seem’, pasar ‘happen’, pesar ‘regret’, placer ‘please’, repugnar ‘disgust’, suceder ‘happen’, tocar ‘be posted, win, be one’s turn, concern’ Some of these verbs also have a subject–direct object construction whose semantic configuration is no longer stimulus–experiencer, but agent–patient. If the direct object is compulsorily inanimate, both constructions correspond to two clearly distinct meanings, as shown by the examples in (4.72) and (4.73): (4.72)

(4.73)

a. Los abogados han convenido las condiciones del contrato. [DO] ‘The lawyers have agreed the conditions of the contract.’ b. Al abogado [IO] le convienen las condiciones del contrato. ‘The conditions of the contract are good for the lawyer.’ a. El gobierno importa alimentos [DO] de Australia. ‘The government imports food from Australia.’

102

Gustar-Type Verbs

b. Al gobierno [IO] no le importa tu opinion. ‘Your opinion does not matter to the government.’ In other instances, the direct object can be animate, like the indirect one, but there are also differences of meaning between the two: (4.74)

a. Voy a extrañar a Mariana [DO], por ser la última mujer de mi vida. (HISTORIAS: 56, 4) ‘I am going to miss Mariana for being the last woman of my life.’ b. A él [IO] también le extraña que, de repente, lo tan ansiado parezca recuerdo de cosa ya olvidada. (SONRISA: 232, 17) ‘He also finds strange that, suddenly, that which was so craved seems a memory of something already forgotten.’

There are some other verbs that seem to reveal differences in their semantics that go hand in hand with their use with an indirect (or direct) object, although mention of both constructions does not always appear in the dictionaries. For example, the verb encantar ‘love’ exhibits two, clearly differentiated meanings, shown in (4.75)–(4.76):24 (4.75) Someter [a alguien o algo] a una acción que sobrepasa lo natural por medio de la magia. (Seco, 1999, entry for encantar) ‘To subject [somebody or something] to an action that goes beyond the natural [world] by means of magic.’ (4.76) Gustar o complacer extraordinariamente [a alguien]. (Seco, 1999, entry for encantar) ‘To like or please [someone] very much.’ Both uses are listed as transitive, contrary to the common tendency in Spanish. Nevertheless, in the same dictionary (Seco et al., 1999), there is an example with an accusative clitic for the sense in (4.75): al escupirles en la cara logra encantarlos, y quedan como muertos por el día ‘when s/he spits in their faces, s/he manages to enchant them, and they remain as if dead for the day’; and another example with a dative clitic for the meaning in (4.76): Le encanta la música ‘S/he loves music’. The possibility of leísmo for the latter example is out of the question, since the dative clitic can have a feminine referent, as in a María le encanta la música ‘María loves music.’ According to the BDS data, of a total of 96 appearances of encantar with the meaning in (4.76) above, 26 exhibit a dative clitic (le/les, whether alone or accompanied by a phrase introduced by a) and 70 have first- and second-person clitics, that is, syncretic forms that neutralize the distinction between accusative and dative. In light of these data, we can claim that in its meaning of

Victoria Vázquez Rozas 103

predicate of affection, encantar displays the inverse construction linked to the GTVs. Significantly more complex is the behaviour observed in a large set of verbs that has the possibility of appearing with the accusative or the dative, without there being any clear semantic difference between them. Among these verbs, we cite the following: (4.77) abrumar ‘overwhelm’, aburrir ‘bore’, admirar ‘admire’, afectar ‘affect’, afligir ‘grieve’, alegrar ‘make happy’, angustiar ‘distress’, apasionar ‘fascinate’, apenar ‘sadden’, asombrar ‘amaze’, asustar ‘scare’, atemorizar ‘frighten’, aterrar ‘terrify’, aterrorizar ‘terrorize’, atormentar ‘torment’, atraer ‘attract’, avergonzar ‘shame’, cansar ‘tire’, complacer ‘please’, consolar ‘console’, convencer ‘convince’, decepcionar ‘disappoint’, deleitar ‘delight’, desanimar ‘dishearten’, descontentar ‘displease’, desconsolar ‘distress’, desesperar ‘exasperate’, disgustar ‘disgust’, distraer ‘amuse’, ‘distract’, divertir ‘amuse’, emocionar ‘move’, ‘touch’, entretener ‘amuse’, ‘entertain’, entristecer ‘sadden’, entusiasmar ‘love’, escandalizar ‘scandalize’, espantar ‘scare away’, estorbar ‘bother’, exasperar ‘exasperate’, fascinar ‘fascinate’, fastidiar ‘annoy’, favorecer ‘favour’, halagar ‘flatter’, impresionar ‘impress’, incomodar ‘inconvenience’, inquietar ‘unsettle’, interesar ‘interest’, intranquilizar ‘worry’, intrigar ‘intrigue’, irritar ‘irritate’, maravillar ‘amaze’, molestar ‘bother’, obsesionar ‘obsess’, ofender ‘offend’, pasmar ‘astonish’, perjudicar ‘be bad for’, preocupar ‘worry’, reconfortar ‘comfort’, satisfacer ‘satisfy’, seducir ‘seduce’, sorprender ‘surprise’, tranquilizar ‘calm down.’ Given that these are verbs that appear in dictionaries with the ‘transitive’ label, one would expect them to take accusative (direct) objects in nonleista dialects. However, the grammarians’ accounts about the fluctuation in casemarking of these verbs are not only recent. With respect to the verb admirar, Bello (1847: §749) states that it projects accusative case in un objeto nos admira ‘an object strikes us as odd’, while Cuervo (1886–93, s.v. admirar) claims that if we interpret the admirar ‘cause admiration’, as in Bello’s example, this entry takes the dative, since the accusative is associated with the meaning ‘consider with admiration’. Other verbs for which Cuervo (1886–93) postulates a dual case marking are agradar25 ‘please’, desagradar ‘displease’, contentar ‘please’, cansar ‘tire’, descontentar ‘displease’, and disgustar ‘annoy’. More recently, several scholars have attempted to identify the factors that determine the selection of accusative or dative case marking among verbs belonging to this class.26 The factors that have been alluded to are related to the nature of the subject and object participants, as well as to the aspectual nature of the predicate as a whole. The features that have been mentioned fit into

104

Gustar-Type Verbs

the predictions of Hopper and Thompson’s (1980) Transitivity Hypothesis (with certain modifications regarding the characteristics of the object participant; see our earlier discussion of ‘Affectedness and Individuation of O’), in the sense that the accusative object is associated with highly transitive clauses, whereas the dative object corresponds to clauses of low Transitivity. With regard to the aspectual configuration of the predicate, we argue that clauses with direct object marking tend to denote dynamic and telic events, whereas indirect object marking corresponds to stative and atelic events. The following examples with the verb atraer ‘attract, appeal’ clearly illustrate how direct object marking appears in a clause that expresses an action of a physical nature (4.78), while indirect object marking appears in a clause expressing a psychic reaction on the part of the experiencer (4.79):27 (4.78) Los rugidos del león atrajeron al cazador, quien, sacándolo del fondo del foso, lo metió en una gran jaula y se lo llevó (1INFAN: 21, 26) ‘The roaring of the lion attracted the hunter, who took it out from the bottom of the pit, put it in a big cage, and took it with him.’ (4.79) El libro estaba encuadernado en piel y tenía el canto dorado, pero a ella no le atraía. (SUR: 104, 30) ‘The book was bound in leather and had a gilded spine, but did not appeal to her.’ The aforementioned differences in telicity can be observed in the following examples with convencer (4.80) in its telic sense, and (4.81) in its atelic one: (4.80) La ha convencido y se casarán en cuanto arreglen los papeles. (SONRISA: 329, 23) ‘He has convinced her and they are going to get married as soon as they straighten the paperwork out.’ (4.81) Veo que a usted le gusta el ambiente porque veranear por ahí en un poblacho, eso no le convence. (MADRID: 31, 24) ‘I see that you like the atmosphere, because spending the summer in a dump, that does not convince you.’ For Di Tullio (1998: 256), the verbs in question ‘can denote complex events as well as states in each of their alternative forms, the accusative and the dative’. In complex events, the activity carried out by the subject triggers a change in the psychological state of the object. As Di Tullio (1998: 258) points out, the activity implies dynamism and control on the part of the subject, and the resulting event is telic since it entails culmination or outcome. On the other hand, the sentence with a dative-marking verb implies neither

Victoria Vázquez Rozas 105

dynamism nor control, nor is it delimited, since it does not entail an outcome. The impossibility of the imperative and the progressive constitutes proof of the stative nature of the predicate with dative marking. Following Croft (1986), Dowty (1991) points out that the psychological predicates that code the stimulus as subject (please, frighten, and so on) can be interpreted as inchoative or stative (unlike verbs such as like, fear, and others, whose subject is experiencer, and are interpreted only as stative). The inchoative interpretation implies a change of state in the experiencer, ‘coming to experience an emotion or a new mental state’ (Dowty, 1991: 580), such that this argument would have a Proto-Patient implication absent from the stative interpretation. Consequently, the experiencer in the inchoative interpretation would be a ‘better’ Patient (ibid.) than the experiencer in the stative interpretation. This would explain, as Ackerman and Moore (1999: 24) note, the distribution of object marking in Spanish: accusative (prototypical direct object) in the inchoative predicate, and dative (non-prototypical object) in the stative predicate. The nature of the subject argument determines the marking of the object. If the subject is animate and possesses the will to act consciously, the likelihood that the experiencer/patient is coded as an accusative object increases notably (see 4.82 below). However, if the subject is inanimate it is more likely that the experiencer will be dative (see 4.83 below): (4.82) No buscó a Diego, no quería molestarlo. (DIEGO: 59, 6) ‘S/he did not seek out Diego. S/he did not want to bother him.’ (4.83) También le molestaba mucho que Agus se metiera el dedo en la nariz. (TERNURA: 90, 26) ‘It also bothered him a lot that Agus picked his nose.’ Strictly speaking, it cannot be claimed that all the apparently (animate) subjects can be interpreted as agents with the verbs under consideration in this article. Di Tullio (1998) distinguishes those cases in which the animate subject denotes an individual with agentive potentialities, like the implicit subject of molestar in (4.82) above, from other situations in which the animate reference is understood as a summation of properties, body parts, characteristics or behaviours (Di Tullio, 1998: 257), as shown by the subject of atraer in 4.84. (4.84) Los [estudiantes] más comprometidos en la lucha seguían en la Facultad. A David le atraían. (JÓVENES: 156, 39) ‘The students that were most committed to the struggle continued on in the Department. David was attracted by them.’

106

Gustar-Type Verbs

The subject of human reference in (4.84) behaves like the subjects of GTVs, which, as we pointed out earlier in the chapter, lack volition and control over the situation, independently of their subcategorization features. With regard to the characteristics of the object, it seems that the way in which it participates in the event can also condition the functional selection. The less active the object argument is, and the more directly affected by the situation it is, the more likely it is to be coded as a direct object. In particular, when the affectedness is physical, there is a tendency to use the accusative, whereas psychic affectedness tends to correlate with the dative. In this respect we can adduce, for example, Hurst’s (1951: 76) data on the verb irritar, which favours the use of the dative in a sequence such as a ella le irritan mis atenciones ‘my affections irritated her (dative)’ vis-à-vis the physical meaning of la irritaba el roce de la cinta ‘the rubbing of the tape [on her skin] irritated her (accusative)’. By the same token, the possibilities of passivization are reduced for the psychological uses of the verbs listed in (4.77). Compare, for instance, the sentences in (4.85) to their counterparts in (4.86): (4.85)

(4.86)

a. Al ladrón lo sorprendió la policía en el interior de la vivienda. ‘The police surprised the burglar in the inner part of the house.’ b. El ladrón fue sorprendido por la policía en el interior de la vivienda. ‘The burglar was surprised by the police in the house.’ a. Al ladrón no le sorprendió la actuación de la policía. ‘The police intervention did not surprise the thief.’ b. *El ladrón no fue sorprendido por la actuación de la policía. ‘The thief was not surprised by the police intervention.’

Ackerman and Moore (1999: 9), following Treviño (1992), explain the contrast between ‘direct affectedness’ and ‘non-direct affectedness’ using the examples in (4.87)–(4.88): (4.87) (4.88)

Los perros lo molestan siempre que llega ebrio. ‘The dogs harass him (DO) every time he comes home drunk.’ Los perros le molestan (*siempre que llega ebrio). ‘The dogs bother him (IO) (every time he comes home drunk).’

In (4.87), we have an object that is more prototypically patient than the one in (4.88), which accounts for its coding as a direct object. In (4.88), however, the object participant does not undergo a change of state, and is not directly affected, but rather is the experiencer of a new stative situation. Again, one can test this: only (4.87) allows passivization. With respect to the dialectal extension of the accusative–dative alternation of the verbs in (4.77), we have to point out that there are instances of

Victoria Vázquez Rozas 107

interference due to the phenomenon of leísmo. In such circumstances, the alternation only has an effect on the feminine form, with the exception of those varieties that exhibit both leísmo and laísmo. Apart from that, it has been pointed out that in certain varieties (for example those in Argentina, Chile, Peru), accusative case marking has been generalized for the verbs listed in (4.77), and for some GTVs (see Fernández-Ordóñez, 1999: 1,325). The example with concernir ‘concern’ illustrates this solution: (4.89) Algo que lo concernía tan íntimamente quizá lo descalificara para juzgar. (HISTORIAS: 93, 2) ‘Maybe something that concerned him so intimately disqualified him from judging.’ Thus, the alternation between accusative and dative case marking for verbs of affectedness is a phenomenon that itself varies somewhat in the Spanishspeaking domain. Although the factors that we have pointed out in this section undoubtedly play a role in the syntactic coding of the object participant, it must be interpreted not as a set of rules that are applied rigidly, but rather as general tendencies that account for a large number of cases, but that do not exclude the existence of examples that go against the predictions. There is one more factor that contributes to the alternation of object marking in verbs of emotion: the relatively low frequency with which speakers are forced to choose between one or the other construction. Emotion verbs express an internal process of the experiencer in such a way that a speaker can more easily appreciate his/her own feeling when s/he is the one affected, but has only indirect evidence of the process when the experiencer is another person.28 This is why in discourse there are more cases in which objects are first-person pronouns than those in which objects are third-person pronouns. Since the syncretism of the Spanish pronominal system excludes the possibility of making a case distinction between direct and indirect object in all but the third person, speakers are freed from having to choose between the two forms when using the first (and second) person pronoun. We have provided statistical analyses in our data that seem to confirm this line of argumentation. Since it is commonly understood that conversational oral language is the most representative of the spontaneous language use, we have limited the statistical analyses to data from oral texts and the theatrical texts of the data Arthus (see note 1). We have also limited the verbs under our investigation to a small sample: alegrar ‘please’, asombrar ‘surprise’, disgustar ‘displease’, divertir ‘enjoy’, interesar ‘interest’, molestar ‘bother’, preocupar ‘worry’, sorprender ‘surprise’ and tranquilizar ‘calm’. Results indicate that there were 21 cases of third-person pronouns while there were 133 cases of first person pronouns (86.4%). Needless to say, a more extensive quantitative study needs to be done, but the data discussed in our study still provide an

108

Gustar-Type Verbs

important indication for the relative frequency of the pronominal forms in real discourse. Why, then, is it considered that a low frequency of use favours variation? The answer to this question is based on the assumption that linguistic structures result from conventionalization of frequently used discourse forms. That is, following claims by Givón (1979), Du Bois (1985) and Hopper (1998), we believe that grammar emerges from discourse and thus a frequency factor is essential to account for the emergence and evolution of the syntactic structures. As pointed out by Du Bois (1985: 363), ‘grammars code best what speakers do most’. Thus, we believe that the accusative–dative alternation in emotive verbs cannot be accounted for either way – generalizing either accusative or dative for all the cases of the same verb – since its frequency of use is not sufficient to force grammaticalization. We do note tendencies for some of these verbs to be used in a fixed construction in order to express a certain meaning (cf. example 4.76 with the verb encantar ‘like’), but the majority of verbs have not lexicalized the accusative or the dative construction, which leaves ample semantic space in which the two variants of construction coexist.

Conclusion In this chapter, we have carried out an analysis of the GTVs within the framework of the Transitivity Hypothesis proposed by Hopper and Thompson (1980). First, we argued that, together with DOs and subjects, IOs should be considered part of the core argument structure of GTVs. After examining the characteristics of the constructions containing verbs and paying special attention to the parameters proposed by Hopper and Thompson (1980), we demonstrated that constructions containing GTVs exhibit low Transitivity. We also analysed the syntax and semantics of the GTVs as compared to transitive constructions with emotion verbs. In contrast to formalist approaches that defend assigning the same semantic configuration to both types of construction, we have seen how the functionalist approach leads us to make a semantic distinction between the two structures. In support of this argument, we put forward typological data which corroborate the existence of a cognitive basis for the constructional contrast. Finally, we examined the properties of those verbs that fluctuate between the construction with IO and the construction with DO. We found that the alternation between the two structures follows perfectly from the predictions made in the Transitivity Hypothesis: sentences with accusative object tend to be dynamic and telic, and tend to have an animate, agentive subject, and their objects tend to be inactive and physically affected. By contrast, sentences with dative objects are usually stative, atelic, have no agentive subject, and their objects are affected psychologically. The choice between accusative and dative objects is, thus, an indication of the degree of Transitivity of the sentence as a whole.

Victoria Vázquez Rozas 109

In any case, we have to remember that speakers ultimately choose the most appropriate construction for each case, depending on the meaning they intend to convey. If high Transitivity is coded by means of the accusative construction, we are able to say only that as a sentence becomes increasingly prototypically less transitive, the greater the possibility is of the dative construction appearing. It is, however, difficult to determine exactly at which point such a change will systematically occur. In this area we are dealing with statistically significant tendencies, not categorical rules.

Notes 1 One should also consider as a typical construction of ‘verba sentiendi’ that one found in clauses of pronominal and prepositional object construction, exemplified in Me alegraba de tenerlo a mi lado (Historias: 156, 13) ‘Having him beside me made me happy’; Jano se olvidó de Betina (Carta: 111, 13) ‘Jano forgot about Betina’; Se encaprichó con la chiquilla (Sonrisa: 301, 22) ‘S/he took a liking to the little girl’. 2 For further information on the database, the reader is directed to http://www. bds.usc.es. 3 Nevertheless, in the oral language the use of the relative is frequent without preposition: Me han hablado muy bien de la obra; ahora, que hay gente …, que hay gente que [a la que] le gusta mucho y otra que [a la que] no le ha convencido demasiado (Madrid: 371, 27) ‘I have been told very good things about the plau; now, that there are people …, that there are people that like it very much and other that has not been convinced enough’. 4 Hopper and Thompson (1980: 271) observe that ‘if the Aspect is perfective, the interpretation – other things being equal – has properties allowing the clause to be classified as more transitive; but if the Aspect is imperfective, the clause can be shown on independent grounds to be less transitive’. 5 Dixon’s (1979: 85) representation of the hierarchy is the following:

1st p. pron. 2nd p. pron. 3rd p. pron.

Human Animate Inanimate proper nouns common nouns

<- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Likelihood of functioning as transitive agent 6 These examples are variants of those presented by Givón (1976: 170): La guerra le sorprendió mucho ‘The war surprised him a lot’ and La guerra lo sorprendió en París ‘The war surprised him in Paris.’ 7 With respect to this type of verb–noun combinations in English, Thompson and Hopper (2001: 34) state: ‘These V–O compounds are low in Transitivity because it is difficult to maintain that O is individuated or affected. In fact, for clauses with V–O compounds, it is not clear whether they should even be considered two-participant clauses at all.’ 8 ‘The core A, S and O arguments are more likely to be systematically and morphologically unmarked than are the other arguments’ (Thompson, 1997: 64) 9 Likewise, Comrie (1981: 121) argues that ‘the most natural kind of transitive construction is one where the A is high in animacy and definiteness, and the P is lower in animacy and definiteness.’

110

Gustar-Type Verbs

10 The analysis of these authors refers to examples such as The music pleases me and I like the music. The degree of parallelism between Spanish and English is high enough to assume the same transformational rules for the Spanish examples. 11 It cannot be said, for example (4.52)*Me falta café despierta, (4.53) *Me pasa algo sola, or (4.54) *Me consta segura que eres el mejor. 12 Examples taken from Jespersen (1924: 160). 13 In the sixteenth century, we find old constructions such as (i), as well as constructions with prepositional objects, as shown in (ii): (i)

(ii)

14

15 16 17 18

19

20 21 22 23

… si ya no gustas que la discreción y sciencia de Tirsi y de Damón te alumbren de la ceguedad en que estás … (From LA GALATEA by Miguel de Cervantes, 1,585) ‘if you don’t like that the discretion and science of Tirsi and Damon enlight your blindness in which you are …’ Su padre de ella gusta de su casamiento. (From LOS HECHOS DE GARCILASO by Lope de Vega, 1,579–83) ‘Her father likes [the idea of] her marriage.’

For its part, Continental Portuguese still has the construction found in (ii). Therefore, one cannot accept Rotaetxe’s (1999: 27) generalization that in their evolution, languages often pass from a dative to a direct object construction, and never the other way round. Although the changes undergone by English like seem to support such a statement, Spanish gustar, along with verbs such as faltar, admirar, and so on illustrate the opposite situation (see Whitley, 1998: 128). With regard to this information, we should also mention the position attributed to the experiencer on the control continumm proposed by Comrie (1981: 53–6), a position that brings the experiencer closer to the agent, but separates it from the patient. ‘The entity that a b[asic]-subject refers to (if any) exists independently of the action or property expressed by the predicate’ (Keenan, 1976: 312–13). These quantitative data have been obtained form the BDS (see note 1). The total frequency of the subject is slightly higher than that of the indirect object due to the existence of clauses without an indirect object. Both terms have been used extensively in Relational Grammar, although authors from other schools have also employed similar terminology. For instance, Bossong (1997: 260) contrasts the terms ‘généralisation’ and ‘inversion’, whereas Whitley (1998) distinguishes ‘direct construction’ from ‘reverse construction’. Proof independent of the more agentive character of recordar is the possibility of the causative construction with this verb, as in Juan le recordó la hora de la cita a María ‘Juan reminded María the time of the appointment’, which is not possible with olvidar (*Juan le olvidó la hora de la cita a María). This sequence would be possible as a reflexive passive, not as an inverse construction. ‘We tend to think of ideas as objects that enter and leave our minds and of emotions as waves that overwhelm us and sweep us away’ (Dhbrowska, 1997: 42). A. Andrews (1985: 68) defines ‘primary transitive verbs’ as ‘the class of twoargument verbs taking an Agent and a Patient (for example kill, eat, smash)’. ‘Contributing properties for the Agent Proto-Role: a. volitional involvement in the event or state b. sentence (and/or perception)

Victoria Vázquez Rozas 111 c. causing an event or change of state in another participant d. movement (relative to the position of another participant) e. exists independently of the event named by the verb) Contributing properties for the Patient Proto-Role: a. b. c. d. e.

undergoes change of state incremental theme causally affected by another participant stationary relative to movement of another participant does not exist independently of the event, or not at all’. (Dowty, 1991: 572)

24 Repeler ‘repel, repulse’ exhibits a similar behaviour, since it presents a transitive pattern in all its meanings, except in the sense of ‘make somebody feel disgust or aversion’ (Diccionario de la Real Academia Española, entry for repeler), which undoubtedly favours the dative: A usted, amiga mía, le repele visceralmente el tipo humano al que pertenece Alvar (Renglones) ‘You, my friend, are viscerally repulsed by the human type Alvar belongs to’. 25 According to Cuervo (1886–93, entry for agradar), at that time the Diccionario de la Real Academia considered this verb as transitive. In its current edition, both agradar and desagradar are regarded as intransitive. 26 Examples of particularly relevant studies are: Hurst (1951), García (1975), Roldán (1975), García and Otheguy (1977), Levy (1980), Orange (1982), Uber (1986), Whitley (1995, 1998), di Tullio (1995) and Roegiest (1995). 27 Each example is associated with a meaning different from that in Seco et al. (1999). Atraer in (4.78) means ‘(of a person or thing) to make somebody or something (DO) come closer to him/her/it or to where s/he it is by means of a physical force or impulse of an instinctive or voluntary nature’; whereas in (4.79) it means ‘(of a person or thing) to bring it about that somebody or something (DO) have a feeling of affection, inclination or desire towards him/her/it’. 28 As Mithun (1991: 522) notes, ‘Speakers do not claim to feel what another individual is feeling’. See also Melis (1999).

References Ackerman, F. and J. Moore (1999) ‘Syntagmatic and Paradigmatic Dimensions of Causee Encodings’, Linguistics and Philosophy, vol. 22, pp. 1–44. Andrews, A. (1985) ‘The Major Functions of the Noun Phrase’, in T. Shopen (ed.), Language Typology and Syntactic Description. Vol. I: Clause Structure. Cambridge: Cambridge University Press, pp. 62–154. Bello, A. (1847) Gramática de la lengua castellana destinada al uso de los americanos. Valparaíso. Bossong, G. (1997) ‘Le marquage de l’expérient dans les langues d’Europe’, in J. Feuillet (ed.), Actance et valence dans les langues de l’Europe. Berlin: Mouton de Gruyter, pp. 258–94. Campos, H. (1999) ‘Transitividad e intransitividad’, in I. Bosque and V. Demonte (eds.), Gramática descriptiva de la lengua española, Vol. 2. Madrid: Espasa Calpe, pp. 1,519–74. Comrie, B. (1976) ‘The Syntax of Causative Constructions’, in M. Shibatani (ed.), Syntax and Semantics, Vol. 6: The Grammar of Causative Constructions. New York: Academic Press. —— (1981) Language Universals and Linguistic Typology. Oxford: Basil Blackwell.

112

Gustar-Type Verbs

Croft, W. A. (1986) ‘Surface Subject Choice of Mental Verbs’, Paper presented at the Annual Meeting of the Linguistic Society of America, New York. Cuervo, R. J. (1886–93) Diccionario de construcción y régimen de la lengua castellana, Paris: A. Roger and F. Chernoviz. –––– (1874) Notas a la Gramática de la lengua castellana de Andrés Bello, ed. de Niceto Alcalá Zamora. Buenos Aires: Sopena, 1973. Dhbrowska, E. (1997) Cognitive Semantics and the Polish Dative. Berlin: Mouton de Gruyter. Danes, F. (1968) ‘Some Thoughts on the Semantic Structure of the Sentence’, Lingua, vol. 21, pp. 55–69. de Miguel, E. (1999) ‘El aspecto léxico’, in I. Bosque and V. Demonte (eds), Gramática descriptiva de la lengua española, Vol. 2. Madrid: Espasa Calpe, pp. 2,977–3,060. Di Tullio, A. (1996) ‘Verbos psicológicos en español’, Signo and Seña, Vol. 5, pp. 219–38. –––– (1998) ‘Alternancia acusativo-dativo en verbos psicológicos del español’, in Atti del XXI Congresso Internazionale di lingüística e filologia romanza, Vol. 2. Tübingen: Max Niemeyer, September 1995, pp. 255–60. Dik, S. C. (1989) The Theory of Functional Grammar. Part I: The Structure of the Clause. Dordrecht: Foris. Dixon, R. M. W. (1979) ‘Ergativity’, Language, Vol. 55, pp. 59–138. Dowty, D. (1991) ‘Thematic Proto-Roles and Argument Selection’, Language, Vol. 67, pp. 547–619. Du Bois, J. (1985) ‘Competing Motivations’, in J. Haiman (ed.), Iconicity in Syntax. Amsterdam: John Benjamins, pp. 343–65. DRAE. (1992) Diccionario de la Real Academia Española. Madrid: Espasa Calpe. Fernández-Ordóñez, I. (1999) ‘Leísmo, laísmo y loísmo’, in I. Bosque and V. Demonte (eds), Gramática descriptiva de la lengua española, Vol. 1. Madrid: Espasa Calpe, pp. 1317–97. Fernández-Soriano, O. (1999) ‘Two Types of Impersonal Sentences in Spanish: Locative and Dative Subjects’, Syntax, Vol. 2, pp. 101–40. Fernández Ramírez, S. (1960) Gramática española. Los sonidos, el nombre y el pronombre. Madrid: Revista de Occidente. Fillmore, C. (1968) ‘The Case for Case’, in E. Bach and R. T. Harms (eds), Universals in Linguistic Theory. New York, Holt: Rinehart and Winston, pp. 1–88. García, E. (1975) The Role of Theory in Linguistic Analysis: The Spanish Pronoun System. Amsterdam: North Holland. —— and R. Otheguy (1977) ‘Dialect variation in leísmo: A semantic approach’, in R. W. Fasold and R. W. Shuy (eds), Studies in Language Variation, Washington, D.C.: Georgetown University Press, pp. 65–87. Givón, T. (1976) ‘Topic, Pronoun, and Grammatical Agreement’, in C. N. Li (ed.), Subject and Topic. New York: Academic Press, pp. 149–188. —— (1979) On Understanding Grammar. New York: Academic Press. —— (1983) ‘Topic Continuity in Discourse: An Introduction’, in T. Givón (ed.), Topic Continuity in Discourse. A Quantitative Cross-language Study. Amsterdam: John Benjamins, pp. 5–41. Halliday, M. A. K. (1985). An Introduction to Functional Grammar. London: Edward Arnold. Herslund, M. (1988) Le datif en français. Louvain-Paris: Peeters. Hopper, P. J. (1998) ‘Emergent Grammar’, in M. Tomasello (ed.), The New Psychology of Language. Cognitive and Functional Approaches to Language Structure, Mahwah. New Jersey: Laurence Erlbaum, pp. 155–75.

Victoria Vázquez Rozas 113 –––– and A. T. Sandra (1980) ‘Transitivity in Grammar and Discourse’, Language, Vol. 56, pp. 251–99. Hurst, D. A. (1951) ‘Influence of Subject and Connotation of Force’, Hispania, Vol. 34, pp. 74–8. Jespersen, O. (1924) The Philosophy of Grammar. London: Allen and Unwin. Keenan, E. L. (1976) ‘Towards a Universal Definition of “Subject” ’, in C. N. Li (ed.), Subject and Topic. New York: Academic Press, pp. 303–33. Lakoff, G. (1970) Irregularity in Syntax. New York: Holt: Rinehart and Winston. Langacker, R. (1991) Concept, Image, and Symbol. The Cognitive Basis of Grammar. Berlin: Mouton de Gruyter. Lazard, G. (1994) L’Actance, Paris: PUF. Levy, P. (1980) ‘Una peculiar oposición entre le y lo en el español de México’, Anuario de Letras, Vol. 18, pp. 263–68. —— (1994) ‘Verbos con sentido causativo en la construcción transitiva’, in A. Alonso, B. Guarza and J. A. Pascual (eds), II Encuentro de lingüistas y filólogos de España y México. Salamanca: Junta de Castilla y León and Universidad de Salamanca, pp. 347–66. Marantz, A. P. (1984) On the Nature of Grammatical Relations. Cambridge, MA: MIT Press. Marín Gálvez, R. (2000) ‘El componente aspectual de la predicación’, Doctoral dissertation. Universidad Autónoma de Barcelona. Melis, C. (1999) ‘Variación sintáctica con los verbos de emoción’, Español Actual, Vol. 71, pp. 49–62. Mithun, M. (1991) ‘Active/Agentive Case Marking and Its Motivations’, Language, Vol. 67, pp. 510–46. Orange, J. A. (1982) ‘Contextual Constraints on the Use of le and lo in Spanish’, Word, Vol. 33, pp. 201–28. Postal, P. M. (1971) Cross-Over Phenomena. New York: Holt, Rinehart & Winston. Roegiest, E. (1995) ‘Accusatif, ergatif et datif en espagnol moderne’, in Atti del XXI Congresso Internazionale di lingüística e filologia romanza. Tübingen: Max Niemeyer, Vol. 2, pp. 739–49. Roldán, M. (1975) ‘The Great Spanish le-lo Controversy’, Linguistics, Vol. 147, pp. 15–30. Rotaetxe, K. (1999) ‘Tipología lingüística: dativo y datividad’, Revista Española de Lingüística, Vol. 29, pp. 1–33. Seco, M., A. Olimpia and R. Gabino (1999), Diccionario del español actual. Madrid: Aguilar. Tesnière, L. (1959) Éléments de syntaxe structurale. París: Klincksieck. Thompson, S. A. (1997) ‘Discourse Motivations for the Core-Oblique Distinction as a Language Universal’, in A. Kamio (eds), Directions in Functional Linguistics. Amsterdam: John Benjamins, pp. 59–82. Thompson, S. A. and J. H. Paul (2001) ‘Transitivity, Clause Structure, and Argument Structure: Evidence from Conversation’, in J. L. Bybee and P. J. Hopper (eds), Frequency and the Emergence of Linguistic Structure. Amsterdam: John Benjamins, pp. 27–60. Treviño, E. (1992) ‘Subjects in Spanish Causative Constructions’, in P. Hirschbühler and K. Koerner (eds), Romance Languages and Modern Linguistic Theory. Amsterdam: John Benjamins, pp. 309–24. Uber, D. R. (1986) ‘Actions and Activeness in Spanish Clitic Selection’, in O. Jaeggli and C. Silva-Corvalán (eds), Studies in Romance Linguistics. Dordrecht: Foris, pp. 153–65.

114

Gustar-Type Verbs

Whitley, M. S. (1995) ‘Gustar and Other Psych Verbs: A Problem in Transitivity’, Hispania, Vol. 78, pp. 573–85. –––– (1998) ‘Psych Verbs: Transitivity Adrift’, Hispanic Linguistics, Vol. 10, pp. 115–53. Wierzbicka, A. (1999) ‘Emotional Universals’, Language Design, Vol. 2, pp. 23–69.

Abbreviations used for the original texts cited Narrative ATC García Márquez, G., El amor en los tiempos del cólera, Bruguera, Barcelona, 1985. BMA Puig, M., El beso de la mujer araña, Seix Barral, Barcelona, 1976. JÓVENES Aldecoa, J.R., Porque éramos jóvenes, Seix Barral, Barcelona, 1986. HISTORIAS Bioy Casares, A., Historias desaforadas, Alianza, Madrid, 1986. CARTA Colinas, A., Larga carta a Francesca, Seix Barral, Barcelona, 1986. GLENDA Cortázar, J., Queremos tanto a Glenda, Alfaguara, Madrid, 1981. CRÓNICA García Márquez, G., Crónica de una muerte anunciada Mondadori, Madrid, 1987. SUR García Morales, A., El sur (seguido de Bene), Anagrama, Barcelona, 1985. TERNURA Martínez de Pisón, I., La ternura del dragón, Anagrama, Barcelona, 1988. LABERINTO Mendoza, E., El laberinto de las aceitunas, Seix Barral, Barcelona, 1982. DIEGO Poniatowska, E., Querido Diego, te abraza Quiela y otros cuentos, Alianza/Era, Madrid, 1987. SONRISA Sampedro, J.L., La sonrisa etrusca, Alfaguara, Madrid, 1985. Theatre Buero Vallejo, A., Caimán, Espasa-Calpe, Madrid, 1981. Diosdado, A., Los ochenta son nuestros, Antonio Machado, Madrid, 1990. COARTADA Fernán Gómez, F., La coartada, Antonio Machado, Madrid, 1987. 1INFAN Olmo, L. y P. Enciso, Teatro infantil I, Antonio Machado, Madrid, 1987. CINTA Reina, M.M., La cinta dorada, Antonio Machado, Madrid, 1989. CAIMÁN

OCHENTA

Oral speech Esgueva, M. y M. Cantarero (eds), El habla de la ciudad de Madrid. Materiales para su estudio, CSIC (Miguel de Cervantes), Madrid, 1981.

MADRID

5 Primary and Secondary Object Marking in Spanish* J. Clancy Clements

Introduction The goal of this chapter is to examine the patterns in Spanish object marking, and in grammatical relation marking in general, in order to determine what type of marking pattern is predominant. Our findings suggest that Spanish has developed signs of an ergative-absolutive marking pattern in grammatical relation marking, but most strikingly object marking, which following Dryer (1986) we will refer to as primary-and secondary-object marking. Such a state of affairs is not necessarily expected from a typological perspective given that Latin, as well as the sister languages French, Italian and Portuguese, do not exhibit this type of marking pattern. Following Comrie (1981: 119), we argue that this object marking pattern found in Spanish developed for reasons of economy and, moreover, that it is sensitive to animacy and definiteness, features also key to the ordering pattern of the subject relative to the verb.1 Before discussing primary v. secondary object marking, we first address some general considerations about split accusative v. ergative marking in languages, followed by a discussion of the ergative-absolutive marking pattern in Spanish intransitive clauses. We then examine the ergative-absolutive marking pattern in direct and indirect objects, followed by a discussion of the presence of this type of pattern in rule domains, before some concluding remarks.

* This study is an augmented, updated, and revised version of ‘Ergative patterning, in Spanish’, which appeared in 2001 in Current issues in Spanish syntax and semantics (ed. Javier Gutiérrez-Rexach and Luis Silva-Villar), Berlin: Mouton de Gruyter (271–90) and in 1998 in Perspectives in Hispanic Linguistics (vol. 3) (ed. José del Valle and Javier Gutierrez-Rexach) Columbus, OH, Ohio State University (20–40). 115

116 Primary and Secondary Object Marking in Spanish

Split nominative-accusative v. ergative-absolutive marking In some languages, such as various Austronesian languages (Silverstein, 1976), we find that a certain part of the case-marking system in a given language follows nominative-accusative marking, whereas another part of the system follows ergative-absolutive marking. The demarcation point between the one and the other marking pattern differs from language to language, but, according to Silverstein, a given language has one point of demarcation. This is shown schematically in Figure 5.1. For instance, a given language will mark the second-person form as accusative (marked in Figure 5.1. as ‘’) and all other forms as ergative (marked as ‘’); or it will mark both the first- and second-person forms as accusative and the rest as ergative, and so on. Below, we suggest that Spanish does not have a demarcation point, but rather a demarcation range for accusative v. ergative marking (shown in Figure 5.1), which includes almost all (human) and some (animate), (human) direct objects (DOs). What is different about the case-marking system in Spanish is that it not only depends on syntactic and semantic features (grammatical relations and animacy, respectively), but also on discourse-related features such as definiteness and topicality. In that sense, the ergative case-marking patterns to be examined are somewhat different than those found in the Austronesian languages discussed by Silverstein (1976). Specifically, while ergative marking patterns in Austronesian languages depend largely if not entirely on the morphosyntax of the languages, in Spanish they are sensitive to discourse-pragmatic notions, which in turn influence syntax. This is clearly so in the case of subject order relative to the verb in Spanish, the focus of the next section.

Subject order relative to the verb We will discuss three views regarding subject order in Spanish. One view posits verb–subject (VS) order for Spanish, the second argues for indeterminant

Acc

| Erg |________ ‘pronouns’ + tu | – tu |________ _ _ _ _ _ _ _| _+_ego_ |_–_ego _ _|________ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ + proper | – proper |_________ + human | – human | __________ ‘nouns’ + animate | – animate |-- Spanish --| ...

Figure 5.1 Possibilities for a simple lexical split of case marking: two two-way subsystems, ‘accusative’ v. ‘ergative’ Note: The vertical lines mark successive division of accusative v. ergative case-marking, only one in a given language.

J. Clancy Clements 117

subject–verb/verb–subject SV/VS order in intransitive clauses and (subject)– verb–object ([S]VO) order in transitive clauses, and the third accounts for marked tendencies in SV, VS, and SVO order from a discourse and semantic perspective. We will argue here that the discourse and semantic reasons for SV and VS word order in Spanish underlie a syntactic ergative-absolutive patterning. In nominative-accusative languages, intransitive-clause and transitive-clause subjects are marked in the same way. For example, in English the default position for all subjects, regardless of whether they are in a transitive or an intransitive clause, is sentence-initial (e.g. John travels, John loves Mary). In some morphologically rich nominative-accusative languages, e.g. Spanish, it is common that both transitive-clause and intransitive-clause subjects agree with their corresponding verb, as in yo canto ‘I sing’, ellos cantan ‘they- MASC sing’, where yo ‘I’ and ellos ‘they-MASC’ agree with the suffixes –o ‘1sg.PRES.IND’ and –n ‘3pl.PRES.IND’, respectively. In ergative-absolutive languages, intransitive-clause subjects and transitiveclause direct objects are marked identically. This marking is most often coded morphologically, but there is a priori no reason why such ergative marking cannot be coded in word order, such as subjecthood in English, which is coded using word order. The marking differences between nominativeaccusative and ergative-absolutive languages are illustrated in Table 5.1, where the terms in bold indicate which arguments in each system are marked identically (the subjects in the nominative-accusative system and the subject and direct object in the ergative-absolutive system). We know that Spanish marks subjects through subject–verb agreement and that this morphological marking follows the nominative-accusative pattern. However, it is also known that languages such as Basque, for example, display morphological ergative-absolutive marking and syntactic nominative-accusative marking (Croft, 1990: 151). For Spanish, we claim that the opposite is true; specifically, that Spanish displays morphological nominative-accusative marking but syntactic ergative-absolutive marking and, furthermore, that this syntactic ergative-absolutive pattern emerges due to discourse-pragmatic considerations (see Chapter 2, pp. 28–32 for related discussion). For Spanish word order, essentially three general analyses have been advanced. The first posits VSO or VOS basic word order as the default order. The second makes a four-way distinction involving combinations of two-slot Table 5.1 Nominative-accusative v. ergative-absolutive marking

Transitive clause Intransitive clause

Nominative marking

Accusative marking

Ergative marking

Absolutive marking

subject subject

direct object

subject

direct object subject

118 Primary and Secondary Object Marking in Spanish

options (VS and VO [verb-initial], SV & VO [≈SVO], SV and OV [verb-final], VS and OV [≈OVS (rare)]) advanced by Dryer (1997). Under this second analysis, Spanish would be a SV/VS & VO language. The third analysis posits an SVO order based on the argument of frequency. That is, intransitive clauses exhibit a slight predominance of VS (53%) over SV (47%) order (GarcíaMiguel, 1995: 50; personal communication), and transitive clauses with two full NPs exhibit SVO order around 95 per cent of the time (see Ocampo, 1995; Silva-Corvalán, 1983; Liceras, 1994). Thus, the sum of frequencies of SV and SVO occurrences indicate a predominance of SVO order (see Chapter 2).2 We now discuss the reasoning for each analysis. The arguments for VSO or VOS order go back at least to Gili Gaya (1961), and have been taken up directly by Groos and Bok-Bennema (1986), Contreras (1976), among many others. The general idea is that in a sentence of one breath unit, the verb appears either in sentence-initial or in second position, but in no other. So, the sentences in (5.1 a–d) are syntactically well-formed, while (5.1 e–f) are syntactically ill-formed if uttered in one breath unit: (5.1)

a. [Mi padre] compró [una casa] (SVO) my father bought a house ‘My father bought a house’. b. Compró [mi padre] [una casa] (VSO) c. Compró [una casa] [mi padre] (VOS) d. [Una casa] compró [mi padre] (OVS) e. *[Mi padre] [una casa] compró (SOV) f. *[Una casa] [mi padre] compró (OSV)

Moreover, since in a one-breath unit both arguments of a mono-transitive clause can appear postverbally, and either a subject or an object can appear preverbally, it is argued (for example by Groos and Bok-Bennema) that there is only one slot in preverbal position. This slot can remain empty, or can be occupied by either a subject or an object, but not by both. Given this and the fact that there are more ordering options in which the subject appears postverbally (5.1 b, c, d), it is claimed in this analysis that the canonical position of the subject is postverbal.3 Another argument in favour of default postverbal subject order is that in non-finite clauses (that is, in infinitive and gerund clauses), preverbal arguments are not allowed, as shown in (5.2–5.3), taken from Groos and Bok-Bennema: (5.2)

a. Diciéndomelo telling-me-it ‘Telling it to me.’ b. *Me lo diciendo me it telling

J. Clancy Clements 119

(5.3)

a. Por decirme estas cosas la abuela … because-of tell-me these things the grandmother ‘Because gramma told me these things …’ b. *Por la abuela decirme estas cosas … because-of the grandmother tell-me these things

In the majority of generative approaches for Spanish since Rizzi, the subject is generated postverbally (that is the VP-internal hypothesis), and the appearance of subjects in preverbal position is a consequence of movement. The advantages of the VP-internal hypothesis are the following: that it accounts for so-called subject inversion phenomena related to argument fronting, wh-word or otherwise (but see endnote 3), it accounts for the impossibility of preverbal arguments for non-finite verbs, and finally it accounts for the fact that for intransitive verbs the default subject position is postverbal. However, the most serious problem for the VS analysis is that in transitive clauses forming part of a narrative, SVO is found between 90 and 95 per cent of the time, depending on the genre considered (see Liceras, 1994; López-Meirama, 1997b; Ocampo, 1995; Silva-Corvalán, 1983).4 As mentioned, the second analysis was advanced by Dryer (1997), who proposes doing away with the 6-way word-order typology, shown in (5.4a), and replacing it with a 4-way word-order typology, shown in (5.5): (5.4)

a. 1. SOV; 2. SVO; 3. VSO; 4. VOS; 5. OVS; 6. OSV b. 1. VS & VO (verb-initial) 2. SV & VO (≈SVO) 3. SV & OV (verb-final) 4. VS & OV (≈OVS [rare])

He offers eight reasons in favour of his proposal. Three of these involve the collapsing of VSO and VOS orders because such orders are unstable and languages with such orders have an indeterminant status because they exhibit no marked preponderance of one order over another. The other five reasons deal with the advantages of calculating orders using the two-place rather than the three-place option. Crosslinguistically, the two-place option (that is, SV, VS, VO, OV) is filled much more frequently than the three-place option (that is, SOV, SVO, VSO, VOS, OVS, OSV). The two-place option is more applicable to those languages that display such flexible word order as to render the three-place option unsuitable for purposes of classification. Moreover, the two-place option is preferable for isolating object order relative to the verb. Finally, the three-place option overlooks the order of the subject and verb in intransitive clauses, ‘even though the order in such clauses is occasionally different from the order of subject and verb in transitive clauses, and even though intransitive clauses containing a noun subject are

120 Primary and Secondary Object Marking in Spanish

much more common than transitive clauses containing a noun subject’ (Dryer, 1997: 70). On this view, Spanish is a SV/VS and VO language, based on frequency of occurrence of intransitive clauses and transitive clauses respectively. The drawbacks of this analysis are two: if VO order is assumed for transitive clauses, we cannot account for the fact that Spanish has obligatory postverbal subjects in gerund and infinitive clauses. Second, by separating the calculation of subject order relative to the verb in transitive and intransitive clauses, the important fact is lost that SV order is by far the most frequent over all given that VO orderings with a subject are SVO 90–95 per cent of the time and that just under half of overt subjects in intransitive clauses (47%) have SV order. In a discourse-pragmatic approach, the third analysis, the phenomenon illustrated by the examples in (5.3) would be considered the grammaticalization of a topicless construction, if we adopt López-Meirama’s notion of topic (Chapter 2, this volume) as a phrasal element that appears preverbally. Regarding the distribution of intransitive-clause SV v. VS, López Meirama (Chapter 2, this volume) isolates definiteness as a strong predictor of SV or VS order: SV order is at approximately 85 per cent if subjects are relatively high in definiteness (those with a proper noun, a personal pronoun or a deictic determiner), while VS order is around 90 per cent if subjects are relatively low in definiteness (noun phrases without determiner). This correlation between the degree of definiteness and SV or VS order supports her argument that subject order relative to the verb is not first and foremost a matter of verbal semantics, but rather of discourse-pragmatics (definiteness). And this leads her to propose that for Spanish it makes more sense to speak in terms of unergative (SV) and unaccusative (VS) construction types than unergative and unaccusative verbs. López-Meirama’s account also addresses the 90–95 per cent occurrence of preverbal subjects in VO clauses: preverbal subjects are topics and overwhelmingly definite. In Casielles (1996, 1998), we find some illustrative examples involving definiteness and subject order relative to the verb that highlight LópezMeirama’s findings. In Spanish, transitive verbs can take bare plurals and mass nouns as direct objects, as illustrated in (5.4). In the se passive, however, such bare plurals only occur in postverbal position, as shown in (5.5). That is, they are ill-formed in preverbal position (see 5.5b): (5.4)

(5.5)

Julia invitó a participantes de México. Julia invited-3sg ACC participants from Mexico ‘Julia invited participants from Mexico.’ a. Se invitaron participantes de México. PASSIVE invited-3p1 participants from Mexico ‘Participants from Mexico were invited.’ b. *Participantes de México se invitaron. Participants from Mexico PASSIVE invited-3p1

J. Clancy Clements 121

Analogously, with intransitive verbs bare plural subjects also appear postverbally, but rarely if ever in preverbal position, as illustrated by the examples in (5.6) and (5.7):

(5.6)

(5.7)

a. Vinieron participantes de México. came-3pl participants from Mexico ‘Participants from Mexico came.’ b. *Participantes de México vinieron. a. Llegaron jugadores para celebrar la victoria. arrived-3p1 players for to celebrate the victory ‘Players arrived to celebrate the victory.’ b. *Jugadores llegaron para celebrar la victoria.

Casielles (1998) points out that postverbal bare plural subjects have only an existential (and narrow scope) reading. Such an interpretation is compatible with focus material but not with topic material. As mentioned, topics constitute old information and are marked as definite NPs in discourse, and this allows only a wide-scope reading. Bare plural subjects, on the other hand, are foci and cannot appear in sentence-initial position. Thus, the ergativeabsolutive pattern found in the default word order of intransitive subjects (postverbal) and mono-transitive objects (also postverbal) is accounted for by appealing to definiteness, which is intricately linked to the distinction in discourse between topics and foci. The second set of supporting data for the third hypothesis involves solely transitive clauses. The data suggest that SVO order is the canonical order of Spanish transitive clauses, something for which we already find overwhelming support in the aforementioned frequency counts. Koontz-Garboden (2002) offers evidence that object doubling is found substantially more frequently in Argentine and Castilian Spanish when the order is not SVO. That is, direct object doubling is found more often in noncanonical orders, as shown by the examples in (5.8): (5.8)

(SVO) a. Juan los ve a sus hermanos a menudo. Juan them-see-3sg ACC his brothers often ‘Juan sees his brothers often.’ b. A sus hermanos los ve Juan a menudo. c. Los ve Juan, a sus hermanos a menudo. d. Los ve a menudo a sus hermanos, Juan.

(OVS) (VSO) (VOS)

e. Juan, a sus hermanos, los ve a menudo. f. A sus hermanos, Juan los ve a menudo.

(SOV) (OSV)

其

Object doubling less often

}

Object doubling more often or obligatory

122 Primary and Secondary Object Marking in Spanish

Similarly, in ditransitive clauses, if the order differs in any way from S–V–DO–IO order, IO doubling is found virtually always. The conclusion to be drawn from the evidence is that the unmarked order is the one that triggers clitic doubling least. This is evidence in favour of SVO order as the default in Spanish. In sum, in this section we have argued that the SV/VS order in Spanish intransitive clauses is best accounted for by appealing to the discoursepragmatic notion of definiteness. By appealing to definiteness it is also possible to account ergative-absolutive pattern (see Table 5.1 above) apparent in subject and direct object order relative to the verb. The subjects and direct objects that are not definite are marked identically in that they appear in postverbal position. By contrast, definite subjects appear preverbally. In direct and indirect object marking, we also find an ergative-absolutive pattern.

Primary-secondary object marking in mono- and di-transitive clauses Analogous to the ergative-absolutive marking in transitive and intransitive clauses, shown schematically in Table 5.1 above, Dryer (1986) argues that the same pattern exists in the marking of objects proper. That is, in nominativeaccusative languages objects of mono- and ditransitive verbs are marked respectively always as DOs and IOs, with accusative and dative case marking, respectively. In ergative-absolutive marking, by contrast, monotransitive DOs and ditransitive IOs would be marked identically, as opposed to the ditransitive DO. The principle is, however, the same: since monotransitive DOs and ditransitive IOs never occur in the same clause together, they can be marked identically. Dryer coined the term Primary Object (PO) marking to refer the identical marking of monotransitive-clause DOs and ditransitiveclause IOs, and Secondary Object (SO) marking to refer to the marking of the ditransitive DO. This is shown schematically in Table 5.2. In Spanish, we find strong evidence for this marking pattern in both full NP objects and object pronouns. We show that the key features underlying the PO–SO distinction are primarily semantic (animacy) and, secondarily discourse-pragmatic (specificity).5 Table 5.2 Direct–indirect v. primary–secondary object marking

Ditransitive clause Monotransitive clause

Direct object

Indirect object

Secondary object

Primary object

DO DO

IO

DO

IO DO

Note: The items in bold on each side of the table indicate that they are marked identically.

J. Clancy Clements 123

Full NP Objects It is well-known that Spanish marks the DO with personal a if it is animate (5.9) or if there is a need to disambiguate (cf. 5. 10) or to highlight, as in (5.11). (5.9) (5.10) (5.11) (5.12)

Juanita visitó a su hermano. ‘Juanita visited A her brother.’ La virtud vence al odio. ‘Virtue conquers A hate.’ Los americanos quieren mucho a sus coches. ‘Americans love A their cars a lot.’ El tío Segundo daba dinero mensualmente a la iglesia. ‘Uncle Segundo gives money monthly A the church.’

In these cases, DO marking is identical to IO marking. As the example in (5.12) shows, it is marked with a just as the DO is marked with a in (5.9)–(5.11). In terms of POs and SOs in the examples just seen, a marks POs. There is one large set of counterexamples that weakens considerably the PO–SO analysis of Spanish object marking: inanimate DOs are not marked with a unless disambiguation is necessary or a speaker wants to use a to highlight a DO. However, this is in our view a stronger set of data that argues in favour of the PO–SO analysis. Company Company (2001, 2002) cites research by Ortiz (2001) who finds in a large diachronic Spanish corpus (14 texts from the thirteenth to the twentieth centuries) 3,075 ditransitive sentences with both DO and IO animate full NPs. Ortiz reports only 33 cases (33/3,075, or 1.07% illustrated by 5.13b–5.13c) where both object NPs are marked with a. In the rest of the cases (3,042/3,075, or 98.9%), the DO is left unmarked and IO marked with a (illustrated by 5.14) Company Company (2001: 21, 2002: 50–51) also concurs that object marking follows a PO–SO pattern: in monotransitive clauses with an animate DO, it is marked with a, in ditransitive clauses with two animate objects, the IO is marked with a. (5.13)

a. Oye, mañana vienen los padres de Julio. ‘Hey, tomorrow Julio’s parents are coming.’ b. Él tendrá que presentar a su padre a los compañeros, y He will-have to introduce his father to the friends-MASC, and ‘He’ll have to introduce his father to his guy friends, and.’ c. Yo me encargaré de presentar a su madre a las compañeras. I REFL take-charge of introduce his mother to the friends-FEM ‘I’ll take care of introducing his mother to my girl friends.’ (5.14) A la niña se le presentó un protector rico. (LA BUSCA; [Baroja, 1961: 43]) A the girl PASSIVE her presented a protector rich ‘A rich protector was introduced to the girl.’

124 Primary and Secondary Object Marking in Spanish

Other data suggest that native speakers today avoid the construction altogether. This is reflected in the Arthus corpus of modern Spanish, in which of 362 cases of the ditransitive verb presentar ‘introduce’ only 22 (6%) correspond to the S–V–DO–IO order with all animate arguments. No example was found with two full object NPs.6 Thus, animate full NPs in both object positions of ditransitive verbs are not common, but where they are found the evidence for a PO–SO marking pattern is exceedingly strong. We suggest that the key feature that underlies this marking pattern is animacy: animate full NP DOs are marked with a and all IOs carry a marking. As we will see in the next section, however, the marking patterns are somewhat different for object pronouns in mono- and ditransitive clauses. Ergative-absolutive marking patterns in the pronominal system The Spanish pronominal system currently found in many countries in Latin America maintains the etymological DO–IO distinction (see Table 5.3). However, in some dialects of Castilian Spanish (as well as in some South American dialects, such as those in Paraguay and Uruguay), an innovative pattern has been gradually emerging. Various studies (such as Lapesa, 1968; Marcos Marín, 1978; García, 1975, 1986; Flores Cervantes, 2002) have examined the phenomena of leísmo, laísmo, and/or loísmo over time.7 In a corpus spanning from the twelfth to the nineteenth century, Flores Cervantes (2002: 115–45) found a total of 7,642 occurrences of innovative use of Spanish pronouns, only 974 (13%) of which correspond to the innovative uses. Nevertheless, she established that the innovative uses increased from around 5 per cent in the twelfth century to roughly 30 per cent in the nineteenth century, and largely favouring the singular. Using older and present-day texts, Klein-Andreu (1992: 171) shows that leísmo in certain varieties of Peninsular Spanish is virtually complete for masculine animate direct objects and very advanced for masculine inanimate direct objects (see Table 5.4, taken from Klein-Andreu 1992: 171). From the sixteenth century texts of Saint Teresa to modern-day speech of professional women and rural speakers, the use of le to mark animate DOs is virtually complete at 96 per cent and close to 70 per cent for inanimate masculine NPs.

Table 5.3 The etymological Spanish pronominal system Gender

Masc.

Masc.

Fem.

Fem.

Number Direct object Indirect object

sg. lo le (se)

p1. los les (se)

sg. la le (se)

p1. las les (se)

J. Clancy Clements 125

If leísmo in the varieties 2 and 3 in Table 5.4 were carried to completion, the result would be a PO-SO pronominal marking system in those Castilian varieties. That is, a pronominal system would arguably emerge such as that displayed in Table 5.5. Although these varieties of Castilian Spanish may be close to a PO–SO pronominal system, Klein-Andreu (2000) demonstrates that the degree of variation in pronominal systems is high and is sensitive to semantics (for example animacy), as well as to location (she studied five major areas in north-central Spain), sociolinguistic factors (for example education, socioeconomic class, gender), grammatical categories (for example number), and function (pronominalization of count v. noncount NPs). Klein-Andreu also found evidence of what she called intermediary systems, or hybrid systems, in which both leísmo and either laísmo or loísmo are found. We find that such hybrid systems have existed for centuries. Marcos Marín (1978: 214–18) comments on the one found in the eighteenth century Moratín

Table 5.4 Frequency of le as a function of the referents’ animacy in (1) writings of St Teresa; (2) speech of presentday Castilian professional women; (3) speech of presentday Castilian rural speakers (n.b. only count NPs were considered) Animates

1. St Teresa 2. Professional women 3. Rural speakers Total

Inanimates

le

lo

% le

le

lo

% le

83 67 36

2 6 0

98 92 100

22 16 22

3 21 6

88 43 76

186

8

96

60

30

67

Source: Klein-Andreu (1992), p. 171.

Table 5.5 A probable result of leísmo in a variety of the Castilian pronominal system Gender

Masc.

Masc.

Fem.

Fem.

Number

sg.

pl.

sg.

Pl.

Primary object (monotrans. DO and ditrans. IO)

le

les

le

les

Secondary Object (ditrans. DO)

lo

los

la

las

126 Primary and Secondary Object Marking in Spanish Table 5.6 General pronominal system found in the Moratín play El sí de las niñas

Direct object Indirect object

Masculine

Feminine

le/s le/s

la/s la/s

play El sí de las niñas, where there is evidence of both leísmo and laísmo. Masculine monotransitive DOs and ditransitive IOs are consistently marked with le/les and feminine monotransitive DOs and ditransitive IOs are consistently marked with la/las. That is, we have the system as shown in Table 5.6. The pronoun lo is found, but in 76 per cent of the cases (111/147) it is used as a neuter. We find one instance of double pronominalization in the play, shown in (5.15). This suggests that in double pronominalization le/s marks the IO and lo (and perhaps la) the DO. Thus, the pattern of pronominal use in Moratin’s play points to a PO–SO pattern that in addition marks gender.

(5.15)

‘Sí, no lo dudo; pero el saber que la merezco alguna inclinación, oyéndoselo decir con aquella boquilla tan graciosa que tiene, sería para mí una satisfacción imponderable.’ EL SÍ DE LAS NIÑAS: (183–84) ‘Yes, I don’t doubt it; but knowing that I deserve from her some sign of affection, hearing her say it with that sweet little mouth of hers would be for me of profound satisfaction.’

In other words, we find a nominative-accusative system that for noncount DOs has the neuter pronoun lo and for noncount IOs uses the pronoun le; for countable objects the system distinguishes grammatical gender without distinction between direct and indirect objects except in IO–DO double pronominalization, in which case se ( le/s) marks the IO. This is illustrated in Table 5.7. To sum up this section, the evidence gathered and presented here makes a strong case for the existence of a PO–SO system in Spanish full NP object marking, as well as a pronounced development in certain varieties of Castilian Spanish toward a PO–SO marking pattern. In addition, there is also strong evidence (see Klein-Andreu 2000) that various hybrid systems exist in Castilian Spanish. The factors that contribute to variation in virtually all these pronominal systems are semantic (animacy), geographic, sociolinguistic (education, socioeconomic class, gender), grammatical (gender, number), as well as functional (pronominalization of count v. noncount NPs) in nature.

J. Clancy Clements 127 Table 5.7 Probable object-marking system based on pronoun use in Moratin’s El sí de la niñas Gender

Masc.

Masc.

Fem.

Fem.

Neuter

Number

sg.

pl.

sg.

pl.

sg.

Single pronominalization (DOs and IOs)

le

les

la

las

IOle DO lo

IO se

IO se

IO

Double pronominalization IO se IO se

DO lo DO los DO la DO las DO lo

Evidence of nominative-accusative or ergative-absolutive marking in rule applications In this section, we examine three rules, namely, passivization, object doubling and clitic ordering, to see to what extent they may be sensitive to the ergativeabsolutive marking pattern already apparent in subject order relative to the verb and in object marking. Passivization If it were the case that passivization were sensitive to PO–SO marking, we should find that Spanish allows the passivization of ditransitive IOs. That is, we would expect POs (monotransitive DOs and ditransitive IOs) to be passivizable and SOs (ditransitive DOs) not to be. This, however, is not the case as it is clear that in no variety of Spanish is it possible to passivize any argument but the DO. For this reason, passive sentences such as that in (5.16b) are impossible for all varieties of Spanish: (5.16)

a. Marta le dió el libro a Juanita. ‘Marta gave the book to Juanita.’ b. *Juanita fue dada el libro (por Marta). ‘Juanita was given the book (by Marta).’

Thus, the passivization rule in Spanish is sensitive to only DO–IO marking, and not to PO–SO marking. Object doubling The facts on object doubling suggest a sensitivity to the PO–SO distinction. If we assume that Spanish has a PO–SO system and it has as part of its object doubling rule the hierarchy SO pronominalization  PO pronominalization, the grammaticality or ungrammaticality of the sentences in (5.17) becomes transparent. The hierarchy SO pronominalization  PO pronominalization means that in a PO–SO language that allows object doubling, if the SO can be doubled, then the PO must be able to be doubled. Assuming Spanish to be

128 Primary and Secondary Object Marking in Spanish

a PO–SO language, sentence (5.17a, b) contain instances of PO doubling. Sentence (5.17c) displays a case of PO and SO doubling, and the sentence is grammatical. However, sentence (5.17d) has SO doubling but no PO doubling, and it is ungrammatical. These facts are seamlessly accounted for if we assume the hierarchy SO pronominalization  PO pronominalization for Spanish.8 (5.17)

a. Lei ví [a Juan]i 3SG-PO saw-1SG A Juan ‘I saw Juan.’ b. Lei dí [el libro] [a Juan]i 3SG-PO gave-lSG the book A Juan ‘I gave the book to Juan.’ c. Sei lok dí [el libro]k [a Juan]i 3SG-PO 3SG-SO gave-lSG the book A Juan ‘I gave the book to Juan.’ d. *Lok dí [el libro]k [a Juan] 3SG-SO gave-lSG the book A Juan ‘I gave the book to Juan.’

Clitic ordering If we assume Spanish to be a PO–SO language, clitic ordering falls out directly from the assumption. In the ordering of referential pronominal clitics (as opposed to the non-anaphoric clitic se) in monotransitive clauses, the PO occupies the first and only clitic slot, as in (5.18a). We know that it is the first slot because of the example in (5.18b), a ditransitive clause, in which the PO occupies the first slot and the SO the second. These clitic ordering facts fall out automatically by appealing to PO–SO marking: (5.18)

a. Quiere verle. wants see-INF-3SG-PO ‘S/he wants to see him.’ b. Quiere entregárselo wants deliver-INF-3SG-PO – 3SG-SO ‘S/he wants to give him to him/her/them.’

In sum, while passivization clearly follows DO–IO marking, the facts about object doubling and clitic ordering are most easily accounted for by appealing to the PO–SO marking pattern.

J. Clancy Clements 129

Conclusion In this chapter, we have argued that in addition to the more commonly known nominative-accusative pattern found in Spanish morphology, Spanish also exhibits an ergative-absolutive marking patterns in a number of ways. First, the evidence involving the ordering of the intransitive-clause subjects and transitive-clause DOs shows that these arguments are most often in postverbal position, especially if they constitute foci of a sentence. By contrast, due to its status as a topic in the vast majority of cases, the transitive-clause subject is overwhelmingly found in preverbal position. This ergative-absolutive pattern in syntax follows a discourse-pragmatic ordering: intransitive-clause subjects and DOs, constituting most often foci, new information and being non-specific/indefinite, appear in clause-final position; transitive-clause subjects, being given/old, animate, and definite information, appear in topic (that is clause-initial) position. We have provided striking evidence that in object marking an ergativeabsolutive pattern is also present and is sensitive to features commonly found in topics, that is those of animacy and specificity/definiteness. That is, animate DOs and IOs are marked identically in Spanish, with personal a. By contrast, inanimate DOs are rarely marked with personal a.9 In those sentence with two animate object NPs, the IO is the element marked with personal a in 98.9 per cent of the cases (3,042/3,075) (Ortiz, 2001). Thus, in such cases the PO–SO pattern is exceedingly strong. In the pronominal marking system, we have also provided evidence suggesting that the leísta system is developing toward a PO–SO object pattern, while loísta/laísta pronominal systems display a variety of hybrid patterns, depending on semantic (animacy), geographic, sociolinguistic (education, socioeconomic class, gender), grammatical (gender, number), as well as functional (pronominalization of count v. noncount NPs) factors. We have also showed that while passivization clearly exhibits a nominative-accusative pattern, two other syntactic rules (object doubling and clitic order in double pronominalization) follow the PO–SO marking pattern. Finally, we must ask the question: why do these ergative-absolutive patterns exist in Spanish? With regard to the subject order relative to the verb, it is clear that the patterns are governed by the discourse-pragmatic notions of topic and focus. As for the PO–SO pattern with two animate full NP objects, the key factor in the relevant leísta dialects that has led to the spread of le/s as a PO and not only an IO pronoun is semantic, more specifically, animacy. However, Klein-Andreu (2000) demonstrates that in several varieties of leísta as well as laísta dialects, a number of factors influence object pronoun use. We state, then, that these pronominal systems are in flux and indeed probably have been for centuries. Given that this degree of ergative-absolutive marking and this degree of variability in grammatical relation marking is

130 Primary and Secondary Object Marking in Spanish

only found in Spanish of all the major Romance languages, one must ask how such a hybrid system emerged. This line of research has not been explored and would provide ample scope for synchronic and diachronic work in the future.

Notes 1 The notion of animacy is clear. Our definition of definiteness is as follows: an NP is definite if it contains a definite, deictic, or possessive determiner referring to something in the context or discourse. 2 For a discussion of unergative and unaccusative verbs, see Chapter 2 of this volume. 3 The phrase structure posited by Groos and Bok-Bennema (1986) is as in (i): (i)

INFL⵮ INFL⬙

Spec

INFL⬘ XP INFL

Vmax V⵮

V⬙ V⬘ V

NP (subject) Spec

PP (indirect object) NP (direct object)

The XP of INFL is an argument position and can accommodate any one argument of the verb. The Spec of INFL  is for adjuncts and accounts for the fact that sentences such as (ii) are: grammatical. (ii)

¿Por qué Juan no cena esta noche? why Juan NEG dine-pres-3sg this evening ‘Why isn’t Juan having dinner this evening?’ Por qué is Spec of INFL  and Juan is in XP position. The problem is that (i) would allow the sentences in (iii), which are ungrammatical in Spanish. (iii) a. ¿*Cuándo Juan va al cine? when Juan go-pres-3sg to-the movies ‘When is Juan going to the movies?’ b. ¿*Dónde Luisa come los sábados? where Luisa eat-pres-3sg the-pl Saturdays ‘Where does Luisa eat Saturdays?’ ¿*Cuánto Mario pesa? how-much Mario weigh-pres-3sg ‘How much does Mario weigh?’

It appears that those wh-words that typically have a response in the form of a proposition (i.e. a clause or more), such as cómo ‘how’ and por qué ‘why’ may appear with non-inverted subjects, as in (iii) above. In contrast, those wh-words that have as a response less than a clause, i.e. a phrase (NP or PP), trigger subject inversion.

J. Clancy Clements 131 Thus, wh-words linked to phrasal adjuncts and arguments form one group, while wh-words linked to clausal adjuncts make up a distinct group. The reason why the wh-words corresponding to adjuncts do not display homogeneous behaviour is not clear and requires further research. 4 In a differently oriented study using the Arthus database, García-Miguel (1995: 48) finds that in mono-transitive clauses with full NP participants the following percentages: (i) (Mono) transitive clauses: SV / VS VO / OV 78%/22% 97%/2.5% 5 We adapt our definition of specificity from that given by Von Heusinger (2002): an NP is specific if the referent is identifiable for the relevant party. In the example in (i), for instance, the reader being referred to is not readily identifiable by the relevant party (the author of the piece of work), whereas in (ii) the friend called Hugo Sosa is readily identifiable by the relevant party (Ruben Danieli). Thus, the underlined NP in (i) is non-specific and the underlined NP in (ii) is specific. (i) La obra abierta sería, en contraste, aquella que busca un lector más imaginativamente activo ante el estímulo de la lectura. ‘The open piece of work would be, in contrast, that which seeks a more imaginatively active reader faced with the stimulus of the reading material.’ (ii) Ruben Danieli vive en Australia desde 1971 y busca a un amigo llamado Hugo Sosa, que por 1975 tenía una empresa de copias en Santiago de Chile … ‘Ruben Danieli has lived in Australia since 1971 and is looking for a friend named Hugo Sosa, who around 1975 had a copying business in S.d.C. …’ 6 Thanks to Scott Schwenter for informing me about the data from Ortiz (2001) and to Victoria Vázquez (personal communication) for providing me with these data on presenter. 7 The term leísmo refers to the extension of use of the indirect object pronouns le and les in monotransitive clauses in the direct object domains of the masculine animate, the masculine inanimate, and, increasingly, the feminine. Laísmo refers to the extension of la to the indirect object function and loísmo to the extension of lo to indirect object function. 8 Thanks to José Ignacio Hualde for this observation. 9 To give an example of the frequency with which indefinite DOs are not marked with personal a, we carried out a search of the VP chunk busca un/una ‘looks for a(n)’ in the CREA database (it contains 1,400,000 words). The search yielded 212 tokens, of which only six (3%) had an animate DO: only one (0.5%) was specific, the remaining five (2.5%) were non-specific.

References Anderson, S. (1976) ‘On the Notion of Subject in Ergative Languages’, Subject and Topic. New York: Academic Press, pp. 1–24. Casielles, E. (1996) ‘On the Misbehavior of Bare Nouns in Spanish’, in C. Parodi, C. Quicoli, M. Saltarelli and M. L. Zubizarreta (eds), Aspects of Romance Linguistics. Washington, DC: Georgetown University Press, pp. 135–48. –––– (1998) ‘Preverbal Subject vs. Dislocated Phrases’, in J. Gutiérrez-Rexach and J. del Valle (eds), Perspectives on Spanish Linguistics. Columbus, OH: Ohio State University, pp. 1–20. Comrie, B. (1981) Language Universals and Linguistic Typology. Chicago: University of Chicago Press.

132

Primary and Secondary Object Marking in Spanish

Company Company, C. (2001) ‘Multiple Dative-Marking Grammaticalization: Spanish as a Special Kind of Primary Object Language’, Studies in Language, Vol. 25, pp. 1–47 —— (2002) ‘Reanálisis en cadena y gramaticalización. Dativos problemáticos en la historia del español’, Verba, Vol. 29, pp. 31–69. Contreras, H. (1976) A Theory of Word Order with Special Reference to Spanish. Amsterdam: North Holland. Croft, W. (1990) Typology and Universals. Cambridge: Cambridge University Press. DeLancey, S. (1980) An Interpretation of Split Ergativity and Related Patterns. Bloomington, Indiana: Indiana University Linguistics Club. Dryer, M. (1986) ‘Primary Object, Secondary Objects, and Antidative’, Language, Vol. 62, pp. 808–45. —— (1997) ‘On the Six-Way Word Order Typology’, Studies in Language, Vol. 21, pp. 69–103. Flores Cervantes, M. (2002) Leísmo, laísmo y loísmo. Sus orígenes y evolución. Ciudad de México: Universidad Nacional Autónoma de México. García, E. (1975) The Role of Theory in Linguistic Analysis: The Spanish Pronoun System. Amsterdam: North Holland. —— (1986) ‘The Case of Spanish Gender’. Neuphilologische Mitteilungen, Vol. 87(2), pp. 165–84. García-Miguel, J. M. (1995) Transitividad y complementación preposicional en español. Universidad de Santiago de Compostela (Verba, anexo 40). Gili Gaya, S. (1961) Curso superior de sintaxis española. Barcelona: Vox. Groos, A. and R. Bok-Bennema (1986) ‘The Structure of the Spanish Sentence’, in H. Contreras, I. Bordelois and K. Zagona (eds), Generative Studies in Spanish Syntax. Dordrecht: Foris, pp. 67–80. Hopper, P. J. and S. A. Thomas (1980) ‘Transitivity in Grammar and Discourse’. Language, Vol. 45, pp. 251–99. Klein-Andreu, F. (1992) ‘Understanding Standards’, in G. W. David and G. K. Iverson (eds), Explanation in Historical Linguistics. Amsterdam: Benjamins, pp. 167–78. —— (2000) Variación actual y evolución histórica: Los clíticos le/, la/s, lo/s. Munich: Lincom Europa. Koontz-Garboden, A. (2002) ‘A Quantitative Analysis of Spanish Indirect Object Doubling’, in J. F. Lee, K. Geeslin and J. C. Clements (eds), Structure, Meaning, and Acquisition in Spanish. Somerville, MA: Cascadilla, pp. 193–211. Lapesa, R. (1968) ‘Sobre los orígenes y evolución del leísmo, laísmo y loísmo’, in Festschrift Walther von Wartburg. Tübingen: Max Niemeyer, pp. 523–51. Liceras, J. (1994) ‘La teoría sintáctica y los juicios de gramaticalidad: la posposición del sujeto en español’, Revista Canadiense de Estudios Hispánicos, Vol. 18, pp. 219–55. Lipski, J. (1991) ‘In Search of the Spanish Personal Infinitive’, in D. Wanner and D. A. Kibbee (eds), New Analyses in Romance Linguistics. Amsterdam: Benjamins, pp. 175–94. López M. B. (1997a). La posición del sujeto en la cláusula monoactancial en español (Lalia, Series Maior, no. 7). Santiago de Compostela: Universidad de Santiago de Compostela. —— (1997b) ‘Aportaciones de la tipología lingüística a una gramática particular: el concepto orden básico y su aplicación al castellano’. Verba, Vol. 24, pp. 45–82. Marcos Marín, F. (1978) Estudios sobre el pronombre. Madrid: Gredos. Moratín, L. F. de (1968) La Comedia Nueva / El Sí de las Niñas, ed. J. Dowling and R. Andioc. Madrid: Clásicos Castalia. Ocampo, F. (1995) ‘The Word Order of Two-Constituent Constructions in Spoken Spanish’, in P. Downing and M. Noonan (eds), Word Order in Discourse. Amsterdam: Benjamins, pp. 425–47.

J. Clancy Clements 133 Ortiz, R. M. (2001) ‘La bitransitividad en español. Estructura y diacronía’, unpublished PhD dissertation. Mexico City: Universidad Nacional Autónoma de México. 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. New York & London: Academic Press, pp. 117–40. Torrego, E. (1984) ‘On Inversion in Spanish and Some of its Effects’. Linguistic Inquiry, Vol. 15, pp. 103–29.

6 Null Direct Objects in Spanish* J. Clancy Clements

Introduction The phenomenon of null objects is, as Schwenter (2005) shows, variable across varieties of Spanish, depending on the factors of animacy and specificity. Null objects are found to varying degrees in Spanish varieties from Mexico, the Andes, Paraguay, and even in Brazilian Portuguese. Schwenter provides us with an excellent discussion of the phenomenon in several varieties of Spanish, first noted by Kany (1945). Our purpose in this chapter is to examine null objects in the standard variety of Castilian Spanish. That is, we address the question of why an overt object pronoun is not required in sentences such as (6.1b) but is necessary in sentences such as (6.2c), without which it is ill-formed. The phenomenon in (6.1b) has been referred to as object drop in the literature (Campos, 1986). We will call it null direct object (DO) pronominalization and will justify this name below. (6.1)

(6.2)

a. ¿Compraste cafe? ‘Did you buy coffee?’ b. Sí, compré. ‘Yes, I bought (some).’ a. ¿Compraste el libro? ‘Did you buy the book?’ b. *Sí, compré. ‘Yes, I bought.’ c. Sí, lo compré. ‘Yes, I bought it.’

* The present study is a revised, refocused, and expanded version of a study that appeared as ‘Notes on topicalization and object drop in Spanish’ in Mazzola (1994: 219–36). 134

J. Clancy Clements 135

Another manifestation of this same phenomenon is found, we argue, in the type of topicalization shown in (6.3b) in which no redundant pronoun is required, while its counterpart in (6.4b) does require it, as is apparent by comparing (6.4b) to (6.4c), which is ill-formed: (6.3)

(6.4)

a. No tomas ni té ni coca cola. ‘You don’t drink tea or coca cola.’ b. Café tampoco tomo. ‘Coffee I don’t drink either.’ a. ¿No bebes esta taza de té? ‘Won’t you drink this cup of tea?’ b. No, pero [el vaso de agua]i sí me loi bebo. ‘No, but the glass of water, yes, I will drink it.’ c. *No, pero [el vaso de agua]i sí me bebo.

We argue that the lack of DO doubling is linked to the semantic nature of the DO. That is, non-referential and non-count DOs (that is, mass nouns and bare plurals) are those that allow the absence of the DO pronoun, as illustrated in (6.1b) and (6.3b). Moreover, we suggest that the cases in question involve a type of Spanish partitive construction analogous to the partitive constructions with en and ne in French, Catalan and Italian.1 Indeed, only indefinite and quantified DOs are affected, as one would expect if one assumes the existence of a Spanish partitive construction. Supporting evidence for the partitive analysis is provided from three other Romance languages. We show that in Catalan, French and Italian, the respective partitive pronouns en, en and ne (in non-adverbial use) occur in precisely the same environments in which Spanish allows null DO pronominalization and topicalization without object doubling. We argue that null DO pronominalization of indefinite DOs is predicted by Hopper and Thompson’s (1980) Transitivity Hypothesis and the Definiteness Hierarchy as proposed by Lazard (1984) and Comrie (1979). The first two authors demonstrate that transitive clauses with indefinite DOs pattern crosslinguistically with intransitive clauses. Accordingly, the expectation for Spanish is that indefinite DOs rather than definite DOs would show no overt marking of pronominalisation because of the lower degree of Transitivity exhibited by clauses containing indefinite DOs. This, in fact, turns out to be the case for Spanish. The data, however, turn out not to be entirely uniform in that the type of construction for expressing the pronominalization of indefinite DOs in Spanish is not obligatory. One finds indefinite DOs pronominalized by what generally are assumed to be definite object pronouns. Badia Margarit (1962: 175, 182) notes, for instance, that the (a) examples of (6.5)–(6.7) may appear without a pronoun, but also can appear with a definite pronoun, as illustrated

136 Null Direct Objects in Spanish

by the corresponding (b) examples of the same sets: (6.5)

(6.6)

(6.7)

a. ¿Tienes dinero? Sí tengo. ‘Do you have (any) money? Yes, I have some.’ b. ¿Tienes dinero? Sí LO tengo. ‘Do you have (any) money? Yes, I have it.’ a. Dame pan si aún tienes. ‘Give me (some) bread if you have (any).’ b. Dame pan si aún LO tienes. ‘Give me (some) bread if you have it.’ a. ¿Has encontrado lana fina? No he encontrado. ‘Have you found (any) fine wool? I haven’t found (any).’ b. ¿Has encontrado lana fina? No LA he encontrado. ‘Have you found (any) fine wool? I haven’t found it.’

We claim that this use of Spanish definite pronouns to pronominalize indefinite DO’s is not ruled out in Lazard’s particular version of the Definiteness Scale. To account for the identical pronominalization of definite and generic NPs in Spanish, we appeal to Givón’s (1984: 407) distinction between the universe of tokens, pertaining to definite (and referential) NPs, and the universe of types, corresponding to generic NPs. Based on data gleaned from CREA, we suggest that the definite-determiner pronominalization of indefinite DOs (understood here as mass-noun and bare-plural DOs) happens within the same speaker’s utterances in one turn in discourse, while null pronominalization of such DOs occurs in utterances of different speakers in discourse. We now turn to the discussion of the partitive construction in four Romance languages and null pronominalization in Spanish.

Marking the partitive in Catalan, French, Italian and Spanish Catalan, French and Italian possess the partitive pronouns en, en and ne respectively, all of which have a wider use than just in the partitive construction. In the purpose at hand, we will concern ourselves with these pronouns only in so far as they pronominalize DOs.2 All three languages make an obligatory distinction between pronominalization of a definite DO and that of an indefinite DO. When speaking of pronominalization of an indefinite DO in the present context, we are referring only to the pronominalization of a mass noun or a bare plural, as in Catalan, or of a PP introduced by de or di that contains a mass noun or bare plural, as in French and Italian.

J. Clancy Clements 137

The partitive pronoun in Catalan may replace an indefinite DO regardless of its number or gender, as the example in (6.8) illustrates: (6.8) Vende en Joan {pasta [fem.sg.]/ pa [masc.sg.]/ peres [fem.pl.]/ libros [masc.pl]}? Si, en vende. (Cat.) ‘Is Joan selling (pasta/ bread/ pears/ books)? Yes, (he) is selling some.’ Similarly, as the examples (6.9)–(6.10) illustrate, both French and Italian partitive pronouns may also replace any DO partitive phrases, regardless of the number or gender of the NP: (6.9) Est-ce que Jean vend {de la glace [fem.sg.]/du beurre [masc.sg.]/des carottes [fem.pl.]/des haricots verts [masc.pl.}? Oui, il en vend. (Fr.) ‘Is Jean selling {ice cream/butter/carrots/green beans}? Yes, he is selling some’. (6.10)

Giovanni vende {della minestra [fem.sg.]/del gelato [masc.sg.]/delle pere [fem.pl.]/dei pomodori [masc.pl.]}? Sì, ne vende. (Ital.) ‘Is Giovanni selling {soup/ice cream/pears/tomatoes}? Yes, (he) is selling some.’

The Spanish analogs to the sentences (6.8)–(6.10) display what we will call a null object, as shown in (6.11): (6.11) ¿Vende Juan {pasta [fem.sg.]/pan [masc.sg.]/peras [fem.pl.]/libros [masc.pl]}? Sí, vende. (Sp.) ‘Is Joan selling {pasta/bread/pears/books}? Yes (he) is selling some.’ Additional evidence is found in the pronominalization quantified DOs.3 Regarding quantified DOs containing numerals, Catalan, French and Italian use their respective partitive pronoun to replace the nonnumeral portion of the NP as in (6.12a–c). In the analogous Spanish example in (6.12d), we find again a null object: (6.12)

a. Si vas a 1’estació a comprar els bitllets, compra’n tres pera nosaltres. (Cat.) b. Si tu vas à la gare pour acheter les billets, achètes-en trois pour nous. (Fr.)

138 Null Direct Objects in Spanish

c. Se vai alla stazione a comprare i biglietti, comprane tre per noi. (Ital.) d. Si vas a la estación para comprar los billetes, compra tres para nosotros. (Sp.) ‘If you’re going to the station to buy the tickets, buy three for us.’ The same correspondence is apparent with a quantified nominal DO. The Catalan, French and Italian sentences in (6.13a–c) contain their respective partitive pronoun, while in Spanish we find again a null object (6.13d): (6.13)

a. Com que avui els pollastres no deuen ser cars, compreu-ne un parell. (Cat.) b. Puisque aujourd’hui les poulets ne doivent pas être chers, achetez-en quelque-uns. (Fr.) c. Poichè oggi i polli non devono essere cari, compratene un paio. (Ital.) d. Como hoy los pollos no deben ser caros, comprad un par. (Sp.) ‘Since today chickens shouldn’t be expensive, buy a couple.’

We encounter the same type of correlation with quantified DOs containing the equivalent of ‘much’, and so on, where the partitive pronoun is used in Catalan, French and Italian (6.14a–c), but where a null object is found in Spanish (6.14d): (6.14)

a. b. c. d.

Tens moltes alumnes? No en tinc gaires. (Cat.) Est-ce que tu as beaucoup d’élèves? Je n’en ai guère. (Fr.) Hai molti alunni? Non ne ho molti. (Ital.) ¿Tienes (a) muchos alumnos? No tengo muchos. (Sp.) ‘Do you have many students? I don’t have many.’

With indefinite pronoun DOs containing the equivalent of English pronominal ‘some’, we find again the same pattern: Catalan, French and Italian exhibit an overt partitive pronoun (6.15a–c), and Spanish displays no overt marker (6.15d): (6.15)

a. Havíeu de coure totes les patates, pero només n’heu cuites algunes. (Cat.) b. Vous deviez faire cuire toutes les pommes de terre, mais vous n’en avez fait cuire que quelques-unes. (Fr.) c. Avevate da fare cuocere tutte le patate, ma non ne avete fatto cuocere che alcune. (Ital.) d. Debíais de cocer todas las patatas, pero sólo habéis cocido algunas. (Sp.)

J. Clancy Clements 139

‘You were to boil all the potatoes, but you only boiled some (of them).’ This very pattern also emerges with DOs containing an adjective phrase, as shown by the sentences in (6.16): (6.16)

a. b. c. d.

Tens cintes blaves? En tinc de grogues. (Cat.) Est-ce que tu as des rubans bleus? J’en ai des jaunes. (Fr.) Hai nastri azzurri? Ne ho di gialli. (Ital.) ¿Tienes cintas azules? Tengo amarillas. (Sp.) ‘Do you have blue ribbons? I have yellow ones.’

Thus, we find strong evidence that, wherever in Catalan, French and Italian we have a partitive pronoun replacing an indefinite DO or a portion of a quantified DO, the corresponding Spanish construction displays null pronominalization. A reasonable deduction based on the above data is that we are dealing in all four languages with partitive pronominalization, the difference being that in Catalan, French and Italian it is overtly marked, whereas in Spanish it is not. These facts lead us to two questions we would like to address: (1) Is the fact that Spanish displays no overt pronominalization for indefinite DOs an isolated phenomenon, or does it fit into a broader picture of how indefinite DOs are marked cross-linguistically? and (2) Why is it that Spanish optionally allows the pronominalization of indefinite DOs with definite object pronouns? That is, why can (6.16d) also appear as (6.17)? (6.17)

¿Tienes cintas azules? Las tengo amarillas. (BADIA MARGARIT, 1962: 182) ‘Do you have blue ribbons? I have yellow ones.’

We offer answers to these questions by appealing to the Transitivity Hypothesis (TH) advanced by Hopper and Thompson (1980), and the Definiteness Scale proposed by Lazard (1984).

Transitivity Hypothesis, Definiteness Scale and the Spanish object pronominalization Defined in the more conventional and traditional way, the notion Transitivity refers to the carrying-over or the transferring of an action from one participant to another. As outlined in Chapter 1, Hopper and Thompson (H&T) show how Transitivity ‘can be broken down into its component parts, each focusing on a different facet of the carrying-over in a different part of the clause’ (1980: 253). Breaking Transitivity down into its different parts allows

140 Null Direct Objects in Spanish

one to characterize clauses in degrees of Transitivity, that is as more or less Transitive, as illustrated in (1.1)–(1.2) in Chapter 1. Using the values ‘high’ and ‘low’ for each of these components, the degree of Transitivity of any given clause can be measured according to how many features of the clause are found in the ‘high’ column in (1.1) or in the ‘individuated’ column in (1.2). The more transitive a clause is, the closer it is to what H&T call Cardinal Transitivity (1980: 253). The general claim of the Transitivity Hypothesis is that high or low values for Transitivity will co-vary within a given clause. That is, ‘[i]f two clauses (a) and (b) in a language differ in that (a) is higher in Transitivity according to any of the features in [(1.1) and (1.2)], then, if a concomitant grammatical or semantic difference appears elsewhere in the clause, that difference will also show (a) to be higher in Transitivity’ (H&T, 1980: 255). With specific reference to their data on indefinite DOs, H&T state that ‘languages which morphologically distinguish between transitive and intransitive clauses, AND between definite and indefinite O[bjects]’s, have a tendency to associate indefinite (that is characteristically unmarked) O’s with INtransitive clauses’ (1980: 259). This quote is relevant for Spanish in the following way. Although Spanish does not distinguish between transitive and intransitive clauses morphologically, it does distinguish them syntactically in terms of word order, as was discussed in some detail in Chapter 5. The position of the subject in the majority of intransitive clauses is postverbal, just as the object of transitive verbs is postverbal. Moreover, if the subject is a bare plural, it may appear in postverbal position, as shown in (6.18a) below, but not in preverbal position, hence the asterisk on sentence (6.18b). Thus, given that DOs in Spanish, and in general, represent new information and are nonreferential considerably more often than transitive-clause subjects, indefinite subjects and direct objects share key defining properties, which would account for the placement of niños ‘children’ in (6.18) (see Chapter 5 for more details): (6.18)

a. Pueden entrar niños al teatro si son acompañados por un adulto. ‘Kids can enter the theatre if they are accompanied by an adult.’ b. *Niños pueden entrar al teatro si son acompañados por un adulto.

Furthermore, both intransitive clauses and transitive clauses with pronominalized indefinite DOs exhibit the same surface structure in terms of overt arguments: without an overt subject both have no overt arguments, with an overt subject both have only one argument. This seems to be a syntactic analog to what is described above by H&T with regard to morphology. That is, Spanish definite DOs are obligatorily and overtly marked (with a

J. Clancy Clements 141

definite article) and are pronominalized obligatorily with overt pronouns. On the other hand, Spanish indefinite DOs (which we repeat are defined here as mass nouns or bare plurals) are not overtly marked with an indefinite pronoun. Thus, they appear either as null objects or are pronominalized with a definite pronoun. According to the Transitivity Hypothesis, obligatory overt marking of definite DOs should co-vary with higher Transitivity, whereas the optional pronominalization of indefinite DOs and the lack of DO with intransitive verbs should co-vary with relatively lower Transitivity. A concrete example bears out these observations and predictions. Below, we have computed the degree of Transitivity for the boldface-typed sentences in (6.19). The intransitive predicate in (6.19a) is a Vendlerian Activity, the transitive predicate in (6.19b) is also an Activity due to the mass term DO, while the transitive predicate in (6.19c) is an Accomplishment due to the count term DO. The relevant values are shown in (6.20) and the results in (6.21):

(6.19)

(6.20)

a. ¿Juan corrió ayer? Sí, (Juan) corrió. ‘Did Juan run yesterday? Yes, (Juan) ran.’ b. ¿Juan compró café ayer? Sí, (Juan) compró. ‘Did Juan buy coffee yesterday? Yes, (Juan) bought (some).’ c. ¿Juan compró el televisor ayer? Sí, (Juan) lo compró. ‘Did Juan buy the television yesterday? Yes, (Juan) bought it.’ (6.19a) (6.19b) (6.19c) A. PARTICIPANTS low high high B. KINESIS high high high C. ASPECT low low high D. PUNCTUALITY low low low E. VOLITIONALITY high high high F. AFFIRMATION high high high G. MODE high high high H. AGENCY high high high I. AFFECTEDNESS OF O low high high J. INDIVIDUATION OF O low low high IND-2 IND-4 NON-IND-4 NON-IND-2 concrete concrete singular singular mass[low] count[high] indefinite[low] def./ref. [high] common common inanimate inanimate

142 Null Direct Objects in Spanish

(6.21)

(6.19a) (w/intransitive verb) 5 high 5 low

(6.19b) (6.19c) (w/indefinite DO) (w/definite DO) 7 high 9 high 3 low 1 low

In terms of Transitivity, (6.19b) is just as close to intransitive (6.19a) as it is to (6.19c), the sentence with a definite DO. In terms of the Vendlerian (1967) predicate types, it is interesting to note as well that although (6.19a) contains an intransitive clause and (6.19b) a transitive one, both are Activities. In other words, both (6.19a) and (6.19b) are semantically compatible with durative contexts and semantically incompatible with telic contexts. This accounts for the well-formedness of the sentences in (6.22) and the ill-formedness of those in (6.23) on an Activity reading: (6.22)

(6.23)

a. Juan pasó tres horas corriendo. ‘Juan spent three hours running.’ b. Juan pasó tres horas comprando café. ‘Juan spent three hours buying coffee.’ a. *Juan tardó tres horas en correr. (on an Activity reading, not an Inchoative reading) ‘It took Juan three hours to run.’ b. *Juan tardó tres horas en comprar café. (on an Activity reading, not an Inchoative reading) ‘It took Juan three hours to buy coffee.’

The two facts – that in terms of Transitivity, transitive clauses with indefinite DOs are situated between transitive clauses with definite DOs and intransitive clauses, and that intransitive clauses and transitive clauses with indefinite DOs are Activities – reflect a pattern that is pervasive across languages: both intransitive clauses and transitive clauses with indefinite DOs are often marked identically. In a survey of 30 languages from 16 different families, Lazard (1984) found many instances in which marking for intransitive verbs and transitive verbs with indefinite DOs is identical.4 We cite here three particularly revealing examples. First, in Hungarian the so-called objective conjugation is used with definite objects while the so-called subjective one is used for indefinite objects (including mass nouns and bare plurals) and intransitive verbs. Crucial here is that the subjective conjugation is thought of as being the conjugation type for intransitive verbs only. Second, in Eskimo, an incorporating language, the grammatical subject in transitive clauses containing a definite DO is marked with a genitive-case suffix and the DO (appearing clause-initially) is morphologically unmarked. In transitive clauses with an indefinite DO, however, the subject is morphologically unmarked, the DO is marked with a model element, and the verb appears as a one-argument verb, agreeing with the subject. This type of grammatical

J. Clancy Clements 143

relation marking is precisely what we find in intransitive clauses, where the intransitive-clause adverbial phrase is marked with a model element, the same way in which the transitive-clause indefinite DO is marked.5 This is shown schematically in (6.24): (6.24) Transitive clause with definite DO: NPDO NPSUBJ-AffixGEN V-AffixSUB-AffixDO Transitive clause with indefinite DO: NPSUBJ NPDO-Model V-AffixSUB Intransitive clause: NPSUBJ ADVP-Model V-AffixSUB The final example involves partitive marking in Finnish, which appears on mass and generic NPs, but not on definite NPs. Based on this and other evidence, Lazard concludes that languages display a certain pattern for marking or not marking definiteness, which is reflected in the Definiteness Scale shown in Table 6.1. The following observation generally holds for the languages surveyed by Lazard: if a point on the scale (for example A, B, C, D, E) is relevant in a given language for marking or not marking a given grammatical or other type of relation, all values (1 through 6) on one side of that point will obtain for marking the pertinent relation and all values from that point on the other side will not obtain for marking it. Referring back to the examples just mentioned, Finnish marks mass and generic nouns with the partitive case, so the crucial point for Finnish is ‘D’. Hungarian marks values 1, 2 and 3 with the subjective conjugation and the values 4 and 5 with the objective conjugation. So, for Hungarian the crucial point on the scale is ‘C’, as it seems to be as well for Eskimo. Interestingly, English mass and generic nouns pattern alike in that only mass nouns, bare plurals and generics (values 5 and 6 on the scale) appear without articles. Thus, the crucial point for English is ‘D’, as with Finnish. For Spanish, the crucial point on Lazard’s Definiteness Scale for null pronominalization of objects is also ‘D’. The Definiteness Scale in Table 6.1 runs into counterexamples involving generic NP marking. In a number of languages, for example the Romance languages and Hungarian, generic NPs and definite NPs are treated identically.

Table 6.1 Lazard’s definiteness scale 1 1st–2nd pronouns A Source:

2

3

4

5

6

3rd pronouns Proper nouns

Definite Human

Indefinite Non-human

Mass

Generic

B

C

D

E

Lazard (1984), p. 283. Reproduced with the permission of Elsevier.

144 Null Direct Objects in Spanish

With specific reference to Spanish, generic NPs pattern like definite NPs in that they are also marked with the definite article, as (6.25a). Moreover, pronominalization of generic NPs also takes place as with definite NPs, as shown in (6.25b): (6.25)

a. Me gustan las manzanas. ‘I like apples.’ b. Me encanta comerlas, sobre todo en otoño. ‘I love eating them, above all in autumn.’

Givón (1984: 407) accounts for this as follows: a referential, definite NP pertains to the universe of Tokens, whereas the generic, definite NP pertains to the universe of Types. That is, a generic NP marked with a definite determiner pertains to a whole category as an abstract type (for example, I like apples, as a type), whereas a definite NP with a definite determiner refers to a set of tokens in the discourse context. Thus, the definite determiner operates on at least two levels of interpretation. In this way, we can also account for definite-pronoun pronominalization of generic as well as definite NPs. There remains, however, the question of pronominalizing mass or bare NPs with a definite pronoun or null pronominalization. The appropriate data for studying this issue are not easy to find. In a database search of indefinite DOs with the verbs beber ‘drink’, comer ‘eat’, comprar ‘buy’, hacer ‘do’, querer ‘do’, and tener ‘have’, we found relevant examples only with tener, which display the following pattern: overt pronominalization with the definite pronoun is sensitive to whether pronominalization takes place within the speech of the same speaker or in a dialogue across speakers. Of the seven examples of DO pronominalization with tener, only two cases, shown in (6.26) and (6.27), display overt pronominalization with the definite article: (6.26) … aunque sí, tengo cáncer de garganta, yo no lo noto … porque Dios está conmigo. ‘although I have throat cancer, yes, I don’t notice it … because God is with me.’ (from CREA, EL MUNDO, 24 September 1994) (6.27) Manuel se sintió herido, pero mantuvo la ecuanimidad: ‘No, tampoco eso. No soy estúpido, Cati. Pero sí esperaba más interés.’ Catalina exclamó, casi a gritos: ‘¡Pues sí tengo interés, claro que lo tengo!’ Manuel felt hurt, but maintained his equanimity: ‘No, it’s not that either. I’m not stupid, Cati. But I did expect more interest from you.’

J. Clancy Clements 145

Catalina exclaimed, almost yelling: ‘But I AM interested, of course I am.’ [lit. yes, I have interest, of course I have it] (from CREA, ROVINSKI, 1993, Herencia de sombras, p. 126) Noteworthy is that in (6.26) the same speaker who is talking about having throat cancer also overtly pronominalizes the relevant NP in the next utterance, and that (6.27) Catalina claims to have interest in something Manuel does, and pronominalizes the relevant NP in her next utterance. In other words, overt pronominalization in these examples occurs within the speech of the same speaker within the same turn in discourse. We conjecture that definite-determiner pronominalization of a mass noun occurs because it is uttered by the same speaker and, secondarily, because of the immediate proximity of the referent and the fact that the referent being pronominalized is the theme of conversation. The key factor seems to be the pronominalization by the same as opposed to a different speaker, given that it does not seem to matter whether the clauses appear adjacently or not. That is, in those cases where two sentences are uttered by two different people in dialogue, we find no definite-determiner pronominalization, even if the sentences are adjacent. The examples illustrating this are shown in (6.28)–(6.29), where the relevant parts are underlined: (6.28)

(6.29)

A: Papá, ¿por qué no tienes pelo en la cabeza? B: Anda, niño, cállate y mira las nubecitas … ¡Y sí tengo pelo! A ¡No tienes! ¡En esta parte no tienes! ¡Se te ve la carne! A: ‘Dad, why don’t you have any hair on your head?’ B: ‘Come on, kid, pipe down, look at the clouds … And I DO have hair’. A: ‘No you don’t. You don’t have any on that part! You can see your skin!’ (From CREA: 2001, Ventajas de ser incompetente y otros monólogos de humor. El club de la Comedia, p. 84) A: ‘¿Tienes tiempo?’ pregunta Andrés pasándole el brazo por la espalda. B: ¿Para qué? A: ¿Tienes o no tienes?6 B: ‘Sí, sí tengo’, dice ella sonriéndose. A: ‘Do you have time’, asks Andrés, putting his hand on her back. B: ‘What for?’ A: ‘Do you or don’t you?’ B: ‘Yes, I do’, she says smiling. (from CREA: MENDOZA, 2002, Satanás, p. 168)

146 Null Direct Objects in Spanish

If definite-determiner pronominalization does not occur between adjacent sentences uttered by different speakers, we would also predict that it will also not be found in those cases of pronominalization across speakers where there are intervening utterances between the indefinite DO and its pronominalization. The data bear the prediction out. In other words, where definitedeterminer pronominalization of an indefinite DO is possible across speakers in dialogue, we find no such overt pronominalization. The relevant examples are given in (6.30)–(6.32), in which the passages are again underlined: (6.30)

(6.31)

(6.32)

Q: ¿Qué es una persona sin sentido del humor? A: Una persona plana, sosa, insípida. Yo creo que sí tengo aunque hay días que me levanto con el pie izquierdo. Q: ‘What is a person without a sense of humor?’ A: ‘A flat, dull, insipid person. I think I have one although there are days when I get up on the wrong side of the bed (lit. with the left foot).’ (from CREA, LA RAZÓN [España], 2 September 2002) Q: Y ¿tiene nietos o tiene … A: Claro que sí tengo. Tengo nueve nietos. ¡Nueve nietos! Tengo cinco hijos. Q: ¿Cuántos hijos? A: Tengo cinco hijos. Q: ‘And do you have grandchildren?’ A: ‘Of course I have. I’ve got nine grandchildren. Nine grandkids! I have five children.’ Q: ‘How many children?’ A: ‘I have five children.’ (from CREA: oral language from Costa Rica in 1992) R: ¿Alguien tiene …? Juanjo ¿tienes desmaquillador? J: No, Rodri, lo siento, no tengo. A mí me lo deja Pilar. M: Yo sí tengo. Espera que me saque el modelito no lo vaya a ensuciar. (from CREA: MIRALLES, 1981, Céfiro agreste de olímpicos embates) R: ‘Does someone have …? Juanjo, do you have any make up remover?’ J: ‘No, Rodri, I’m sorry, I don’t have any. Pilar’s the one who lets me use hers.’ M: ‘I have some. Wait till I get out of the dress so it doesn’t get dirty.’

We note in these examples of inter-speaker null pronominalization, that we find not only mass terms (sentido de humor ‘sense of humour’, tiempo ‘time’, pelo ‘hair’, and desmaquillador ‘make-up remover’), but also an animate bare

J. Clancy Clements 147

plural (nietos ‘grandchildren’). Even the feature of animacy is not strong enough to trigger definite-determiner pronominalization across speakers. In our data, we encountered one apparent counterexample to the pattern we have been discussing, which we give in (6.33). This exchange contains an overt, inter-speaker pronominalization of a topicalized quantified DO (un poquito de vino ‘a little bit of wine’). However, the definite-determiner pronominalization (bébaselo [lit. drink-IMPERATIVE-3SG.REFLEXIVE-it-MASC.SG] ‘drink it down’) does not constitute a bona fide counterexample given that the argument structure of beberse ‘drink up/down’, an ingestion verb with completive aspect, requires an overt, count DO (cf. Clements, Chapter 11, this volume). In other words, in the relevant sentence, T in the dialog does not have the option of saying *¡Bébase a sorbitos! [lit. drink it down in sips] ‘sip a little bit at a time’ because the argument structure of beberse requires the overt pronoun lo in this particular case: (6.33)

a. P: Un poquito de vino en las comidas sí bebo. ‘P: A little bit of wine with meals I do drink.’ b. T: ¡Bébaselo a sorbitos ! ‘Sip it a little at a time.’ (from CREA, OLMO, 1984: 43–4)

To summarize this section, we maintain that it is not coincidental that Spanish transitive clauses with null-pronominalized DOs appear very similar to Spanish intransitive clauses. Indeed, we find evidence of the same phenomenon cross-linguistically, in languages as diverse as Eskimo, Finnish, Hungarian and English. The connection between the structure of intransitive clauses and clauses containing indefinite DO pronominalization is also corroborated by the behaviour of predicates in terms of Vendlerian predicate types. Finally, we have seen that overt or null pronominalization of indefinite DOs (understood here as mass nouns and bare plurals) is sensitive to the distinction between pronominalization within the same speaker’s utterances v. across the utterances of two speakers. The data gleaned from a database analysis suggest that definite-determiner pronominalization of an indefinite DO is found within the utterances of the same speaker, but not in utterances across speakers, where null DO pronominalization is found. In the next section we discuss the connection between topicalization and null DO pronominalization.

Topicalization in Spanish Topicalized DOs, that is sentence-initial DOs, are typically doubled, as shown in (6.34): (6.34)

Esta casai, lai miro cada vez que paso por aquí. ‘This house, I look at it each time I pass by here.’

148 Null Direct Objects in Spanish

However, as Rivero (1980) points out, the lack of doubling of topicalized indefinite DOs, that is mass-term and bare-plural DOs, is common in Spanish. that is, sentences such as (6.33a) and (6.35) are commonly found: (6.35) Y como, aunque leo despacio, buena letra sí tengo, también quiere que pase a limpio las comedias. ‘And as, although I read slowly, I have good penmanship, he wants me to copy the comedies.’ (FERNÁN-GÓMEZ, 1985: 82) In our data collection, we found a number of topicalized DOs with no doubling and all contain an indefinite DO. We mentioned the one in (6.33a); the others are given in (6.40)–(6.41): (6.40) Por eso dirán que soy simple. Sin duda lo soy, pues siempre he evitado las complicaciones. Pero algunos anhelos sí tengo. ‘Because of that, they will say I am simple. I am undoubtedly, because I’ve always avoided complications. But some desires I do have.’ (VARGAS LLOSA, 1988: 183) (6.41) A: Pero [el libro de] Largaespada seguro que lo tiene. Bueno, eso es lo primero que necesitamos. Y también un buen mapa de carreteras. B: Mira, de eso resulta que sí tengo. A: ‘But surely you have [the book by] Largaespada. Well, that is the first thing we need. And also a good road map.’ B. ‘See, it turns out that some of those I do have.’ (SAVATER, 1981: 142) Example (6.40) contains a quantified NP (algunos anhelos ‘some desires’); (6.41) is a partitive DO (de eso ‘of that’), similar to the partitive in French and Italian (see above, 6.9 and 6.10 respectively). These topicalized DOs are most not overtly doubled when they are indefinite NPs. In other words, these cases of null DO pronominalization are analogous to the examples discussed in the previous section and are accounted for in exactly the same way.

Conclusion In some of the generative literature on null direct objects and topicalization, these phenomena have been linked to wh-movement (Rivero 1980 and Campos 1986). In this chapter, we have suggested that both topicalization and null objects are cases of null DO pronominalization. The evidence in favour of positing a null DO pronoun is overwhelming. First, exactly in

J. Clancy Clements 149

those cases where Catalan, French and Italian exhibit an overt partitive DO pronoun, Spanish does not mark anything overtly, which can be reasonably interpreted as a null DO (partitive) pronoun. We have showed that the behaviour of null pronominalization of Spanish indefinite DOs, understood here as mass nouns and bare plurals, is consistent with how indefinites are treated cross-linguistically, in such diverse languages as Eskimo, Hungarian and Finnish. It is thus not surprising that Spanish transitive clauses with pronominalized indefinite DOs and Spanish intransitive clauses have an identical structure given that, in terms of Transitivity features (Hopper and Thompson, 1980), they are close to one another. The similarity in structure between intransitive clauses and clauses containing indefinite DO pronominalization is also corroborated by the behaviour of predicates in terms of Vendlerian predicate types in that both are Activities. Finally, the evidence gleaned from a database search suggests that definite-determiner pronominalization of indefinite DOs is found in utterances of the same speaker within the same turn in discourse, while null pronominalization of such DOs is found across speakers. A more extensive study of corpora would be able to confirm or disconfirm the strength of this tendency. Notes 1 Vallduví (1988) also observes the idiosyncratic nature of topicalization in Spanish and proposes an indefinite pro object to account for it. The present proposal is that certain DOs are not pronominalized and they can be characterized as partitives. Moreover, the phenomenon is part of a broader phenomenon displayed by nondefinite DOs crosslinguistically, which we discuss from a typological perspective in this study. 2 Catalan en can pronominalize not only indefinite objects but indefinite subjects as well, as in (i): (i) Poden entrar nens al teatre? Si, n’hi poden entrar. (Catalan) ‘Can the children go into the theater? Yes, they can.’ In contrast, the pronominalization of indefinite subjects in French (ii) and Italian (iii) is impossible: (ii) *Des enfants peuvent entrer dans le théâtre? Oui, y en peuvent entrer. (French) ‘Can the children go into the theater? Yes, they can.’ (iii) *Dei bambini possono entrare nel teatro? Sì ne possono entrare. (Italian) ‘Can the children go into the theater? Yes, they can.’ The reason for the ungrammaticality of (ii–iii) may well have to do with the presence of de and di v. the absence of it in Catalan and Spanish. The specific difference is still in need of research. 3 We recognize that the examples of quantified DOs in (6.12–6.16) do not constitute cases of null-object proper. Rather, they are examples of the Spanish null partitive. This syntactic difference corresponds to a difference in syntactic behaviour: in (6.11) pronominal reference is possible, whereas in (6.12–6.16) it is not.

150 Null Direct Objects in Spanish 4 The breakdown of languages considered by Lazard (1984) is as follows: Northeast Caucasian languages (3), Equatorial Brazilian languages (1), Indo-Iranian (3), Sino-Tibetan (3), Mayan (1), Austronesian (7), Australian (1), Eskimo (1), Uralic (3), Altaic (1), Wakashan (1), Siberian (1), Andean (1), Semitic (1), Romance (1), Aztecan (1). 5 The use of the term ‘indefinite’ in this particular example refers to NPs with indefinite articles. Nevertheless, the strength of the argument is not therefore diminished. 6 In this example, one could argue that the same speaker null-pronominalizes tiempo. While this may be true, the speaker does so in another turn, not in the same turn.

References Badia Margarit, A. M. (1962) Gramática Catalana. Madrid: Editorial Gredos. Burzio, L. (1986) Italian Syntax. Reidel: Dordrecht. Campos, H. (1986) ‘Indefinite Object Drop’, Linguistic Inquiry, vol. 17, pp. 354–59. Comrie, B. (1979) ‘Definite and Animate Objects: A Natural Class’, Linguistica Silesiana, vol. 3, pp. 13–21. Fernán-Gómez, F. (1985) El viaje a ninguna parte. Madrid: Debate. Givón, T. Syntax. (1984) A Functional-Typological Introduction. Vol. 1. Amsterdam: Benjamins. Hopper, P. and S. Thompson (1980) ‘Transitivity in Grammar and Discourse’, Language, vol. 56, pp. 251–99. Kany, C. (1945) American–Spanish Syntax. Chicago: University of Chicago Press. Lazard, G. (1984) ‘Actance Variations and Categories of the Object’, in F. Plank (ed.), Objects: Toward a Theory of Grammatical Relations. New York: Academic Press, pp. 269–92. Mazzola, M. (1994) Issues and Theory in Romance Linguistics. Washington: Georgetown University Press. Mendoza, M. (2002) Satanás. Barcelona: Seis Barral. Miralles, A. (1981) Céfiro agreste de olímpicos embates (Come y calla, que es cultura). Madrid: Fundamentos. Olmo, L. (1984) Pablo Iglesias. Espectáculo dramático/musical original. Sada, A Coruña: Edición do Castro. Rivero, M. L. (1980) ‘On Left-Dislocation and Topicalization in Spanish’, Linguistic Inquiry, vol. 11, pp. 363–93. Rovinski, S. (1993) Herencia de sombras. San José, Costa Rica: REI Centroamérica. Savatér, F. (1981) Caronte aguarda. Madrid: Ediciones Cátedra. Schwenter, S. (2005) ‘Null Objects Across South America’, in T. L. Face and C. A. Klee (eds), Selected Proceedings of the 8th Hispanic Linguistics Symposium, Somerville, MA: Cascadilla Proceedings Project. Vallduví, E. (1988) ‘Functional Load, Prosody, and Syntax: Left-Detachment in Catalan and Spanish’, Chicago Linguistic Society, vol. 24, pp. 391–404. Vargas Llosa, M. (1988) Elogio de la madrastra. Barcelona: Tusquets. Vendler, Z. (1967) Linguistics in Philosophy. Ithaca: Cornell University Press.

7 Transitivity and the Syntax of Inalienable Possession in Spanish* Richard Winters

Introduction Inalienable possession is the linguistic term used to classify the relation of possession that exists between a possessor and possessum ‘object possessed’ when the later cannot be easily separated, or conceived of apart, from the possessor. Chappell and McGregor (1996) observe that the semantic classes of nouns associated with alienable possession generally denote body parts, clothing, kinship, and tools or weapons closely associated with their possessor(s), with spatial relations also falling into this category. The consideration of these noun classes as inalienable does vary cross-linguistically, they note, with some being included or excluded by particular languages. Kinship terms in some Australian languages are treated as alienable, according to Dixon (1980), while in Ewe, body-part nouns are not treated as inalienable nouns (Ameke, 1996). In Spanish, there is no semantic class of nouns that is always treated as inalienable. In general, however, nouns denoting body parts and clothing are those that are most consistently encoded as inalienable in Spanish, with more variation on the part of other semantic classes of nouns prototypically considered to be inalienable. This variation in encoding does not extend from the semantic classes, but rather from other factors such as verbal semantics and the grammatical function of the possessum. Extralinguistic factors, such as a particular speaker’s perception of an event, may influence the encoding as well (Dumitrescu, 1990). In Spanish, inalienability is typically encoded through the use of a definite or an indefinite determiner in the place of a possessive adjective modifying the inalienable noun. Sometimes there is an accompanying dative clitic co-referent with the possessor. Constructions with this dative clitic have traditionally been labelled dative possessive. The broader term possessor ascension reflects that cross-linguistically there are other grammatical means to encode * This chapter has benefited from helpful comments from Clancy Clements and two anonymous reviewers. All errors remain my own. 151

152 Transitivity and the Syntax of Inalienable Possession in Spanish

this phenomenon, such as incorporation, where the possessor raises (or appears to raise) from its original position to a higher one. As an illustration of the possibilities for encoding inalienability in Spanish, consider the examples in (7.1).1 Note that in every case the possessed object is accompanied by a definite article: (7.1)

a. Susana levantó la mano. the hand Susana raised.3S ‘Susana raised her hand.’ b. Arturo le tocó el brazo a Carmela. Arturo CL.3S.DAT touched.3S the arm to Carmela ‘Arthur touched Carmela’s arm.’ c. Miguel se lavó el pelo. Miguel REFL.3 washed.3S the hair ‘Miguel washed his hair.’ d. Luis me quebró un dedo. Luis CL.1S.DAT broke.3S a finger ‘Luis broke one of my fingers.’ e. *Marta me abrió la puerta.2 Marta CL.1S.DAT opened.3S the door Intended meaning: ‘Marta opened my door.’

In (7.1a), lacking any other context, la mano ‘the hand’ is interpreted as Susana’s hand. In (7.1b), a possessive relation between el brazo ‘the arm’ and the dative clitic me ‘to me’ obtains, yielding the interpretation that it was my arm which was touched. In (7.1c), the dative clitic is also reflexive, with the interpretation that Miguel washed his (own) hair.3 Sentence (7.1d) illustrates that an indefinite article may also participate in this construction, with a partitive reading of the possessed noun in question, as people generally have ten fingers. This sentence exemplifies another property of possessive dative sentences, namely that they also have a benefactive/malefactive, as well as a possessive reading. Thus, in (7.1d) there is also the reading of Arturo broke his finger (and I was affected by that), although the gloss shown for (7.1d) is the one most speakers would have as the default interpretation. Additionally, as Roldán (1972) and García (1975) note, while possessive sentences have a benefactive/malefactive reading, the converse is not necessarily true; that is, a benefactive sentence might not have a possessive reading, which is the case we see in (7.1e). (7.1e) is not acceptable to most speakers as meaning Marta opened my door; rather, it has the interpretation of Marta opened the door for me. This may be attributed to the unlikelihood that door could be considered as something inalienably possessed. The dative possessive in particular poses something of a syntactic puzzle. While the possessor relation in sentences like (7.1a) can be deduced through

Richard Winters 153

knowledge of the world and context, sentences like (7.1b) and (7.1d) appear to have too many arguments, considering the valency of the verbs in question. That is, touch and break typically have two arguments, an agent and a patient, but not a benefactive/malefactive argument. What, then, determines whether and when the ‘extra’ dative argument – with possessive interpretation – may appear? If we examine the data further, more discrepancies arise. Even though the (in)definite article, with or without a dative clitic, is typically used in constructions involving inalienable nouns, this is not always the case. Kliffer (1983) points out that the sentence in (7.2) may have an inalienable reading or not, depending on the context: (7.2)

Mariana se quitó la falda. Mariana REFL.3 took.off.3S the skirt ‘Mariana took off the skirt.’ / ‘Mariana took off her skirt.’

Without additional context, the default interpretation is that the skirt is Mariana’s. However, as Kliffer states, if we know that Mariana is shopping for a skirt and has just tried this one on, then the possessive interpretation does not obtain. Also, the inalienable nature of a noun does not preclude it being accompanied by a possessive adjective instead of a determiner. Note that (7.3a) is pragmatically marked, while (7.3c) is the most neutral way to convey this information: (7.3)

a. Tus ojos son azules. your eyes are.3P blue ‘Your eyes are blue.’ b. *Los ojos son azules. the eyes are.3P blue Intended meaning of (7.3a). c. Tienes los ojos azules have.2S the eyes blue ‘You have blue eyes.’ / ‘Your eyes are blue.’

These, then, are the issues involved in encoding inalienable nouns in Spanish. In the next section I review the factors previous investigations have identified as affecting the encoding of the possessor relation in Spanish, focusing on the alternation between a dative possessor and a possessive determiner. I than discuss how this encoding provides additional evidence in support of Hopper and Thompson’s Transitivity Hypothesis,4 and discuss an empirical analysis of Spanish data that lends further support to a link between the choice of possessive type and Transitivity. In the final section concluding remarks are presented.

154 Transitivity and the Syntax of Inalienable Possession in Spanish

The syntax of inalienability and the dative possessive in Spanish While sentences such as (7.1a) or (7.3c) rely on the context of the discourse or one’s ‘knowledge of the world’ to identify the possessors, in the sentences in (7.1b–d), a dative clitic co-referent with the possessor appears. What is the function of this clitic, especially when it appears in sentences where it may alternate with a possessive determiner? Consider the examples in (7.4): (7.4)

a. Oigo tu voz. hear.1S your voice ‘I hear your voice.’ b. Te oigo la CL.2S.DAT hear.1S the ‘I hear your voice.’

voz. voice

Dumitrescu (1990) observes that these sentences are not interchangeable, although they do communicate the same basic information. In the following section I will discuss factors that influence the choice between these two options of encoding the possessor. Various factors have been identified in previous research as influencing how inalienable nouns are treated in Spanish, as well as in other Romance languages. The referentiality of the part and its autonomy from the whole, the affectedness of the whole, topicality and focus have all been identified as influencing the choice between (in)definite articles and possessive modifiers.5 Vergnaud and Zubizarreta (1992) argue that inalienable nouns accompanied by the definite article in dative possessive sentences are non-referential. A reflex of this in the grammar is that they cannot be modified by nonrestrictive (attributive) adjectives, as first noted in French by Kayne (1975).6 Such nouns may be seen as inseparable from the whole and have no referentiality on their own apart from the whole. Consider the examples given in (7.5): (7.5)

a. *Rosa le lavó la cara sucia a la niña. Rosa CL.3S.DAT washed.3S the face dirty to the girl Intended meaning: ‘Rosa washed the girl’s dirty face.’ b. Rosa le lavó la mano izquierda a la niña. Rosa CL.3S.DAT washed.3S the hand left to the girl ‘Rosa washed the girl’s left hand.’ c. Rosa lavó la cara sucia de la niña. Rosa washed.3S the face dirty of the girl. ‘Rosa washed the girl’s dirty face.’ d. Rosai lavó suj cara sucia. Rosa washed.3S her face dirty ‘Rosa washed her dirty face.’

Richard Winters 155

Example (7.5a), in which the inalienable noun is modified by a non-restrictive adjective, is ungrammatical in this dative possessive construction. The only grammatical reading is the one in which the girl has two or more faces, one of which is dirty. The sentence in (7.5b) contains a restrictive adjective, izquierda ‘left’, which differentiates one of the two possible hands the girl has. In contrast to (7.5a), both (7.5c) and (7.5d) illustrate the possibility of modifying the possessed noun when the possessor is genitive or a possessive adjective. Greater autonomy of the part from the whole is associated with the likelihood of using a possessive modifier. Spanoghe’s (1995) statistical analysis of body-part nouns in Spanish literary works finds that there is a slight preference for use of the possessive when the body-part noun is the subject and an overwhelming preference for the article with the body-part noun is the direct object.7 As a subject, the noun part has greater autonomy from the whole than when it is a direct object, this difference being encoded by the choice between possessive and article; the greater the autonomy, the more likely the use of a possessive adjective. Spanoghe also considers the possessive dative construction to highlight the affectedness of the possessor, a claim also argued for by García (1975) and Kliffer (1973, 1979, 1983, 1987). Along these lines, Dumitrescu (1990), comparing the possessive dative in Spanish and Romanian, suggests that the dative possessive is used to show that the possessor is implicated in the event, that the possessor is an involved party to the event. Dumitrescu also argues that the dative possessive appears when the possessor is topical, and that the possessive modifier is used when the possessum is topical. This concords with Manoliu-Manea’s (1990, 1996) and Popescu-Ramírez and Tasmowski-De Ryck’s (1988) analyses of the possessive dative in Romanian, which encodes topicality of the possessor in a similar fashion to that of Spanish. Popescu-Ramírez and Tasmowski-De Ryck’s analysis of Romanian treats topicality and focus in terms of thematicity and rhematicity, with a rhematic, or focused, status of the possessum resulting in a greater probability that the possessive adjective will be used. In turn, this concurs with similar research by Barnes (1980, 1985) for French and Levine (1984) for Russian. In summary, the syntax of inalienable nouns in Spanish can be linked to the level of referentiality and autonomy of the possessum, the affectedness of the possessor/whole, and the topical or focused nature of the possessor and possessum. A topical and/or highly affected possessor is likely to surface as a dative clitic and a non-referential, low autonomy part is likely to be modified by an article. A focused or topical possessum, referential and highly autonomous, is likely to be modified by a possessive adjective. In the next section we shall see how the notion of Transitivity aides us in predicting whether a possessor is likely to be encoded as a dative argument or as a possessive determiner.

156 Transitivity and the Syntax of Inalienable Possession in Spanish

Transitivity and possessive constructions in Spanish Hopper and Thompson correlate high Transitivity with foregrounding, and thus in discourse foregrounded elements are likely to be in (relatively) highly Transitive constructions, while backgrounded elements are likely to be in constructions with (relatively) low Transitivity. Further, the factors that determine the relative level of Transitivity of a sentence are those that play a role in influencing the choice of a dative possessor or a possessive determiner. Transitivity and choice of possessive type As we have seen in the previous section, the possessive dative construction correlates with a topical possessor, while an inalienable noun modified by the possessive adjective correlates with topicality or focus (thematic or rhematic status); their status also correlates with foregrounding. Following Dumitrescu’s (1990) observation that these two constructions are not interchangeable, we can conclude that Spanish has opted for the dative possessive to foreground the possessor, correlated with relatively higher degree of Transitivity, and modification of the possessum by a possessive adjective to foreground the inalienable part, with a relatively lower degree of Transitivity. Consider the two sentences in (7.6), which illustrate a contrast between the dative possessive and the possessive determiner in Spanish with regard to Transitivity: (7.6)

a. Juana me pisó el pie. Juana CL.1S.DAT stepped.3S the foot ‘Juana stepped on my foot.’ b. Tus dientes penetraron la envoltura de plástico. Your teeth penetrated.3p the wrapping of plastic ‘Your teeth penetrated the plastic wrapping.’

While both sentences are highly Transitive according to Hopper and Thompson’s parameters in Table 1.1 in Chapter 1 (7.6a) is slightly higher in Transitivity than (7.6b). Where the parameters differ is in the number of participants, the affectedness of the object, and the individuation of the object. The differences between these two sentences vis-à-vis the Transitivity parameters emphasize the presence and participation of the possessor, giving it salience. As Manoliu-Manea (1980) points out, our conception of the world is human-centred. Accordingly, it should be expected that [human] elements are more salient than [human] elements. This accounts for the slightly higher Transitivity that (7.6a) has, in comparison to (7.6b). With regard to the individuation of the object, the difference between the two sentences tilts in favour of the prominence of the possessive adjective construction. The object in (7.6a) is less referential than the object in (7.6b), for the reasons discussed above. This can be seen as a way to highlight the autonomy

Richard Winters 157

of the possessed item from its possessor. In (7.6a), the possessor is more salient than the possessum, while the possessum has greater saliency in (7.6b). Empirical evidence To further support the claim that the factors determining the relative level of Transitivity are those that influence the choice of possessive type, Winters (2004) provides a statistical analysis of empirical data that correlates factors yielding relatively higher levels of Transitivity with the factors favouring the use of a dative possessor. This study, which analysed dialogue from eight Spanish-language plays from the latter half of the twentieth century, found several variables that were significantly associated with the choice of one type of possessor or the other, at confidence levels of 95 per cent or greater. Among the variables associated with the choice of a dative possessor were dynamic telic verbs, the possessum as a direct object, and body-part nouns. Variables associated with the choice of a possessive determiner were stative verbs, the possessum as a copula predicate, and nouns referring to abstract possessions (for example, liberty, happiness, love) and interpersonal relationships. Such results are not unexpected, given Hopper and Thompson’s Transitivity Hypothesis. As the dative clitic serves to highlight the possessor’s participation or affectedness, its presence in a sentence increases the relative level of Transitivity. It is not surprising then that dynamic telic verbs, possessa as direct objects, and body-part nouns are associated with the appearance of a dative possessor. Dynamic telic verbs are likely to affect their objects, in turn affecting the possessor of the object. If the possessum is the direct object, then it follows that the possessor will be affected if the possessum is. Moreover, affected body-part nouns are very likely to have affected possessors as well. On the other hand, a possessive determiner serves to reflect an unaffected or non-participating possessor. Neither objects of stative verbs nor copula predicates are affected by the verbs in question, and therefore will not have affected possessors. Actions affecting nouns referring to abstract possessions and interpersonal relationships are less likely to have a (physical) effect on their possessors. Summary The dative possessive construction, serving to foreground the possessor, derives a higher level of Transitivity in the sense of Hopper and Thompson through the addition of a third participant argument, the possessor, while the use of the definite article with the possessed noun lowers its autonomy, referentiality and individuation, resulting in the possessor appearing as an entity both affected by, and involved in, the action. The use of a possessive adjective serves to foreground the inalienable noun by increasing its autonomy, referentiality and individuation and decreasing the possessor’s prominence. The slightly higher level of Transitivity possessed by the dative possessive

158 Transitivity and the Syntax of Inalienable Possession in Spanish

construction can be attributed to the salience accompanying the [human] dative clitic. Therefore, the factors that contribute to the relative level of Transitivity of a clause correspond to those that influence the choice of one possessive type. That this is the case is further shown by the empirical study in Winters (2004), which found that certain variables in semantics of the nouns and verbs in a sentence, and the grammatical function of the possessum are associated at a significant level of confidence with the choice of possessor type, and that these variables correspond to factors determining level of Transitivity.

Conclusion Past researchers have argued that the syntax of inalienable nouns in Spanish encodes the referentiality and autonomy of the possessum, and the topicality or focused nature of the possessor and possessum, factors that correlate with the parameters of Hopper and Thompson’s (1980) notion of Transitivity. In this chapter the correlation of the topicality of the possessor with the use of the dative possessive, and the correlation of the use of the possessive adjective with the topicality or focus of the possessum, has been linked to Hopper and Thompson’s Transitivity parameters. By adding a third argument to the verb in the form of a dative clitic, the dative possessive construction raises the level of Transitivity and foregrounds the possessor, giving salience to its affectedness and involvement as well as increasing the number of participants and decreasing the referentiality of the possessum. The use of the possessive adjective, on the other hand, foregrounds the possessum by increasing its autonomy and referentiality and by decreasing the prominence of the possessor. The syntactic behaviour of these possessive constructions with inalienable nouns provides additional support for Hopper and Thompson’s hypothesis. This claim is further supported by a statistical analysis of empirical data from Winters (2004), showing that factors that are associated with influencing the choice between a dative possessor and a possessive determiner are among those that determine the relative level of Transitivity. Notes 1 The following abbreviations will be used in the glosses in this chapter: 1  1st person S  singular CL  clitic REFL  reflexive 3  3rd person P  plural DAT  dative 2 This sentence is grammatical with the interpretation of ‘Marta opened the door for me.’ 3 Kliffer (1983), among others, observes that reflexive clitics are required in Spanish when the action is performed on or to the possessum (generally body parts and clothing), while actions performed with the possessum do not require reflexive clitics. Following Herschensohn (1975), Kliffer terms the non-reflexive verbs

Richard Winters 159

4 5

6

7

pseudo-transitives, and following Diffloth (1974), he terms the reflexive verbs reflexive body movements. Junker and Martineau (1987) present a similar analysis for French verbs. See Chapter 1 for a detailed presentation on this hypothesis. Stylistics have also been identified as a factor. Hatcher (1944a, 1944b) suggests that in literary French using a possessive modifier where an article is expected breaks away from the cliché-like nature of the article. The use of the possessive modifier give a more literal reading as well. Kliffer (1983) compares contemporary literary usage of articles and possessives with their usage by native speakers, finding that possessives are used much more frequently by authors than in (non-literary) speech, supporting Hatcher’s observations. Stylistics will not be further addressed in this chapter. Spanoghe (1995) finds a few instances of Spanish literary usage where Kayne’s observation does not hold for postnominal nonrestrictive adjectives, but such examples are rare. It may be that these are due to stylistic choices on the part of the authors whose works she analysed. Popescu-Ramírez and Tasmowski-De Ryck (1988) report similar findings in Romanian for the possessum as a direct object; with regard to the possessum as subject, Romanian shows a pronounced preference for the possessive adjective.

References Ameke, F. (1996) ‘Body Parts in Ewe Grammar’, in Chappell and McGregor, op. cit., pp. 783–840. Barnes, B. K. (1980) ‘The Notion of “Dative” in Linguistic Theory and the Grammar of French’, Lingvisticae Investigationes, vol. 4, pp. 245–92. —— (1985) ‘A Functional Explanation of French Nonlexical Datives’, Studies in Language, vol. 9, pp. 159–95. Chappell, H. and W. McGregor (1996) ‘Prolegomena to a Theory of Inalienability’, in Chappell and McGregor, op. cit., pp. 3–30. —— (eds) (1996) The Grammar of Inalienability: A Typological Perspective on Body Part Terms and the Part–Whole Relation. Berlin: Mouton de Gruyter. Diffloth, G. (1974) ‘Body Moves in Semai and in French’, Chicago Linguistic Society, vol. 10, pp. 128–38. Dixon, R. M. W. (1980) The Languages of Australia. Cambridge: Cambridge University Press. Dumitrescu, D. (1990) ‘El dativo posesivo en español y en rumano’, Revista Española de Lingüística, vol. 20, pp. 403–29. García, E. (1975) The Role of Theory in Linguistic Analysis: The Spanish Pronoun System. Amsterdam: North Holland. Hatcher, A. G. (1944a) ‘Il me prend le bras vs. il prend mon bras’, Romanic Review, vol. 35, pp. 156–64. —— (1944b) ‘Il tend les mains vs. il tend ses mains’, Studies in Philology, vol. 41, pp. 457–81. Herschensohn, J. (1975) ‘An Interpretive Approach to Inalienable Possession in French’, ERIC Microfiche no. 123912. Washington: Center for Applied Linguistics. Hopper, P. J. and S. A. Thompson (1980) ‘Transitivity in Grammar and Discourse’, Language, vol. 56, pp. 251–99. Junker, M. and F. Martineau (1987) ‘Les possessions inaliénables dans les constructions objet’, Revue Romane, vol. 22, pp. 194–209.

160 Transitivity and the Syntax of Inalienable Possession in Spanish Kayne, R. S. (1975) French Syntax: the Transformational Cycle. Cambridge, MA: MIT Press. Kliffer, M. D. (1973) The Spanish Dative: Selected Syntactic and Semantic Problems. Ithaca, NY: Cornell University Ph.D. dissertation. —— (1979) ‘Levanto la mano / Me lavo las manos / Me levanto la mano’, Canadian Modern Language Review, vol. 35, pp. 217–26. —— (1983) ‘Beyond Syntax: Spanish Inalienable Possession’, Linguistics, vol. 21, pp. 759–94. —— (1987) ‘Los sustantivos intrínsecamente relacionales: un examen multi-nivelístico’, Revista Española de Lingüística, vol. 17, pp. 283–99. Levine, J. S. (1984) ‘On the Dative of Possession in Contemporary Russian’, Slavic and East European Journal, vol. 28, pp. 493–501. Manoliu-Manea, M. (1980) ‘Pragmatics of Romance Possessives: Romanian as an Affectively Marked Language’, Southeastern Europe, vol. 7, pp. 1–10. —— (1990) ‘Ethno-syntax and Discourse: Inalienability and Topicality in Romanian’, Revue Roumaine de Linguistique, vol. 35, pp. 323–29. —— (1996) ‘Inalienability and Topicality in Romanian: Pragma-semantics of Syntax’, in Chappell and McGregor, op. cit., pp. 711–43. Popescu-Ramírez, L. and L. Tasmowski-De Ryck (1988) ‘Thématicité et possessivité en roumain’, Lingvisticae Investigationes, vol. 12, pp. 303–35. Roldán, M. (1972) ‘Concerning Spanish Datives and Possessives’, Language Sciences, vol. 21, pp. 27–32. Spanoghe, A.-M. (1995) La syntaxe de l’appartenance inaliéable en français, en espagnol et en portugais. Frankfurt am Main: Peter Lang. Vergnaud, J.-R. and M.-L. Zubizarreta (1992) ‘The Definite Determiner and the Inalienable Constructions in French and English’, Linguistic Inquiry, Vol. 23, pp. 595–652. Winters, R. A. (2004) The Syntax of Inalienable Possession in Spanish and Romanian. Bloomington: Indiana University Ph.D. dissertation.

8 Ser-estar in the Predicate Adjective Construction J. Clancy Clements*

Introduction An overall analysis of copula choice in the Spanish ‘copula  predicate adjective’ construction will need to account, at least, for the distribution of ser and estar found in the examples in (8.1)–(8.4): (8.1)

(8.2)

(8.3)

(8.4)

a. El ser humano es /*está mortal. ‘The human being is mortal.’ b. Juanita es /*está inteligente. ‘Juanita is intelligent.’ c. Sus acciones son necesarias. ‘His/her/their actions are necessary.’ d. El exilio es preferible a la muerte. ‘Exile is preferable to death.’ a. El empleado es/está sincero. ‘The employee is sincere.’ b. El empleado es/está cortés.1 ‘The employee is polite.’ a. Juan es/está muy alto. ‘Juan is very tall.’ b. La chica/la casa es/está muy bonita. ‘The girl/the house is very pretty.’ c. La sopa es/está muy sabrosa. ‘The soup is very tasty.’ a. Adriana *es/está levantada. ‘Adriana is standing.’ b. El nene *es/está descalzo. ‘The little boy is barefoot.’

* Special thanks to Kimberly Geeslin for her comments on this study. Of course, all errors and oversights remain mine. 161

162

Ser-estar in the Predicate Adjective Construction

c. La máquina *es/está rota. ‘The machine is broken.’ d. La llegada está prevista para las tres. ‘The arrival was anticipated for three o’clock.’ Ser, but not estar, combines with adjectives such as mortal ‘mortal’, inteligente ‘intelligent’, necesario ‘necessary’, and preferible ‘preferable’; both ser and estar appear with adjectives like sincero ‘sincere’, alto ‘tall’, bonito ‘pretty’, and sabroso ‘tasty’; and estar, but not ser, combines with adjectives such as levantado ‘standing’, descalzo ‘barefoot’, roto ‘broken’, and previsto ‘anticipated’. In this chapter, we ask which semantic and pragmatic factors of the subject referents, the copulas and the different adjective types can be appealed to in order to account for copula distribution in (8.1)–(8.4), and the different interpretations of the copula sentences in (8.2)–(8.3), where both ser and estar are possible. In the following section we provide a semantic characterization of ser and estar, following which we distinguish the two copulas using the feature [aspect]: ser does not carry the feature and estar carries it. We then examine two analyses of adjectives – one based on the distinction between individual- v. stage-level predicates and another taken from Clements (1988) – to test the extent to which each predicts the distribution of adjective types with ser and estar. The distribution of the copulas with second-order subject referents is then discussed, and our conclusions and final remarks are found in the last section.

Semantic characterization of ser and estar2 We begin by discussing ser and estar as main verbs. As a main verb, the core meaning of ser is ‘exist’, and in reference to activities and events, ‘take place’. As the three examples in (8.5) illustrate, ser ‘exist’ appears with one argument (the subject) and no predicative nominal: (8.5)

a. Dios es. ‘God is (exists).’ (FALK, 1979: 25) b. Y el tiempo, una palabra volátil, frágil, etérea y veloz para quienes pertenecen a él, pues para otros el tiempo simplemente es, ni va ni viene. ‘And time, a volatile, fragile, ethereal, and fleeting word for those who belong to it, because for others time simply is, it neither goes or comes.’ (CREA) c. La otra vía llevaba a las cosas mismas, por emplear una repetida fórmula de Husserl. Fue el camino de los escritores del nouveau roman. Para ellos, el mundo ni es absurdo ni está lleno de sentido, simplemente es.

J. Clancy Clements 163

‘The other way carried the things themselves, to use a well-worn formula by Husserl. It was a way of the writers of the ‘nouveau roman’. For them, the world is neither absurd or full of meaning, it simply is.’ (CREA) In the reading of ser as ‘take place’, the subject refers to an event or activity. Illustrative examples are given in (8.6). (8.6)

a. La llegada del tren será a las 21 horas 15 minutos. ‘The arrival of the train will take place at 21 hours 15 minutes.’ b. La celebración es hoy a las cuatro. ‘The celebration is today at four o’clock.’

Ser is also used as a copula to express an intrinsic identification between the referent and an attribute, a substantive, or a pronoun (modified or not). This is shown in (8.7):3 (8.7)

a. Ella es hermosa. ‘She is beautiful.’ b. Ella es profesora. ‘She is a teacher.’ c. La profesora es ella. ‘The profesor is her.’ d. Ella es la que ganó el premio. ‘She is the one who won the prize.’

To understand the semantic content of estar, a brief look into its etymology is useful. Estar derives from Latin stare ‘stand’. According to Falk (1979: 116–17), estar in medieval Spanish had both the meaning hallarse en el espacio en posición vertical ‘be in (a) space in a vertical position’, as well as that of hallarse inmovilizado, persistir en un estado ‘be immobilized, remain in a state’. In modern Spanish, the core meaning of estar is ‘be located’, the essential notion here being located in space. For example, the standard manner in Spanish of asking for someone (for example, Pedro) on the telephone is ¿Está Pedro?, for which the translation in English, German and French would respectively be, ‘Is Pedro there?’, ‘Ist Pedro da?’, ‘Pedro est-il la?’ In these languages, the presence of the locative adverb ‘there’ is necessary, whereas in Spanish its presence is superfluous because, we argue, the semantics of estar contains a feature which we will for now call [locative]. Another piece of evidence supporting the view of a [locative] feature in estar involves the optional nature of an adverbial as a complement of estar as main verb whenever reference to a location is understood from the context, as in (8.8a). However, if the same sort of deletion takes place in an estar

164

Ser-estar in the Predicate Adjective Construction

copula construction, the resulting sentence, (8.8b), is at best strange: (8.8)

a. Hemos buscado por toda la casa y Pedro no está ø. ‘We have looked all through the house and Pedro isn’t (here).’ b. *Estas manzanas están verdes y aquella está también. ‘These apples are green and that one is too.’ c. Estas manzanas están verdes y aquella lo está también. ‘These apples are green and that one is too.’ d. Estas manzanas están verdes y aquella también. ‘These apples are green and that one too.’

Note that for (8.8b) to be entirely felicitous, either the pronoun lo must be inserted, as in (8.8c), referring to the colour being predicated of aquella ‘that one (apple)’, or está has to be deleted, as in (8.8d). The respective gapping restrictions are different for Spanish than for languages such as English, French and German, a fact that we can account for if we assume that estar has as its basic meaning ‘be located’. From a diachronic perspective, the first contexts in which estar is found in the history of Spanish involve, as already mentioned, the idea of location, or a physical situation (cf. Hanssen, 1912; and Bouzet, 1953, in Fernández Leborans, 1999: 2,422–3). Thus, both on synchronic and diachronic grounds, the evidence suggests the presence of the [locative] feature in estar. We assume, therefore, that the semantic make up of estar contains the semantic feature [locative]. As a copula in the ‘estar  adjective’ construction, this verb is used to identify the referent with the predicate complement. Here, the semantics of estar does not necessarily express the location of the referent, (though this is possible), but rather it expresses a connection to another situation involving the referent, either assumed or expected, depending on the discourse situation. (8.9)

a. Las nubes están altas. ‘The clouds are high.’ b. Sus ojos estaban rojos. ‘His/her eyes were red.’

(VAÑO-CERDÁ, 1982: 145) FALK, 1979: 66)

Thus, just as in (8.9a) a connection to the location of the referent (las nubes ‘the clouds’) within its natural frame of reference (the sky) is assumed for a felicitious reading (see Vaño-Cerdá, 1982: 146) (whereby location here is viewed as variable and not a classifying characteristic), so it is in (8.9b) that for a felicitious interpretation a connection to a prior, assumed or anticipated state of the referent is assumed in which a classifying characteristic of the referent is not his/her red eyes. The claim is, then, that the semantics of estar contains something like a [locative] feature that supports this assumption.

J. Clancy Clements 165

We can say that there is, in connection with the semantic feature [locative], the presence of a presupposition of some prior state of affairs. This [locative] feature in the lexical configuration of estar is the distinguishing semantic trait between the core meanings of estar and ser. This difference in the semantics of ser and estar is the basis for certain Pragmatic Presuppositions, defined by Stalnaker (1974: 198) as ‘propositions whose truth [the speaker] takes for granted, or seems to take for granted, in making his statement’. In the ‘copula  adjective’ construction, the [locative] is assumed to be present but as a more general feature. Clements (1988) proposes the pragmatic feature [nexus], suggesting that just as [locative] of estar’s core meaning implies a connection to a location, [nexus] implies a connection to some other situation, assumed or expected. On this analysis, ser would carry no relevant marker or the [nexus] in its lexical feature bundle and estar would have the [nexus] feature. Other scholars, such as Navas Ruiz (1963), Luján (1980, 1981) and Hernanz (1988), adopt the semantic [aspect] to distinguish the two copulas: ser is [aspect] and estar is [aspect]. Schmitt (1992) uses the notion of aspect to distinguish the two copulas, maintaining that ser is aspectually inert, while estar is marked for aspect. All these analyses contain the same essential observation, namely, that estar is the marked copula in the sense that it is marked for aspect and ser is unmarked or negatively marked for the same feature. In our discussion, we assume that ser and estar differ with regard to the [aspect] feature in that estar carries the [aspect] feature and ser does not. Having characterized ser and estar, let us now turn to the semantic characterization of the referent and the adjective in the ‘copula  adjective’ construction.

The role of the subject referent and the adjective in the ‘copula  adjective’ construction With regard to the nature of the subject referent, we will consider the effect of animacy and the pragmatic assumption about whether a given subject referent is able or not to undergo change with respect to a given position in space or a given state. While animacy will seem important in some cases, the pragmatic assumption is probably the most general overarching consideration. For example, if there is a situation (with a lot of contextual support) in which a sentence such as Marta está venezolana ‘Marta is Venezuelan’ is interpretable in some way other than that Marta is a Venezuelan national – for instance as a behaviour typical of Venezuelans – the sentence becomes relatively more acceptable, in spite of the fact that the combination estar  venezolano is rarely found. Although there are specific rules that contravene the pragmatic assumption (for example all derived adjectives in ble appear only with ser), it is a good general principle that has wide use. For the semantic characterization of adjectives, we discuss two analyses: one based on the distinction between inidividual-level and stage-level predicates

166

Ser-estar in the Predicate Adjective Construction

and the other based on whether or not for any given adjective an underlying event or activity can be assumed. Individual-level and stage-level predicates in the Spanish ‘copula  adjective’ construction The now well-known distinction between individual-level predicates (ILPs) and stage-level predicates (SLPs) was first proposed by Milsark (1974, 1977) and Carlson (1977). Here, we follow the latter. According to Carlson, ILPs are predicates of individuals that denote properties of true state, such as an entity being human or an object being wooden. By contrast, SLPs are predicates of stages of entities or objects. They can denote nonstative as well as stative situations; examples of the latter include physical positions of an entity, such as standing, sitting and so on. We adapt Yoon’s (2001: 86) schema of Carlson’s idea, shown in Figure 8.1. Examples of SLPs are the predicate adjectives nice and mean. They do not necessarily refer to inherent properties of an individual in that one can behave in a mean way but in reality be an inherently nice person. Examples of ILPs are the predicate adjectives tall/short or intelligent: these refer necessarily to inherent properties of an individual. If a person is short or intelligent, shortness and intelligence are inherent properties of that person. Carlson devised a number of tests for teasing apart ILPs from SLPs. One involves the progressive construction, according to which SLPs are acceptable in the progressive construction (8.10a), while ILPs are not (8.10b): (8.10) a. Mary was being nice/mean. (SLP) b. *Mary was being tall/intelligent. (ILP) The reason an ILP is illformed in the progressive is that as a true property of an individual it is incompatible with the progressive construction that expresses ongoingness in a dynamic or stative situation. A SLP, on the other hand, is compatible with the progressive construction as it denotes a stage of an entity (a transitory behaviour of Mary), not necessarily a true property of Mary. Predicates

IL

States

SL

Stative verbs (e.g. stand, lie)

Processes

Events

Figure 8.1 The distribution of ILPs and SLPs over states, activities and events

J. Clancy Clements 167

Leonetti (1994) argues that the distinction between ILPs and SLPs is largely sufficient to account for the distribution of adjectives with ser and estar in the ‘copula  adjective’ construction.4 We first note that the acceptability judgments of the English sentences in (8.10) also apply to their Spanish counterparts in (8.11): (8.11)

a. Mary está siendo amable/mala. (SLP) ‘Mary is being nice/mean.’ b. *Mary está siendo alta/inteligente. (ILP) ‘Mary is being tall/intelligent.’

If the ILP–SLP distinction were the sole predictor governing which predicate adjectives appear with ser or estar, one would expect ILPs to be compatible with ser but not with estar, and SLPs to be compatible with estar but not with ser. Taking our initial examples from (8.1) to (8.4), the prediction is borne out in part. That is, ser is compatible with ILPs such as mortal ‘mortal’, inteligente ‘intelligent’, necesario ‘necessary’ and preferible ‘preferable’, but ser is also compatible with SLPs such as sincero ‘sincere’ and cortés ‘polite’, which should only be compatible with estar. We take sincero ‘sincere’ and cortés ‘polite’ to be SLPs based on the criterion that they are acceptable to many native Spanish speakers in the progressive construction, as shown in (8.12): (8.12)

Marta está siendo muy sincera/cortés últimamente. ‘Marta is being very sincere/polite lately.’

Moreover, given that the adjectives sabroso and bonita in (8.3b, c) appear with ser and are illformed in the progressive construction (see, 8.13), we take them to be ILPs and should therefore only be acceptable with ser. However, both adjectives appear with both ser and estar, as shown in (8.3b, c) above. (8.13)

a. *La sopa está siendo sabrosa. ‘The soup is being tasty.’ b. *La chica/casa está siendo bonita ‘The girl/house is being pretty.’

The adjectives that appear in (8.4) above, that is, levantada ‘(gotten) up’ (denoting a transitory state), descalzo ‘barefoot’, roto ‘broken’, and previsto ‘anticipated’, do not appear with ser, regardless of the construction they appear in. That is, these adjectives appear exclusively with estar. Milsark (1974) takes predicates that denote transitory states such as being barefoot or broken, as well as others such as sick, hungry, tired, naked, drunk, open, closed (and by extension enfermo, cansado, desnudo, borracho, abierto, cerrado) to be SLPs. One test that distinguishes these SLPs from ILPs involves small clauses

168

Ser-estar in the Predicate Adjective Construction

(see Yoon, 2001: 88). For example, ILPs in (8.14a) are incompatible in the relevant small-clause construction, but the SLPs in (8.14b, c) are not: (8.14)

a. *Ví a Marta inteligente/mortal/alta. ‘I saw Marta intelligent/mortal/tall.’ b. Ví a Marta cansada/enferma/borracha/desnuda. ‘I saw Marta tired/sick/drunk/naked.’ c. Ví la puerta abierta/cerrada. ‘I saw the door open/closed.’

ILP SLP SLP

This criterion also works for identifying nontransitory resultant states as SLPs: they are wellformed in the relevant small-clause construction, as shown in (8.15): (8.15)

Ví el vaso roto. ‘I saw the glass broken.’

SLP

To sum up the discussion so far, the ILP–SLP distinction makes the right predictions for some adjectives, namely that ILPs such as inteligente ‘intelligent’ and mortal ‘mortal’ appear with ser, and SLPs like enfermo ‘sick’, cansado ‘tired’, descalzo ‘barefoot’, abierto ‘open’, and cerrado ‘closed’ appear only with estar. However, it fails to predict the correct distribution with the copulas for the ILPs sincero ‘sincere’, cortés ‘polite’, alto ‘tall’, sabroso ‘tasty’, and bonito ‘pretty’. The alternative analysis of predicate adjectives that we will consider addresses this shortcoming by differentiating states according to whether or not they imply some underlying process or event, and by appealing to the notion of time stability as it applies to the different types of states underlying adjectives. State types and time stability The manner in which we encode our world through language has to do with how we perceive the parts of our world to be or to function. For example, those parts of our world that have physical or spatial mass, such as mountains, stones, rivers, sky, people and objects in general, are seen prototypically as being more time-stable and are encoded as nouns. By contrast, those parts of our world that are more fleeting, such as the activity of singing or the punctual event of finding an object, are considered prototypically less time-stable and are generally encoded as verbs (Givón 1979, Hopper and Thompson 1984). Between these two extremes is the area of states, coded commonly as adjectives (for example, sick) or as stative verbs (lie, stand, know). This is shown schematically in Figure 8.2.5

J. Clancy Clements 169 Time stability

+ Objects,

–

Physical, mental states

Processes

Events

Animate entities

| -----------------Nouns----------------------------| |---------------------- Verbs -------------------------| |------------- Adjectives ---------------| Figure 8.2 Prototypical coding of the world in language based on time stability

Table 8.1 Characterization of verb classes Dynamic situation States Activities (processes) Accomplishments (durative events) Achievements (punctual events) Source:

   

Durative Telic    

   

Based on Vendler (1967).

Within the area of phenomena coded as verbs or predicates, one distinguishes states, processes and events, using the criteria of whether or not a predicate denotes a dynamic situation (in the sense of Comrie, 1976), whether or not it has duration, and whether or not it contains an intrinsic endpoint, that is whether or not it is telic. Vendler (1967) speaks of dynamic situations in terms of Activites (durative, nontelic), Accomplishments (durative, telic) and Achievements (nondurative, telic). This is shown in Table 8.1. For our purposes, we refer to Activities as processes and Accomplishments and Achievements as events. Here are some illustrative examples of these verb/ predicate classes. A typical stative predicate is know in that knowing something is a mental state rather than a dynamic activity. Typical process-predicates are learn, think and sing in that they take time to do (durative) and do not have an intrinsic endpoint. That is, we can learn, think things, or sing without there being an intrinsic endpoint to the processes. However, we can give such processes endpoints. For example, while sing and learn are processes, sing a song and learn a song are limited by the count NP a song. That is, sing a song and learn a song are durative and have an endpoint: singing or learning a song ends when the song has been sung or learned. Intransitive verbs can also denote durative events. Melt, for example, has an intrinsic endpoint and takes time. Examples of punctual-event predicates are find and arrive in that they take place punctually and are intrinsically telic (see Chapter 11 this volume for discussion of verb classes).

170

Ser-estar in the Predicate Adjective Construction

With particular reference to adjectives, the relevant factors for classification (first discussed by Clements, 1988) are whether or not they have an underlying dynamic situation, and, if so, whether or not the dynamic situation is an underlying process or an underlying event (durative or punctual). Adjectives without any underlying event or process are those such as mortal–inmortal ‘mortal–immortal,’ inteligente ‘intelligent,’ and colombiano ‘Colombian’ (in reference to nationality). Examples of adjectives with an underlying process are nuevo/joven–viejo ‘new/young–old’ (the underlying predicate is envejecer ‘get old’) and sincero–insincero ‘sincere–insincere’ (the underlying predicate is volverse sincero–insincero ‘become sincere/insincere’). Finally, adjectives with corresponding underlying events are those such as adjectives bautizado ‘baptized’ (the underlying event is bautizar ‘baptize’) and levantado–tumbado/ sentado ‘standing-lying/sitting down’ (the underlying events are levantarse– tumbarse/sentarse ‘get up–lie/sit down’).6 In terms of time stability, we propose that adjectives with no underlying process or event are relatively more time-stable than adjectives with an underlying process, and those adjectives with an underlying process are in turn more time-stable than adjectives with an underlying event. This is illustrated schematically in Figure 8.3. Three predictions emerge from this classification. First, given that ser is unmarked for the feature [aspect], it is predicted to be most compatible with adjectives denoting states that have no aspectual properties connected to them in terms of an underlying process or event. This turns out to be true: adjectives such as mortal ‘mortal’ and inteligente ‘intelligent’ have no underlying dynamic situation and the appear vritually only with ser, as shown in (8.1), repeated below: (8.1)

a. El ser humano es /*está mortal. ‘The human being is mortal.’ b. Juanita es /*está inteligente. ‘Juanita is intelligent.’

The second prediction is the converse of the first: since estar is marked with the feature [aspect], it would be most compatible with adjectives denoting states that have an underlying event because events are telic and least timestable. This also turns out to be true. Adjectives such as levantado ‘standing’

+ Adjectives with no underlying process or event

Time stability

–

Process-adjectives

Event-adjectives

Figure 8.3 Time stability of adjectives with no underlying process/event v. adjectives with an underlying process v. adjectives with an underlying event

J. Clancy Clements 171

and bautizado ‘baptized’ are the resultant states of the underlying events levantarse ‘wake up’ and bautizar ‘baptize’, and they appear virtually only with estar, as shown in (8.16): (8.16)

a. Marta *es/está levantada. ‘Marta is (standing) up.’ b. La niña ya *es/está bautizada. ‘The little girl is already baptized.’

The last prediction is that adjectives with underlying processes are compatible with both copulas or with neither. It turns out that process adjectives appear with both copulas, as shown in (8.17): (8.17)

a. Luisa fue/estuvo sincera. (volverse sincero ‘become sincere’) ‘Luisa was sincere.’ b. Juana es/está muy alta. (crecer ‘grow’ is the underlying predicate) ‘Juana is very tall.’ c. Luis es/está muy joven. (envejecer ‘get old’ is the underlying predicate) ‘Luis is very young.’

Thus, applying the notion of time stability to adjectives allows us to make a series of predictions that turn out to be true. However, several questions remain unanswered. For one, why is a resultant-state event-adjective like casado ‘married’, which is predicted to appear only with estar, also found with ser? Moreover, for those adjectives that allow both copulas, what is the difference in interpretation with one and the other copula and is there any way of predicting the different interpretations that arise? To attempt an answer to these questions, we introduce the notion of directionality in the classification of the adjectives. We will also discuss the issue of pragmatically feasible interpretations that follow from the semantics of the ‘copula  adjective’ construction with ser and estar, as well as conventionalization of patterns. Directionality of underlying processes and events according to the animacy of the referent It turns out that if we introduce the notion of directionality into the classification of the adjectives we are dealing with, we can predict with greater accuracy the compatibility of ser and estar with adjectives in the ‘copula  adjective’ construction. For adjectives such as mortal ‘mortal’, in our naïve, every-day understanding of how the world works, there is no directionality between mortal ‘mortal’ and its antonym inmortal ‘immortal’, or between finito ‘finite’ and its antonym infinito ‘infinite’. Directionality is typically linked to an underlying process or event. For example, in reference

172

Ser-estar in the Predicate Adjective Construction

to living entities, alto ‘tall’ assumes growth and growth is a unidirectional process in our naïve understanding of the world. In the world, an animate entity is thought to grow bigger or taller, but not smaller or shorter. A subject referent can be bajo ‘short’, joven ‘young’, or nuevo and over time become become alto ‘tall’ or viejo ‘old’. Both of these cases involve a unidirectional process (UP) (bajo → alto, joven/nuevo → viejo). In the case of the pair sincero ‘sincere’ and insincero ‘insincero’ a bidirectional process (BP) is involved. As Clements (1988) argues, in reference to an inherent trait (as opposed to a behaviour) of an animate referent a person can over time become sincere or insincere (sincero ↔ insincero). Directionality is also apparent in events. The antonym pair soltero ‘unmarried’ and casado ‘married’ involves a unidirectional event (UE) (soltero → casado), whereas the pair levantado–sentado ‘standing–sitting’ involves a bidirectional event (BE). Finally, there are dynamic situations that can be considered as bidirectional events or bidirectional processes, depending on the context. For example, the adjectives bonito–feo ‘pretty–ugly’ in Spanish can have either an underlying event or an underlying process. An event-interpretation of (8.17a) is that Juanita made herself look pretty using dress, make up, and so on. An process-interpretation of (8.17b) is that Juanita may not have been so beautiful in the past, say as a teenager, but having grown out of teenagerhood has become a beautiful woman: (8.17)

a. Ahora Juanita está bonita. ‘Now Juanita is beautiful.’ b. Ahora Juanita es bonita. ‘Now Juanita is beautiful.’

By including directionality in the classification, we distinguish the adjective types given in Figure 8.4 and displayed in terms of features in Table 8.2. Before discussing these adjective classes with animate and inanimate referents, we would like to revisit the notions of individual-level and stage-level predicates in light of the discussion of the adjectives so far. We saw earlier that alto ‘tall’ (and we can now add bajo ‘short’ by extension) is considered an ILP. We also saw that this fact has no bearing on whether alto can appear with ser or estar. We would like to propose that the distinction between ILPs and SLPs is not necessarily a semantic distinction, but rather a pragmatic one that in some languages may conventionalize into a semantic distinction. As suggested by Yoon (2001), for Spanish it is more accurate to speak in terms of IL and SL Readings. As an illustration, we take the examples in (8.18): (8.18)

a. Mary es alta. ‘Mary is tall.’ (synchronic comparison with her peer age group) b. Mary está alta. ‘Mary is tall.’ (compared with an earlier stage of her life)

J. Clancy Clements 173

Adjectives with no underlying dynamic situation

Adjectives with an underlying dynamic situation

Event adjectives

UE ADJs

BE ADJs

Process adjectives

BE/BP ADJs

UP ADJs

BP ADJs

Figure 8.4 Spanish adjective classification

Table 8.2 Examples of Spanish adjective types in terms of features mortal / despierto / muerto / sincero / alto / bonito / —7 Event adjective Process adjective Bidirectional Unidirectional Source:

   

   

   

   

()   

   

   

Based on Clements (1988).

The reading of (8.18a) is that Mary is tall with respect to a class, say, her age group (see Falk, 1979). Following Carlson (1977) and others, the predicate alta ‘tall’ is considered an ILP, where tallness is a defining property of Mary. However, in Spanish the property alta ‘tall’ is only a defining characterisic of the referent if it appears as ser alta. We shall call this an IL Reading. If alta appears with estar, we obtain an entirely different reading. In sentence (8.18b) Mary is tall, not in relation to her peers, but rather as compared to an earlier time at which she herself was shorter. Given that in this reading, tallness is not a defining property of Mary among her peers, it cannot be the case that this interpretation depends solely on the adjective. Rather, it must depend on the combination estar  adjective. Following Yoon (2001), we call the reading that corresponds to the sentence (8.18b) a SL reading. Thus, although the distribution of ser and estar in the ‘copula  adjective’ construction is not accounted for by the ILP–SLP distinction, it is possible to say that the ‘ser  adjective’ construction will tend to correspond to an IL reading, and the ‘estar  adjective’ to a SL reading. However, we will see that there are counterexamples to this generalization.

174

Ser-estar in the Predicate Adjective Construction

One more consideration needs to be mentioned before continuing. Above we saw that introducing the notion of time stability leads to certain predictions about which of the copulas is compatible with which type of adjective. States with no apparent underlying dynamic situation are predicted to prefer ser over estar, states having an underlying event prefer estar over ser, and finally states with corresponding underlying processes can appear with both copulas. We will find that, to a large extent, these predictions turn out to be accurate. Where counterexamples emerge, we will see that they are cases in which certain pragmatic assumptions override conventions based on semantic considerations to yield acceptable interpretations of different copula  adjective combinations. We now turn to this discussion, first with animate and then with inanimate subject referents. ‘Copula  adjective’ construction with an animate referent At the one extreme of the continuum shown in Figure 8.3, a copulative sentence containing an animate subject referent and a predicate adjective with underlying event or process, such as mortal ‘mortal’, venezolano ‘Venezuelan’, or inteligente ‘intelligent’, is not affected by the lack of time stability of any underlying event or process. On this view, adjectives of this type are relatively more time-stable than those with an underlying process or event. We argue that the higher time stability of these adjectives is compatible with the semantically empty nature of ser, which is unmarked for [aspect]. Thus, we expect such adjectives to appear with ser and have an IL reading. This prediction is borne out. We have seen examples of this in (8.1) above and include another example in (8.19):8 (8.19)

El presidente {es/*está} {mortal/venezolano/inteligente}. ‘The president is {mortal/Venezuelan/intelligent}.’

On the other extreme of the continuum are the event-adjectives. We argue that the relatively lower time stability of the underlying events of these adjectives (which are often punctual) determines their incompatibility with ser and their compatibility (both bidirectional levantado ‘standing’ and unidirectional bautizado ‘baptized’ and muerto ‘dead’) with estar (with the corresponding SL reading). This prediction is borne out, as we can see from the sentences in (8.20). Event-adjectives combine with estar and yield a SL reading. This is overwhelmingly the case for the bidirectional adjectives we examined (cansado ‘tired’, despierto ‘awake’, harto ‘fed up’, descalzo ‘barefoot’, enojado ‘angry’):9 (8.20)

a. Luz {*es/está} levantada. ‘Luz is up/standing.’

J. Clancy Clements 175

b. La niña {*es/está} bautizada. ‘The little girl is baptized.’ c. La cucaracha está muerta. ‘The cockroach is dead.’ However, there were exceptions for the unidirectional adjectives. We find that some (not all) of the unidirectional event-adjectives combine with ser and refer to the particular status of an individual, usually a legal status, overriding the effect of the less time-stable underlying event. This reading would arguably correspond to an IL reading. For example, although the adjective casado ‘married’ is found with estar 80 per cent (185/230) of the time in CREA, if the civil status of being a married person is being referred to, ser casado is used. In all of these cases, there are contexts in which routine information regarding family is provided, as in ‘married, with two children,’ or in contexts such as (8.21) in which it is also clear that the legal status of the referent is being referred to: (8.21) Hace algunos días me propuso que formalizáramos nuestra situación, pero la verdad es que no sé qué hacer, porque resulta que él también es casado, separado legalmente y con tres hijos, más o menos de la misma edad de los míos. ‘A few days ago, he suggested to me that we formalize our situation, but the truth is that I don’t know what to do, because it turns out that he also is married, legally separated and with three children, more or less of the same age as mine.’ (CREA) This is also the case for other unidirectional event-adjectives, such as divorciado ‘divorced’. That is, the only case of es divorciado found in CREA is in a context in which civil status and other such type of information is included (8.22): (8.22)

Juan es un abogado de treinta y tres años, es divorciado y tiene una hija llamada Malena. ‘Juan is a lawyer, 33 years old, divorced, and has a daughter named Malena.’

By contrast, está divorciado is found 10 times and not in this context. Moreover, in a metaphorical sense, divorciado appears only with estar and not with ser as illustrated in (8.23). This suggests that the default is, most likely, estar divorciado: (8.23)

Su idealismo, pues, está divorciado de la realidad. ‘His idealism, then, is out of touch with reality.’

176

Ser-estar in the Predicate Adjective Construction

According to Silva-Corvalán (1986, 1994), the unidirectional event-adjective muerto ‘dead’ behaves similarly in that it appears with both ser and estar, both with an SL reading.10 It is interesting to note that its antonym, vivo ‘alive’, has different readings with ser and estar. In reference to animate entities, ser vivo ‘be lively’ refers to a defining quality, with a corresponding IL reading, whereas estar vivo ‘be alive’ has an SL reading and is only understood as the opposite of dead. We see, then, that in the case of unidirectional eventadjectives, discourse considerations may favour the use of ser over estar in some cases, but overall they appear with estar and have an SL reading. With regard to ‘copula  process-adjectives’, the prediction is that they will be compatible with both ser and estar. This will be discussed for both bi- and unidirectional process-adjectives. Compared to the bidirectional event-adjectives, bidirectional process-adjectives as a group appear either entirely with ser, for example (in)cauto ‘(un)wary’, or principally with ser and occasionally with estar.11 In terms of their range of reference, with an animate subject referent these adjectives more often denote a mental, as opposed to a physical trait. As expected, there are exceptions to this general tendency. The adjective (in)capaz ‘(in)capable’ can refer to a mental disposition when appearing with ser. For example, in (8.24a, b) it refers to whether the subject referents are mentally (in)capable to carrying out something and have an IL reading. However, this adjective appears also with estar, as in (8.24c), where reference is made to the physical (in)capacity of procreating. No examples of estar capaz were found in CREA or elsewhere: (8.24)

a. Beckmann es incapaz de adaptarse. ‘Beckmann is incapable of adapting.’ (CREA) b. ¿Usted es capaz de decirle todo lo que piensa a Fidel? ‘Are you capable of telling Fidel everything you think?’(CREA) c. Penélope: Un hombre que no nos gusta, cuanto más viejo, mejor. Eurimena: Si está incapaz … Pero, ¿por qué?, ¿por qué? ‘P: A man who we don’t like, the older, the better. E: But he’s not virile anymore. … But, why?, why?’ (CREA)

The construction ‘copula  sincero’ has a similar distribution. The sentence [ ] es/fue sincero ‘[ ] is/was sincere/was telling the truth’ occurs 60 times in CREA with an animate subject referent. In nearly all instances, reference is made to a particular situation in which the subject referent displays defining property of being sincere, which in Spanish is often interpreted as telling the truth, and is illustrated by the two examples in (8.25): (8.25)

a. Badú fue sincero al decir que tanto Estados Unidos como México son seguros clasificados.

J. Clancy Clements 177

‘Badu was sincere in saying that the U.S. as well as Mexico are sure to be classified [in the World Soccer Cup].’ (CREA) b. John Kerry afirmó que el presidente George W. Bush no fue sincero ante la Organización de las Naciones Unidas (ONU) sobre la verdadera situación de la posguerra en Irak. Kerry criticó el discurso de Bush ante la Asamblea General de la ONU en Nueva York, en el que durante 30 minutos defendió su decision de invader Irak el año pasado. ‘John Kerry affirmed that President George W. Bush didn’t tell the truth before the United Nations Organization (UNO) about the true situation of postwar Iraq. Kerry criticized Bush’s speech to the General Assembly of the UNO in New York, in which he spent 30 minutes defending his decision to invade Iraq last year.’ (Radio Cooperativa, Santiago, Chile, 21 March 2004) However, it can refer to being sincere or genuine, as shown in (8.26). Whereas in the examples in (8.25) reference is made to a particular situation, here reference is not only to a particular situation, but to a general behaviour pattern of the subject referent, that of being always sincere or genuine. (8.26) Bielsa camina todo el tiempo, pero, a veces, se queda quieto. Y como siempre, fue sincero. ‘Bielsa walks the whole time, but at times, he stands still. And as always, he was sincere.’ There are few examples of sincero with estar; the three we found are given in (8.27): (8.27)

a. Y el otro autor, el Alex de la Nuez, pues me cayó bien, estuvo sincero aunque se nota que el pisito se lo paga la SGAE … ‘And the other author, Alex de la Nuez, well I liked him, he was sincere although it’s clear that his digs are being paid by SGAE …’ (www.bandaancha.st/foros.php?temid430982; Raposo, 24 Feburary 2005) b. Fernández Meca en su primero estuvo sincero y peleón, sin conseguir lucimiento. ‘In his first [phase of the bull fight] Fernández Meca was sincere and aggressive without being able to really shine.’ (El País. Madrid, 23 June 2002)

178

Ser-estar in the Predicate Adjective Construction

c. Saulo de Tarso, más tarde conocido como el apóstol Pablo, estaba sincero cuando perseguía la iglesia. ‘Saul of Tarsus, later known at Paul the apostle, was sincere when he pursued the church.’ (Phillip Gray, http://68.153.150.155/condenatorio_es_ preguntar_quien.htm) In all three cases, reference is made, not to telling the truth, but rather to a sincere attitude or behaviour. It is important to note that the examples (8.27a, b) refer to particular instances: in the first the speaker refers to a particular opinion written by Alex de la Nuez; in the second reference is made to the first part of torero Fernández Meca’s performance in a particular bullfight. In both these cases, the preterit is used. This contrasts with the use of the imperfect estaba in (8.27c), where the speaker refers to a rather long interval of time in which Saul of Tarsus was engaged in pursuing followers of Jesus. In these examples, at least, the choice of preterit v. imperfect seems to coincide with the length of the interval being referred to. Other adjectives of this class display a slightly different range of readings because of the nature of their reference. As bidirectional process-adjectives cortés and descortés are predicted to combine with both ser and estar. We find that (des)cortés ‘(im)polite’ primarily appear with ser, as in the case of sincero ‘sincere’, although some speakers admit sentences such as (8.28): (8.28)

Estuviste muy descortés con doña Ramona. ‘You were very impolite with Mrs Ramona.’

Just as with event-adjectives where pragmatic factors support possible interpretations such as the exceptional ‘legal-status’ reading (IL), similarly in the case of descortés pragmatic factors make interpretations such as the ‘transient behaviour’ SL reading possible, apart from the default ‘defining characteristic’ IL reading. Thus, descortés ‘impolite’ in (8.28) makes reference to impolite behaviour displayed by the subject referent towards doña Ramona. In CREA, we find no examples of estar with descortés, but 60 cases of copula  cortés, of which only four cases appear with estar. These are shown in (8.29): (8.29)

a. No por esto don Pío dejó de estar cortés, dentro de lo que su estado le permitía. ‘Not because of that did Don Pio quit being polite, as much as his condition permitted.’ b. Procuré estar cortés y sonreírle. ‘I tried to be polite and smile at him.’

J. Clancy Clements 179

c. No has estado muy cortés, aunque no me extraña… ‘You were not very polite, although it doesn’t surprise me…’ d. Yanguas Messía, está afable, bondadoso y cortés. ‘Yanguas Messia, is affable, kindhearted and polite.’ In each of these cases, the reference is to the behaviour of the subject referent. But we also find a clear behaviour-reading with ser and these adjectives, examples of which are given in (8.30): (8.30)

a. No quiero ser tan descortés. ‘I don’t want to be so impolite.’ (CREA) b. Ni fui descortés con los alemanes de mi contorno, ni, como algunos, cortejé el favour de los ingleses y los norteamericanos. ‘I neither was impolite to the Germans in my surrroundings, nor did I, as some did, court the favour of the English and the Americans.’ (CREA) c. Tratando de ser cortés, grita hacia afuera, ‘¡Señor!’ ‘Trying to be polite, he calls towards outside, “Sir!” ’ (CREA) d. El Cabo procuró ser cortés, pero medido. ‘The private attempted to be polite, but restrained.’ (CREA)

Thus, it seems that while estar with this class of adjective has only the SL reading, ser allows both an inherent-property reading as well as a transientbehaviour reading. This fits within the prediction in the following way: since bidirectional process adjectives in principle should be compatible with both copulas, there is no reason why either an IL or an SL reading would not be possible with both copulas. However, this is constrained by the pragmatic nature of the properties denoted by the adjectives. The trait sincero in Spanish refers in all cases to being (in)sincere and telling the truth (or not), whereas cortés (and its counterpart descortés) encompasses a larger range of references from behaving politely (with and without the desire to do so) to possessing a polite nature. Turning now to unidirectional process-adjectives, to discuss these we need to comment on the nature of unidirectional processes. First, the prediction derived from the continuum in Figure 8.3 is that these adjectives, as bidirectional process-adjectives, should be compatible with both copulas and, in addition, that the subject referent of a unidirectional process-adjective together with discourse factors will determine the exact interpretation in each case. For purposes of illustration, let us take as a representative example the unidirectional process crecer ‘grow’. As predicted, process-adjectives that refer to animate subject referents, such as alto ‘tall’ or bajo ‘short’, may refer to a defining property of the subject referent or a nondefining property. That is, process-adjectives can refer, at least, to a static state, to a dynamic state that implies a set of resultant states along the trajectory of growth, or to

180

Ser-estar in the Predicate Adjective Construction

some other aspect of the growth process. The examples found in CREA bear this out: those in (8.31)–(8.32) refer to a static state (with an IL reading), those in (8.33) to a stage in a dynamic process (with an SL reading): (8.31) (8.32) (8.33)

Mi tío es alto y rubio… ‘My uncle is tall and blond…’ Mi hermana es alta y bastante graciosa. ‘My sister is tall and quite funny.’ a. ¡Qué alto estás! ¡Cómo has crecido! ‘How tall you are! How you have grown!’ b. ¡Estás más alto que nunca! ‘You are taller than ever.’

Alto (and bajo) can also be used in a metaphorical sense, referring not to height, but to a hierarchy, as in (8.34): (8.34)

No hay duda de que Dios es alto y las criaturas son bajas. ‘There’s no doubt that God is high and the creatures are low.’

In (8.34), alto ‘high’ is a defining feature of God relative to his creatures, who are by nature bajas ‘low’ relative to God. When change of state involving the notions of ‘high’ and/or ‘low’ is implied, estar alto is used (with an SL reading), as in (8.35) in which an emotional (manic) state is being referred to and contrasted with its opposite depressive state: (8.35) Cuatro años después, volvía a estar alto, es decir, en otra etapa de euforia. ‘Four years later, he returned to being high, that’s to say, in another state of euphoria.’ The last group to be considered is the bidirectional event/process adjective. This adjective type can freely refer to states independent of or that result from an underlying process or event. Thus, in the copula  bidirectional event/process adjective construction, the semantics of the copula in connection with the specific discourse situation determines whether there is an underlying process or event or no underlying dynamic situation, and consequently, the IL or SL reading. That is, a speaker can refer to someone as bonito ‘pretty’ in terms of a defining trait of the person, as in (8.36a), or as the result of an event or process, as in (8.36b). In the latter case, the possible readings imply either that the referent or her mother has brought about such a state or that she has become so over time: (8.36)

a. Elisabeth es bonita y exhala dulzura y bondad. ‘Elizabeth is pretty and exudes sweetness and goodness.’ (CREA)

J. Clancy Clements 181

b. Mi Nelly es un amorcito y ahora sin la pancita está bonita. ‘My Nelly is a little love and now without her little belly she’s pretty…’ (CREA) Notice that copula choice with the same adjective (in this case bonita ‘beautiful, pretty’) can change within the same conversation, and even in the speech of the same interlocutor (in the case of linda ‘pretty, beautiful’): (8.37) Qué bonita. ¿Y ésta es M? M, ahí sí. Esa no cambia. No, y A M. C A y G. Ella es bonita. … Ellas dos están bonitas también. Sí, ésta era bella cuando estaba cumpliendo quince, dieciséis años, estaba linda, era no era porque era mi hija pero era linda. Es sí, se ve muy bonita. ‘How pretty. And this is M. M, ah yes. That one doesn’t change. No, and A M. C A and G. She is (ser) pretty. Those two are (estar) also pretty. Yes, this one was (ser) beautiful when she was turning 15–16 years old, she was (estar) pretty, it was it wasn’t because she was my daughter but she was (ser) pretty. She is, yes, she looks very beautiful.’ (CREA) Overall, bonita with animate subject referents appears more often with ser than estar (10 instances to one instance in the CREA search of present-tense forms). Thus, copula choice with this adjective is apparently similar to copula choice with the bidirectional process-adjectives like cortés, discussed above. In sum, with animate subject referents, the presence of ser or estar in the ‘copula  adjective’ construction involves the semantic interplay of the referent, the copula, and the adjective within a given discourse situation. We have proposed to capture the difference between ser and estar as unmarked v. marked for [aspect] respectively. Our analysis of adjective semantics invokes the notion of time stability, according to which we have at one extreme those adjectives with no underlying dynamic situation that are arguably more time-stable (for example inteligente ‘intelligent’), and at the other extreme those adjectives with an underlying event (for example cansado ‘tired’), whereby events are known to be the least time-stable dynamic situations (Hopper and Thompson, 1984; Givón, 1984). The prediction following from this analysis is that the adjectives linked to greatest degree of time stability should be most compatible with ser, and least compatible with estar, and conversely, that the adjectives linked to the smallest degree of time stability should display most compatibility with estar, and least with ser. With the exception of the unidirectional event adjectives (for example casado ‘married’), this prediction turned out to be accurate. Between the two extremes of the continuum, we have on the one hand adjectives with underlying bi- or unidirectional processes (sincero ‘sincere’,

182

Ser-estar in the Predicate Adjective Construction

cortés ‘polite’, alto ‘tall’) and, on the other, adjectives that can be construed as having either no underlying dynamic situation, or an underlying process or event (bonito ‘good-looking, pretty’). The prediction for these adjectives is that there would be more fluidity in copula choice and semantic reading, and indeed, this is the case. We saw that, with ser, sincero ‘sincere’ and cortés ‘polite’ can have an IL or an SL reading, depending on the context, and that cortés ‘polite’ with estar has only an SL reading, as would be expected. With such indeterminacy, context plays a major role in establishing the reading for these adjectives in the ‘copula  adjective’ construction. The same is true of adjectives such as bonito ‘good looking, pretty’, which according to context and copula choice ( both copulas are possible) may have no underlying dynamic situation, or an underlying process or event. Finally, where there is a choice between copula in the ‘copula  adjective’ construction, in terms of frequency of occurrence ser seems to be the default for all adjectives except the event-adjectives such as cansado ‘tired’ and casado ‘married’. This follows, we argue, from the semantics of the copula: ser is semantically empty, while estar is marked with the feature [aspect]. If it were true, we might expect the forms of ser overall to be more frequent than the forms of estar, even though both ser and estar each appear in various different constructions. Looking at frequency counts gleaned from two different sources (CREA and Alameda and Cueto’s, 1996, frequency dictionary), we see that, overall, the forms of ser are far more frequent than their corresponding forms of estar. The comparison is shown in Table 8.3, with the form-by-form frequency comparison in Table 8.4. The data suggest that of the two verbs, in general, ser is the default copula. We also note that with the exception of the singular imperfect forms (era and estaba) in Alameda and Cuetos, which are more numerous than in CREA, the percentages largely coincide. Table 8.3 Comparison of commonly occurring verb forms of ser and estar CREA Verb forms es ser son era eran Total está estar están estaba estaban Total

Raw numbers

Alameda and Cuetos Percentage

Verb forms

Raw numbers

Percentage

957,399 232,525 217,181 212,883 45,496 1,665,484 195,934 47,767 81,467 93,156 26,914

57 14 13 13 3 100 44 11 18 21 6

es ser son era eran Total está estar están estaba estaban

14,453 3,467 2,897 5,201 939 26,957 2,050 646 711 1,925 488

54 13 11 19 3 100 35 11 12 34 8

445,238

100

Total

5,820

100

J. Clancy Clements 183 Table 8.4 Form-by-form frequency comparison of different forms of ser and estar CREA Verb forms es/está ser/estar son/están era/estaba eran/estaban Total

Alameda and Cuetos

Raw numbers

Percentage

Verb forms

Raw numbers

Percentage

957,399/195,934 232,525/47,767 217,181/81,467 212,883/93,156 45,496/26,914

83/17 83/17 73/27 70/30 63/37

es/está ser/estar son/están era/estaba eran/estaban

14,453/2,050 3,467/646 2,897/711 5,201/1,925 939/488

88/12 84/16 80/20 73/27 66/34

1,665,484/ 445,238

79/21

TOTAL

26,957/5,820

82/18

In Table 8.4, we see that the singular present-tense and infinitival forms of ser are, apart from being far more numerous, more frequent than the corresponding forms in estar, while the imperfect forms of ser are somewhat less frequent. The two sources differ the most on the plural present-tense form. The main point of this comparison is, however, that ser is undisputedly the default copula in Spanish. ‘Copula  adjective’ construction with an inanimate referent The characterization proposed for adjectives with animate subject referents in the ‘copula  adjective’ construction can be adjusted to discuss inanimate subject referents. In this section, we examine what and where the similarities and differences are. Earlier we stated that the ‘copula  adjective’ construction is most accurately accounted for by taking into consideration the interaction between the semantics of the copula, the semantics and pragmatics of reference, and the discourse situation, as well as the animacy of the subject referent. In the last section, we have seen that an animate subject referent determines in a relatively homogeneous manner the underlying directionality of adjectives and their reference to situations (for example growing is unidirectional and becoming tired/rested is bidirectional). Mutatis mutandis, the same can be predicted of inanimate subject referents in the ‘copula  adjective’ construction. Although there is an obvious difference between animate and inanimate referents, they have in common some key properties. For example, as compared to states, processes and events, both animate as well as inanimate entities – people, animals, plants and things – represent what Lyons (1977: 443) calls first-order entities. On the time stability continuum in Figure 8.3, first-order entities are at the left of the continuum and contrast with second-order entities – states, processes and events. As Lyons (1977: 443) notes, states, processes and events ‘are located in time and … in English, are said to take place, rather than exist’. It is reasonable to assume

184

Ser-estar in the Predicate Adjective Construction

that this also obtains for Spanish. Apart from being second-order entities, states, processes and events are considered first-order properties, which can be ascribed to first-order entities (Lyons 1977: 439). Specifically, adjectives such as alto ‘tall, high’, cansado ‘tired’ or roto ‘broken’ are, as states, secondorder entities, as well as first-order properties in that they can be ascribed to first-order entities. For example, in Juanita es alta ‘Juanita is tall’, Marta está cansada ‘Marta is tired’, and la radio está rota ‘the radio is broken’, the states ‘tall’, ‘tired’ and ‘broken’ are ascribed through predication to the entities Juanita, Marta and la radio respectively. If the subject referent denotes an inanimate first-order entity, such as an object, the issue becomes whether, in our naïve understanding of the world, the denoted entity can undergo alteration or not with respect to a given state or a given spatial position. If we are talking about the moon, for example, we acknowledge that its state of visibility changes; in the case of a road, it can be made wider or narrower, and so on. Given that the directionality of the states predicated of inanimate subject referents hinges upon whether they are considered alterable or not with respect to a given spatial position or a given state, this feature is pragmatically assessed by the speaker within a given situation for each subject referent. Consequently, the interpretation of a given adjective depends on the speaker’s assessment of the situation and the nature of the elements involved in the utterance. Adjectives predicated of an inanimate subject referent that have no imaginable underlying event or process, such as infinito ‘infinite’, are assumed to behave like their counterparts mortal ‘mortal’ and inteligente ‘intelligent’ in their compatibility with ser and lack thereof with estar, and in their semantic reading (they have an IL reading). This turns out to be the case, as illustrated by the example in (8.38) from CREA. In our database search, we found no cases of estar with infinito ‘infinite’ or finito ‘finite’. (8.38) El presente es infinito. El futuro no tiene más realidad que la esperanza actual. ‘The present is infinite. The future doesn’t have any more reality than current hope.’ Event adjectives with inanimate subject referents generally behave as they do with animate subject referents, although, as just mentioned, directionality depends on the semantic nature of the subject referent and discourse considerations, including our understanding of the world. The examples in (8.39) are bidirectional event-adjectives that denote the resultant state of their corresponding event: (8.39)

a. abrir–cerrar ‘to open–to close’

→

abierto–cerrado ‘open–closed’

J. Clancy Clements 185

b. tapar–destapar ‘to cover–to uncover’ c. encender–apagar ‘to turn on–to turn off’

→ →

tapado–destapado ‘covered–uncovered’ encendido–apagado ‘on–off’

Due to the lower time stability of the underlying events of these adjectives (they are punctual), they are predicted to be incompatible with ser and compatible with estar, and have an SL reading.12 This prediction is borne out. As we can see from the sentences in (8.40), event-adjectives combine with estar and yield an SL reading: (8.40)

a. La puerta *es/está abierta/cerrada.13 ‘The door is open/closed.’ b. La botella *es/está tapada/destapada. ‘The bottle has its lid on.’ c. La luz *es/está encendida/apagada. ‘The light is turned on/off.’

With inanimate subject referents, some event-adjectives can be either unior bidirectional, depending on the nature of the subject referent they appear with. For example, given that typically we understand the breaking of an object, such as a cup, to be unidirectional, the sentence (8.41a) is considered unacceptable, or at best odd. The sentence (8.41b), however, is well-formed given our understanding of how radios work and how they can be repaired again and again to a functioning state: (8.41)

a. *?La taza está siempre intacta/rota. ‘The cup is always intact/broken.’ b. La radio está siempre intacta/rota. ‘The radio is always intact/broken.’

The adjective intacto ‘intact’ is an anomaly because there is no event necessarily underlying the state, but rather a lack of such an event, one that may be expected to have happened but did not. This is apparent in the examples in (8.42) taken from CREA. (8.42a) is said of an animal that underwent an operation. In this case, one expectation may have been that the swallowing reflex would stop functioning because of the severity of the operation, yet it did not: (8.42)

a. El reflejo de tragar está intacto. ‘The swallowing reflex is intact.’ b. El poder atómico de Rusia está intacto. ‘Russia’s atomic power is still intact.’

(CREA) (CREA)

186

Ser-estar in the Predicate Adjective Construction

In (8.42b), there may have been an expectation that Russia’s atomic power would not be intact because of the dissolution of the Soviet Union, yet it remains. Thus, one argument that intacto appears obligatorily with estar and not with ser is that there is a potential but unrealized underlying event. However, perhaps a simpler account is that intacto–roto ‘intact–broken’ are considered an antonymic pair and thus would share the same syntactic constraints. In other words, they would appear in the same constructions, such as in the ‘copula  adjective’ construction. This would constitute a more local explanation of the behaviour of intacto. Properties denoted by adjectives that are typically considered to have an underlying event, such as limpio ‘clean’ in (8.43a) (limpiar el apartamento ‘clean the apartment’ is the underlying, durative event) and sucio ‘dirty’ in (8.43b) (ensuciarse el escritorio ‘the desk to become dirty’ is the underlying, durative event), (8.43)

a. Después me acerco al escritorio que está limpio, quiero decir, no hay nada sobre él, salvo una lata de té Mazawattee, vacía, usada para guardar lápices, un marcador rojo, una regla, una goma de borrar, un broche de metal. ‘Then I approach the desk, which is clean, I mean, there’s nothing on top of it, except an empty tin of Mazawatee tea, used to hold pencils, a red marker, a ruler, an eraser, a metal brooch.’ (CREA) b. El apartamento está sucio, yo apurada limpiando cuando llama él. ‘The apartment is dirty, and I harried cleaning when he calls.’ (CREA)

can also be understood as defining characteristics of an object, independently of any underlying event. On such as reading, they can refer to how an object is made or to an inherent property of an object. In (8.44a), the use of ser with sucio and limpio here expresses that coal is dirty or not clean by its very nature. In (8.44b), the train is perceived as clean from the perspective of how it is furnished and finished, both of which are defining characteristics of the train: (8.44)

a. Los detractores del carbón dicen que este combustible es sucio. El carbón no es sucio y de hecho se está empleando hasta en industrias alimentarias: es su manipulación la que puede no ser limpia, si no se utilizan las técnicas adecuadas. ‘Detractors of coal say that this solid fuel is dirty. Coal is not dirty and in fact it is even used in the food industry. It is its use that has not been clean if the suitable technology is not employed.’ (CREA)

J. Clancy Clements 187

b. Adentro, el tren es limpio, con asientos tapizados en azul y rojo, con ventanas grandes y claras. ‘Inside, the train is clean, with seats upolstered in blue and red, with large and clear windows.’ (CREA) With inanimate subject referents, adjectives such as the antonymic pair nuevo–viejo ‘new–old’ is of the unidirectional process-adjective type, analogous to alto ‘tall’ with animate subject referents. Such adjectives generally refer to the relative age span or to the aging of a given object, and correspond to the underlying process of aging or developing. Adjectives having such an underlying process may or may not be perceived as denoting resultant states, although the semantics of their underlying predicates do not have an endpoint coded in their semantics. Just as in the case of the unidirectional process crecer ‘grow’ discussed above, the process of an object aging can be viewed as a series of various resultant states along the trajectory of aging or as a continuous development. Thus, we would expect that unidirectional process-adjectives, situated as they are in the middle on time stability continuum, would be compatible with both ser and estar. The examples in (8.45) bear this out: (8.45)

a. El coche es viejo. ‘The car is old.’ b. El hecho de cambiar de coche, ¿tú crees que se debe al hecho de que aumenta la familia, primariamente?, ¿o tal vez al hecho de que el coche está viejo, y por eso se cambia? ‘The fact of getting another car, do you think it’s because of the fact that the family is getting bigger, primarily, or perhaps because of the fact that the car is getting old, and because of that it’s being exchanged?’ (CREA)

The developmental process of entities generally considered slightly animate to inanimate (for example fruit, vegetables, meat, and so on) is unidirectional in the sense that it follows one of at least two developmental paths: immature → mature → rotten/dried or fresh → rotten. One would expect adjectives denoting entities that undergo one of these processes to also behave like unidirectional process-adjectives such as alto ‘tall’, and this is in fact the case. To denote the defining colour of an apple, for instance, ser is used, independently of the developmental process it undergoes (cf. 8.46a). The same sentence with estar, given in (8.46b), implies a comparison with another state of the apple in the same way that in sentence (8.33) alto ‘tall’ implies a comparison with another state of the referent: (8.46)

a. La manzana es roja/verde. (a defining characteristic) ‘The apple is red/green.’

188

Ser-estar in the Predicate Adjective Construction

b. La manzana está roja/verde (stage of maturation). ‘The apple is red (ripe)/green (unripe).’ The notion of an edible object being tasty or delicious may be reasonably assumed to be a unidirectional process or event. That is, an adjective such as sabroso ‘tasty, delicious’, mentioned in the introduction to the chapter, is considered, among other things, the result of its development over time (process, with ser and estar), or as the result of being cooked (event, with estar). We find, however, that there are other factors involved, such as whether the objects being ascribed the property sabroso ‘tasty’ are understood as generic or specific in discourse. For example, in the sentence in (8.47a) the meat spoken of is understood to be a generic reference. The sentence contains a condition event (the ‘fattening up’ event), which would result in the turkey meat being delicious. The sentence in (8.47a) is a general, timeless, conditional statement. This is also true of (8.47b), in which the meat (with generic reference) would be delicious (the result) if it is prepared in a certain manner (the preparation event). Again, it is a general, timeless, conditional statement. The timelessness and generality of these statements present them as general truths and therefore sabroso appears with ser and has an IL reading: (8.47)

a. La carne de los guajolotes es sabrosa y aun exquisita si se ha cebado cuidadosamente. ‘Turkey meat is tasty and even exquisite if they have been fattened up carefully.’ b. Si la carne se cocina cuando el agua ya está hirviendo la superficie se coagula y se impide en cierta medida la salida de jugos, esta carne es sabrosa, jugosa y conservará su poder nutritivo y sabor … ‘If the meat is cooked when the water is already boiling the surface coagulates and to a certain extent blocks the juices from getting out, this meat is tasty, juicy, and it will conserve its nutritional value and its flavour.’ (CREA)

This adjective appears with estar when reference is made to a particular instance in which the objects themselves are definite in discourse, in the sense of being identifiable by the speaker. For instance, in (8.48) the discourse participant A offers discourse participant B a mandarin orange in the first turn of the conversation provided here. B accepts the offer, and A then refers back to ‘a mandarin’ with the definite pronoun, indicating that it is identifiable: (8.48)

A. ¿Quiere una mandarina? B. Sí, por favour. A. Tómesela, está sabrosa … ‘A. Do you want a mandarin orange? B. Yes, please. A. Take it, it’s tasty.’ (CREA)

J. Clancy Clements 189

This case of está sabrosa ‘it is tasty’ is remarkable because there is no implication of an underlying process or event. We argue that it appears with estar because the statement refers to a definite object identifiable in a particular discourse situation. With estar sabroso, a personal experience, implied or real, of an edible item is expressed. In this particular case, we assume that the speaker A is speaking of the batch of mandarins which she has sampled and one of which she offers to her interlocutor B. Turning to bidirectional process/event-adjectives, an adjective such as bonito ‘pretty, beautiful’ with inanimate subject referents, as with its animate counterparts, can refer to a defining property of an entity or to a phase or stage of the object, either with an underlying event or process. In a conversation, given in (8.49), we find both es bonito ‘it is beautiful’, as well as está bonito ‘it is beautiful’. The difference is that the first use refers to a place (Chorrera) being beautiful as a defining property of the place (corresponding to an IL reading). In the second use, reference is made to the fact that the place has undergone restoration. That is, there is a restoration event underlying the state denoted by está bonito and it has a SL reading. (8.49)

A. La Chorrera allá en Bailadores es muy bonito … . ‘The Chorrera there in Bailadores is very beautiful … .’ B. Hace mucho tiempo. ‘Quite a while back.’ A. Eso es una belleza, eso lo han arreglado mucho. Ahorita está muy aseado eso. ‘It’s a beauty, they’ve fixed it up a lot. Now it all very clean.’ B. Sí está bonito. ‘Yes, it is beautiful.’ (CREA)

In (8.50), we find that está bonita makes reference to a similar situation as in (8.49b), linking the fact that the school has been cleaned to the observation that it is pretty, with the corresponding SL reading: (8.50)

Pues está bonita la escuela. Y qué limpia. ‘Well the school is pretty. And how clean it is.’

(CREA)

Some unidirectional process adjectives with animate subject referents, such as alto ‘tall, high’ and bajo ‘short, low’, behave with an inanimate subject referent like bidirectional event/process adjectives. Even if reference is made to spatial relations, objects can be spoken of with ser, as shown in (8.51), arguably because the location of the place is a defining property of the place of the object: (8.51)

a. El lugar es alto, ventoso, misérrimo como toda región otomí. ‘The place is high, windswept, wretched as all Otomi regions.’ (CREA)

190

Ser-estar in the Predicate Adjective Construction

b. … un Cristo antiguo de madera, de tamaño natural y expresión severa y dulce, inclina la cabeza, resignada y tristemente, como queriendo implorar el perdón. La cruz es alta, de color verde claro, y en su parte posterior dos escaleras llegan hasta sus brazos. ‘An old wooden crucifix, life-size and showing a harsh and sweet expression, bows its head, resigned and sad, as if wanting to implore forgiveness. The cross is high, light green in color, and behind it two stairways go up to its arms.’ (CREA) Even in reference to something as fluctuating as the height of the water in a river, both alto and bajo appear with ser and estar. In the example in (8.52), the topic is the mass extinction that took place millions of years ago and that is recorded in the fossil record. We include a larger portion of the text to show that it is a general statement, and it may to due to this that alto and bajo appear with ser: (8.52) Pregunta: ¿se debió esta extinción a un suceso repentino producto de una catástrofe, o se trató de un evento gradual? Imagine que usted deja caer gotas de tinta roja a un ritmo constante, digamos una gota por minuto, en un río, y que en una estación a un kilómetro río abajo realiza muestreos del agua una vez al día. Si el caudal del río es alto, la tinta se diluirá más que si el caudal es bajo, como por ejemplo durante una sequía. ‘Question: Is this extinction due to a sudden event such as a catastrophe, or does it involve a gradual event? [Answer:] Imagine that you squeeze drops of red ink at a constant rate, say one drop per minute, into a river and, that one kilometer downstream from that place we have our testing station set in and we carry out our water samplings once a day. If the volume of the river is high, the ink will become diluted more than if the volume is low, as for example during a drought.’ (CREA) However, this is not a steadfast rule. In (8.53), for example, we have a general statement involving a water levels in the sea where alta appears with estar: (8.53) Los pulpos son muy tímidos dice, todo les asusta, permanecen acurrucados a la puerta de sus casitas, cuando la mar está alta se les puede ver allí desde la lancha … ‘Octopuses are very timid he says, everything scares them, they stay curled up at the door of their little houses, when the sea is high, they can be seen there from the boat …’ (CREA)

J. Clancy Clements 191

An example that portrays the concrete situation of low water volume in the Amazon as a process expressed with estar, is shown in (8.54): (8.54) La navegación se hace cada vez más lenta. El río está muy bajo y se debe andar con mucho tino para evitar quedar encallados. ‘Navigating [the Amazon] is increasingly slow. The river is very low and we have to proceed with very sound judgment to avoid running aground.’ (CREA) The location of an object May be referred to in relation to its surroundings, and in (8.55) the use of estar expresses its location with regard to the background of the valley in which it is situated: (8.55) La casa está muy baja (situada en un valle). ‘The house is (situated) very low (in the valley).’ (VAÑO-CERDÁ, 1982: 146) Given what has been discussed so far, we can now expect objects that change or are alterable with respect to their physical characteristics to be compatible with both ser and estar. This is true with rivers and seas, we expect a certain amount of fluctuation with their water volume based on our understanding of the world. If we apply our understanding of the world to something commonly thought to fluctuate less, such as the height of buildings, we see that referring to the height of a building does not allow construal with both ser and estar, even though we know that buildings can be made taller or shorter. Most native speakers find the sentence in (8.56) to be unacceptable with estar, and no examples of this were found in CREA: (8.56)

Con la adición de otros diez pisos, este edificio ahora es/*está alto. ‘With the addition of another ten floors, this building is now tall.’

This leads us to conclude that our knowledge of the world dictates what may or may not be conceived of as having an underlying dynamic situation. And this, in turn, most likely depends on frequency of occurrence. That is, from a relative and impressionistic perspective, buildings are less frequently made taller or shorter than, for example, rivers become higher or lower, or roads are made wider or narrower. Although CREA has no examples of (8.57b), native speakers consulted found this example acceptable: (8.57)

a. La carretera es ancha ahora. (defining characteristic) ‘The road is wide now.’ b. ¡Qué ancha está la carretera! (resultant state) ‘How wide the road is!’ (FALK, 1979: 73)

192

Ser-estar in the Predicate Adjective Construction

In both sentences, the comparison involves knowledge on the part of the speaker about a previous state of the road. In sentence (8.57a), the speaker expresses that a defining trait of the road is now its wideness. With (8.57b), the speaker expresses that the road in question is wider than before, although there is no implication that for this reason a defining characteristic of the road is its wideness. Hence, a sentence such as (8.58), in which it is expressed that the road is wider but not wide, is entirely acceptable: (8.58) Ahora la carretera está más ancha que antes pero todavía no es muy ancha. ‘Now the road is wider than before but it is still not very wide.’ Thus, in determining the possibility of ser v. estar in the ‘copula  adjective’ construction, not only the semantics of the copulas and the adjectives comes into play, but also pragmatic knowledge about how we perceive our world to work, which in turn seems to be a function of frequency. To summarize this section, adjectives on the left end of the time stability continuum in Figure 8.3, that is, those adjectives without any underlying dynamic situation, such as infinito–finito ‘infinite–finite’, behave identically with inanimate or animate subject referents: they appear with ser, not with estar, and they have an IL reading. Similarly, those adjectives on the right end of the time stability continuum, such as the clear-cut event-adjective cerrado ‘closed’, and so on, behave like their counterparts with animate subject referents: they appear with estar, rarely if ever with ser, and have an SL reading. In the centre of the continuum, for inanimate subject referents we find no process-adjectives equivalent to sincero ‘sincere’ or cortés ‘polite’ for animate subject referents. What we have are unidirectional process-adjectives such as nuevo–viejo ‘new–old’ and event/process-adjectives, that is, those that can have no underlying dynamic situation or an underlying process or event, such as bonito ‘pretty, beautiful’. As expected, these appear with both ser and estar, with the respective IL and SL readings. The interpretation of the adjectives in this class depend on the semantic nature of the subject referent, but possibly more on discourse-related considerations and how we perceive our world to function. In turn, our understanding of the world is, at least in part, a function of frequency of occurrence of certain situations and how we conventionalize the encoding of them over time.

‘Copula  adjective’ construction with a second-order entity as subject referent In the previous section the distinction was mentioned between first- and second-order entities, following Lyons (1977: 439–43). Recall that first-order entities include people, animals, plants and things, while second-order

J. Clancy Clements 193

entities are states, processes and events more likely to be located in time and to take place rather than exist. Second-order entities are typically encoded in Spanish as adjectives and verbs. However, adjectives and verbs can be nominalized through derivational morphological processes. For example, the adjectives alto ‘high, tall’, cansado ‘tired’, and roto ‘broken’ can be nominalized respectively as altura ‘height’, cansancio ‘tiredness’ and rotura ‘breaking’, although these nominalizations do not always correspond exactly to their respective adjectival counterparts. Leaving the syntactic and word-class differences aside, while el cansancio de Marta ‘Marta’s tiredness’ is roughly equivalent to Marta está cansada ‘Marta is tired’, la altura de Juan ‘Juan’s height’ is not equal to Juan es alto ‘John is tall’ because Juan es bajo ‘Juan is short’ also corresponds to la altura de Juan ‘Juan’s height’. Similarly, the phrase la rotura del plato ‘the breaking of the plate’ is not equivalent to el plato está roto ‘the plate is broken,’ because roto refers to the resultant state of the event romper ‘break’, whereas rotura ‘breaking’ refers to the event itself. Even though such differences exist between adjectives and their corresponding derived nominals, as long as the derived nominal refer to a state, event or activity, we maintain that they are second-order entities. The qualification regarding the reference to states, events or activities is necessary because deverbal and deadjectival nominalizations can come to refer to first-order entities. Dokulil (1968), Lüdke (1978) and Clements (1979) discuss nominalizations that change or expand reference from second- to first-order entities. They note that it is not uncommon for nouns derived from verbs to come to denote objects or individuals. Nouns from intransitive verbs, for example, can come to refer to the subject of the underlying verb. So, the deverbal noun congregación ‘congregation’ from congregar ‘congregate’ can refer not only to the action of congregating, but also to those individuals who have congregated, which is a first-order entity. Similarly, nouns derived from transitive verbs can come to refer to the object of the underlying verb. For instance, the deverbal noun creación ‘creation’ from crear ‘create’ can refer not only to the act or action of creating something, but also to the resultant object of the act of creating, which again is a first-order entity. In addressing the question of compatibility of the copulas with secondorder entities, we refer to those nominalizations that exclusively denote a state, process, or event. For this discussion, we have chosen the nominalization conocimiento ‘knowledge’ from the stative verb conocer ‘know, be acquainted with’, the nominal form acciones ‘actions’, which has no clear underlying verb but is semantically a second-order entity, and two nominalizations from punctual verbs, muerte ‘death’ (from morir ‘die’) and llegada ‘arrival’ (from llegar ‘arrive’). The nominalization conocimiento appears with both copulas.14 It appears with ser when the property denoted by the adjective is to be understood as

194

Ser-estar in the Predicate Adjective Construction

being a defining property of conocimiento and has the corresponding IL reading. In (8.59a), the speaker compares knowledge to imagination and intuition, concluding that knowledge is limited by its own nature. Interesting here is the appearance of ser with a participial adjective, but it is important to note that limitado ‘limited’ is not necessarily an event-adjective because limitar can be a stative verb in the sense of a statement such as knowledge limits imagination. Assuming stative limitar, ser with stative-verb participial adjectives is not uncommon, as evidenced by examples such as es sabido que ‘it is known that’ or Marta es muy querida ‘Marta is very beloved’. (8.59)

a. La imaginación y la intuición son más importantes que el conocimiento. El conocimiento es limitado. La imaginación abarca al mundo entero. ‘Imagination and intuition are more important than knowledge. Knowledge is limited. Imagination encompasses the whole world.’ (CREA) b. La universidad es un cimiento arcaico. Es un instrumento arcaico, nacido hace muchos siglos, cuando los conocimientos humanos se limitaban a cinco o seis disciplinas. Hoy, el conocimiento es increíblemente extenso. ‘The university is an archaic foundation. It is an archaic instrument, born many centuries ago, when human knowledge was limited to five or six disciplines. Today, knowledge is incredibly extensive.’ (CREA) c. Todo conocimiento es cultural y, por tanto, adquirido. ‘All knowledge is cultural and, therefore, acquired.’ (CREA) d. El conocimiento es radical. ‘Knowledge is radical.’ (CREA)

The example in (8.59b) is interesting because the state of general knowledge in different eras is being compared. Whereas in (8.59a) knowledge is being compared to other entities – something akin to a class comparison – in (8.59b) knowledge is being compared to itself at different stages. That is, the development of knowledge is being addressed. Given our discussion of the unidirectional process-adjectives such as alto ‘tall, high’, we might expect knowledge accumulation to be largely a unidirectional process and that conocimiento should appear with estar in a context in which such accumulation is implied. This is, however, not the case. This is one indication that second-order entities behave differently than first-order entities with respect to copula choice. In the example (8.59c), knowledge is being defined as cultural and thus appears with ser. In (8.59d), the property ‘radical’ is being attributed to knowledge as a defining trait.

J. Clancy Clements 195

The noun conocimiento ‘knowledge’ also appears with estar, as shown by the examples in (8.60): (8.60)

a. El conocimiento está intrínsecamente vinculado a la comunicación. ‘Knowledge is intrinsically linked to communication.’ (CREA) b. El conocimiento está afectado por el reduccionismo. ‘Knowledge is affected by reductionism.’ (CREA) c. El conocimiento está disperso, extendido, es múltiple en el seno de la naturaleza. ‘Knowledge is scattered, widespread, it is multifarious in the bosom of nature.’ (CREA)

In (8.60a, b), estar appears with the event-adjectives vinculado ‘linked’ and afectado ‘affected’, which is predicted since such adjectives appear almost exclusively with estar. In (8.60c) there is the additional feature of implied space in the adjectives disperso ‘scattered’ and extendido ‘widespread’, and one would expect them to appear with estar, even though the implication to spatial relations is metaphorical. The next set of examples involves the noun acciones ‘actions’. In each of the cases in (8.61)–(8.62), acciones ‘actions’ is clearly a second-order entity. In (8.61a), ser appears because the adjective is a derived in ble. All derived adjectives in ble appear solely with ser. The adjective necesario ‘necessary’ in (8.61b) can be interpreted as a defining feature of acciones ‘actions’ because of its semantics, regardless of whether there is an underlying process/event or not: (8.61)

(8.62)

a. Estas acciones son denunciables ante los Tribunales ordinarios. ‘These actions are reportable before the ordinary courts.’ (CREA) b. Indicó que estas acciones son necesarias en una sociedad que va camino al desarrollo. ‘He indicated that these actions are necessary in a society that is in the process of development.’ (CREA) a. Pero de hecho tales acciones estaban dirigidas, bajo cuerda, por dirigentes sindicales de la empresa o de fuera de ella. ‘But in fact such actions were directed, secretly, by union leaders of the company or outside of it.’ (CREA) b. Manifestó que dichas acciones están orientadas ‘para liberar al país de los lazos de la dependencia interna y externa.’ ‘He revealed that said actions are designed “to liberate the country from the ties of internal and external dependence.’ (CREA)

196

Ser-estar in the Predicate Adjective Construction

c. Ahora sus acciones estaban determinadas por su conversión a la iglesia protestante. ‘Now his actions were determined by his conversion to protestantism.’ (CREA) In (8.62a, b, c), the event-adjectives dirigidas ‘directed’, orientadas ‘designed, oriented’ and determinadas ‘determined’ denote resultant states and as such they appear with estar. Another clear example of a second-order entity is muerte ‘death’ in reference to the event of dying. Here, we find that if the property denoted by the adjective is a defining feature, such as a characteristic of death in general or the death of a particular individual, as in (8.63a, b, c), or if the adjective is derived in ble, as in (8.63d), it takes ser: (8.63)

a. Su muerte es injusta y sobre todo, ilegal. ‘His death is unjust and above all, illegal.’ b. Esa muerte es política- y moralmente estúpida. ‘That death is politically and morally stupid.’ c. La muerte es gradual. ‘Death is gradual.’ d. La muerte es preferible al exilio. ‘Death is preferable to exile.’

(CREA) (CREA) (CREA) (CREA)

If, however, spatial location is implied by the semantics of the adjective, as in (8.61a, b), or if there is a participial adjective, as in (8.64c), estar appears: (8.64)

a. La muerte está cercana. ‘Death is near.’ b. Sin embargo, la muerte está presente. ‘However, death is present.’ c. La muerte está programada en los genes. ‘Death is programmed in the genes.’

(CREA) (CREA) (CREA)

Lastly, an event such as llegada ‘arrival’ can be afforded properties that define it for a particular person in a given situation. In (8.65a), in which the speaker sees the arrival of a person as hermoso ‘beautiful’ or ritual, as in (8.65b), regardless of any underlying process or event (Spanish ritual is always an adjective; the noun is rito ‘ritual’): (8.65)

a. Su llegada es hermosa. ‘His arrival is beautiful.’ b. Su llegada es prácticamente ritual. ‘Her arrival is practically ritual.’

(CREA)

J. Clancy Clements 197

We find llegada ‘arrival’ occurs with estar in the presence of event-adjectives such as prevista ‘anticipated’ or repleta ‘filled, full, replete’; the latter a predicate like llenar mucho ‘stuff full’. Examples are shown in (8.66): (8.66)

a. La llegada estaba prevista para la media tarde de ayer. ‘The arrival was anticipated for the middle of the afternoon yesterday.’ (CREA) b. La llegada estuvo repleta de emociones. ‘The arrival was filled/full/ with emotions.’ (CREA)

To sum up this section, if the subject referent is a second-order entity, such as a state (for example conocimiento ‘knowledge), a process (for example acciones ‘actions’) or an event (muerte ‘death’, llegada ‘arrival), ser appears when the states denoted by the adjectives are taken as defining properties of the subject referent. In these cases, there is an IL reading. We also have IL readings and ser with all adjectives derived in –ble. By contrast, those adjectives that imply a spatial relation, such as cercano ‘near’ or disperso ‘scattered’, and event-adjectives, almost always encoded as participial adjectives like vinculado ‘linked’, appear with estar. We also saw one case of participial adjective, limitado ‘limited’, derived from what we assume to be the stative verb limitar ‘limit’ that appears with ser. As noted, this is not uncommon with such adjectives given that they do not have a corresponding underlying event or process. We saw also that second-order entities behave somewhat differently than first-order entities. While event-adjectives and adjectives without an assumed underlying dynamic situation seem to behave in the expected manner, the adjective extenso ‘extensive’, with an implied process of knowledge accumulation (in 8.59b), appears not with estar, but with ser. Thus, copula choice with second-order entities in the ‘copula  adjective’ construction may be differently constrained than with first-order entities, and is an area in need of further study.

Concluding remarks In this chapter, we have argued that in order to understand the presence of ser or estar in the Spanish ‘copula  adjective’ construction one must take into consideration the semantic nature of the subject referent, the copulas, and the adjectives, as well as the discourse-pragmatic information such as referentiality, generic v. definite readings, and our naïve understanding of the world. We used the feature [aspect] to differentiate the copulas: ser does not carry the feature and estar does. For subject referents, we made the distinction between first-order entities (animate as well as inanimate entities) and second-order entities (nominalized states, processes and events). Further, we characterized the semantics of the adjectives by appealing to the

198

Ser-estar in the Predicate Adjective Construction

notion of time stability (Givón, 1979; Hopper and Thompson, 1984). Here, we distinguished adjectives with no underlying dynamic situation from those that have an underlying process or event. In the case of animate subject referents, for example, mortal ‘mortal’ and infinito ‘infinite’ have no underlying dynamic situation but sincero ‘sincere’, alto ‘high, tall’ and viejo ‘old’ have an underlying process and casado ‘married’, cansado ‘tired’ and cerrado ‘closed’ have an underlying event. Following Clements (1988), we further distinguished unidirectional (crecer ‘grow’: bajo ‘short’ → alto ‘tall’) from bidirectional (cansar ‘tire’: cansado ‘tired’ ↔ descansado ‘rested’) adjectives. The reason for this distinction is that it allows us greater precision is stating compatibility constraints between copulas and adjectives. We summarize our findings for first-order entities (animate and inanimate subject referents) in Table 8.5. In general, the only major differences found between animate and inanimate subject referents in the ‘copula  adjective’ construction are (1) for inanimate subject-referents there is no equivalent to the processadjective class, such as sincero ‘sincere’ for animate subject referents; and Table 8.5 Copula-adjective combination according to adjective type and reading for both animate and inanimate subject referents Copula

Adjective type

Reading (IL and/or SL) and combination with copula

a. ser

Adjective with no underlying dynamic situation (animate: mortal ‘mortal’; inanimate: infinito ‘infinite’)

IL

b. ser (estar)

Bidirectional process-adjectives (animate: sincero ‘sincere’; inanimate: no such adjective with inanimate subject referents)

IL, with ser SL, with ser and estar

c. ser/estar

Unidirectional process-adjectives (animate: alto ‘tall’; inanimate: viejo ‘old’)

IL, with ser SL, with estar

d. ser/estar

Bidirectional event/process-adjectives (animate: bonito ‘beautiful, pretty’; inanimate: bonito ‘beautiful, pretty’)

IL, with ser SL, with estar

e. estar (ser)

Unidirectional event-adjectives (animate: casado ‘married’; inanimate: roto ‘broken’)

SL, with estar IL, with ser (only animate subject referents)

f. estar

Bidirectional event-adjectives (animate: cansado ‘tired’; inanimate: abierto–cerrado ‘open–closed’)

SL, with estar

Note: The parentheses around estar and ser (b) and (e) indicate that with the corresponding adjective type the particular copula is attested far less than the other.

J. Clancy Clements 199

(2) with inanimate subject referents the specific nature of the referent seems to determine directionality more often than with animate referents. For example, we saw in the sentences in (8.41) that a cup breaking is more conceivable as a unidirectional event than a radio breaking, although they both constitute examples of the event breaking. This overall similarity between animate and inanimate subject referents in the ‘copula  adjective’ reflects some findings in the literature on the acquisition of Spanish copulas. Geeslin (2002) found no statistically significant difference in the acquisition of Spanish by native English speakers. However, as just reported we do find two major differences between animate and inanimate referents in our construction. Interestingly, in another study, Geeslin (2004, ch. 5) found there to be a significant interaction between animacy and adjective class in second-language acquisition. Thus, just as we have identified similarities and differences in the interaction between copulas and adjective class, Geeslin has also reached different results in separate studies. Although second-order entities also seem to follow the same behaviour as first-order entities on the extremes of the time stability scale, there is considerable variability in the centre of the continuum, which seems to depend on how the speaker chooses to express him/herself. That is, there seems to be greater freedom for the speaker to choose an IL reading by selecting ser or a SL reading by selecting estar, although the data in Tables 8.3 and 8.4 strongly suggest that ser can reasonably be considered the default copula On a more general level, for all subject referents copula choice appears to be most predictable with those adjectives linked to the extremes of the time stability scale in Figure 8.3. In the middle of the continuum is where we find most variability and where discourse-related considerations and our naïve understanding of the world play the biggest role in meaningfully combining copulas with adjectives in the Spanish ‘copula  adjective’ construction.

Notes 1 An asterisk indicates that the form is unacceptable to the native speakers consulted and not found in the database consulted (Corpus de Referencia del Español Actual [CREA] of the Real Academia Española, a 140 millon-word corpus, at www.rae.es). A question mark followed by asterisk indicates that the copula  adjective combination was not found in CREA but found by some native speakers to be acceptable. 2 This section is an updated version of section 2.1 in Clements (1988). 3 For an in-depth treatment of ser, see Fernández Leborans (1999). 4 Leonetti notes that the ILP–SLP distinction fails to account for sentences such as la sopa es/está sabrosa ‘the soup is tasty’. Other scholars, such as Fernández Leborans (1999: 2,438–41), present evidence suggesting that the estar-to-SLP and ser-to-ILP mapping is unable to account for other cases as well, which we address below. 5 For our purposes, we take objects, animate entities, etc. to be what Lyons (1977: 443) calls first-order entities: ‘[f]irst-order entities are such that they may be

200

Ser-estar in the Predicate Adjective Construction

referred to, and properties may be ascribed to them’. We take states, activities and events to be what Lyons (1977: 443) calls second-order entities: ‘By secondorder entities we shall mean events, processes, state-of-affairs, etc., which are located in time and which, in English, are said to occur or take place, rather than to exist.’ 6 We note that this classfication is not categorical. For example, given the appropriate context, such as a comparative context in (ia) or an iterative context in (ib), event adjectives may also have underlying processes and process adjectives may also have underlying events: (i) a. Juan es/está más gordo que antes. ‘John is fatter than before.’ b. Juan tardó dos años en volver a crecer. ‘It took John two years to begin to grow again.’ Although the verb engordar ‘become fat’ is arguably a durative-telic event, the sentence in (ia) can be interpreted either as the result of John getting fat or as a gradual process. Similarly, in (ib) crecer is the underlying process of the adjectives bajo ‘short’ and alto ‘tall’, but it can be interpreted as an iterative event when an implied endpoint is involved. 7 See Clements (1988) for a detailed discussion and motivation of the semantic traits of the adjectives. In that study, as well as in the present one, we found no adjectives that are both uni- and bidirectional and that simultaneously correspond to both processes and events. 8 In such types of examples, we found only rare and marginal cases in which we there is an SL reading. These will be discussed below. 9 For some of these bidirectional event-adjective adjectives, there are no cases of es  adj in the CREA database. For example, enojado ‘angry’ appears with está 27 times but zero times with ser. In CREA, está cansado ‘s/he is tired’ appears 127 times and es cansado five times, shown below: (i) Lo que estoy es cansado y harto de tus cambios de humor y de tus apasionamientos. ‘What I am is tired and fed up with your mood swings and your passions.’ (ii) Aunque es cansado estar de gira, compensa, porque hay que montarse sobre una tarima y mostrarle al público lo que sabes hacer, que es cantar y comunicarte. ‘Although it is annoying to be on tour, it’s worth it because you’ve got to get up on stage and demonstrate to the public what you know how to do, which is sing and connect with them.’ (iii) Esto es cansado, triste, melancólico a tope. ‘This is annoying, sad, melancholic to the max.’ (iv) Mirar y saber es cansado. ¿Es que cambian las cosas? No, nada. ‘To watch and to know is annoying. Do things really change? No, not a bit.’ The case in (i) is a cleft construction that always takes ser, while the other three an inanimate subject and cansado has the meaning ‘annoying’. In CREA, the adjective despierto ‘awake’ appears with es twice and with está 35 times. In those cases with es, the adjective refers not to the awake state (corresponding to an SL reading), but rather to an inherent state of being bright (with an IL reading), as shown by the example in (v). This is a clear example of the discourse

J. Clancy Clements 201 context allowing an interpretation that contravenes the semantics of the adjective and its underlying event: (v) Pedro Rimales es despierto, avispado y mañoso, siempre listo para burlarse de los demás y engañarlos, sean ellos sus enemigos, sus amigos o sus parientes. ‘Pedro Rimales is bright, sharp, and good with his hands, always ready to make fun of others and cheat them, whether they are his enemies, his friends, or his relatives.’ 10 In CREA, we found examples to support Silva-Corvalán’s assertion. In (i), an old man is mentioned who is now dead. It clearly has an SL reading, although it appears with ser: (i) Aquí había un viejito, que ya es muerto, llamado M. ‘Here there was a little old man, who now is dead, named M.’ 11 A detailed argumentation motivating the category of uni- and bidirectional process adjectives is presented in Clements (1988). 12 These adjectives appear with ser in the passive construction, as in la puerta es abierta por el empleado a las ocho cada día ‘the door is opened by the employee at eight o’clock every morning’. This, however, is not a ‘copula  adjective’ construction. 13 With animate subject referents, abierto ‘open’ and cerrado ‘closed’ can readily refer to defining traits of a person, as illustrated by the examples in (i), where ‘open’ and ‘closed’ are used metaphorically. With inanimate subject referents, we found no such use. (i) a. Como persona, Luis es muy abierto/cerrado. ‘As a person, Luis is very open/closed.’ 14 We disregard those constructions with ser that can be considered passive, as in (i): (i) El conocimiento es valorado como el factor de mayor incidencia. ‘Knowledge is valued as the factor of greatest impact.’

References Alameda, J. R. and F. Cuetos (1995) Diccionario de frecuencias de las unidades lingüísticas del castellano. Oviedo, Spain: Universidad de Oviedo, Departamento de Psicología. Bouzet, J. (1953) ‘Orígenes del empleo de estar. Ensayo de sintaxis histórica’, in Estudios dedicados a Menéndez Pidal (vol. 4), pp. 3–58. Carlson, G. (1977) ‘Reference to Kinds in English’, unpublished PhD dissertation, University of Massachusetts, Amherst MA. Clements, J. A. (1979) ‘Die ATA-Bildungen im Spanischen’, unpublished Universität Tübingen dissertation. Tübingen, Germany. Clements, J. C. (1988) ‘The Semantic and Pragmatics of the Spanish COPULA  ADJECTIVE Construction’, Linguistics, vol. 26, pp. 779–822. Comrie, B. (1976) Aspect. Cambridge: Cambridge University Press. Dokulil, M. (1968) ‘Zur Theorie der Wortbildung’, Wissenschaftliche Zeitschrift der Karl-Marx Universität Leipzig. Gesellschafts- und Sprachwissenschaftliche Reihe, vol. 17, pp. 203–12. Falk, J. (1979) SER y ESTAR con atributos adjetivales. Anotaciones sobre el empleo de la cópula en catalán y castellano. Stockholm: Almqvist & Wiksell.

202

Ser-estar in the Predicate Adjective Construction

Fernández Leborans, M. J. (1999) ‘La predicación: las oraciones copulativas’, in I. Bosque and V. Demonte (eds), Gramática descriptiva de la lengua española. Madrid: Espasa Calpe, vol. 2, pp. 357–460. Geeslin, K. (2004) Crossing Disciplinary Boundaries to Improve the Analysis of Second Language Data: A Study of Copula Choice with Adjectives in Spanish. Munich: Lincom Europa. Geeslin, K. (2002) ‘The Second Language Acquisition of Copula Choice and its Relationship to Language Change’, Studies in Second Language Acquisition, vol. 24, pp. 419–51. Givón, T. (1979) On Understanding Grammar. New York: Academic Press. Hanssen, F. (1912) ‘La pasiva castellana’, Anales de la Universidad de Chile, pp. 2–28. Hernanz, M. L. (1988) ‘En torno a la sintaxis y la semántica de los complementos predicativos en español’, Estudi General, vol. 8, pp. 7–29. Hopper, P. J. and S. A. Thompson (1980) ‘Transitivity in Grammar and Discourse’, Language, vol. 56, pp. 251–99. –––– (1984) ‘The Discourse Basis for Lexical Categories in Universal Grammar’, Language, vol. 60, pp. 703–52. Leonetti, M. (1994) ‘Ser y estar: estado de la cuestión’, Barataria (U. de Alcalá de Henares), vol. 1, pp. 182–205. Lüdke, J. (1978) Prädikative Nominalisierungen mit Suffixen im Französischen, Katalanischen im Spanischen. Tübingen: Max Niemeyer. Luján, M. (1980) Sintaxis y semántica del adjetivo. Madrid: Cátedra. –––– (1981) ‘The Spanish copulas as aspectual indicators’, Lingua, vol. 54, pp. 165–210. Lyons, J. (1977) Semantics. Cambridge: Cambridge University Press. Martín, R. J. (1981) Caballos desbocaos. Madrid: Ediciones Cátedra. Milsark, G. L. (1974) ‘Existential Sentences in English’, PhD dissertation, MIT. –––– (1977) ‘Toward an Explanation of Certain Peculiarities of the Existential Construction in English’, Linguistic Analysis, vol. 3, pp. 1–29. Mourelatos, A. (1978) ‘Events, Processes, and States’, Linguistics and Philosophy, vol. 2, pp. 415–34. Navas, R. R. (1963) Ser y estar. El estudio sobre el sistema atributivo del español. Salamanca. Schmitt, C. (1992) ‘Ser and Estar: A Matter of Aspect’, Northeastern Linguistic Society, vol. 22, pp. 411–26. Silva Corvalán, C. (1986) ‘Bilingualism and language change: The extension of estar in Los Angeles Spanish’, Language, vol. 62, pp. 587–608. —— (1994) ‘The Gradual Loss of Mood Distinctions in Los Angeles Spanish’. Language Variation and Change, vol. 6, pp. 255–72. Stalnaker, R. (1974) ‘Pragmatic Presuppositions’, in M. K. Munitz and P. K. Unger (eds), Semantics and Philosophy. New York: New York University Press, pp. 197–214. Van Valin, R. and R. LaPolla. (1997) Syntax: Structure, Meaning and Function. Cambridge: Cambridge University Press. Vaño-Cerdá, A. (1982) Ser y estar ⫹ adjetivo. Tübingen: Narr. Verhaar, J. (1990) ‘How Transitive is Intransitive?’ Studies in Language, vol. 14, pp. 93–168. Yoon, J. (2001) ‘Small Clauses in Spanish: A Semantic-Functional Approach’, unpublished Indiana University dissertation, Bloomington, IN.

9 Spanish Adjective Position: Differences between Written and Spoken Discourse Richard J. File-Muriel

Introduction Little is known about the interaction between competence-based and performance-based constraints, but production and processing of speech are clearly influenced by both. Although the present study does not address this interaction, it does add to the growing body of literature on the role of performance constraints in constituent ordering. The aim of this chapter is to provide an account of attributive adjective (A) position in Spanish based on performance constraints. We argue that (1) adjective position in spoken Spanish may be explained in terms of word length and semantic constraints, and (2) show that adjective position has become partially grammaticalized due to word length. In addition, we provide evidence that shows adjective realization differs greatly across written and spoken registers and argue that it is inadequate to depend solely on written language to make claims about language in general.

Word length In much of the literature on adjective position in Spanish, there has been brief discussion of a ‘heaviness shift rule’, according to which the syllabically heavier element in a phrase consisting of a noun (N) and an A is postposed. For example, in the sentence No debemos someternos a falsasA modestiasN ‘we should not submit ourselves to false modesties’, the preposed A falsas ‘false’ has two syllables, while the N modestias ‘modesties’ has three. In the phrase La prohibición de las fuerzasN armadasA ‘the prohibition of the armed forces’, the first element fuerzasN ‘forces’ has two syllables, while the heavier element armadasA ‘armed’ is postposed with three. Some dismiss the heaviness shift rule as a significant factor for predicting adjective position, arguing that the phenomenon is merely the result of the fact that long descriptive adjectives 203

204 Spanish Adjective Position: Differences between Written and Spoken Discourse

tend to follow the noun because they represent an important proposition (Terker, 1985). Almost all agree that semantic values cannot be ignored, pointing out that if word length were the key determiner governing adjective position, other factors would be irrelevant and semantics would not play any role. In other words, the adjective would be the same whether it is preposed or postposed (Prado, 1980). This phenomenon of the heavier element being postposed is not unique to Spanish. Standard grammars of French often state that the phonological form of the attributive adjective influences its position relative to the noun that it modifies. The claim is that if the adjective has fewer syllables than the noun, the adjective is preposed; if the adjective has more syllables than the noun, it is postposed (Wilmet, 1980).1 Fernández (1951) was one of the first to look at Spanish adjective position with respect to weight and rhythmic laws. He analysed two constructions in literary prose containing the preposition con: (1) con  indefinite article  AN/NA(for example, con un buen chico ‘with a good boy’ v. con un chico bueno lit. ‘with a boy good’), and (2) con  AN/NA (for example con pura agua ‘entirely with water’ v. con agua pura ‘with pure water’). He found 40 examples of the con un(a)  NA/AN construction: 30 cases were realized as NA and 10 were AN. In the constructions with con  determiner, the final element was either equal or syllabically heavier than the preceding element. In the second construction involving con with no determiner, Fernández found that 60 per cent of the cases reflected AN order compared to 40 per cent NA; the syllabic rule failed to apply 100 per cent of the time. He concluded that Spanish demonstrates a tendency to postpose the adjective, but if it is shorter than the noun, the adjective is preposed. Prado (1980) conducted a more general analysis not limited to any one particular preposition. Based on data gleaned from four novels, he found that in AN constructions, the length of the element did not correlate highly with the position of the adjective. For example, the preposed adjective was lighter or equal to the noun 62 per cent, 75 per cent, 58 per cent and 70 per cent of the time across the four novels, whereas in NA constructions, the postposed adjective was heavier or equal to the noun 80 per cent, 83 per cent, 80 per cent and 83 per cent in the four novels. Although Prado notes that there is a tendency for lighter adjectives to precede the noun, he concludes that there is not sufficient evidence to support a heaviness principle and that semantic values are the crucial factor in determining adjectival position. Others, such as Terker (1985), take a much stronger position against the influence of syllable length on AN/NA order. Terker claims that adjective position should not be dealt with in terms of tendencies, giving several counterexamples, presumably from a written text, that blatantly violate the rule of ‘heaviness shift’ (for example el distinguidoA senadorN ‘distinguished senator’). Although Terker argues against syllabic heaviness as a possible determiner of adjectival position, he considers syntactic complexity a factor,

Richard J. File-Muriel 205

showing that complex adjective phrases must always be in postnominal position.2 The claim that the phonological component may have influence at the syntactic level is quite controversial. The standard assumption has been a top-down flow of information from a syntactic plan to a phonetic one (Garrett, 1982). This model of speech production is summarized by Bock (1987: 120) and outlined in (9.1): (9.1) 0. Message level: production begins at the message level with a non-linguistic representation of the speaker’s intention. 1. Functional level: semantically specified lexical items are assigned syntactic roles, which control the creation of a representation at the next linguistic level. 2. Positional level: phonologically specified words (primarily open class or content words) are inserted into a planning frame, which represents the hierarchical constituent organization, various closed-class elements (functional words and inflectional affixes), and the prosodic structure of the sentence. 3. Phonetic level: the resulting positional level structure – an ordered representation of phonologically specified morphemes integrated with the planning frame – then guides the development of the utterance’s detailed phonetic form. This model holds that there is a complete absence of feedback from the positional to the functional level due to the unidirectional flow of information (that is, functional to positional). In other words, processing and the representation of information at the positional level should have no impact on the selection of morphemes or their assignment to grammatical roles. This unidirectional information flow is argued for by Miller, Pullum and Zwicky (1997), who propose the Principle of Phonology-Free Syntax (PPFS) as a principle of Universal Grammar (UG) that prohibits reference to phonological information in syntactic rules or constraints. The claim is that rules of syntax make no reference to phonology. They consider four phenomena of French that were previously used to argue against the PPFS. For their part, they claim that any statement regarding weight as a determiner of word order does not represent a rule of grammar at all, but merely a statistical tendency observed in usage. They further claim that although speakers or writers may often choose a given order because it sounds better than the alternative, these choices do not imply the existence of a grammatical constraint: ‘The typical effect of a skilled writer or speaker’s choices in a certain context may be to create a tendency for words with heavier syllabic

206 Spanish Adjective Position: Differences between Written and Spoken Discourse

weight to follow words of lighter weight, but that does not mean that there is a grammatical constraint enforcing this ordering’ (1997: 74). In contrast, Bock (1987) argues against the claim that the flow of information is only from the higher, functional level to the lower, positional level. She claims that factors related to the retrieval of phonological forms of words do, in fact, modify the structure of the sentences in which they appear. Evidence, she states, can be found in the phonological effects on word exchange errors: ‘Dell and Reich (1981) found that words which undergo exchanges, anticipations, and perseverations share more phonological features than would be expected by chance’ (1987: 121–2), indicating that there is some feedback between the phonological and syntactic levels. She claims that ‘sentences tend to be structured in such a way that phonologically accessible word forms precede less accessible forms, other things equal’(1987: 122). For his part, Hawkins (1994) finds that processing constraints affect word order. Arguing that sentences need to be both produced and understood in real time, he explores what role such processing constraints impose on grammars. Specifically, he claims that processing, rather than discourse or pragmatics, better explains the properties and constraints of word order and constituent structure. The role of an innate Universal Grammar is, in his view, greatly reduced in favour of processing mechanisms, which are ultimately innate as well. He proposes a set of processing explanations that he calls the assumption of absolute unprocessability: ‘Some absolute limit on human processing resources is proposed within a given domain, and a grammatical response is then explained on the grounds that this limit would be exceeded by structures of a certain type’ (1994: 8). Hawkins believes words and constituents occur in an order that allows syntactic groupings and their Immediate Constituents (IC) to be recognized, comprehended and produced both rapidly and efficiently as possible in on-line language performance; different constituent orderings result in a more, or less, rapid IC recognition. In other words, complex structures that are difficult to process are avoided and thus not incorporated into the grammar. According to his model, which he calls Early Immediate Constituent (EIC), grammars are moulded by processing constraints on speakers and listeners during rapid and efficient production and comprehension of language. Immediate Constituents (ICs) can be recognized on the basis of a proper subset of the words and categories dominated by that phrase. According to Hawkins, processing favours Constituent Recognition Domains (CRD)3 for Verbal Phrases (VP) and Sentences (S), which allow for the most efficient recognition of the ICs contained within them.4 Hawkins’ model of performance is concerned mainly with processing, that is with the listener’s perspective. By contrast, Wasow (1997a) argues that the role of the speaker is paramount and that weight effects exist primarily to

Richard J. File-Muriel 207

facilitate utterance planning and production, as speakers take advantage of syntactic flexibility to postpone hard-to-produce constituents until later in the utterance (cf. also Bock, 1987). Wasow (1997b: 360), however, points out that ‘language use is characteristically a joint activity. If we wish to explain aspects of language structure in terms of performance, we should look at the tasks facing both participants, and at how the participants carry out those tasks.’ In a subsequent study, Arnold, Ginstrom, Losongco and Wasow (2000) concentrate on production in examining heaviness in terms of syntactic complexity. They hypothesize that new and heavy constituents are likely to be harder to produce than given, light ones, and following Levelt (1989) they posit three stages for language production: (a) conceptualizing a message, (b) formulating grammatical characteristics of the message, and (c) articulating it. Given that online processing occurs rapidly, ‘discourse status and syntactic weight could influence production at any of these stages’ (2000: 46).5 They note that postposition of a heavier and newer sentential element may be due to speaker and listener concerns. Speakers often begin to produce an utterance before the proposition has been fully planned. They argue that postposition of the heavier element favours processing for the listener, saving ‘the hardest for last’. From the speaker’s point of view, the postposition of the heavier element functions to facilitate performance concerns such as planning and production. They give a baseball example to illustrate how the postposition functions to give the play-by-play announcer time to check the score card (for example, ‘That brings to the plate … Ozzie Smith’). The two experiments carried out by Arnold et al. (2000) focus on production rather than perception and investigate the roles of discourse status (that is, new v. old information) and weight in constituent ordering. The first study examines two constructions in the Align-Hansard6 corpus: Heavy Noun Phrase Shift (HNPS) and Dative Alternation (DA). Heaviness was measured as the number of words in the theme minus the number of words in the goal.7 Measuring the relative weight between two constituents allows one to observe the graded nature of weight effects (Wasow, 1997a).8 The second experiment carried out was an elicitation experiment focusing on production of DA constructions. The authors were interested in what word order the participants would produce and how this would be affected by the discourse status of the referent and the weight/complexity of the NP used to denote it. As expected, they found both heaviness and newness to be significant factors in constituent order choice. In both studies, the heavier, newer constituent tended to follow the lighter, old one. Arnold et al. (2000: 46) point out that ‘lexical retrieval and phonological processing should be more difficult for long, complex phrases than shorter ones simply because they contain more linguistic material’. Hawkins (1994)

208 Spanish Adjective Position: Differences between Written and Spoken Discourse

and Arnold et al. (2000) base their models on syntactic weight (in terms of number of words) in determining constituent order. This methodology is easily extendable to include the word level with relative syllable counts as the unit of measure. Looking at weight in terms of words, nodes or phrasal nodes may overlook effects stemming from word length only. That is, one would expect lexical retrieval and phonological processing to be more difficult for longer words than shorter ones because they contain more linguistic material. We argue that performance considerations favour the postnominal position of the attributive adjective in spoken Spanish because it is generally longer than the noun it modifies. Moreover, we claim that the prenominal placement of attributive adjectives in spoken discourse is reserved for light, frequently occurring, non-relational adjectives. Finally, we offer evidence suggesting that adjective position in Spanish may be in the process of grammaticalization due to performance considerations based on word length.

Methodology The present study compares oral and written Costa Rican Spanish. The speech data come from transcribed semi-informal interviews with five university-educated males from San José, Costa Rica, between the ages of 27–45. Each interview lasted 15–30 minutes and was conducted on the topic ‘the absence of a military in Costa Rica’. The writing samples, which consisted of seven letters/editorials taken from San José’s prominent newspaper, La Nación, were taken between 4–21 August 1999. In collecting the data, an effort was made to select articles that were near to the previously mentioned topic. Articles on other topics were not considered due to the possibility that they could represent translations from another language. Each article and transcribed interview was scanned for instances of nouns with corresponding attributive adjectives. A total of 231 spoken tokens and 287 written tokens were found and coded, using the variables given in (9.2): (9.2)

ORDER: whether adjective was pre or postposed (NA or AN order) WEIGHT: difference in syllable count between A and N SEMANTIC CLASS: semantic class of A STYLE: whether the token came from a spoken or written source STRESS: whether or not an adjective has syllable-final stress

Adjective position, pre v. postposed, was the dependent variable, while all others constituted the independent variables. For each token, the difference in word length between the A and N was calculated by subtracting the number of syllables of the A from the number of syllables of the N. For example, in the phrase buenaA ocupaciónN ‘good occupation’, the A buena has two syllables while the N ocupación has four; the adjective would be scored 2, that is

Richard J. File-Muriel 209

two syllables lighter than the noun. The semantic class of the adjective was coded following Delbecque’s (1990) classification, included in (9.3): (9.3)

Class 1: Colour: e.g. amarillo Class 2: Relational: Affiliation/association/attachment to a. geographic entity e.g. colombiano ‘Colombian’ b. ethnic group e.g. celta ‘Celtic’ c. socio-political or socio-political grouping e.g. liberal ‘liberal’ d. professional or scientific discipline e.g. científico ‘scientific’ e. symbolic or philosophical movement e.g. intelectual ‘intellectual’ Class 3: Temporal markers e.g. moderno ‘modern’ Class 4: Shape and space markers e.g. interno ‘internal’ Class 5: Physical property markers e.g. gordo ‘fat’ Class 6: Modality markers e.g. posible ‘possible’ Class 7: Evaluation markers e.g. bueno ‘good’ Class 8: Personality/disposition markers e.g. responsable ‘reliable’

As mentioned, the style variable refers to whether the token was taken from a spoken or written source. Finally, adjectives were also coded for whether they had syllable-final stress or not (for example syllable-final stress: militar ‘military’, leal ‘loyal’, real ‘real’, actual ‘current’; non syllable-final stress: científico ‘scientific’, capacitado ‘with the capacity’, armada ‘armed’). This variable was tested just in case it turned out that word-level stress assignment is a factor in the position of an A relative to the N it modifies.

Results The corpus was submitted to a binary logistic regression analysis using SPSS 11.5. Adjective position relative to the noun was entered in the regression as the dependent variable and all the other factors listed in (9.2) were entered as covariates. The independent variable semantic class was entered as categorical. The results of the binary logistic regression, given in Table 9.1, show that style, weight and semantic class are all significant determiners (.000) of whether or not an A is preposed, whereas stress was not significant (.092). The Exp(B) values in the last column of Table 9.1 tell us the number of times more likely that a given token will be preposed than not. Note that weight is the strongest predictor of N–A order, with an Exp(B) value of 2.311. That is, based on the variable of the relative weight (measured in syllables) between the N and its modifying A, A position is 2.311 times more likely to be correctly predicted than by the null hypothesis. The Wald test measures

210 Spanish Adjective Position: Differences between Written and Spoken Discourse

the goodness of fit of the model. The Wald values of the variables ‘style’ (36.575), ‘weight’ (36.575), ‘semantic class’ (40.793) are high, indicating that they account for a goodness of fit of the model. With specific regard to weight, Table 9.2 shows the distribution of A position with respect to the difference in syllabic weight between the A and N (for example 2 means that the A was two syllables lighter than the N, and so on). Tables 9.3 and 9.4 have the same function but they separate the spoken from the written data respectively.

Table 9.1 Spoken and written style

Step 1a

style weight stress semantic constant

B

S.E.

Wald

df

Sig.

Exp(B)

1.429 .838 .742 .472 4.584

.365 .139 .441 .074 .513

15.300 36.575 2.839 40.793 79.885

1 1 1 1 1

.000 .000 .092 .000 .000

.239 2.311 2.101 .624 97.877

a

Variable(s) entered on step 1: style, weight, stress, semantic class.

Table 9.2 Position of As to Ns for the all the data according to their respective weights Weight Preposed Postposed

3 or 

2

1

0

1

2

3 or

5 1

13 22

21 60

22 124

11 85

1 86

1 30

Table 9.3 Position of As to Ns in the spoken data according to their respective weights Weight Preposed Postposed

3 or 

2

1

0

1

2

3 or

1 0

3 9

4 28

4 74

0 34

0 50

0 19

Table 9.4 Position of As to Ns in the written data according to their respective weights Weight Preposed Postposed

3 or 

2

1

0

1

2

3 or

4 1

10 13

17 32

18 50

11 51

1 36

1 11

Richard J. File-Muriel 211 Table 9.5 Preposed adjectives in spoken discourse Token un gran espíritu una gran cosa muy malos ojos una bajísima escolaridad una nueva constitución la gran paz una muy buena ventaja los grandes edificios al libre comercio una mejor Costa Rica mala influencia un alto grado de

Weight

Semantic class

3 1 0 1 (3) 2 0 1 2 1 2 1 0

Shape & space Shape & space Evaluation Shape & space Temporal Shape & space Evaluation Shape & space Shape & space Evaluation Evaluation Shape & space

Given that in the spoken data there were only 12 cases of a preposed A, it would be useful to list them all, and these are given in Table 9.5. The findings for heaviness are strong: preposed adjectives tend to be equal or lighter than the noun: 83.8 per cent overall and in written discourse, and 100 per cent in spoken discourse, In spoken discourse, all AN pairs respected a word-length principle; that is, the As were either syllabically equal to or lighter than the Ns they preceded. These frequencies were supported by a comparison of mean weights of the A across written and spoken discourse. Postposed As in both the spoken and written corpus’ were substantially heavier (3.76 and 3.81) than the preposed As (2.0 and 2.74). Preposed As in the written corpus had a mean weight of 2.74, while As in the spoken corpus had a mean weight of only 2.0. In this regard, there is a clear difference between written and spoken data. A logical explanation is performance; performance constraints would be more visible in unplanned discourse (that is, spoken) than in planned discourse (that is, written).

Discussion Throughout the literature, it seems that the analysis of Spanish adjective position faces a written language bias. Itkonen (1991) notes that linguists tend to analyse written rather than spoken sentences, which has the negative consequence that units resulting from such grammatical analyses are characteristic of written language only. We share this concern and show that attributive adjectives in Spanish vary across written and spoken discourse in terms of frequency of position (whether the adjective appears in prenominal or postnominal position), distribution of semantic class, and adherence to word-length constraints. The richness and variety of adjective use is greater in written discourse. Table 9.6 indicates that the preplacement of the A happens with much

212 Spanish Adjective Position: Differences between Written and Spoken Discourse

greater frequency in written discourse than it does in spoken discourse, 24.2 per cent to 5.3 per cent respectively.9 Why would such a great difference exist between written and spoken discourse? Clearly, the preplacement of A represents a greater burden on performance and or processing. Table 9.7 shows that written discourse allows multiple (2) adjectives to be linked to one noun more frequently than does spoken discourse (8.3% v. 1.5% respectively). Table 9.8 suggests that the distribution of semantic class of the adjective differs greatly across spoken and written discourse. As shown by the Chi-square test in Table 9.9, this distribution is nonrandom and statistically significant. These findings support the claim that one cannot depend solely on written language to make claims about language in general. This is especially true when examining performance-based constraints. Clearly, performance constraints will be more active in on-line speech production and processing than in off-line performance (that is, reading and writing). The differences noted across written and spoken language seem to reflect a difference in ease

Table 9.6 Adjective position

Adjective preposed (AN) Adjective postposed (NA)

Spoken (226)

Written (256)

12 (5.3%) 214 (94.7%)

62 (24.2%) 194 (75.8%)

Table 9.7 One v. two adjectives modifying same noun

1 adj. el carro rojoA 2 adj. el nuevoA carro rojoA

Spoken (231)

Written (287)

226 (97.8%) 5 (2.2%)

256 (89.2%) 31 (10.8%)

Table 9.8 Distribution of semantic categories Semantic classification Relational Shape & space Evaluative Modality Personality Temporal Physical

Total

Spoken (226)

Written (256)

282 52 47 36 32 28 5

156 (69.0%) 26 (11.5%) 12 (5.3%) 5 (2.2%) 17 (7.5%) 7 (3.1%) 3 (1.3%)

126 (49.2%) 26 (10.2%) 35 (13.7%) 31 (12.1%) 15 (5.9%) 21 (8.2%) 2 (0.8%)

Richard J. File-Muriel 213 Table 9.9 Chi-square test for Table 9.8 data (semantic class)

Pearson Chi-square Likelihood ratio Linear-by-linear assoc. No. of valid cases

Value

df

Asymp. sig. (2-sided)

38.833a 41.630 14.231 482

6 6 1

.000 .000 .000

a

2 cells (14.3%) have expected counts less than 5. The minimum expected count is 2.34.

of production and processing between the two registers. If a particular structure is said to be difficult to process, the hypothesis could be advanced that it should also occur with less frequency in spoken samples. Comparisons between spoken and written data could also form the basis for hypotheses about planned v. unplanned discourse. A comparison of written texts from different time periods may also help determine patterns of diachronic grammaticalization in adjective position. The findings in the present study regarding journalistic prose are strikingly similar to those in Prado (1980) based on the literary works Crónica de una muerte anunciada by Gabriel García Marquez (1981) and Cobra by Severo Sarduy (1974). The journalistic prose sample in the present study (AN 24.2% and NA 75.8%) shares the same AN and NA distribution with the data gleaned from the two twentieth-century novels (23%–27% AN and 73%–77% NA). Prado points out that the averages were quite different 100 years ago, examining Pepita Jiménez by Juan Valera (1886) and Doña perfecta by Benito Pérez Galdós (1876), where the distribution is 55 per cent–58 per cent AN and 42 per cent–45 per cent NA. If we trace A position back even farther, we see a noticeable preference for A preplacement. Grimsley (2000), for example, examines three plays written by Calderón dating back to 1648: La segunda esposa y Triunfar muriendo, Andromeda y Perseo, and El indulto general, finding that the allocation of the A in prenominal position was predominant at 62.9 per cent, as opposed to postnominal position 37.1 per cent. Penny (1994) notes that in Latin the descriptive adjective could be both pre- and postposed. It appears, then, that adjective position in written Spanish has increasingly migrated towards postnominal position, as illustrated by the percentages in the aforementioned text, chronologically ordered in Table 9.10. This development suggests that the postplacement of the Spanish adjective may be grammaticalizing due to the processing constraints on word length. As Hawkins (1994: 14) points out in reference to VO word order in English, grammatical data should correlate with degrees of processing ease in performance, providing evidence for its influence on grammars: ‘The degree of processing complexity will determine which structures are selected for grammaticalization from each domain.’ Structures that are easier to produce and process will grammaticalize before more difficult structures of the

214 Spanish Adjective Position: Differences between Written and Spoken Discourse Table 9.10 Diachronic distribution of A position in texts Adjective position Texts

Pre

Post

3 plays (Calderón 1648–09) Doña perfecta (Galdós, 1876) Pepita Jiménez (Valera, 1886) La sangre y la ceniza (1965) Cobra (Sarduy, 1974) Crónica de una muerte anunciada (Marquez, 1981) Letters/editorials in la Nación (1996–2000)

63 58 55 45 23 27 26

37 42 45 55 77 73 74

Table 9.11 Mean word length of As by semantic class Semantic class

A

pre

pos

%preposed

relational evaluation personality modality temporal physical shape & space

3.93 3.55 3.5 3.25 3.11 2.8 2.56

0 22 6 11 10 2 23

282 25 26 25 18 3 29

0.0 46.8 18.8 30.6 35.7 40.0 44.2

relevant type. Processing and production ease and efficiency can determine the relative ordering of categories. The commonly held notion about Spanish adjectives is that there is a large number of adjectives that cannot appear in prenominal position. These include adjectives of relational nature, which are affiliated with geographic entities, ethnic groups, social or political groups, professional or scientific disciplines, or symbolic or philosophic movements (Delbecque, 1990). Mean word lengths were recorded for nouns and adjectives in the entire corpus, as well as adjectives by semantic class (Table 9.11). Overall, the adjective was heavier with a mean of 3.61 to the noun at 3.19. It is not surprising, then, that the relational semantic class, whose members do not appear in prenominal position, is syllabically the heaviest adjective class in our corpus at 3.91. In fact, with one exception (evaluation) a shorter mean A length corresponds exactly to an increase in the preposition of the A. At first glance, the semantic class evaluation (3.55) seems to pose a problem to the notion of light-before-heavy, as it was the second heaviest A and the most frequently preposed (46.8%). A closer examination of the data, however, shows that the adjectives in the semantic class evaluation did obey a heaviness principle. The mean weights for the As in this class were 3.55 overall, 3.88 for postposed As, and 3.18 for preposed.

Richard J. File-Muriel 215

Conclusions By comparing written and spoken discourse, one can advance claims about planned (for example written) v. unplanned (for example spoken) discourse types. Planned discourse may violate certain constraints on production and processing ease that are more strictly adhered to in spoken/unplanned discourse. As far as the writer is concerned, there is no on-line time constraint in production in that s/he does not face the on-line pressures typical in conversational discourse and can write, erase/delete and rewrite. Likewise, the reader can read and reread. This would explain the stricter observance of the weight constraint by the spoken data found in the present study. We have argued that length in general, not just syntactic complexity (Hawkins, 1994; Arnold et al., 2000), should be considered when dealing with constraints on performance. The present study has shown that a constraint on word length disfavours syllabically heavy adjectives in prenominal position because there is more linguistic material to process. That is to say, word length (measured in syllables) is a performance constraint which influences structure. When lighter adjectives are postposed, it is because they belong to a semantic class (relational) that has grammaticalized (that is, it categorically occurs in postnominal position) due to issues of the word length of its class. This study presents more evidence against the notion of a phonology-free syntax. It’s essential that linguists not dismiss tendencies as purely stylistic choices made by speakers that have no subsequent effects on the grammatical structure of language. The present study shows that these tendencies exist due to performance constraints and ease of processing and that tendencies can ultimately lead to grammaticalization. The total absence of violations of the word-length constraint in our spoken corpus suggests that most of the violations of length (that is, when the heavier adjective is preposed) cited in the literature come from written texts or created examples not representative of spoken discourse. As far as adjective use is concerned, it has been demonstrated in this study that written data is an inadequate representation of spoken language and one should be cautious when making claims based on written texts only. In order to fully appreciate the role that performance constraints play in constituent ordering, one must necessarily deal with spoken rather than written data. Future research could attempt to extend these findings across different registers. We propose that as registers go from informal to formal, there may be less adherence to processing principles. That is, informal speech will be marked by an accelerated tempo, thus, a higher faithfulness to processing constraints would be necessary for both the speaker’s planning and production, and taking into consideration the processability by the listener. By contrast, formal or careful speech would be marked by a slower tempo, more closely representing planned speech and written discourse.

216 Spanish Adjective Position: Differences between Written and Spoken Discourse

The claim here is not that weight is the sole determiner of adjective placement in Spanish. Clearly, there is a variety of competing factors in determining its position (weight, semantics and perhaps frequency). As suggested by Clements (personal communication), the frequency of use and length of a word may be linked. On the one hand, frequency may be the driving force behind short forms of adjectives (buen-bueno, gran-grande). On the other hand, it may be that frequent preposing is the reason for reduction. Furthermore, it may be that high frequency is the actual driving force behind the selection of words for preposition. Future research should look at frequency as a possible performance constraint. Certainly, one would expect that more frequently accessed lexical items, regardless of length, would be easier to produce and process than less frequently accessed words. Alameda and Cuetos (1995) have compiled a frequency dictionary for written Spanish, and, interestingly, all the preposed adjectives in our spoken corpus were among the most frequently occurring words in their dictionary (see Appendix A), the average occurrence being 1,357 times for each word. By contrast, the postposed adjectives in our spoken corpus occurred with much lower frequency in the dictionary.10 There is also an interaction with semantic class and frequency. Preposing an evaluative A such as mal is quite common in both spoken and written Spanish. It should also be noted that the number of possible evaluative As is comparatively limited to the number of relational adjectives and that many evaluative adjectives (for example bueno ‘good’, mal ‘bad’, mejor ‘better’ and so on) are among the most frequently occurring adjectives in the language. Thus, it is difficult to tell whether the driving force is semantic class or frequency of occurrence.

Appendix A Adjectives preposed in spoken corpus and the number of occurrences in Alameda and Cuetos (1995): gran, grande(s) mal (a/o/as/os) bajo (a/o/as/os) nuevo (a/as/os) buen (a/as/o/os) libre (s) mejor (es) alto (a/o/as/os)

2,147 1,228 1,250 1,760 1,868 358 1,348 895

Notes 1 Benjamin Hebblethwaite (personal communication 10 November 2002) points out that Haitian Creole is even more restrictive about allowing heavy adjectives in prenominal position.

Richard J. File-Muriel 217 2 a. El joven ancioso de cantar se puso nervioso. ‘The child anxious to sing became nervous.’ b. *El ancioso de cantar joven se puso nervioso. ‘The anxious to sing child became nervous.’ c. Las rutas difíciles de ascender son las más estimadas. ‘The routes difficult to climb are the most esteemed.’ d. *Las difíciles de ascender rutas son las más estimadas. ‘The difficult to climb routes are the most esteemed.’ 3 A CRD (Constituent Recognition Domain) is a type of structural domain consisting of the constituents that make the phrase-structure recognition possible online. Hawkins defines it as follows: ‘A CRD for a mother node M consists of the terminal and non terminal nodes that must be parsed in order to recognize M and all immediate constituents (IC)s of M, proceeding from the terminal node that constructs the first IC on the left, to the terminal node that constructs that last IC on the right, including all intervening terminal nodes and the non terminal nodes that they construct’ (1994: 58–9). 4 The following sentence with its possible orderings illustrates how the CRD works. The CRDs for S and VP are calculated by the number of ICs to non-ICs. In other words, the number of words the listener has to parse before all ICs are recognized: [SRicardo [VPllevó] a su] novia S 2/2 (100) VP 2/2 (100) (100) [[S VPLlevó Ricardo] a su] novia S 2/2 (100) VP 2/3 (66) (83) [VPA su novia [SRicardo llevó]] S 2/2 (100) VP 2/5 (40) (70) 5 Arnold et al. (2000) calculate syntactic weight in terms of ‘number of words’. Since they do not refer to grammatical categories, it is not clear whether heaviness is being calculated syntactically or phonologically. 6 The Align-Hansard corpus consists of transcriptions of debates in the Canadian parliament. 7 Wasow (1997a) notes that weight effects can be measured quite accurately by counting words, nodes or phrasal nodes and concludes that these measures are indistinguishable as predictors of order. 8 Wasow (1997a) found that weight phenomena are sensitive, not only to the weight of one constituent, but rather to the relative weights of the constituents whose ordering is at issue. 9 Analogously, Hawkins (1994: 452–3) predicts that spoken data will conform to the Early Immediate Constituent, just like written data. ‘If anything, we expect that spoken data will be more (or equally) in conformity with EIC, because the processing load associated with less efficient Constituent Recognition Domain will not be helped by the additional source of visual information that comes with written material.’ 10 In the words surveyed, most of the postposed adjectives in my corpus occur less than 100 times in the frequency dictionary. The only exceptions constitute those words that function in multiple lexical categories. For example, the word general ‘general’ functions not only as an adjective, but also as a noun (military general) and as part of the adverbial phrase por lo general ‘in general.’ Even so, these words fall far short in frequency compared to the preposed adjectives in my corpus.

218 Spanish Adjective Position: Differences between Written and Spoken Discourse

References Alameda, J. R. and F. Cuetos (1995) Diccionario de frecuencias de las unidades lingüísticas del castellano. Oviedo: La Universidad de Oviedo. Arnold, J., T. Wasow, A. Losongco and R. Ginstrom (2000) ‘Heaviness vs. Newness: The Effects of Structural Complexity and Discourse Status on Constituent Ordering’, Language, vol. 76, no. 1, pp. 28–55. Bever, T. G. (1970) ‘The Cognitive Basis for Linguistic Structures’, in J. R. Hayes (ed.), Cognition and the Development of Language. New York: Wiley, pp. 279–362. Bock, J. K. (1987) ‘An Effect of the Accessibility of Word Forms on Sentence Structures’, Journal of Memory and Language, vol. 26, pp. 119–37. Delbecque, N. (1990) ‘Word Order as a Reflection of Alternate Conceptual Construals in French and Spanish: Similarities and Divergences in Adjective Position’, Cognitive Linguistics, vol. 1, pp. 349–416. Dell, G. S. and P. A. Reich (1981) ‘Stages in Sentence Production: An Analysis of Speech Error Data’, Journal of Verbal Learning and Verbal Behaviour, vol. 20, pp. 611–29. Erdmann, P. (1988) ‘On the Principle of “Weight” in English’, in C. Duncan-Rose and T. Vennemann (eds), On Language, Rhetorica, Phonologica Syntactica: A Festschrift for Robert P. Stockwell from his Friends and Colleagues. London: Routledge, pp. 325–39. Fernández, S. (1951) Gramática Española. Madrid: Revista de Occidente. Garrett, M. F. (1982) ‘Production of Speech: Observations from Normal and Pathological Language Use’, in A. Ellis (ed.), Normality and Pathology in Cognitive Functions. London: Academic Press, pp. 19–76. Gili Gaya, S. (1961) Curso superior de sintaxis española. Barcelona: Biblograf. Grimsley, S. (2000) ‘Quantitative Study of Historical Spanish Adjective Placement’, Paper presented in the 4th Hispanic Linguistic Symposium. Hawkins, J. A. (1994) A Performance Theory of Order and Constituency. Cambridge: Cambridge University Press. Itkonen, E. (1991) Universal History of Linguistics: India, China, Arabia, Europe. Amsterdam: John Benjamins. Klein-Andreu, F. (ed.) (1983) Discourse Perspectives on Syntax. New York: Academic Press. Levelt, W. (1989) Speaking. Cambridge, MA: MIT Press. Levelt, W. and B. Maassen (1981) ‘Lexical Search and Order of Mention in Sentence Production’, in W. Klein and W. Levelt (eds), Crossing the Boundaries in Linguistics. Dordrecht: Reidel. 221–252. Miller, P. H., G. K. Pullum and A. M. Zwicky (1997) ‘The Principle of Phonology-Free Syntax: Four Apparent Counterexamples in French’, Journal of Linguistics, vol. 33, pp. 67–90. Pajunen, A. (1998) ‘Adjectives in Spoken Language’, Word, vol. 49, pp. 341–68. Penny, R. (1994) History of the Spanish Language. Cambridge: Cambridge University Press. Prado, M. (1988) ‘The Semantics of Adjective Position in Spanish’, Selecta, vol. 9, pp. 1–8. Terker, A. (1985) ‘On Spanish Adjective Position’, Hispania, vol. 68, pp. 502–10. Wasow, T. (1997a) ‘Remarks on Grammatical Weight’. Language Variation and Change, vol. 9, pp. 81–105. Wasow, T. (1997b) ‘End-Weight from the Speaker’s Perspective’, Journal of Psycholinguistic Research, vol. 26, pp. 347–61. Wilmet, M. (1980) ‘Antéposition et postposition de l’épithète qualificative en français contemporain: Matériaux’, Travaux de linguistique, vol. 7, pp. 179–201.

10 Adjective Placement and Noun Semantics in Spanish Mariche García-Bayonas

Introduction This chapter addresses the issue of adjective placement in Spanish and, more specifically, it investigates how adjective placement is related to semantic interpretation of both the noun it modifies and the noun phrase (NP) where it is contained. Adjectives in Spanish are said to have a restrictive or nonrestrictive interpretation depending on their position relative to the noun they modify (Luján, 1980). The restrictive interpretation of a noun-adjective (NA) or an adjective-noun (AN) pair is a complex matter. Delbecque (1990: 350) notes that the semantic interpretation of adjectives such as Spanish grande ‘big’ – when preposed it refers to what the entity denoted by the noun does or has done and when postposed to a physical characteristic of the person (for example gran hombre ‘great man [because of his deeds]’ v. hombre grande ‘big man’) – is valid for a restricted number of adjectives and in conjunction with a restricted number of nouns that usually refer to humans. So, for example, the noun phrases in (10.1) can be interpreted to be a person who knows and performs music well or a good person who is a musician. The one or the other interpretation would be determined, ultimately, by context: (10.1)

a. buen músico ‘good musician’ b. músico bueno musician good ‘good musician’

Another example shows, however, that adjective position may more important for the interpretation of the noun phrase (NP) than previously thought. In the example in (10.2), the NP in (10.2a) refers not to a civil servant who is tall, but rather to a civil servant who is highly ranked in his profession. 219

220 Adjective Placement and Noun Semantics in Spanish

The most common interpretation of (10.2b), by contrast, is a civil servant who is physically tall: (10.2)

a. alto funcionario tall/high civil servant ‘high-ranking civil servant’ b. funcionario alto civil servant tall/high ‘tall civil servant’

A restrictive interpretation of an adjective, then, can refer to the general semantics of the noun it modifies (for example for músico such a feature might be [human]), or to its more specific semantic features (for example for músico, these might be [profession], [music performance], and so on), with the context often providing the key disambiguating material if necessary. In some NA/AN pairs, such as those in (10.2), there is a clearer reference in AN order to the more specific semantic features of the noun, and in NA order to its more general semantic features. In other words, the semantic interpretations of the AN order in (10.2a) and that in (10.2b) are closer to being conventionalized in the sense that the AN and NA orderings are being given a certain interpretation, independently of the actual lexical items in the construction.

Spanish adjectives with professions In this study, we limit ourselves to discussing general v. specific semantic features of nouns that denote human professions or vocations. For instance, the general semantic features of two nouns used in the present study, poeta ‘poet (masculine)’ and azafata ‘flight attendant’, would be roughly those shown in (10.3): (10.3)

POETA [masculine] [animate] [human] [countable] …

AZAFATA [feminine] [animate] [human] [countable] …

An incomplete characterization of the specific semantic make up of these nouns, in terms of semantic features, would be roughly those given in (10.4): (10.4)

POETA AZAFATA [profession/vocation] [profession]

Mariche García-Bayonas 221

[writer] [poems] …

[waitress] [airplane] …

In Spanish, such adjective-noun combinations as buen poeta, poeta bueno ‘good poet’, or gran azafata, azafata grande ‘great/big airhostess’ are common. The possible restrictive semantic readings for these constructions are given in (10.5) and (10.6) respectively: (10.5)

(10.6)

buen poeta/poeta bueno: a. a poet who is good as a poet, referring to the specific semantics of the noun. b. a poet who is a good person, and may or may not be a good poet, referring to the general semantics of the noun. gran azafata/azafata grande: a. a flight attendant who is great as a flight attendant, referring to the specific semantics of the noun. b. a flight attendant who is physically big, referring to the general semantics of the noun.

Although there are at least two possible restrictive readings for each AN or NA combination in (10.5) and (10.6), there seems to be a contextindependent tendency to give the specific reading in (10.5a) to buen poeta. Similarly, gran azafata generally receives a specific reading, while azafata grande refers to the general semantics of the noun. The question we pose in this study is: to what extent is this the case for other AN and NA combinations? The question of specific v. general reference of AN or NA combinations is hardly touched upon in the literature on the semantic interpretation of Spanish adjective placement. Traditionally, the distinction is made that postplaced adjectives restrict the scope of the noun (edificios hermosos ‘beautiful buildings’ is a subset of the set of edificios ‘buildings’), whereas preplaced adjectives do not. Research by Klein-Andreu (1983) shows that pragmatic factors play a larger role in the semantic interpretation of AN and NA combinations than previously thought. Other research, such as that by Luján (1980), focuses on the semantic interpretation of more than one postplaced adjective. However, neither scholar addresses the question of specific v. general semantic interpretation. Delbecque (1990: 375, 402–3), on the other hand, does address this issue, noting that in AN combinations the A represents the Landmark and the background element to which the domain of the N (the Trajector and foreground element) is assigned. The result is that the A modifies the whole semantic domain of the N, the specific as well as the general semantic traits. By contrast, in NA combinations the N is the Trajector and background element and the A the Landmark and foreground

222 Adjective Placement and Noun Semantics in Spanish

element, which allows, among other things, that the adjective may be assigned to only part of the semantic domain of the noun (see example [10.2] above). The question we ask in this study is to what extent this is the case, independently of contextual or discourse cues.

Research questions In this chapter we address two specific questions: what role (if any) does preor postplacement of an adjective play in the semantic interpretation of the noun it modifies, independently of context? And, if pre- or postplacement of the adjective does play a role in the semantic interpretation of the noun it modifies, why is this so and does the grammaticalization of the AN or NA construct play a role in the semantic interpretation of NA/AN combinations? To attempt to respond to these questions, a study was designed to test whether there is a significant correlation between specific semantics reference and preplacement on the one hand, and general semantics reference and postplacement, on the other.

The study The instrument A list of NA/AN combinations (40 in all, 27 of which were used for the present study; see Appendix) was given to each of the 106 participants in the study (20 adults and 86 adolescents).1 The adjectives buen ‘good’, terrible ‘terrible’, alto ‘tall, high’, pequeño ‘small, low’, grande ‘big’, genial ‘brillant’, malo ‘bad’ and magnífico ‘magnificent’ were used four times, twice pronominally and twice postnominally, as attributive modifiers of nouns referring to professions (compositor ‘composer’, estudiante ‘student’, funcionario ‘civil servant’ [used twice], poeta ‘poet’, empresario ‘business person’ [used twice], filósofo ‘philosopher’, arquitecto ‘architect’, economista ‘economist’, médico ‘doctor’, abogado ‘lawyer’,azafata ‘flight attendant’, campesino ‘peasant’, maestro ‘teacher’, gimnasta ‘gymnast’, recepcionista ‘receptionist’, cura ‘priest’, actor ‘actor’, presidente ‘president’ [used twice], cómico ‘comedian’, fotógrafo ‘photographer’, músico ‘musician’, alumno ‘pupil’, escritor ‘writer’, matemático ‘mathmatician’, aparejador ‘surveyor’, contable ‘accountant’, naturópata ‘naturopath’, notario ‘notary public’, piloto ‘pilot’, granjero ‘farmer’, professor ‘teacher, professor’, tenista ‘tennis player’, telefonista ‘telephone operator’, monja ‘nun’ and actriz ‘actress’. The general goal of the task was to ascertain the extent to which native speakers of Spanish of different ages display a significant tendency to match up adjective preplacement with what we have called the specific semantic interpretation of the noun and postplacement with the general semantic interpretation of the noun. To afford a comparison, some of the nouns

Mariche García-Bayonas 223

have been selected as pairs with regard to their lexical meaning. For instance, buen compositor ‘good composer’ corresponds to músico bueno ‘good musician’. As indicated in the list, three nouns (funcionario ‘civil servant’, empresario ‘business person’, presidente ‘president’) were included twice in the instrument with the same adjective, once prenominally and once postnominally. The rationale for this was to compare responses of participants to possible lexicalized expressions (for example, alto funcionario ‘high-level civil servant’ v. funcionario alto ‘tall civil servant’; pequeño empresario ‘small-time business person’ v. empresario pequeño ‘small business person’, and so on) as opposed to possibly less lexicalized expressions, such as mal presidente ‘bad president’ v. presidente malo ‘president who is a bad person’. Moreover, the adjective pobre ‘poor’, though excluded from the general study, was included in order to test the extent to which its commonly held pronominal interpretation, ‘unfortunate, miserable’, and postnominal interpretation, ‘financially poor’, would hold for the set of participants of the study. The order of appearance of combinations on the list is random. This, along with the fact that the test was designed so that adjectives appeared twice in the same position, was meant to function as a test of reliability among the participants. Thus, if they interpret funcionario alto to be a case of reference of the adjective to the general semantics of the noun, as in ‘a tall civil servant’, the same could be expected to hold for azafata alta as well. It will become apparent that this is not always the case. As already mentioned, the combinations were given to the participants devoid of any context. The participants were given the option of marking whether the meaning of the AN or NA combination referred to the general semantics of the noun, its specific semantics, or another option in case they did not agree with either of the other possible answers offered. Participants The test was administered to two groups of participants: an adult group (part one) and a preadolescent group (part two). Part one of the study took place in Marbella, Spain. The adult subjects (n  20) were selected randomly, 10 males, and 10 females, in order to prevent possible gender differences. All participants were native speakers of Spanish living in Andalucía, Spain, their ages ranging from 18 to 65 (mean age of 32.4). Part two of the study took place in Madrid, Spain. Children (n  86) were selected randomly from a primary school for this part of the study. Their ages ranged from 10 to 14 (mean of 11.28). All members of this group were also native speakers of Spanish. The dependent variable in the study was pre- or postplacement of the adjective. The independent variables coded for were age, gender, number of books reported to have been read in the preceding year (for children), and specific-general-other semantics of the noun.

224 Adjective Placement and Noun Semantics in Spanish

Hypotheses The null hypothesis for this study was that there would be no correlation between pre-/postplacement of an adjective relative to the noun it modifies and the age, gender, number of books read by the participants (children), or the semantics of the noun. However, based on the semantic readings commonly assigned to alto funcionario ‘high-ranked civil servant’ and funcionario alto ‘tall civil servant’ (cf. 10.2 above), we expected that there might be a tendency for a preposed adjective to receive a semantic interpretation applying to the specific semantics of the noun, and a postposed adjective to the general semantics of the noun. Moreover, given the observation by Delbeque (1990) that the semantic interpretation of an AN or NA combination is ultimately determined by context, we expected that participants would display vacillation between different semantic readings because of the lack of context. That is, we expected there to be randomness. Finally, given that adjective use develops later in language (cf. Gili Gaya, 1961), we also expected that there might be more indecisiveness, reflected in higher variation, in the preadolescents’ responses relative to those of their adult counterparts.

Results and discussion The null hypothesis held regarding the correlation between the effect of age, gender and number of books reportedly read by the participants (children) on pre- v. postplacement of an adjective. That is, we found no statistically significant correlation between adjective placement relative to the noun it modifies on the one hand and age, gender and number of books read by the participants (children) on the other. Before moving on to address the frequencies and distributions of the two groups at large, we will first discuss the results of individual items in the instrument. As mentioned above, various pairs tested with the instrument may be partially or entirely. The pair funcionario alto and alto funcionario is a case in point. The semantic interpretation of the AN ordered phrase is typically ‘high-level civil servant’, referring to the specific semantics of the noun, whereas the semantic interpretation of the NA ordered phrase is ‘tall civil servant’, referring to the general semantics of the noun. Table 10.1 displays the frequency of responses from both the adults and children for alto funcionario v. funcionario alto. The responses from both groups are comparable. Although it is by no means categorical, both groups preferred the general semantic interpretation for the NA order (funcionario alto), with the children displaying a stronger preference. This could be because among younger people this phrase is at a different degree of lexicalization than with the adults. A more detailed study of this would be useful to determine if this is in fact the case. For alto funcionario, the percentages of the two groups are almost identical in overwhelmingly preferring the specific semantic interpretation. This

Mariche García-Bayonas 225

suggests that the AN phrase is more lexicalized that its NA counterpart. Again, a more detailed study could make a more definitive statement as to whether or not this is true. In light of the results in Table 10.1, the pairs azafata alta and alto piloto, shown in Table 10.2, are remarkable: while the children’s responses for these pairs are comparable to those in Table 10.1, the adults overwhelmingly preferred the general interpretation of the NA combination azafata alta (95%), but only slightly more than half (55%) preferred the specific interpretation in the AN phrase alto piloto. By contrast, only 53.5 per cent of the children preferred the general interpretation of azafata alta but overwhelmingly preferred the specific interpretation of alto piloto. Again, this may have to do with generational differences which an in-depth study would better determine. There might also be an analogical influence of alto funcionario in the children’s language. That is, this phrase may be serving as a type of exemplar, or prefab, for children to form other similar conventionized collocations with an analogous semantic interpretation (see Bybee, 2005). By contrast, the results of the adults’ preferences do not show this possible effect.

Table 10.1 Interpretations of alto funcionario and funcionario alto in adults and children Adults

Children

Funcionario alto

Specific General Other Total

9 (45%) 10 (50%) 1 (5%) 20 (100%)

23 (27%) 50 (58%) 13 (15%) 86 (100%)

Alto funcionario

Specific General Other Total

18 (90%) 1 (5%) 1 (5%) 20 (100%)

77 (90%) 9 (10%) — 86 (100%)

Table 10.2 Interpretations for azafata alta and alto piloto in adults and children Adults

Children

Azafata alta

Specific General Other Total

1 (5%) 19 (95%) — 20 (100%)

46 (53.5%) 31 (36%) 9 (10.5%) 86 (100%)

Alto piloto

Specific General Other Total

11 (55%) 8 (40%) 1 (5%) 20 (100%)

77 (90%) 8 (9%) 1 (1%) 86 (100%)

226 Adjective Placement and Noun Semantics in Spanish

Another phrase that we perceived to possibly be lexicalized is pequeño empresario ‘small business owner’. In Table 10.3, we compare this to its NA counterpart empresario pequeño ‘little (short-statured) business owner’, and to two other phrases that could have been influenced by pequeño empresario by analogy. For pequeño empresario ‘small business owner’, both the adults and children display the same strong preference toward a specific interpretation. This would suggest that the phrase is highly lexicalized. Moreover, the similar split in preference for empresario pequeño between the general and specific readings apparent in both groups points to a lack of lexicalization, similar to that found for funcionario alto discussed in Table 10.1. The preferred semantic interpretations of preposed v. postposed pobre (‘miserable’ v. ‘poor’, respectively) and simple (‘mere’ v. ‘simple, stupid’, respectively) are generally perceived to be somewhat lexicalized in colloquial speech. That is, pobre filósofo is more often thought to refer to a philosopher who is down on his luck rather than one who is economically not well-off. A monja pobre, on the other hand, is generally considered a nun without economic means rather than one who is down on her luck. Similarly, for a simple poeta the perceived preferred interpretation would be a ‘mere poet’, whereas gimnasta simple would be a simple-minded gymnast. Although we did not include these examples in the overall data gleaned from the groups, we wanted to see to what extent the preferences of the adults and children in our study coincided with the preferences we perceived. The results are given in Tables 10.4 and 10.5. Table 10.3 Semantic interpretations for N-pequeño and pequeño-N in adults and children Adults

Children

Pequeño empresario (lit. ‘small bus. owner’)

Specific General Other Total

19 (95%) — 1 (5%) 20 (100%)

77 (90%) 8 (9%) 1 (1%) 86 (100%)

Empresario pequeño (lit. ‘small bus. owner’)

Specific General Other Total

12 (60%) 6 (30%) 2 (10%) 20 (100%)

59 (69%) 26 (30%) 1 (1%) 86 (100%)

Pequeño granjero (lit. ‘small farmer’)

Specific General Other Total

18 (90%) 2 (10%) — 20 (100%)

66 (77%) 18 (21%) 2 (2%) 86 (100%)

Campesino pequeño (lit. ‘small farmer’)

Specific General Other Total

6 (30%) 14 (70%) — 20 (100%)

58 (68%) 27 (31%) 1 (1%) 86 (100%)

227 Table 10.4 ‘Miserable’ v. ‘poor’ interpretation for preposed v. postposed pobre Adults

Children

Pobre filósofo

‘miserable’ ‘poor’ Other Total

10 (50%) 1 (5%) 9 (45%) 20 (100%)

38 (44%) 20 (23%) 28 (33%) 86 (100%)

Pobre cura

‘miserable’ ‘poor’ Other Total

16 (80%) 3 (15%) 1 (5%) 20 (100%)

54 (63%) 21 (24%) 11 (13%) 86 (100%)

Matemático pobre

‘miserable’ ‘poor’ Other Total

2 (10%) 13 (65%) 5 (25%) 20 (100%)

19 (22%) 47 (55%) 20 (23%) 86 (100%)

Monja pobre

‘miserable’ ‘poor’ Other Total

3 (15%) 15 (75%) 2 (10%) 20 (100%)

13 (15%) 66 (77%) 7 (8%) 86 (100%)

Table 10.5 ‘Mere’ v. ‘simple-minded’ interpretation for preposed v. postposed simple Adults

Children

Simple poeta

‘mere’ ‘simple-m,’ Other Total

3 (15%) 12 (60%) 5 (25%) 20 (100%)

4 (5%) 78 (90%) 4 (5%) 86 (100%)

Simple tenista

‘mere’ ‘simple-m,’ Other Total

2 (10%) 13 (65%) 5 (25%) 20 (100%)

7 (8%) 72 (84%) 7 (8%) 86 (100%)

Gimnasta simple

‘mere’ ‘simple-m,’ Other Total

11 (55%) 7 (35%) 2 (10%) 20 (100%)

33 (38%) 48 (56%) 5 (6%) 86 (100%)

Escritor simple

‘mere’ ‘simple-m.’ Other Total

9 (45%) 8 (40%) 3 (15%) 20 (100%)

18 (21%) 61 (71%) 7 (86%) 86 (100%)

228 Adjective Placement and Noun Semantics in Spanish

The positional interpretations of pobre appear to be largely lexicalized. That is, in preposed position its interpretation is ‘miserable’, postposed its interpretation is ‘poor, having little money and few possessions’. With respect to preposed v. postposed simple, preposed ‘simple-minded’ appears also to be largely lexicalized for the children’s group, but less so for the adult group of informants. Analogously, gimnasta/escritor simple is less lexicalized than its AN counterpart for the adult group, but more so for the younger group. Interestingly, both semantic interpretations of both pobre and simple refer, not profession-specific semantic features, but to the general semantic features of the person being denoted by the respective adjective. Chi-square analyses were calculated for the data of the two groups in order to test the statistical significance of preposed v. postposed As, and how they relate to the general or specific semantics of the Ns. The dependent variable is the position (whose values are pre v. postposed position of the adjective). The independent variable is the semantic interpretation of the phrase containing the adjective (whose values are reference to the general v. specific semantics of the noun). As a point of departure, we adopt the null-hypothesis. That is, independently of whether an adjective is preposed or postposed, there should be no correlation between adjective position and the general v. specific semantic interpretation of the phrase in which the adjective appears. In other words, the phrases buen músico ‘good musician’ and músico bueno [lit. musician good] should not correlate with the general v. specific semantic interpretation, that is ‘good person who is a musician’ (general) or ‘good musician’ (specific). The results of the Chi square tests are given in Tables 10.6 (adults) and 10.7 (children).

Table 10.6 Chi-square results for the adults

Preposed Postposed Totals

General

Specific

Other

Totals

30 (12%) 85 (30%) 115

224 (86%) 184 (66%) 408

6 (2%) 11 (4%) 17

260 (100%) 280 (100%) 540

Degrees of freedom: 2, Chi-square  31.08, p  0.0001. The distribution is significant.

Table 10.7 Chi-square results for the children

Preposed Postposed Totals

General

Specific

Other

Totals

184 (16%) 306 (25%) 490

902 (81%) 853 (71%) 1,755

32 (3%) 45 (4%) 77

1,118 (100%) 1,204 (100%) 2,322

Degrees of freedom: 2, Chi-square  30.8, p  0.0001. The distribution is significant.

Mariche García-Bayonas 229

Based on the high significance of the Chi-square test, the null hypothesis is rejected. While there is a very strong overall preference for the specific semantic interpretation independently of the position of the adjective, there is also a statistically significant preference for the general semantic interpretation in conjunction with the postposed adjective. From another perspective, the preference for the specific semantic interpretation is greater in a statistically significant way with the preposed adjective. In sum, independently of any context whatsoever, the position of the adjective influences whether a general v. specific semantic reading of an AN or NA pair is preferred. One possible criticism of the instrument is ordering of the options the informants had to choose from. In the 27 examples we took for this study, the options were always given as in (10.7) (see Appendix): (10.7)

a. Specific semantic interpretation b. General semantic interpretation c. Other

To test whether the specific–general or general–specific ordering in the instrument had an effect on the choices made by the informants, four examples were included that had the ordering shown in (10.8): (10.8)

a. General semantic interpretation b. Specific semantic interpretation c. Other

Using the tokens for the most chosen option (the specific semantic interpretation) because their number was more robust, we performed a Chi-square test on the specific–general v. general–specific ordering in the instrument. The results are shown in Table 10.8. The ordering presented to the informants, whether that in (10.7) or that in (10.8), had no influence on the options chosen by them in this study.

Table 10.8 The significance of specific–general and general–specific ordering in the instrument, calculated with tokens from the specific semantic interpretation in adult and child groups

Adults Children Totals

Spec v. gen

Gen v. spec

Totals

408 (85%) 1,755 (86%) 2,163

71 (15%) 290 (14%) 361

479 (100%) 2,045 (100%) 2,524

Degrees of freedom: 1, Chi-square  0.130 (critical value is 3.84). The distribution is not significant.

230 Adjective Placement and Noun Semantics in Spanish

Another possible reading of the result of Table 10.8 would be that the tokens are too few to get a significant result. However, given that the number of tokens is fairly robust, this possibility can be ruled out.

Conclusion The central question in this study is whether, devoid of context, the position of the adjective relative to the noun it modifies has an influence on the general v. specific semantic interpretation of the noun, as predicted by the analysis advanced by Delbecque (1990). Concretely, we wanted to know whether the position of an A such as buen(o) ‘good’ relative to a noun such as músico ‘musician’ would influence whether an informant preferred a general semantic interpretation of the phrase, that is, ‘a good person who is a musician’ v. ‘a good musician’. Our conclusion, based on a Spanish native-speaker informant base of 20 adults (18–65 years) and 86 children (10–14 years), is that the general v. specific semantic interpretation of such phrases is influenced by position, shown to be significant by Chi-square tests. We also examined the interpretation of semi-fixed expressions such as alto funcionario ‘high-level civil servant’ and pequeño empresario ‘small business owner’ to measure the extent to which they may be lexicalized. We found they were indeed on their way to being lexicalized, and that there may be some exemplar effect carrying over from these to phrases such as alto piloto ‘high-rank’ or pequeño granjero ‘small-scale farmer’. However, for the corresponding phrases with postposed adjective (funcionario alto ‘tall civil servant’, azafata alta ‘tall air hostess’ and campesino pequeño ‘small farmer’) we found little evidence for a lexicalized interpretation. Thus, the AN combinations may lend themselves more so than the NA combinations to the process of phrasal lexicalization. The preferences of interpretation of the phrases containing pobre ‘miserable/ poor’ and simple ‘mere/simple-minded’ by our informants also suggest that the semantic interpretation of these is linked to their position relative to the noun they modify, independently of context. That is, preposed pobre refers to the person denoted as hapless, though not necessarily poor, while postposed pobre refers to the financial state of the person denoted, just as preposed simple refers to the person denoted as a mere being, while postposed simple refers to the mental capacity of the person denoted. These are arguably cases prefabs in that pobre  N and simple  N have a particular semantic interpretation, which is systematically different than N  pobre and N  simple, respectively. Returning to the initial issue of the link between specific semantic interpretation and adjective preplacement v. general semantic interpretation and adjective postplacement, the question needs to be posed as to why this is so. Delbecque (1990) argues that it has a cognitive basis. Following a similar line

Mariche García-Bayonas 231

of reasoning, Clements (2004) examines phrase-initial v. phrase-final position of sentences, DPs and NPs, arguing that the phrase-initial position is cognitively salient and is often used to refer to whole or known entities, whereas the phrase-final position is most commonly used to code partitive or contrastive meaning, or indicate new information. This cognitive interpretation accommodates Spanish adjective position in the following way: preposed adjectives in Spanish are preferred to refer to the specific semantics of an entity denoted by a noun, which logically also include the general semantics of the noun, as well. By contrast, postposed adjectives are preferred to refer to the general semantics of an entity denoted by the noun, that is, only part of the semantic features of the noun. In this context, it is relevant to note an observation by King and Suñer (1999): preposed adjectives can imply that the entity denoted by the corresponding noun is taken to be a whole class, whereas postposed adjective do not have this reading. For example, the phrase perezosos alumnos ‘lazy students’ in sentence (10.9a) refers to a whole set of students, whereas the phrase alumnos perezosos [lit. students lazy] in (10.9b) refers to a subset of the whole set of students (examples are taken from King and Suñer, 1999: 170): (10.9)

a. Los perezosos alumnos de esta clase no sacan buenas notas. ‘The lazy students [all are lazy] in this class do not get good grades.’ b. Los alumnos perezosos de esta clase no sacan buenas notas. ‘Those students in this class who are lazy do not get good grades.’

We argue, in consonance with Debecque (1990) and Clements (2004), that the findings of this study and the semantic interpretations of the sentences in (10.9) are related and have to do with the cognitive nature of phraseinitial and phrase-final positions. The scope for research on this is as vast as it is important. We hope that our study has contributed a part of the answer to the larger question of the cognitive significance of phrase-initial and phrasefinal positions.

Appendix Items used in study 3. Funcionario alto a. prestigioso b. hombre alto c. otro/a 5. Pequeño empresario a. empresario con capital/negocio pequeño b. hombre de estatura pequeña c. otro/a

232 Adjective Placement and Noun Semantics in Spanish 7. Arquitecto grande a. arquitecto famoso/conocido/bueno b. hombre de gran estatura c. otro/a 10. Abogado magnífico a. referido a su profesión b. referido a su cualidad humana c. otro/a 11. Azafata alta a. alta en la escala profesional b. persona de estatura alta c. otro/a 12. Campesino pequeño a. hombre que posee pocas tierras b. hombre de estatura pequeña c. otro/a 13. Buen maestro a. maestro que es bueno b. persona que es buena c. otro/a 15. Terrible recepcionista a. referido a la profesión b. referido a la persona c. otro/a 17. Genial actor a. actor que es muy bueno b. persona que es muy buena c. otro/a 18. Presidente malo a. referido a la profesión b. referido a la persona c. otro/a 19. Magnífico cómico a. referido a la profesión b. referido a la persona c. otro/a 20. Gran fotógrafo a. buen profesional b. personal grande c. otro/a 21. Músico bueno a. referido a la profesión b. referido a la persona c. otro/a 22. Terrible alumno a. mal estudiante b. mal chico c. otro/a 23. Alto funcionario a. funcionario de alto cargo

Mariche García-Bayonas 233

25.

27.

28.

29.

30.

31.

32.

33.

35.

37.

38.

39.

b. funcionario de alta estatura c. otro/a Empresario pequeño a. empresario con poco negocio b. persona de baja estatura c. otro/a Gran aparejador a. aparejador famoso/bueno b. aparejador de gran tamaño c. otro/a Contable genial a. referido a la profesión b. referido a la persona c. otro/a Naturópata malo a. mal profesional b. mala persona c. otro/a Magnífico notario a. gran profesional b. gran persona c. otro/a Alto piloto a. piloto de alto cargo b. persona de estatura alta c. otro/a Pequeño granjero a. granjero con pocas posesiones b. granjero de baja estatura c. otro/a Profesor bueno a. buen profesor b. buena persona Telefonista terrible a. mala profesional b. mala persona c. otro/a Actriz genial a. actriz que es muy buena b. persona que es muy buena c. otro/a Mal presidente a. presidente que no es bueno b. persona que no es buena c. otro/a Presentador magnífico a. buen profesional b. buena persona c. otro/a

234 Adjective Placement and Noun Semantics in Spanish

Items not used in the study 1. Buen compositor a. buena persona b. buen profesional c. otro/a 2. Estudiante terrible a. terrible referido a la persona b. terrible referido al estudiante c. otro/a 4. Simple poeta a. sólo un poeta b. poeta tonto c. otro/a 6. Pobre filósofo a. hombre sin dinero b. hombre que da pena c. otro/a 8. Genial economista a. hombre genial b. profesional genial c. otro/a 9. Mal médico a. hombre malo b. médico de mala calidad c. otro/a 14. Gimnasta simple a. gimnasta que es simple b. persona que es sólo un gimnasta c. otro/a 16. Pobre cura a. cura que da pena b. cura sin dinero c. otro/a 24. Escritor simple a. escritor tonto b. persona simplemente escritora c. otro/a 26. Matemático pobre a. persona sin dinero b. persona que da pena c. otro/a 34. Simple tenista a. tenista tonto b. persona que sólo es tenista c. otro/a 36. Monja pobre a. persona sin dinero b. persona que da pena c. otro/a

Mariche García-Bayonas 235 40. Cámara grande a. un cámara reconocido/famoso b. persona de gran tamaño c. otro/a

Note 1 Of the 13 combinations not included, four contained the adjective simple ‘simple, only’, four contained the adjective pobre ‘poor, miserable’, both of which tend to have a lexicalized meaning in preposed position. Five examples were excluded because they contained a different ordering of the questions from the other 27.

References Beard, R. (1991) ‘Decompositional Composition: The Semantics of Scope Ambiguities and “Bracketing Paradoxes” ’, Natural Language and Linguistic Theory, vol. 9(2), pp. 195–229. Bello, A. (1847) Gramática de la lengua castellana destinada al uso de los americanos. Santiago de Chile. Bybee, J. (2005) ‘The Impact of Use on Representation: Grammar is Usage and Usage is Grammar’, Presidential address to the Linguistics Society of America annual meeting. Oakland, CA, 6–9 January. Clements, J. C. (2004) ‘Salience Effects of Initial (and Final) Position in Clauses, Phrases, and Words’, Paper presented at the High Desert Linguistic Society biennial meeting. University of New Mexico, Albuquerque, NM, 4–6 November. Delbecque, N. (1990) ‘Word Order as a Reflection of Alternate Conceptual Construals in French and Spanish and Divergences in Adjective Position’, Cognitive Linguistics, vol. 1(4), pp. 349–416. Gili Gaya, S. (1961) Curso superior de sintaxis española. Madrid: Vox. King, L. D. and M. Suñer (1999) Gramática española. Análisis y práctica. Boston: McGraw Hill. Klein-Andreu, F. (1983) ‘Grammar in Style: Spanish Adjective Placement’, in F. KleinAndreu (ed.), Discourse Perspectives on Syntax. London: Academic Press, pp. 143–79. Luján, M. E. (1980) ‘La estructuración e interpretación de secuencias de adjetivos’, Boletin de la Biblioteca de Menendez Pelayo, vol. 56, pp. 387–99.

11 Transitivity and Spanish Non-Anaphoric se J. Clancy Clements*

Introduction In the languages of the world, it is not uncommon to find reflexive pronouns taking on other functions. Indeed, the reflexive pronoun is a widespread source for passive-voice constructions in several language families (Semitic, Uto-Aztecan, Athabaskan and Carib languages; see Givón, 1990: 602 and references therein). In Romance and non-Romance European languages (for example the Germanic languages), we find the reflexive pronoun as a marker of the passive and/or middle voice (Abraham, 1995; Fagan, 1992; Vater, 1988).1 In Spanish, the reflexive pronoun marks not only middle and passive voice, but also appears as a marker in impersonal and antipassive constructions (Masullo, 1992), serves as an aspectual marker on verbs of various sorts (see Hernández, 1966: 50), and accompanies certain other verbs where it co-varies with markers of definiteness and foregrounded material. And although it seems that this wide variety of non-anaphoric functions of Spanish se is quite disparate, this chapter presents a unified analysis of nonanaphoric se. With ‘unified’ we mean here that by appealing to a single notion, that of Transitivity as proposed by Hopper and Thompson (1980), the basic function of non-anaphoric se can be accounted for.2 That is, the various functions of Spanish non-anaphoric se are all, essentially, a function of Transitivity. And although the notion of Transitivity, and the focus of this study, are largely functionalist in nature, we refer to notions from the formalist perspective, such as the notions of binding, case assignment and absorption, in order to give a broader characterization of the effects of se on the Transitivity of the clause in which it appears. * Thanks to José Camacho, Edelmiro Salas, Liliana Sanchez and several anonymous reviewers for their comments and suggestions. Thanks also go to Sonia Colina and Phillip Elliott for their comments on an earlier version of this study, and to Richa Pauranik Clements for assistance in proofreading. 236

J. Clancy Clements 237

The non-anaphoric functions of Spanish se have developed from its function as a reflexive pronoun. Reflexivity and reciprocity in Spanish are typically expressed with clitic pronouns me ‘myself’, te ‘yourself’, se ‘himself, herself, themselves’, nos ‘ourselves’, os ‘yourselves’. Some examples are given in (11.1–11.2): (11.1)

(11.2)

Luisai sei lava. Luisa REFL washes ‘Luisa washes herself.’ Juani y Luisaj sei-j aman. Juan and Luisa REFL love ‘Juan and Luisa love each other.’

In order for sentences such as (11.1–11.2) to be interpretable on a propositional level (in Givon’s 1984 sense), certain general semantic/syntactic requirements, given in (11.3) and (11.4), must be met: (11.3)

(11.4)

Reflexive pronouns typically must have an antecedent in their local domain (that is, within a clause containing a finite verb) (≈ Principle A in a PP-type framework) Arguments must be interpretable in terms of case (for example nominative, accusative or dative, and so on), and semantic role (such as agent, experiencer, source, theme and so on).

Apart from these requirements, there is another, expressed in (11.5), which sets a key condition for a felicitous pragmatic interpretation of a reflexive construction: (11.5)

Condition for the felicitous interpretation of the reflexive pronoun: The participant expressed as the lexical subject of the verb is assumed by the speaker of the utterance to have the ability to bring about the action referred to by the verb.

(11.5) has been adapted from one of Searle’s (1990: 123) two preparatory conditions, the sincerity condition, elaborated initially to account for the commissive speech act of promising: If speaker S promises to do proposition p, ‘the speaker believes it is possible for him to do the act’. Accordingly, if S utters that someone does something to him/herself, S assumes that the person being referred to is able to carry out the action referred to by the verb.3 Applying (11.3)–(11.5) to the sentences in (11.1–11.2), we see that the local domain in each case is the whole sentence and both reflexives are bound within their respective local domain. Moreover, the subjects in (11.1–11.2) have nominative case, and the reflexive pronoun se is a direct

238 Transitivity and Spanish Non-Anaphoric se

object (accusative case) of transitive verbs lavar and amar respectively. Both verbs have agentive subjects and patient DOs. Finally, both subjects are human and understood as being able to carry out the event/activity denoted by the respective verbs. Thus, the basic syntactic, semantic, and pragmatic requirements for the proper interpretation of the reflexives are met. According to Cennamo (1993), the development of Spanish se from a reflexive/reciprocal to an intransitive impersonal traced the path in (11.6). Both reflexive-verb passives (for example se venden muebles) and impersonals (for example se vende muebles) are found as early as the ninth to eleventh centuries (see Cennamo, 1993, in Maldonado, 1996: 97). From the thirteenth to the nineteenth century these forms became increasingly more frequent, first gradually, then by the seventeenth century substantially more common. In part, this increase had to do with the stabilization of the use of personal a (Barry, 1985: 210–13). The impersonal construction for intransitive clauses represents the end of the development.4 (11.6)

Category REFLEXIVE/ RECIPROCAL

⇓ MIDDLE VOICE

⇓ ‘EMPHATIC’ SE

⇓

Example a. Juan se lava. Juan REFL washes ‘Juan washes himself.’ b. Juan se levanta por la mañana. Juan REFL gets-up in the morning ‘Juan gets up in the morning.’ c. Juan y María se venden muebles. Juan & María REFL sell furniture ‘Juan and María sell each other furniture.’ ⇓ d. Juan se despierta por la mañana Juan REFL wakes-up in the morning ‘Juan gets up in the morning.’ e. Luisa se cansa fácilmente. Luisa REFL tires easily ‘Luisa becomes tired easily.’ ⇓ f. Luisa se va mañana. Luisa REFL goes tomorrow ‘Luisa is leaving tomorrow.’ g. El abuelito se murió ayer. the grandpa REFL died yesterday ‘The grandpa died yesterday.’ ⇓

J. Clancy Clements 239

PASSIVE VOICE/ IMPERSONAL PASSIVE (with transitive verbs) ⇓ IMPERSONAL (with intransitive verbs)

h. Se venden muebles. ‘REFL’ sell-3pl furniture-3pl ‘One sells furniture.’ ‘Furniture are sold.’ i. Se vende muebles.5 ‘REFL’ sells furniture-3pl ‘One sells furniture.’ ‘Furniture is sold.’ ⇓ j. Se trabaja mucho aquí. ‘REFL’ works a lot here ‘People work a lot here.’

From a cognitive perspective, the reflexive and the middle voice overlap somewhat, as is apparent in the examples (11.6b) and (11.6d)–(11.6e). Recall that here middle voice is considered a conceptual domain that includes utterances like Juan se acuesta [lit. Juan lies (himself) down] ‘Juan lies down’where agentivity is present, as well as those such as Juan se durmió ‘Juan fell asleep’, in which the subject is an experiencer and se is an aspectual marker (see section Increasing Transitivity late in the chapter).6 These two sentences are comparable to precisely those in (11.6b) and (11.6d)–(11.6e) respectively: Juan in (11.6b) is an agent, whereas Juan in (11.6d) and Luisa in (11.6e) are experiencers. Depending on the context, there are cases in which a given sentence containing se has several possible readings. Givón (1990: 604) cites the example reproduced in (11.7), for which he gives the possible interpretations in (11.8): (11.7)

Se

curaron los brujos. cured-3pl the sorcerers a. The sorcerers cured themselves. (reflexive or reciprocal interpretation) b. The sorcerers got well. (middle voice interpretation) ( Givón’s ‘middle/intransitive’) c. The sorcerers were cured. (passive interpretation)

REFL

(11.8)

A context for (11.8a) would be, for example, a group of ailing sorcerers, in which each sorcerer would cure him/herself (reflexive reading). But (11.8a) could also correspond to a situation in which each of two groups of sorcerers were suffering from a different ailment, and one group provided a remedy for the other group, and vice versa (a reciprocal reading). In both cases, the sorcerers are the agents of the change as well as the patients, either reflexively or reciprocally. The reading in (11.8b) corresponds to a group of sorcerers that recovered from the ailment over a period of time, without necessarily doing anything proactively. They are the experiencers of healing. Finally, the reading for (11.8c) is one in which the sorcerers are the patients,

240 Transitivity and Spanish Non-Anaphoric se

being cured by an outside person.7 One could make a similar case for the reflexive/reciprocal or, marginally, the passive reading for the sentence (11.6h). Thus, the discourse situation often determines the exact interpretation of such sentences as those in (11.6h) and (11.7). In other cases, the interpretation of se is more predictable. As we will see below, an impersonal or an aspectual interpretation of a sentence containing dormir and se depends on factors having to do with Transitivity, a characterization of which is given in the next section. In this characterization, the notion of Transitivity is then extended to include, as well, typically impersonal constructions of intransitive verbs.

Transitivity Traditionally, transitive clauses are those whose verbs take a DO. Cast more broadly, however, it can be said that Transitivity involves the carrying-over of an action (or the transferring of energy) from one participant to another (Hopper and Thompson, 1980). Hopper and Thompson argue convincingly that Transitivity ‘can be broken down into its component parts, each focusing on a different facet of the carrying-over in a different part of the clause’(1980: 253). By breaking Transitivity down into its component parts, it is possible to characterize clauses, not as transitive or intransitive, but rather as more or less transitive. On this view, Transitivity becomes a question of degree. The factors making up the notion of Transitivity, adapted from Hopper and Thompson (1980), are given in (11.9–11.10): (11.9)

Component parts of Transitivity High a. PARTICIPANTS 2 or more participants, A(gent) and O(bject) b. KINESIS action c. ASPECT telic d. PUNCTUALITY punctual e. VOLITIONALITY volitional f. AFFIRMATION affirmative g. MODE realis h. AGENCY A high in potency i. INDIVIDUATION OF S(UBJECT) S highly individuated j. AFFECTEDNESS OF O O totally affected k. INDIVIDUATION O highly OF O individuated

Low 1 participant

non-action atelic non-punctual non-volitional negative irrealis A low in potency S nonindividuated O not affected O nonindividuated

J. Clancy Clements 241

(11.10) Component parts of Individuation (11.9i and 11.9k) Individuated Non-Individuated proper common human, animate inanimate concrete abstract singular plural count mass referential, definite non-referential, indefinite Note that item (11.9i), highlighted, is an addition to Hopper and Thompson’s original proposal, an extension originally proposed in a similar framework by Verhaar (1990: 94–7), who distinguishes between what he calls intransitive verbs of actorhood, high on the Transitivity scale such as the equivalents of ‘swim’, ‘talk’ and ‘run’, and intransitive verbs of undergoing, low on the Transitivity scale such as equivalents of ‘fall’, ‘melt’ and ‘faint’. By adopting this extension, it is possible to include intransitive clauses on the Transitivity scale as well. Thus, with the help of the values ‘high/low’in (11.9) and ‘individuated/non-individuated’in (11.10), one can measure the degree of Transitivity of any given clause. Within this definition of Transitivity, Spanish se functions in two distinct ways; it can have the function of reducing the Transitivity of a clause either by decreasing the valency of the verb by one argument, from a valency of two to a valency of one, or by ‘impersonalizing’the verb of an intransitive clause such that the presence of se disallows nominal or pronominal realization of the subject. This results in the subject argument of the intransitive clause receiving an indefinite and non-referential interpretation. However, the impersonal construction in transitive clauses (that is, se vende casas) does not receive such a reading. Rather, the impersonalization effected by se here is tantamount to a se passive construction, the sole difference being that the DO does not become the grammatical subject. Revealingly, both of these types appear at the same time in their historical development. The valency-reduction function of se has been the subject of many formalist analyses. In a Government-and-Binding or Principles-and-Parameters framework, the decrease in Transitivity is accounted for by attributing se the function of a case absorber, neutralizer or suspender which deprives the verb of assigning inherent case (see Belletti, 1982: 5; Otero, 1986: 87–95; Cinque, 1988: 575; Suñer, 1990; and Mendikoetxea, 1994).8 This will be addressed in more detail below. Se can also have the function of increasing the Transitivity of the clause. In this function, the claims in (11.11), which follow from Hopper and Thompson’s Transitivity Hypothesis, will obtain for the presence or absence of se: (11.11) Claims: a. Presence of se co-varies with relatively higher Transitivity. b. Absence of se co-varies with relatively lower Transitivity

242 Transitivity and Spanish Non-Anaphoric se

These functions of se will be taken up in the following sections.

Decreasing Transitivity9 Se as a valency-reducer: middle and passive se Above, in (11.6), we showed just how the functions of Spanish se are historically related. The proximity between the reflexive and the middle interpretation is illustrated by Givón’s examples in (11.7–11.8a, b). Such cases are not uncommon. Consider the examples in (11.12) containing the verb mojarse ‘drench oneself, get wet’; (11.12a) is clearly reflexive, indicated by the adverbial con entusiasmo ‘with enthusiasm’, while sentence (11.12b), however, cannot be reflexive and must have something like a patient semantic role since the subject el cuadro ‘the painting’ is inanimate. Sentence (11.12c) has an animate, experiencer subject, which has a middle interpretation: (11.12)

a. … Roque Lacentella se mojó con entusiasmo, dio una leve y jocosa morisqueta. … ‘Eso es lo que se llama un baño, ¿eh?’ ‘… Roque Lacentella drenched himself with enthusiasm, made a light and humorous face … “That’s what’s called a bath, uh?” ’ (from CREA: Ayerra, 1984: 307)10 b. … y tiene un cuadro que fue de su abuela y viene una un temporal y se mojó el cuadro y se puso feo, pues lo bota. ‘ … and she has a painting that was her grandmother’s and a, a storm comes and the painting got wet and it got ugly, then she throws it out.’ (from CREA, oral, woman, Puerto Rico, 1990) c. Como no llevaba paraguas, Juan se mojó en el chubasco tormentoso. ‘As he was carrying an umbrella, Juan got wet in the stormy downpour.’

Such ambiguity in interpretation is found in greater or lesser measure in many other verbs, such as acercarse ‘pull oneself toward, approach’, alejarse ‘move oneself away from, get away from’, and moverse ‘move oneself, move’. In all these cases, the prerequisites for a felicitous reflexive interpretation, stated in (11.3)–(11.5), may or may no be met, depending on the animacy and agency of the subject, as well as on the context. Other verbs with se, such as enamorarse ‘fall in love’in (11.13), and similar verbs as aburrirse ‘become bored’, ‘become tired’, confundirse ‘become confused’, despertarse ‘wake up’, divertirse ‘have fun’, enojarse ‘get angry’, espantarse ‘become scared’, hartarse ‘become fed up’, ofenderse ‘become offended’, preocuparse ‘worry, become worried’, allow only a MIDDLE interpretation.11

J. Clancy Clements 243

(11.13)

Luisa se enamora fácilmente. Luisa REFL falls-in-love easily ‘Luisa falls in love easily.’

Although structurally se displays reflexive behaviour by having its antecedent in its local domain (see 11.3 above), se cannot be said to be the direct object or the patient of the verb enamorar, as would be necessary according to (11.4); rather, the addition of se to enamorar makes it an intransitive verb. In some formalist accounts, se is said to absorb accusative case and the corresponding patient semantic role. Other formalist accounts of such verbs (for example Masullo, 1992) consider them antipassives. Both analyses of enamorarse-type verbs disregard the complex issue of thematic roles. As a transitive verb, enamorar ‘kindle love in the person’ (Moliner, 1984, our translation) is a causative, transitive verb with an agentive subject and a patient DO. By contrast, enamorarse ‘begin to feel love for a person of the opposite sex’ (ibid., our translation) is an intransitive verb with an experiencer rather than an agentive subject. No absorption mechanism in the formalist literature deals adequately with the change of semantic roles between enamorar and enamorarse and verbs like them. Regarding case assignment, however, se does reduce the valency by one. However, the condition in (11.5) is not met: in the manner in which we understand our world to generally function, the subject participant Luisa in (11.12) logically does not have the ability to make herself fall in love. Thus, the se in enamorarse is non-anaphoric. According to (11.9a), a verb with two participants is higher in Transitivity than one with only one participant. This is precisely the affect of se here: it lowers the Transitivity of enamorar by reducing its valency from two arguments to one. In addition, it converts a verb with an agentive subject into one with an experiencer subject. In terms of agency (11.9h), enamorar is high in potency while enamorarse is low in potency. In the PASSIVE construction, the function of se is also that of a valencereducing mechanism which lowers Transitivity (cf. 11.9a). In addition, se also affects the semantic roles by leaving the patient (or ‘theme’ in formalist accounts) as the remaining semantic role. Consequently, the subject of a se passive construction is non-volitional (11.9e), and non-agentive (11.9h), thus reducing the Transitivity of the clause even further. This lowering of Transitivity is captured in the formalist literature by appealing again to the notion of passive se being a case/semantic-role ‘absorber’ (Belletti, 1982; Jaeggli, 1986; Cinque, 1988; Suñer, 1990; Mendikoetxea, 1994). Thus, in (11.14b) se is said to absorb accusative case and the agentive thematic role. The inanimate object (los) muebles, then, receives the nominative case assignment intended for the NP subject slot.12 The end result is, again, that se reduces the valency of the verb by one.

244 Transitivity and Spanish Non-Anaphoric se

(11.14)

a. Juan vende (los) muebles. ‘Juan sells (the) furniture.’ b. Se venden (los) muebles. SE sell-3PL (the) furniture-3PL ‘(The) furniture is sold.’

Note that in (11.14b), the object-turned-subject is inanimate. If the DO is animate, but indefinite and non-referential (bare plural), valency reduction also takes place (11.15a), in which se also renders the verb intransitive and does away with the agentive semantic role. With an animate DO which is referential and/or definite, as in (11.15b) – one of the so-called IMPERSONAL constructions – valency reduction does not take place; it is nevertheless lower in Transitivity in that the understood subject is lower in individuation properties (cf. 11.10): it is non-referential and indefinite (arbitrary or expletive in the formalist perspective [see Cinque, 1988, and Suñer, 1990]). (11.15)

a. Se invitaron a cenar participantes del congreso. SE invited-3p to dine participants of-the conference ‘Participants of the conference were invited to dinner.’ b. Se invitó a cenar {a los participantes del congreso/a Marta y a Pepe} REFL invited-3s to dine {to l. p. of the conf. / to M. and to P.} {The participants of the conference /Marta and Pepe} were invited to dinner.’

Although the examples in (11.15) are considered the default constructions in Spanish, there is some variation. A search of the CREA corpus shows that aside from the expected cases, which are comparable to (11.15a) and (11.15b) and shown in (11.16a.c) and (11.16b) respectively. (11.16)

a. En los inicios de la Reforma Agraria se advirtieron en el Caquetá condiciones especiales para impulsarla y ‘In the beginnings of the Agrarian Reform special conditions were indicated in Caquetá to stimulate it and b. se invitó a los colonos para que sembraran caucho y palma africana, the tenant farmers were invited in to plant cork trees and African palms, c. se les entregaron tierras y créditos especiales para ganados y especies menores, pero el gobierno no les hizo adecuado seguimiento, abandonó los programas y casi todo se fue a la ruina.

J. Clancy Clements 245

they were given land and special credits for livestock and secondary species, but the government did not carry out an adequate follow up strategy, it abandoned the programs and almost everything went to ruin.’ (From CREA: El Tiempo, 13 September 1996, Colombia) we also encounter examples in which the ‘se  verb-3pl’ construction co-occurs with referential subjects without personal a (in 11.17), as well as with referential subjects with personal a (in 11.18 and 11.19): (11.17)

(11.18)

(11.19)

El cardenal nicaraguense, Miguel Obando y Bravo, recomendó ayer que de realizarse el debate entre los principales aspirantes a la presidencia de Nicaragua, el liberal Arnoldo Alemán y el sandinista Daniel Ortega. . . . Alemán y Ortega, se invitaron el martes a un debate televisivo. ‘The Nicaraguan cardinal, Miguel Obando y Bravo, recommended yesterday that upon staging a debate between the main presidential candidates the liberal Arnoldo Alemán and the Sandinista Daniel Ortega. . . . Alemán and Ortega were invited Tuesday to a televised debate.’ (From CREA: La Nación, Nicaragua, 12 September 1996) Es obvio que a usted se le ha dado la información equivocada’, indica Sol Bang a los diputados. A la licitación, según lo refiere CEL, se invitaron a las siguientes firmas: BZM/Barclays Bank/Ernst & Young/Ahorromet/Nera, Citibank N.A. . . . ‘“It is obvious that you were given erroneous information,” indicates Sol Bang to the members of parliament. To the tender, according to CEL, the following enterprises were invited: BZM/ Barclays Bank/Ernst & Young/Ahorromet/Nera, Citibank N.A . . .’ (From CREA: El Salvador Hoy, El Salvador, 4 March 1997) Se hará una presentación de los peloteros que participarán en esta serie final, se invitaron a las glorias del béisbol. ‘A presentation will be made about the players who will participate in the final series, the stars of baseball were invited.’ (From CREA: La Prensa, Nicaragua, 29 November 1996)

This variation is found in Latin American Spanish varieties. We return to this question below in our discussion of impersonal constructions. Before that, however, we will address the issue of the so-called unaccusative and antipassive constructions. As already alluded to, even though they are structurally identical, the se middle construction (11.12b–11.13) and the se passive construction (11.14b–11.15a) differ in their respective thematic roles. In the middle

246 Transitivity and Spanish Non-Anaphoric se

construction, the subject has either an experiencer semantic role (animate subject) or a patient role (inanimate subjects). In the passive se construction, by contrast, the subject can only have a patient semantic role. In terms of both structural and semantic role properties, the passive se and the UNACCUSATIVE se constructions in Spanish are identical. An example of an unaccusative construction is given in (11.20b): (11.20)

a. Juan rompió la ventana/rama. ‘John broke the window/branch.’ b. Se rompió la ventana/rama REFL broke the window/branch ‘The window/branch broke.’

The term unaccusative is used here to refer to telic, intransitive verbs.13 If we extend the definition to include derived intransitive constructions as well, such as the passive se and unaccusative constructions, the identicalness of these in Spanish can be accounted for in a straightforward manner, given that both constructions exhibit only one argument slot, the other having been eliminated by se. Note also that with both the passive and the unaccusative se, the sole argument receives the patient role. In terms of Transitivity, what was applicable for passive se constructions above applies here as well: Transitivity is lowered by the reduction of the number of participants from two to one (11.9a), and by the non-agentive (11.9h), and thus non-volitional (11.9e), nature of the subject. The ANTIPASSIVE construction, typically found in ergative-absolutive languages, contrasts with the passive construction in key aspects. Whereas the passive makes a subject and/or topic out of a non-subject argument and codes the original subject/topic with an adposition or non-nominative case marking (for example John was kissed by Mary), the antipassive construction retains the original subject and recodes the object (Givón, 1990: 624ff; 1995: 77ff). Masullo (1992) identifies the construction in (11.21b) as the Spanish antipassive: (11.21)

a. Juan confiesa sus pecados. ‘Juan confesses his sins.’ b. Juan se confiesa de sus pecados. Juan REFL confesses OBL his sins ‘Juan confesses his sins.’

Note that there is no crucial semantic distinction between (11.21a) and (11.21b). By demoting the DO to an oblique clause, and introducing se, (11.21b) puts the focus on the subject and the event. In both sentences, Juan is the agent of the confessing event. However, for some speakers there may

J. Clancy Clements 247

be a slight difference in degree of agentivity and/or intentionality between the sentences, (11.21a) possibly being slightly more agentive than (11.21b). The greater degree of agentiveness in (11.21a) co-varies with the higher Transitivity, and with the fact that (11.21a) has a DO, whereas (11.21b) does not have a comparable DO (cf. 11.9a). Masullo also includes other, fundamentally different verbs in the set of those that may undergo the antipassive construction, such as alegar(se) ‘be glad, gladden’, lamentar(se) ‘lament’, olvidar(se) ‘forget’, preocupar(se) ‘worry’, sorprender(se) ‘surprise’, temer(se) ‘fear’, and so on. Whereas lamentar(se) and olvidar(se) function roughly like confesar(se) in that there is no substantial semantic difference with or without se, this is not the case for remaining verbs just mentioned. For example, there is a substantial difference in interpretation between (11.22a) and (11.22b): (11.22)

suj vida. a. Juani alegra Juan gladdens his/her/their life ‘Juan makes his (*own)/her/their life happy.’ b. Juani se alegra de sui/j vida. Juan REFL gladdens OBL his life ‘Juan is happy about his/her/their life.’

In (11.22a), Juan can be the agent or the source of another’s life being happy and Juan and su are disjoint in reference. By contrast, in (11.22b) Juan is the experiencer of happiness and the source thereof is his own life or that of another or others. Given that, mutatis mutandis, the same also applies for the other verbs just mentioned, it becomes apparent that the group of verbs in question is heterogeneous in nature. They constitute no distinct class or type and only share the trait that they enter into what looks like an antipassive construction, which is unproductive in Spanish, whereas in ergative languages it is productive.14 Based on these considerations, it seems reasonable to include the verb type illustrated by alegrar(se) in (11.22) in the group of verbs that can appear in the middle construction. Recall that a typical middle construction is that in (11.6d), repeated as (11.23b) below: (11.23)

a. Juan despierta a su hermano. Juan wakes-up A his brother ‘Juan wakes up his brother.’ b. Juan se despierta por la mañana. Juan REFL wakes-up in the morning ‘Juan gets up in the morning.’

Compare (11.23a), where Juan is the agent of his brother awaking, to (11.23b), in which Juan is the experiencer of the event, analogous to the

248 Transitivity and Spanish Non-Anaphoric se

example in (11.22b), and to the case of enamorarse discussed in (11.13) above. Treating cases such as (11.21b–11.22b) as cases of middle se would have the advantage of doing away with the notion of an antipassive construction for Spanish, which by most accounts is a forced application. The observation on Transitivity regarding (11.21) above applies to the example sets in (11.22) and (11.23) as well: the former are transitive clauses, whereas the latter are intransitive. Impersonal se Traditionally, IMPERSONAL constructions are said to involve either transitive verbs, such as the example in (11.6i) and (11.15b) (repeated as 11.24a and 11.24b below), or intransitive verbs, as illustrated by the example in (11.25b): (11.24)

(11.25)

vende muebles.15 REFL sells furniture-3pl ‘One sells furniture.’ ‘Furniture are sold.’ b. Se invitó a cenar {a los participantes del congreso/a Marta y a Pepe} REFL invited-3s to dine {to l. p. of the conf. / to M. and to P.} ‘{The participants of the conference /M. and P.} were invited to dinner.’ a. Marcos trabaja en la fábrica. ‘Marcos works in the factory.’ b. Se trabaja mucho en la fábrica. REFL works much in the factory ‘One works a lot, i.e. a lot of work goes on, in the factory.’ a. Se

From the viewpoint of Transitivity, the presence of se in the examples in (11.24) lowers Transitivity because it renders the subject of the clause indefinite and non-referential (cf. 11.10 above), although the verb retains a DO. In (11.25), since trabajar is an intransitive verb, se cannot reduce its valency. However, it does affect the interpretation of the subject: se in (11.25b) is indefinite and non-referential, as in the cases of se in (11.24). In addition (11.25b) has an obligatory generic or habitual reading. Given that there is a difference in interpretation between the impersonal constructions with transitive (11.24) and intransitive (11.25b) verbs – the latter necessarily having a generic or habitual reading, but the former not necessarily so – it is reasonable to question whether constructions such as (11.24b) constitute an impersonal variant of the passive se construction. At issue is the fact that in cases such as (11.24), the DO does not become the subject, as in (11.6h) or (11.14b) above. If this could be otherwise accounted

J. Clancy Clements 249

for, we would be able to consider them a type of passive construction. There is a historical consideration. The passive and the impersonal construction of transitive verbs both developed as early as the ninth to eleventh centuries, whereas the impersonal construction of intransitive verbs, illustrated by (11.25b), emerged much later (Cennamo, 1993, in Maldonado, 1996). Moreover, cases such as (11.24) show interesting variation but only among transitive verbs. Above, we saw data gleaned from Latin American Spanish sources that suggest that the construction in (11.24b) with personal a and not verb–subject agreement has variants in which we find agreement between the verb and an overtly marked DO (for example se invitaron a las siguientes firmas ‘the following enterprises were invited’ [11.18], se invitaron a las glorias del béisbol ‘the baseball stars were invited’ [11.19]), just as we also find cases in which there is agreement between the verb and an unmarked, definite and referential NP, as in Alemán y Ortega, se invitaron el martes a un debate televisivo ‘Alemán and Ortega were invited Tuesday to a televised debate’ (11.17). This range of variation only happens among transitive verbs, which suggests that (11.24b) may more accurately be considered one of the impersonal variants of the passive se construction. Summary Various constructions involving the non-anaphoric reflexive pronoun have been examined (referred to in the literature as middle voice se, passive se, unaccusative se, antipassive se, and impersonal se). In all constructions except those such as (11.24) and (11.25b), se lowers the Transitivity of the clause in which it appears by reducing the valency of the verb by one argument. It was pointed out that there is in principle no reason, either semantic or syntactic, to treat unaccusative se and passive se as distinct. Nor is there any compelling reason to separate the so-called antipassive se from middle voice se, given the similarity of the evidence in (11.13) and (11.21b) on the one hand, and (11.22b) and (11.23b) on the other. Consequently, in this set of valency-reducing constructions, we need distinguish for Spanish only middle voice se and passive/unaccusative se. Finally, we have also argued that the passive construction in (11.15a) is linked historically with that illustrated in (11.24b). Due to this, and to the fact that the variation in this construction only involves transitive verbs, (11.24b) may more accurately be considered a case of passive se. The impersonal se from intransitive verbs (cf. 11.25b) is clearly distinct from the valency-reducing constructions. There is no reduction in valency of the verb given that this verb type has a valency of one. The distinguishing trait with the impersonal construction is the decrease in overall Transitivity of the clause. In terms of individuation of the subject in impersonal constructions (cf. 11.9i), the subject in a sentence such as se trabaja

250 Transitivity and Spanish Non-Anaphoric se

mucho aquí ‘a lot of work is done here’ is non-individuated, as illustrated in (11.26): (11.26)

Individuation of subject: S highly individuated proper animate concrete singular count referential, definite

S non-individuated common inanimate abstract plural mass non-referential, indefinite

Does not apply n.a.

n.a. n.a.

Given that the subject in impersonal constructions is abstract, non-referential and indefinite, and also that the construction has a generic or habitual interpretation, its Transitivity is lower relative to the same sentence without se, such as Juan trabaja mucho aquí ‘Juan works a lot here’. Regarding what component of the grammar would deal with passive, middle and impersonal se constructions, these are syntactic phenomena, although they have, naturally, an important semantic and pragmatic aspect to them as well (cf. 11.13 above and subsequent discussion, and 11.8). An exception to this would be those verbs, such as arrepentirse ‘repent’, which have se as part and parcel their lexical make up. Such items would be entered as such in the lexicon. In what we have seen so far, non-anaphoric se affects the Transitivity of clauses by lowering it in one of two ways: by reducing the valency of the verb or by limiting the interpretation of a subject to indefinite and non-referential with an accompanying generic or habitual reading of the process denoted by the verb. However, se affects Transitivity in the opposite way as well; that is, its presence co-varies with heightened Transitivity, which we now examine.

Increasing Transitivity In this section, we argue that the predictions in (11.11), repeated as (11.27) below, hold unambiguously for Spanish: (11.27)

Claims: a. Presence of se co-varies with relatively higher Transitivity b. Absence of se co-varies with relatively lower Transitivity

In what follows we will see se as it functions as a marker of verbal aspect and a highlighter of foregrounded material and a marker of definiteness of different types.

J. Clancy Clements 251

Se as an aspectual marker Vendler (1967) treats Aktionsart, or inherent verbal aspect, in terms of Achievements, Accomplishments and Activities. Clements (1985: 178–81) adds the category ‘states’, characterizing the four categories in terms of dynamic v. non-dynamic situations, duration and telicity, as shown in Table (11.1). To illustrate briefly what these terms refer to, states comprise, among many others, mental states such as creer ‘believe’, querer ‘want’, saber ‘know’, and so on, and resultant states such as estar despierto ‘be awake’, estar cansado ‘be tired’, and so on. One key criterion for distinguishing states from dynamic situations is that states do not typically appear in the progressive aspect. Thus, the sentences in (11.28) are ill-formed: (11.28)

a. *Luisita está sabiendo francés. ‘Luisita is knowing French.’ b. *Elena está {creyendo/queriendo} estar en casa. ‘Elena is {believing/wanting} to be at home.’ c. *Dolores está estando {despierta/cansada}. ‘Dolores is being {awake/tired}.’

Dynamic situations are distinguished by two criteria: duration and telicity. One can test dynamic situations by using tests for duration and telicity, taken from Clements (1985). Verbs and VPs that have an endpoint, such as llegar ‘arrive’ and construir una casa ‘build a house’, will be grammatical in a telic context, that is a context that demands an endpoint.16 In the telic context given in (11.29–11.30), Vs and VPs with an endpoint will be well-formed (11.29), while Vs/VPs without an endpoint will be ill-formed (11.30): (11.29)

a. Luis tardó tres meses en llegar. ‘It took Luis three months to arrive.’ b. Ignacio tardó tres meses en construir una casa. ‘It took Ignacio three months to build a house.’

Table 11.1 Characterization of Vendlerian Aktionsart categories

States Achievements Accomplishments Activities

Dynamic situation

Durative

Telic

   

   

   

252 Transitivity and Spanish Non-Anaphoric se

(11.30)

a. *Forest tardó tres meses en correr. ‘It took Forest three months to run.’ b. *Lola tardó tres horas en construir casas. ‘It took Lola three hours to build houses.’

By contrast, in a durative context, non-durative telic Vs/VPs will be illformed (11.31), while durative telic Vs/VPs will be well-formed (11.32): (11.31) (11.32)

*Luis pasó tres meses llegando. ‘Luis spent three months arriving.’ a. Ignacio pasó tres meses construyendo una casa. ‘Ignacio spent three months building a house.’ b. Forest pasó tres meses corriendo. ‘Forest spent three months running.’ c. Lola pasó tres meses construyendo casas. ‘Lola spent three months building houses.’

Verbs such as llegar are Achievements because they are telic and nondurative. VPs such as construir una casa are Accomplishments because they are telic but durative. Finally, Vs/VPs which are durative but atelic are Activities, such as correr and construir casas. Note that the sole difference between the Accomplishment constuir una casa and the Activity constuir casas lies in that the former has a count NP (which constitutes an endpoint), whereas the latter has a bare plural NP (non-count and cannot constitute an endpoint). Having reviewed the Vendlerian categories, we now turn to the link between se, Activities (durative and atelic) and Achievements (non-durative and telic, hence punctual). In the characterization of Transitivity in (11.9–11.10) aspect and punctuality are two of its component parts. Thus, the difference between dormir ‘sleep’and dormirse ‘fall asleep’ is easily captured: (11.33a), being an Activity, is atelic and non-punctual; (11.33b), an Achievement, is telic and punctual: (11.33)

a. Marta durmió en casa de su amiga. ‘Marta slept at her friend’s house.’ b. Marta se durmió en casa de su amiga. ‘Marta fell asleep at her friend’s house.’

In all other aspects of Transitivity, except in aspect and punctuality, dormir and dormirse are non-distinct. Thus, due to this crucial difference in the Aktionsart of the two verbs, dormirse is higher, and dormir lower in overall

J. Clancy Clements 253

Transitivity. This is illustrated in (11.34): (11.34) ASPECT PUNCTUALITY

high dormirse telic punctual

low dormir atelic non-punctual

In the case of dormirse, the question of ambiguity arises in that se in some cases could theoretically be interpreted either as impersonalized, or as an aspectual marker. As already discussed, the impersonal reading of se occurs in generic or habitual context, such as in (11.35). In this sentence, se could only be interpreted as an aspectual marker if the equivalent of ‘one’(uno) were added, as shown in (11.36): (11.35)

(11.36)

Aquí se duerme bien. here REFL sleeps well ‘One sleeps well here.’ Aquí uno se duerme bien. here one REFL sleeps well ‘One falls asleep well here.’

In (11.36), uno is the subject and se interpreted as an aspectual marker. However, with the appropriate adverbial and contextual support, an aspectual interpretation of se could still be possible without uno, as in the sentence (11.37) uttered in a context in which falling asleep is the topic of conversation, as long as there is an antecedent in the discourse: (11.37)

Aquí se duerme en seguida/rápidamente. right away/quickly here REFL sleeps ‘One falls asleep right away/quickly here.’

In sum, the more definite the context is, with adverbials and contextual support, the less likely it is to obtain an impersonal interpretation of se with dormir. In other words, the impersonal and aspectual functions of se appear to occur in mutually exclusive contexts and there is little or no scope of ambiguity in the interpretation of se. Motion verbs such as ir ‘go’ and marchar ‘march’ have counterparts with se that have over time acquired aspectually distinct meanings: irse ‘go, leave’ and marcharse ‘leave’ (see De Molina Redondo, 1974: 48–49, 53). Both ir ‘go’ and marchar ‘march’ are Activities, as they are durative, and yield ill-formed sentences in a telic context (cf. 11.38). (11.38)

*Luisa tardó una hora en ir/marchar. ‘Luisa took an hour to go/march.’

254 Transitivity and Spanish Non-Anaphoric se

By contrast, irse and marcharse are Achievements and therefore yield wellformed sentences in telic context. (11.39)

Luisa tardó una hora en irse/marcharse. ‘Luisa took an hour to leave.’

Thus, the verbs in (11.39) are higher in Transitivity than their counterparts in (11.38) in the crucial areas of aspect and punctuality, just as in the case of dormir-dormirse above in (11.34). (11.40)

ASPECT PUNCTUALITY

high irse marcharse telic punctual

low ir marchar atelic non-punctual

As De Molina Redondo (1974: 47–56) illustrates with key examples, the same aspectual differences exist between the motion verb pairs venir-venirse ‘come’, salir-salirse ‘leave, go out’, entrar-entrarse ‘enter’, caer-caerse ‘fall’, volver-volverse ‘return’, and escapar-escaparse ‘escape’. In his view, se with verbs of motion functions either as a marker of an implied source (origin) or as a marker emphasizing the source. In both cases, se adds the element of telicity. For instance, in declaring ya me voy (lit. already REFL go-1sg/pres-ind) ‘I leave/I’m leaving’, the source de aquí ‘from here’ is implied, though not overtly expressed. Without me in the sentence, the implication is not present. With many of these motion verbs, the presence of se also co-varies with elements higher in Transitivity. A case in point is the pair caer ‘fall’ and caerse ‘fall’. Moliner (1984) discusses the difference between caer and caerse, essentially restating De Molina Redondo’s position. Especially revealing, however, are her examples, in which the use of caer occurs in indefinite, non-referential (often generic) contexts (11.41), whereas caerse is used in definite, referential contexts (11.42):17 18 (11.41)

a. Caen las hojas en otoño. (GENERIC, NON-REF) ‘Leaves fall in the fall.’ b. La lluvia cae de las nubes. (GENERIC, NON-REF) ‘The rain falls from the clouds.’ c. Los cuerpos caen con movimiento acelerado. (GENERIC, NON-REF) ‘Bodies fall with accelerated movement.’

J. Clancy Clements 255

(11.42)

d. Ha caído un aerolito al mar. ([NON]-REF, INDEF) ‘A meteorite has fallen into the sea.’ a. Se están cayendo todas las hojas del rosal. (DEF, REF) ‘All the leaves of the rose bush are falling off.’ b. Se han caído todas las naranjas del árbol. (DEF, REF) ‘All the oranges on the tree fell off.’

A related function of se to that of marking the origin of movement of a motion verb, is that of a highlighter of foregrounded material. In a narrative, the foreground is that portion of the text that carries the story, that is, it is ‘the material which supplies the main points of the discourse’ (Hopper and Thompson, 1980: 280). We use examples taken from two older texts because they offer a particularly striking illustration of the point at hand. However, the function of se as a marker of foregrounding is common place in Spanish today. The first set of examples (11.43–11.45) is taken from a legal document written in 1578 (García Carrillo, 1988: 66,119,128), in which the speaker is highlighting, with the use of se, a foregrounded portion of the text, that is the escape of the prisoner and the fact that he is heading toward the church: (11.43)

(11.44)

oyó dar bozes hazia la Cárçel Real de corte a vn hombre, que dezían que era alcaide, diziendo: ‘que SE huye el preso! … ’ (p. 128, ls. 854–6) ‘he heard the yells directed to the Royal Jail … by a man who they said was the mayor, and he was yelling, “the prisoner SE is getting away!” ’ Martín Vélez, que allí estaua, le dixo: ‘Guillén es, que SE a salido de la cárçel y SE viene a la yglesia … ’ (p. 119, ls. 330–1) ‘and Martín Vélez, who was there, said to him: “It’s Guillén who just SE got out of jail and SE is coming to the church …” ’

The second set of examples is taken from the confessions of a nun who lived and wrote in the early eighteenth century. The text describes a mystical encounter with Christ (Myers and Powell, 1999): (11.45)

estando un dia en orasion vide a nr sr crusificado cerca de mi y vide como se desclabo bajandoSE de la [crus] y SE vino para mi la crus se quedo parada y su mgd llego donde io estaba y SE entro dentro de mi y abraso a mi [alma] … (volume II, Folios 175 (verso) - 176) ‘One day during prayer I saw our Lord, crucified, near me and I watched him as he took out the nails, lowered himself off the

256 Transitivity and Spanish Non-Anaphoric se

cross and SE came to me. The cross stayed upright and He approached where I was and SE entered within me and embraced my soul.’ In the examples se viene a la yglesia (11.44) and in se vino para mi and se entro dentro de mi (11.45), it is not readily as apparent that an origin is necessarily being implied, as it is apparent that these movements form part of the foreground, and are being highlighted by the use of se. Hopper and Thompson’s (1980) study on Transitivity makes clear that the main discourse function of higher Transitivity is that of foregrounding, with which se here co-varies. Thus, these excerpts serve as prime examples of this discourse function of se. Transitivizing se co-varies with definiteness and/or referentiality With optionally transitive verbs of ingestion, such as comer ‘eat’ and beber ‘drink’, the addition of se makes them not only obligatorily transitive, but also sensitive to the definiteness of the object as well. Taking comer as an illustration, with bare plurals and mass nouns, comerse is odd to unacceptable (cf. 11.46b) (see also Zagona, 1996): (11.46)

a. Emilia comió manzanas/arroz. ‘Emilia ate apples/rice.’ b. ?*Emilia se comió manzanas/arroz. ‘Emilia REFL ate apples/rice.’

This fact is linked to the question of Aktionsart, not of the verb alone, but of the entire VP in this case. The VP comer manzanas is an Activity, that is durative and atelic. We know it is atelic because it is ill-formed in contexts that demand an endpoint to be interpretable, such as (11.47). The English gloss of this is also ill-formed for the same reason (see 11.30b above and accompanying discussion): (11.47)

*Juan tardó tres horas en comer manzanas. ‘Juan took three hours to eat apples.’

Given that a bare plural in the VP renders the NP non-countable, the NP provides no endpoint for the VP. If a count NP is substituted for the bare plural, it provides an endpoint and also forces the presence of se in most cases.19 That is, with count nouns, such as la manzana ‘the apple’, a single sentence is wellformed only as long as se is present. Without se, it is generally odd to ill-formed: (11.48)

a. Emilia se comió la manzana. Emilia REFL ate the apple ‘Emilia ate the apple.’ b. ?*Emilia comió la manzana.

J. Clancy Clements 257

The foregoing generates the following prediction: any partitive construction with comer would appear obligatorily without se. This turns out to be true, as the sentences in (11.49) illustrate: (11.49)

a. Emilia comió de la manzana. Emilia ate of the apple ‘Emilia ate some of the apple.’ b. *Emilia se comió de la manzana.

In terms of Transitivity, various components, shown in (11.50), are involved: aspect and affectedness/individuation of the object. Note that the presence of se always co-varies with higher Transitivity, while its absence co-varies systematically with lower Transitivity: (11.50)

high comerse ASPECT telic AFFECTEDNESS O O totally affected INDIVIDUATION O individuated count referential, definite

low comer atelic (optionally telic) O not totally affected non-individuated mass (and bare plural) non-referential, indefinite

In the case of se with the parecer, its absence covaries with indefinite, nonreferential predicate NPs, as illustrated in (11.51): (11.51)

Roberto parece un boxeador. ‘Roberto looks like a boxer.’

With se, the corresponding predicate NP is referential (either definite or indefinite), and it must be accompanied by a personal a, as shown in (11.52): (11.52)

a. Roberto se parece a uno de esos boxeadores en la tele. (INDEF, REF) Roberto REFL looks-like A one of those boxers in the television. ‘Roberto looks like one of those boxers on television.’ b. Roberto se parece a ese boxeador/Mohammed Ali. (DEF, REF) Roberto REFL looks-like A that boxer/Mohammed Ali ‘Roberto looks like that boxer/Mohammed Ali.’

Note that with a referential predicate NP, the absence of se results in an illformed sentence, and with a non-referential predicate NP, the presence of se

258 Transitivity and Spanish Non-Anaphoric se

yields an ill-formed sentence as well: (11.53)

a. *Roberto parece uno de esos boxeadores en la tele. (INDEF, REF) b. *Roberto parece ese boxeador/Mohammed Ali. (DEF, REF) c. *Roberto se parece un boxeador. (INDEF, NON-REF)

This set of examples involves only the individuation of the predicate NP. Parecerse appears with referential NPs, parecer with non-referential NPs: (11.54)

Individuated parecerse referential

Non-individuated parecer non-referential

Summary Whereas in the previous section se affects for the most part the valency of the verb and is a syntactic phenomenon, in the examples in this section se either alters the Aktionsart of the verb or co-varies with definite NPs. Given that in these functions se is found rather idiosyncratically and that its effect is within the area of verbal semantics, it is uncontroversially considered a lexical phenomenon. Consequently, there would be different lexical entries for verb pairs such as comer–comerse, dormir–dormirse, ir–irse, parecer–parecerse, and so on.

Conclusion In this study, we have attempted to show that Spanish non-anaphoric se has one broad function: it affects the Transitivity of the clause in which it appears. It does this in two general ways. First, se reduces the valency of a given verb by one participant. This function accounts for middle, passive, unaccusative and antipassive se. We have argued that, semantically (that is in terms of semantic roles) and syntactically, there is no difference between unaccusative, passive and impersonal passive se in Spanish. There is, therefore, no need to treat them separately. In the same vein, given the evidence in (11.13), (11.22b), (11.23b) and the corresponding discussion, there is no reason to treat middle and antipassive se as separate either. The valencyreducing se can also lower the Transitivity of an already intransitive clause by making the clause impersonal, with a habitual or generic reading (see 11.25b). In terms of Transitivity in (11.9–11.10), any impersonal se construction is lower in Transitivity than its non-impersonalized counterpart (11.25a). The other, second, general function of se is that of increasing or heightening the Transitivity of a clause. This function accounts for aspectual differences between verb pairs such as dormir–dormirse ‘sleep–fall asleep’ and

J. Clancy Clements 259

marchar–marcharse ‘march–leave’, for the presence of se with count NPs as in Juan se comió una/la manzana ‘Juan ate up a/the apple’, the absence thereof with bare plurals as in *Juan se comió manzanas, ‘Juan ate apples’, and a similar type of phenomenon in the pairs caer–caerse ‘fall’, parecer–parecerse ‘look like’, and quedar–quedarse ‘remain’. We also saw that se co-varies with foregrounded motion verbs, which is not surprising given that one always finds increased Transitivity in the foregrounded events of a narrative (see Hopper and Thompson, 1980: 285). As to where these two different functions of se are treated in a grammar, valency-reducing se is a syntactic phenomenon, but involves, of course, semantic and pragmatic issues as well. Se as increaser of Transitivity is treated as a lexical phenomenon, best handled in the lexicon. Finally, the presence of se with foregrounded motion verbs would be accounted for in the pragmatic component of the grammar, with a clear link between this component and the lexicon. In fact, the general function of non-anaphoric se as a marker of Transitivity may be seen as evidence that the Transitivity links at least the lexicon, syntax, semantics and pragmatics, and probably morphology as well. Notes 1 Traditionally, the middle voice is defined as it was used in Ancient Greek, where its sense was broadly reflexive. For English, this would include transitive sentences such as John bought himself a house, as well as intransitive ones such as John shaved. In both sentences, there is a reflexive meaning involved (see Matthews, 1997: 226). For Spanish, Maldonado (1999: 392) considers the domain of the middle voice to refer to similar types of sentences as those just mentioned, all of which involve se: Juan se lava las manos ‘Juan washes his (own) hands’, Juan se afeita ‘Juan shaves’, Juan se acuesta ‘Juan lies (himself) down’, and Juan se durmió ‘Juan fell asleep’. In this study, the notion ‘middle voice’ is used to refer to this range of senses. 2 When used in Hopper and Thompson’s sense, we capitalize the term, as in Transitivity. 3 Comrie (1984: 59) adopts this same preparatory condition for directive speech acts. 4 Reflexive and middle interpretations of se are common from Latin. Moreover, ‘emphatic se’ is found in Peregrinatio Aeteriae: ‘En “Eteria” aparece vadent se [‘se van’], intransitivo con forma reflexiva, indicando la voluntariedad en la acción’ (Terracini, 1945, in Hernández, 1966: 49). As mentioned, the passive se and the se in transitive-clause impersonal constructions developed together, while the intransitive-clause impersonal construction developed later. For Spanish, then, Givón’s (1990: 605) scale, based on Marín (1989a, 1989b), of Reflexive ➔ Middle/ Intransitive ➔ se Passive with subject agreement ➔ se Passive without subject agreement, would for Spanish need to be revised by collapsing the last two phases of the development, and adding the development of the intransitive, impersonal construction. 5 The Castilian Spanish informants consulted found se vende muebles ill-formed. That is, if the DO is inanimate, in Castilian Spanish there must be subject–verb agreement as in se venden muebles. However, for most of the speakers of Latin American

260 Transitivity and Spanish Non-Anaphoric se

6

7

8

9

10

11

Spanish dialects consulted, (11.6i) was acceptable, despite the lack of subject–verb agreement. The semantic role ‘experiencer’ is understood here as corresponding to an animate entity who experiences some sensation. For example, in Juan tiene un dolor ‘Juan has a pain/an ache’, the subject is the experiencer and the DO the sensation experienced. In Juan disfruta sus vacaciones, ‘Juan enjoys his vacaciones’, Juan is the experiencer, while the DO sus vacaciones is the source. It is important to observe that native speakers we consulted ruled out (11.8c) as a possible interpretation for (11.7). For a passive interpretation, one needs the so-called impersonal construction such as se curó a los brujos. This will be taken up in the discussion of example (11.15) and below. Generally, se is considered to have the function of ‘absorbing’ case. Cinque (1988: 575), however, distinguishes a [argument] si (the Italian counterpart of Spanish se) and a [argument] si, each of which has different properties: the former absorbs the external theta-role and nominative case. The latter suspends the external theta-role and accusative case. In speaking of the valency-reducing function of se, we will use at times the terms ‘absorb’ and ‘absorption’ to refer to both [arg] and [arg] se alike. Apart from those just cited, others have also done work in this area, including García (1975), Arce Arenales (1992), Maldonado (1999) and Vann (1993), to mention just a few. Many of the examples in the study have been taken from Corpus de Referencia de Español Actual (CREA) made available by the Real Academia Española (www.rae.es). Although uncommon, some of these verbs can receive a reflexive, or pseudoreflexive, interpretation, as in: (i)

Juanita se aburre a propósito para no tener que quedarse en la fiesta. ‘Juan REFL gets bored purposely so she won’t have to stay at the party.’ (ii) Juanita se enoja a propósito para salirse con la suya. ‘Juanita REFL gets-angry purposely to get her own way.’ (iii) Juanita se cansa a propósito para dormir mejor. ‘Juanita REFL tires purposely to sleep better.’ The first two examples correspond to situations in which the subject feigns boredom or anger as a way of getting her way. As such, they are what may be called pseudo-reflexives. In fact, these and the other aforementioned verbs customarily receive a middle voice interpretation. Sentence (iii) can have a reflexive interpretation, though, again, it is not common practice that one gets tired purposely. (iii) is a case somewhat similar to that in (11.12), in that cansarse is ambiguous and can have a reflexive or a middle interpretation, depending on the context. Given that much more context is necessary in order to obtain a reflexive interpretation of cansarse than the middle reading, the latter is reasonably the default interpretation. 12 A sentence such as se vende los muebles is also possible. For an account of this, see examples in (11.15) and related discussion. 13 The term unaccusative (also ergative), referring to verbs, was coined by Perlmutter (1978) to refer to a subclass of intransitive verbs with certain syntactic properties. Haspelmath (1985: 157) points out that two properties of the verbs in this subclass are telicity and that their participle can be used adjectivally, for example a broken glass. (We follow here Haspelmath’s definition.) By contrast, the term

J. Clancy Clements 261 unergative refers to intransitive verbs that are atelic and whose participles cannot be used adjectivally, for example *a danced girl. See De Miguel (1992: 44–8; 63–96) for an in-depth discussion of the semantics of unaccusative verbs in Spanish. 14 Masullo (1992: 183) considers the construction idiosyncratic. Given that in ergative languages the antipassive is commonplace, productive, and typically used with indefinite, non-referential NPs, it is questionable whether we are dealing here with a bona fide antipassive construction. The verbs mentioned here could also be handled as cases with middle-voice interpretation. 15 In our search of CREA, we found clear examples of se vende  NP-3pl only in Latin American sources. We include some examples below. (i)

El minicentro comercial está compuesto por más de una docena de pequeñas tiendas donde se vende perfumes, carteras … ‘The mini-comercial center is made up of more than a dozen small stores where perfumes, purses … are sold.’ (From CREA: El Nuevo Herald, US, 3 February 1997) (ii) Indicaron que esto le sucede con varios productos que venden en Jamaica donde estos se vende, aunque no en la cantidad que desearían. ‘They indicated that this happens with various products that they sell in Jamaica where these are sold, although not in the quantity that they would like.’ 16 Here we are using the notions ‘telic’ and ‘endpoint’ in the broad sense, although in certain cases, one could speak more concretely in terms of a starting point. For instance, the stative verb conocer ‘know, be acquainted with’ is well-formed in telic contexts, as in (i): (i)

María tardó algún tiempo en conocer a su sobrino. ‘María took a long time to get to know her nephew.’

Strictly speaking, however, the telic aspect of (i) is the initial point of María knowing her nephew, although one could also argue that the point of knowing also is the culminating point of a process leading up to María knowing her nephew. The fact remains, nonetheless, that the notions ‘telic’ and ‘endpoint’ are not to be understood in this paper in a narrower, but rather in a wider sense. 17 Maldonado (1988) frames this difference in terms of expectations of the participants in the discourse context. In his analysis of Mexican Spanish, he notes that (i) is unnatural with se because the event is expected, whereas (ii) is unnatural without se because the event is unexpected. Thus, se, in Mexican Spanish at least, seems to code the degree of expectedness of the event occurring: (i)

La pelota (*se) cayó de la mesa como era esperado. ‘The ball fell off the table, as was expected.’ (ii) La pelota se (*(Ø) cayó de la mesa inesperadamente. ‘The ball unexpectedly fell off the table.’ The link between Maldonado’s analysis and the present one is this: generic statements, in which caer would be more likely used, are generally expected events. By contrast, falling events involving definite and referential grammatical subjects would more often be one-time occurrences, and more often than not unexpected. 18 Although it is not a movement verb, but rather a stationary verb, the presence of se in quedarse ‘stay, remain’ also co-varies with elements higher in

262 Transitivity and Spanish Non-Anaphoric se Transitivity: quedarse appears with animate, and mostly referential subjects (i–a,b), but can also appear with definite inanimate subjects, as in (i–c). (i) a. Se quedó un muchacho en la fiesta.(ANIMATE, INDEF, [NON]-REF) REFL stayed one boy at the party ‘One boy stayed at the party.’ b. Se queda aquí Juanita/la mujer del pelo largo. (REF, DEF, ANIMATE) REFL stays here Juanita/the woman with the long hair. ‘Juanita/the woman with the long hair stays/is staying here.’ c. El sofá se queda aquí. The sofa REFL stays here ‘The sofa stays here.’ By contrast, with quedar ‘remain, be left over’, one finds in the vast majority of cases – if not exclusively – inanimate and largely partitive subjects. In fact, definite, referential, non-partitive subjects are not possible with quedar, as is the case in the English equivalent: (ii)

a. Quedan las manzanas de ayer. (DEF, PARTITIVE) left-over-are the apples from yesterday ‘Yesterday’s apples are left over.’ b. Quedan manzanas. (INDEF, NON-REF) left-over-are apples ‘There are apples left.’ c. *Quedan las manzanas. (GENERIC) left-over-are the apples *‘There are the apples left.’

Thus, the presence of se co-varies with definiteness and referentiality, its absence co-varies with indefiniteness and non-referentiality. Consequently, quedar is lower, and quedarse is higher in Transitivity, confirming predictions in (11.11) regarding the presence-absence of se. See Bull (1940) for a detailed study of quedar–quedarse. 19 The presence of se with count VPs is not an automatic process, but can be overridden by a context, in which the focus is mainly on the event denoted by the verb instead of on the whole VP. In the examples in (i), (i) Emilia (se) comió la bocata y (se) bebió el vino. ‘Emilia (REFL) ate the sandwich and (REFL) drank the wine.’ the focus is the sequence of events more so than what was eaten or drunk. In such cases, se is optional and adds an aspectual quality of completion. However, devoid of any particular context, a single sentence with comer ‘eat’ or beber ‘drink’ accompanied by a count noun will trigger the presence of se.

References Abraham, W. (1995) ‘The Middle, Particularly in West-Germanic: What does Reflexification have to do with Valency Reduction’, in W. Abraham, T. Givón and S. Thompson (eds), Discourse Grammar and Typology. Amsterdam and Philadelphia: Benjamins, pp. 3–47. Arce Arenales, M. A. (1989) Semantic Structure and Syntactic Function: The Case of Spanish ‘se’. Boulder, Colorado: University of Colorado dissertation.

J. Clancy Clements 263 Ayerra, R. (1984) La lucha unútil. Madrid: Debate. Barry, A. (1985) ‘The Rise of the Impersonal-SE Construction’, Hispanic Journal, vol. 6, pp. 209–19. Bull, W. E. (1953) ‘Quedar and Quedarse: A Study of Contrastive Ranges’, Language, vol. 26, pp. 467–80. Cennamo, M. (1993) The Reanalysis of Reflexives: A Diachronic Perspective. Napoli: Liguoi Editore Napoli. Clements, J. A. (1985) Verb Classification and Verb Class Change in Spanish. Seattle, Washington: University of Washington dissertation. Comrie, B. (1984) ‘Reflections on Subject and Object Control’, Journal of Semantics, vol. 4, pp. 47–65. De Miguel, E. (1992) El aspecto verbal en una gramática generativa del español. Madrid: Universidad Autónoma de Madrid. De Molina Redondo, J.A. (1974) Usos de se : Cuestiones sintácticas y léxicas. Madrid: Sociedad General Española de Libreria, S.A. Fagan, S. (1992) The Syntax and Semantics of Middle Constructions: A Study with Special Reference to German. Cambridge: Cambridge University Press. García, E. (1975) The Role of Theory in Linguistics Analysis: The Spanish Pronoun System. Amsterdam: North-Holland Publishing. García Carrillo, A. (1988) El español en México en el siglo XVI: estudio lingüístico de un documento judicial de la Audiencia de Guadalajara (Nueva España) del año 1578. Sevilla: Ediciones Alfar. Givón, T. (1984,1990) Syntax: A Functional-Typological Introduction. Vols. 1 and 2. Amsterdam and Philadelphia: Benjamins. –––– (1995). Functionalism and Grammar. Amsterdam and Philadelphia: Benjamins. Haegemann, L. (1994) Introduction to Government and Binding Theory, 2nd edn. Oxford, UK: Blackwell. Haspelmath, M. (1994) ‘Passive Participles across Languages’, in B. Fox and P. Hopper (eds), Voice: Form and function. Amsterdam and Philadelphia: Benjamins, pp. 351–77. Hernández, C. (1966) ‘Del se reflexivo al impersonal’, Archivum Oviedo, vol. 16, pp. 39–66. Hopper, P. J. and S. A. Thompson. (1980) ‘Transitivity in Grammar and Discourse’. Language, vol. 56(2), pp. 251–99. Jaeggli, O. (1986) ‘Three Issues in the Theory of Clitics: Case, Doubled NPs, and Extraction’, in H. Borer (ed.), The Syntax of Pronominal Clitics (Syntax and Semantics, Vol. 19). New York: Academic Press, pp. 15–42. Levin, B. (1993) English Verb Classes and Alternations. Chicago: University of Chicago Press. Maldonado, R. (1988) ‘Energetic Reflexives in Spanish’, in S. Axmaker, A. Jaisser and H. Singmaster (eds), Berkeley Linguistics Society: Proceedings of the Fourteenth Annual Meeting. Berkeley: Berkeley Linguistics Society, pp. 153–65. –––– (1996) ‘Review of “The Reanalysis of Reflexives: A Diachronic Perspective” by M. Cennamo’, in Sprachtypologie und Universalienforschung, vol. 49(4), pp. 94–8. —— (1999) A media voz. Problemas conceptuales del clítico se. México D.F: Universidad Nacional Autónoma de México. Marin, M. (1989a) ‘Functions of se in the El Poema de Mio Cid’. Eugene: University of Oregon (ms.). —— (1989b) ‘Functions of se in El Quijote’. Eugene: University of Oregon (MS). Masullo, P. J. (1992) ‘Antipassive Constructions in Spanish’, in K. Koerner and P. Hirschbuehler (eds), Romance Languages and Modern Linguistic Theory. Amsterdam and Philadelphia: Benjamins, pp. 175–94.

264 Transitivity and Spanish Non-Anaphoric se Mazzola, M. (ed.) (1994) Issues and Theory in Romance Linguistics. Selected Papers from the Linguistic Symposium on Romance Languages XXIII. Washington, DC: Georgetown University Press. Mendikoetxea, A. (1994) ‘Impersonality in Non-Finite Contexts: the Spanish se Construction in Control and Raising Environments’, in M. Mazzola (ed.), op. cit., pp. 385–401. Moliner, M. (1984). Diccionario del uso del español. Madrid: Gredos. Myers, K. (ed.) and Powell, A. (transl.) (1999) A Wild Country out in the Garden: The Spiritual Autobiography of María de San José. Bloomington: Indiana University Press. Otero, C. (1986) ‘Arbitrary Subject in Finite Clauses’, in I. Bordelois, H. Contreras and K. Zagona (eds), Generative Studies in Spanish Syntax. Dordricht: Foris, pp. 81–109. Perlmutter, D. (1978) ‘Impersonal Passives and the Unaccusative Hypothesis’, Proceedings of the Berkeley Linguistic Society, vol. 4, pp. 157–89. Searle, J. R. 1990 [1965] ‘What is a Speech Act?’ in A. P. Martinich (ed.), The Philosophy of Language, 2nd edn. Oxford: Oxford University Press, pp. 115–25. Suñer, M. (1988), ‘The Role of Agreement in Clitic-Doubled Constructions. Natural Language and Linguistic Theory vol. 14(2), pp. 309–24. Terracini, B. (1945) ‘Sobre el verbo reflexivo y el problema de los orígenes románicos’. Revista de Filología Hispánica, vol. 7, pp. 1–22. Vann, R. (1994) ‘Middle Voice, “No-Fault SE”, and the Anticausative: Ergativity in Spanish’, in Michael Mazzola (ed.), op. cit., pp. 493–508. Vater, H. (1988) ‘Mittelkonstruktionen em Englischen, Danischen und Deutschen’, in Pavica Mrazovic and Wolfgang Teubert (eds), Valenzen im Kontrast: Urich Engel zum 60. Geburtstag. Heidelberg: Groos. Vendler, Z. (1967) Linguistics in Philosophy. Ithaca, NY: Cornell University Press. Verhaar, J. (1990) ‘How Transitive is Intransitive?’, Studies in Language, vol. 14(1), pp. 93–168. Zagona, K. (1996) ‘Compositionality of Aspect: Evidence from Spanish Aspectual se’, in C. Parodi, C. Quicoli, M. Saltarelli and M. L. Zubizarreta (eds), Aspects of Romance Linguistics. Washington, DC: Georgetown University Press, pp. 475–90.

12 ‘Juan salió contento’: Semantic Constraints on Small Clauses in Adjunct Position* Jiyoung Yoon

Introduction Most analyses of small clauses (SCs), illustrated in (12.1), have been mainly syntactic rather than semantic. Only a few studies such as Demonte (1992), McNally (1994) and Rapoport (1991, 1993b, 1999) have included semantic factors in the syntactic accounts they provide: (12.1)

a. SC in complement position: Considero inteligente a Juan. ‘I consider Juan intelligent.’ b. SC in adjunct position: Juan llegó al parque cansado. ‘Juan arrived at the park tired.’

These semantic analyses, in turn, focus on one of two areas: (a) the analysis of the semantics of adjunct predicates to find out why certain types of predicates are allowed to occur with a particular verb or verb phrase,1 whereas others are disallowed – this type of analysis has been done in terms of stage – versus individual-level predicates (SLP and ILP, respectively) (Rapoport, 1991, 1993a, 1993b; McNally, 1994); or (b) the semantic analysis of matrix

* I would like to thank two anonymous reviewers and Clancy Clements who provided me with helpful comments on the empirical data, the theoretical background, and the organization of an earlier version of this chapter. I would also like to express my gratitude to the audiences at the 4th Hispanic Linguistics Symposium, the 55th annual Kentucky Foreign Language Conference, and the 31st Annual Meeting of the Linguistic Association of the Southwest for their suggestions and comments on the partial and modified versions of this chapter presented at each conference. All errors are my own. 265

266 Semantic Constraints on Small Clauses in Adjunct Position

verb phrases (stative [STA], activity [ACT], accomplishment [ACC], or achievement [ACH]), to see whether the lexical aspect of the verb phrase (that is, Aktionsart) correlates with the allowance or disallowance of a particular type of predicate (Demonte, 1992; Rapoport, 1999). Nevertheless, few studies have considered the interaction of these constraints, that is, the interaction of the semantics of the matrix verb phrase and the adjunct predicate, let alone the semantics of direct objects and subjects themselves. In addition, the distinction of lexical aspect (based on Vendler, 1967, and Dowty, 1979) seems to be too general to capture the idiosyncractic semantic nature of different verb phrases, as shown in examples (12.2–12.5), which show that STA, ACT, ACC and ACH verb phrases do not behave consistently with respect to allowing the subject host, suggesting that lexical aspect may not be the only factor that determines the allowance of the adjunct predicate:2 (12.2)

(12.3)

(12.4)

(12.5)

STA verb phrases:3 a. Algún día -pensó desanimado. (RAE)4 ‘One day-he thought discouraged.’ b. *María sabía la verdad confusa/preocupada. ‘Mary knew the truth confused/worried.’ ACT verb phrases: a. María corre cansada. ‘María runs tired.’ b. ?? María corre incapaz. ‘María runs incapable.’ ACC verb phrases: a. María escribió el poema inspirada/alegre. ‘María wrote the poem inspired/happy.’ b. ? Juan construyó una casa animado. ‘John constructed a house animated.’ ACH verb phrases: a. La escena en que aparece borracho … no fue ficticia. (RAE) ‘The scene in which he appears drunk … was not fictitious.’ b. ?? María encontró el libro inesperada. ‘Mary found the book unexpected.’

As the examples in (12.2–12.5) show, the fact that two different verb phrases belong to the same semantic aspectual category – STA, ACT, ACC or ACH – does not necessarily mean that these verb phrases behave in the same way with respect to the (dis)allowance of the subject host to the adjunct predicates. In (12.2), both pensar ‘to think’ and saber la verdad ‘to know the truth’ are STA verb phrases, but only the first allows the subject host. Likewise, in (12.3a), the ACT verb correr ‘to run’ allows the adjunct predicate, whereas in (12.3b) the

Jiyoung Yoon 267

same verb with a different adjunct predicate sounds less acceptable. The same is true for ACC verb phrases: the adjunct predicate in (12.4a) with escribir el poema ‘write the poem’ can freely refer to the subject host María while (12.4b) with construir una casa ‘to construct a house’ is less acceptable. ACH verbs in (12.5) do not behave consistently either: aparecer ‘to appear’ freely allows for the adjunct predicate, but the same is not true for encontrar el libro ‘to find the book.’ All of these facts indicate that a mere distinction of the semantics of the main verbs or verb phrases – as STA, ACT, ACC or ACH – is too general to capture all of the various factors that affect the behaviour of the adjunct predicates. This suggests then that we need to take into account not only the aspectual class of the main verb, but also its other semantic traits, as well as the semantics of the other elements of the sentence. In this chapter, we argue that the licensing of the adjunct predicate in SC constructions is not dependent on a single factor but rather on the interaction among the semantics of verbs, predicates and objects. As a starting point, we analyse the lexical aspects of verb phrases (that is, STAs, ACTs, ACCs, ACHs) in relation to the licensing of the adjunct predicate. The semantic factors of the adjunct predicate are also examined in terms of individual-level versus stage-level predicates (ILP v. SLP). In addition, we examine both the affectedness and the definiteness of the object, and summarize all of the factors in terms of the Transitivity Hypothesis (Hopper and Thompson, 1980; Thompson and Hopper, 2001).

Stage-level predicates/verbs v. individual-level predicates/verbs Background Let us consider the following examples of SLPs and ILPs with respect to the lexical aspect of each verb phrase: (12.6)

(12.7)

(12.8)

STA verb phrases: a. María está en el cuarto emocionada. (SLP) ‘Mary is in the room excited.’ b. *María está en el cuarto inteligente. (ILP) ‘Mary is in the room intelligent.’ ACT verb phrases: a. Ese día trabajábamos nerviosos. (SLP) ‘That day we were working nervous.’ b. *Trabajaba incapaz en la empresa de su padre. (ILP) ‘He was working incapable in the company of his father.’ ACC verb phrases: a. Alzaba los brazos tras gritar emocionado la experiencia de Victoria … (SLP)(RAE)

268 Semantic Constraints on Small Clauses in Adjunct Position

(12.9)

‘He raised his arms after yelling excited the experience of Victory … .’ b. *José escribió inteligente una novela interesante. (ILP) ‘José wrote an interesting novel intelligent.’ ACH verb phrases: a. Afirmó haber llegado muy cansado por el largo viaje. (SLP) (RAE) ‘He admitted having arrived very tired because of the long trip.’ b. *Manolo la besó optimista. (ILP) ‘Manolo kissed her optimistic.’

These examples show that not every adjunct predicate can be added to a transitive or intransitive sentence. Following Rapoport (1991) and McNally (1994), we suggest that this contrastive behaviour can be accounted for based on the distinction of SLPs and ILPs, in which only SL adjunct predicates attribute a temporary property to the referent of their host NP. In other words, semantic properties of the adjunct predicate play an important role in determining the (un)acceptability of a given adjunct predicate. In particular, the distinction between SLPs and ILPs is a crucial factor in accounting for the contrasting behaviour in (12.6–12.9). SLPs like emocionda ‘excited,’ nervioso ‘nervous,’ and cansado ‘tired’ as in the (a) examples in (12.6–12.9) are predicates that imply a transitory state, whereas ILPs like inteligente ‘intelligent,’ incapaz ‘incapable,’ and optimista ‘optimistic’ as in the (b) examples in (12.6–12.9) assume a permanent state (Carlson, 1977; Milsark, 1974, 1977).5 We propose that the SL/IL distinction can be reduced to the characterization of the feature [telicity], lexical aspectual notions indicating whether a given dynamic situation contains an endpoint as a result of an action or an event, and whether it denotes a duration of that action or event. SLPs can be taken to be [telic] in that they describe characteristics of individuals that hold in space and time, which necessarily entail a duration and an endpoint of such characteristics of an event (we call them ACC-like predicates), whereas ILPs are [telic] since they do not denote an endpoint of such an event (we call them ACT-like predicates).6 According to Hopper and Thompson’s Transitivity Hypothesis, telic events are considered to be higher in Transitivity than atelic events, thus, we analyse SLPs as predicates higher in Transitivity than ILPs.7 (See Chapter 1 for detailed discussions on the Transitivity Hypothesis.) In the following section we extend the SL/IL distinction to verbs. We show that just as only SL adjunct predicates are semantically compatible with a small clause construction, only SL verbs allow adjunct predicates (cf. Milsark, 1977; Carlson, 1977; Rapoport, 1991). The stage-level restriction on verbs In Carlson’s (1977) framework, IL verbs are characterized as ‘true’ statives while SL verbs can be stative or eventive. In other words, STA verbs (that is

Jiyoung Yoon 269

telic verbs) can be either SL or IL as shown in the examples in (12.10): (12.10)

a. Juan está en el cuarto. ‘Juan is in the room.’ b. Juan posee muchos libros. ‘Juan possesses many books.’

(SL verb) (IL verb)

In spite of this contrast, one can find more IL verbs than SL verbs in the STA verb class. Being IL verbs, these STA verbs do not allow any type of predicates – regardless of whether they are ILPs or SLPs – in adjunct position. The following examples illustrate subject-hosted adjunct predicates in IL verbs: (12.11) (12.12) (12.13)

(*SLP/*ILP) María sabe la teoría *{confusa/dudosa}/*inteligente.8 ‘María knows the theory {confused/doubtful}/intelligent..’ María posee el libro ??contenta/??curiosa. (??SLP/??ILP) ‘María possesses the book happy.’ María ama a sus nietos ??emocionada/??sincera. (??SLP/??ILP) ‘María loves her grandchildren excited/sincere.’

As the examples in (12.11–12.13) illustrate, most STA verbs, such as cognitives (saber ‘know’), possessives ( poseer ‘possess’) and emotives (amar ‘love’), disallow all adjunct predicates with subject hosts. Note, however, that with the emotivefactive verbs such as lamentar ‘to regret,’ sospechar ‘to suspect,’ and dudar ‘to doubt’ in (12.14–12.16), which presuppose some fact in the past that is assumed to be true, the sentences with SL adjunct predicates are more acceptable than those in (12.11–12.13). This may imply that this group of STA verbs is closer to SL verbs. As the label ‘emotive-factive’ implies, the phrase lamentar la situación ‘to regret the situation’ in (12.14) expresses an emotion regarding some presupposed event in which there was a problem. As such, the predicate is bound to an eventive space and time. In (12.15), the phrase sospechar a su colega ‘to suspect her colleague’ also expresses an emotion regarding a presupposed belief/suspicion that, for example, a person was deceptive. The following minimal pairs show that the IL/SL distinction is still effective, that is, the sentences containing these emotive-factive verbs are less acceptable with ILPs than with SLPs: (?SLP/*ILP) (12.14) María lamentaba (?)triste/*pesimista la situación.9 ‘María regretted the situation sad/pessimist.’ (12.15) María sospechaba (?)enojada/*inteligente a su colega. (?SLP/*ILP) ‘María suspected her colleague angry/intelligent.’ (12.16) María dudaba (?)nerviosa/*pesimista que vaya a pasar el examen. (?SLP/*ILP) ‘María doubted nervous/pessimist that she would pass the exam.’

270 Semantic Constraints on Small Clauses in Adjunct Position

When applying the IL/SL distinction to non-STA verbs, that is, ACT, ACC and ACH verbs, it becomes clearer that only SLPs can be added to sentences as an adjunct predicate. In such cases, unlike the objects in STA verbs, the objects in ACT, ACC and ACH verbs are more likely to be affected objects in Hopper and Thompson’s (1980) sense to be defined below (cf. Basilico, 1996). The examples are given in (12.17–12.19):10 (12.17) Mi hermano bebe cervezas contento. ‘My brother drinks beers happy.’ (12.18) María comió el pedazo de pizza desanimada. ‘María ate the piece of pizza distressed.’ (12.19) Juan la besó nervioso. ‘Juan kissed her nervous.’

(ACT) (ACC) (ACH)

In these examples, the objects cervezas ‘beers,’ el pedazo de pizza ‘the piece of pizza,’ and la ‘her’ undergo change, in other words, are ‘affected’ as a result of the action of the verb. As explained in Hopper and Thompson (1980), the affectedness of objects is determined by ‘the degree to which an action is transferred to a patient’ (1980: 252). In other words, it is ‘a function of how completely that patient is affected; it is done more effectively in, say I drank up the milk than in I drank some of the milk’ (ibid.: 253). For example, the object la casa ‘the house’ in destruir la casa ‘destroy the house’ would be more completely affected than the same object in tener la casa ‘have the house,’ in that the former undergoes change due to the action of the verb. Because affected objects are taken to be higher in Transitivity than nonaffected objects, and because SL verbs (ACH, ACC, ACT and some STA verbs) are also considered to be higher in Transitivity than IL verbs (for more detailed discussion of the Transitivity Hypothesis, see Chapter 1, this volume), we would predict that if Transitivity in a sentence (that is, the Transitivity features of its verbs and objects) is high, adjunct predicates are more likely to be allowed in a sentence. In the following section, we examine another factor that affects the licensing of predicates: the definiteness of objects.

Definiteness of objects We propose in this section that a higher degree of definiteness of the object (and thus a higher degree of Transitivity) makes a sentence more likely to be compatible with adjunct predicates. The following examples show the contrast between objects with indefinite and definite articles: (12.20) a. #~?? Juan cavó un hoyo animado/cansado. ‘Juan dug a hole animated/tired.’

Jiyoung Yoon 271

b. ? Juan cavó el hoyo animado/cansado. ‘Juan dug the hole animated/tired.’ (12.21) a. #~?? María tejió un suéter alegre/agotada. ‘María knit a sweater happy/exhausted.’ b. ? María tejió el suéter alegre/agotada. ‘María knit the sweater happy/exhausted.’ (12.22) a. ?? Juan construyó una casa animado. ‘Juan constructed a house animated.’ b. ? Juan construyó la casa animado. ‘Juan constructed the house animated.’ (12.23) a. *~?? María creó un partido político animada. ‘María created a political party animated.’ b. ? María creó el partido político animada. ‘María created the political party animated.’ All of the (b) examples with definite articles have a higher rate of acceptability than the (a) examples with indefinite articles for the native speakers of Spanish consulted. One of the impressionistic comments from my informants was that when they hear an object NP with a definite article, they tend to expect the adjunct predicate not to refer to the object since the object is already definite without a need for further modification, and thus they find less ambiguity for determining whether the adjunct predicate is subjecthosted or object-hosted.

Transitivity hypothesis and adjunct predicates We have examined so far how different elements of a sentence such as a verb, a predicate and an object can affect the licensing of adjunct predicates semantically. We propose that all of these factors are closely interrelated and that they can fit into the Transitivity Hierarchy proposed in Hopper and Thompson (1980). We have claimed that adjunct predicates are more likely to be acceptable in sentences that are higher in the Transitivity Hierarchy. First, we have argued that SLPs and SL verbs, which are characterized by their [telicity] property, allow for predicates in adjunct positions. Telic predicates (that is SLPs) can be taken to be higher in Transitivity than ILPs, which are not perfective in the Transitivity Hierarchy, since SLPs and SL verbs are spatio-temporally bound to an event, which entails some outcome or result. Therefore, telic predicates have an endpoint, as a consequence of the result entailed by the action or event. This corresponds to the [telic] feature in the Transitivity Hypothesis. Accordingly, we can affirm that adjunct predicates are more likely to be compatible if the verbs or the predicates are higher in Transitivity with respect to telicity.

272 Semantic Constraints on Small Clauses in Adjunct Position

On the other hand, SL verbs exhibit a high probability of entailing affected objects (that is, more patient-like objects in high Transitivity), while IL verbs low in Transitivity tend to entail non-affected objects also low in Transitivity. We have shown along the same lines that affected objects (normally accompanied by SL verbs) are more likely to be compatible with adjunct predicates. In addition, we have claimed that the more definite the object is, the more likely the adjunct predicate is to be compatible with the sentence. Therefore, verbs and objects higher in Transitivity are more likely to be compatible with adjunct predicates. Table 12.1 summarizes the discussion of the semantic factors licensing adjunct predicates in relation to the Transitivity Hypothesis. As Table 12.1 shows, adjunct predicates are not licensed solely by one part of the sentence. It is not simply the semantics of verbs or predicates that impose restrictions on adjunct predicates as suggested by most of the linguists discussed here (Demonte, 1992; Rapoport, 1999; McNally, 1994, among others). Instead, the semantics of all of the elements of a sentence (for example verbs, adjunct predicates, objects) interact in such a way as to contribute to a higher or lower Transitivity of the sentence as a whole, which in turn allows or disallows adjunct predicates. Table 12.1 shows that if the features of all of these elements correspond to higher Transitivity, the sentence is more likely Table 12.1 Licensing factors of adjunct predicates in relation to Transitivity Sentence part Verbs

Adjunct predicates

Objects

Higher Transitivity

Lower Transitivity

Licensing factors

SL verbs (more likely with an affected object; but a non-affected object is also possible)

IL verbs (with a non-affected object)

Examples

María destruyó el juguete enojada.

*María posee el libro contenta.

Licensing factors

SLPs ([telic])

ILPs ([telic])

Examples

María estudia contenta. María usó la lavadora sucia. ( María la usó sucia.)

*María estudia inteligente. *María usó la lavadora cara. ( *María la usó cara.)

Licensing factors

Affected objects ([patient]) and some non-affected objects

Only non-affected objects ([patient])

Examples

María destruyó el juguete enojada. (affected object) María busca un apartamento desesperada (non-affected object)

*María necesitaba el coche desesperada.

Licensing factors

[definite] objects

[definite] objects

Examples

?María tejió el suéter

??~*María tejió un suéter

agotada.

agotada.

Jiyoung Yoon 273

to allow adjunct predicates. With respect to verbs, SL verbs, which are usually non-stative, are more likely to be compatible with adjunct predicates, as in María destruyó el juguete enojada ‘Mary destroyed the toy angry’ versus *María posee el libro contenta ‘Mary possesses the book happy’. For adjunct predicates, SLPs (i.e. [telic] predicates), which are spatio-temporally bound to the universe and also entail a result or change of state as a consequence of the action of the verb, allow for adjunct predicates, as shown in María estudia contenta ‘María studies happy’ (contenta as SL predicate) versus *María estudia inteligente ‘María studies intelligent’ (inteligente as IL predicate). Objects, on the other hand, also contribute to Transitivity: Affected and definite objects are higher in Transitivity and non-affected and indefinite objects are lower. Thus, if the Transitivity features of the object are higher, it is more likely that subject-hosted adjunct predicates will be compatible with the sentence, as in María destruyó el juguete enojada ‘María destroyed the toy angry’ (affected and definite object) versus *María necesitaba un coche desesperada ‘María needed a car desperate’ (non-affected and indefinite object). It is important to point out, however, that Table 12.1 does not provide clear-cut criteria or an exact dichotomy for determining the (dis)allowance of a given adjunct predicate. Rather, semantic licensing should be understood as taking place on a continuum of high to low Transitivity. In other words, what Table 12.1 implies is that a more prototypical transitive sentence, in which the subject is more agent-like, the object more patient-like, and the verb/predicate bound to space and time (that is telic), is more likely to allow for adjunct predicates, imposing fewer restrictions on them: (12.24) a. María destruyó el juguete enojada (more prototypical transitive sentence) ‘María destroyed the toy angry.’ [volitional subject, telic verb, affected and definite object] b. ? María encontró el libro inesperada. ‘María found the book unexpected.’ [non-volitional subject, telic verb, definite object] c. *Este reloj necesita pilas roto (less prototypical transitive sentence) ‘This watch needs batteries broken.’ [non-volitional subject, atelic verb, indefinite plural object] The sentences in (12.24) represent different degrees of acceptability of adjunct predicates based on corresponding degrees of higher or lower Transitivity. Sentence (12.24a) is considered a more prototypical transitive sentence in that the subject is volitional, and the object undergoes change (that is, is an affected object) and is a definite noun, thus allowing for a higher chance of licensing the adjunct predicate enojada ‘angry.’ In contrast, sentence (12.24c) represents a less prototypical transitive sentence since the

274 Semantic Constraints on Small Clauses in Adjunct Position

subject is non-volitional, and the object does not undergo change and is a plural noun. As predicted, this sentence does not allow licensing of the adjunct predicate roto ‘broken.’ Sentence (12.24b) represents a medium between the two in terms of Transitivity: the subject is non-volitional, the verb is telic, and the object is definite but less affected than that of (12.24a). But (12.24b) is still less acceptable than (12.24a), because the Transitivity features of the subject and object are lower. This all suggests that a sentence with higher Transitivity does not guarantee the licensing of the adjunct predicate. Instead, it results in a greater probability that an adjunct predicate will be compatible with the verb or verb phrase in the sentence.

Conclusion In this chapter, we have claimed that the licensing of the adjunct predicate is not dependent on any single factor, as has been proposed by many linguists, but rather is based on the interaction among the semantics of verbs, adjunct predicates, and objects. We have argued that neither Vendlerian verb typologies nor the IL/SL distinction alone can account for the licensing of adjunct predicates unless other variables such as the affectedness and definiteness of the object are taken into consideration. We have claimed that there is an interrelation between the licensing of adjunct predicates and the Transitivity Hypothesis (Hopper and Thompson, 1980; Thompson and Hopper, 2001): The higher in Transitivity a sentence is, that is, the closer a sentence is to the prototypical transitive sentence, the higher the likelihood that the sentence allows adjunct predicates. Thus, licensing is not a clear-cut dichotomy that a formal syntactic analysis might be based on, but rather a continuum of probability for adjunct predicates to be allowed. This is what is proposed in the functionalist framework, namely, the notion of ‘prototypicality’ (Givón, 1984). The functional approach typically does not treat grammar as autonomous formal rules and discrete categories, but rather allows partial categoriality (Givón, 1984: 14). For instance, we have observed that most prototypical SL verbs require affected objects in order for an adjunct predicate to be added to a sentence (for example Comía tacos contenta ‘She was eating tacos happy’), while it is still possible for other SL verbs (for example some STA verbs and some ACT verbs) with nonaffected objects to allow adjunct predicates (for example Juan miraba la foto deprimido ‘Juan looked at the picture depressed’). In other words, although we have claimed that an adjunct predicate is more likely to be allowed in a prototypical high-Transitivity sentence (that is, one with high-Transitivity features in all parts of the sentence), this does not imply that a less prototypical transitive sentence (for example a sentence with a non-affected object) cannot allow an adjunct predicate at all. The continuum of probability for allowing an adjunct predicate should be understood as the extent to which ‘typical’ properties of a prototypical transitive sentence

Jiyoung Yoon 275

(for example a combination of a highly volitional subject, an affected and definite object, and a telic or SL verb) overlap. Thus, the functionalist view that we adopt replaces the exclusive, all-or-nothing approach of the formalist with the notion of prototypicality. Notes 1 Lexical aspect is determined not only by a verb itself but also by a verb phrase, that is, a verb and its accompanying elements such as adverbs and prepositional phrases. A well-known example is the difference between Juan corrió ‘Juan ran’ (ACT) and Juan corrió una milla ‘Juan ran a mile’ (ACC). 2 For more recent discussions about the typology of lexical aspects (termed ‘states of affairs’), see Van Valin and La Polla (1997: ch. 3–4), Groot (1983), and Dik (1997). In these studies, the parameters that distinguish the four states of affairs are [static], [telic], [punctual] and [control] (this last feature does not appear in Van Valin and La Polla). The four types of states of affairs are, then, situation, event, process and action, which correspond to the lexical aspects (that is, Aktionsart) state, achievement, accomplishment and activity, respectively. 3 Some of my examples were created for the purpose of testing their grammaticality/ acceptability with respect to the different lexical aspects. They were judged by native speakers of Spanish who participated in this study. 4 Some data in this study were collected from the Real Academia Española (RAE) database at http://www.rea.es. Some parts of the authentic examples were omitted (indicated as …) due to space limitations. 5 The notion of ILP and SLP can be referred to as ‘ILP/SLP reading’, which implies that the ILP/SLP distinction is in many cases not absolute but rather depends on the context (compare discussion of the ILP/SLP distinction in Chapters 8 and 13, this volume). 6 The contrast between ILPs and SLPs holds for object hosts as well as subject hosts, as shown in the following examples: (i)

(ii)

a. Juan lo pintó enorme. ‘Juan painted it enormous.’ b. *Juan lo pintó caro. ‘Juan painted it expensive.’ a. María leyó roto un libro. ‘María read a book torn.’ b. *María leyó interesante un libro. ‘María read a book interesting.’

(SLP) (ILP)

(SLP) (ILP)

7 See Bosque (1990) and Clements (1988) for more discussion of the classification of adjective predicates. 8 Note that the use of the preterite for the verb saber ‘to know’ (which means ‘found out’) makes the sentence sound better, as shown in the following contrast: (i)

a. *María sabía la verdad angustiada/preocupada. ‘María knew the truth distressed/worried.’ b. ?María supo la verdad angustiada. ‘María found out the truth distressed.’

Sentence (ib) is paraphrased in Spanish as María ya estaba angustiada cuando supo la verdad ‘María was already distressed when she found out the truth.’ Note that, with the use of the preterite, the meaning of the verb saber changes from STA to a punctual

276 Semantic Constraints on Small Clauses in Adjunct Position VP, that is, the beginning of knowing. It seems that this can be accounted for in terms of high/low Transitivity (Hopper and Thompson, 1980) since supo is punctual while sabía is stative. 9 The preferred word order for sentence (12.14) for most native speakers is María, triste, lamentaba la situación ‘María, sad, regretted the situation’. The sentence with an ILP, however, still does not allow this word order (*María, pesimista, lamentaba la situación ‘María, pessimistic, regretted the situation’). 10 Note that ACT verbs do not necessarily entail affected objects as shown in (i): (i) Manolo miraba la casa orgulloso. ‘Manolo looked at the house proud.’ (ii) Marisol busca un apartamento desesperada. ‘Marisol looks for an apartment desperate.’ In such examples, objects are not affected by the action of the ACT verbs. Nevertheless, almost all the ACC and ACH verbs (which exhibit higher Transitivity than ACT verbs) normally entail affected objects (for example destruir la casa [ACC] ‘destroy the house,’ esconder el regalo [ACC] ‘hide the present,’ dibujar una manzana [ACC] ‘draw an apple’; tocar la pared [ACH] ‘touch the wall,’ perder el cheque [ACH] ‘lose the check’). However, the correlation between the lexical aspect of the verb and the affectedness of the object is not absolute, but should be understood instead as a probability.

References Basilico, D. (1996) ‘Asymmetric Object Positions’, Eastern States Conference on Linguistics ‘95, pp. 1–12. Cornell University. Bosque, I. (ed.) (1990) Tiempo y aspecto en español (Tense and Aspect in Spanish). Madrid: Ediciones Cátedra. Carlson, G. (1977) ‘Reference to Kinds in English’, PhD dissertation, University of Massachusetts, Amherst. Clements, J. C. (1988) ‘The Spanish COP  ADJ Construction’, Linguistics, vol. 26, pp. 779–822. Demonte, V. (1992) ‘Temporal and Aspectual Constraints on Predicative Adjective Phrases’, in H. Campos and F. Martínez-Gil (eds), Current Studies in Spanish Linguistics. Washington D.C.: Georgetown University Press, pp. 165–200. Dik, S. C. (1997) ‘States of Affairs and Semantic Functions’, in K. Hengeveld (ed.), The Theory of Functional Grammar, vol. 1. Berlin: Mouton de Gruyter, pp. 105–26. Dowty, D. (1979) Word Meaning and Montague Grammar: The Semantics of Verbs and Times in Generative Semantics and in Montague’s PTQ. Dordrecht; Boston: D. Reidel, Publishing Company. Givón, T. (1984) Syntax: A functional-Typological Introduction, vol. 1. Amsterdam and Philadelphia: John Benjamins. Groot, C. de. (1983) ‘Typology of States of Affairs’, in H. Bennis and W.U.S. van Lessen Kloeke (eds), Linguistics in the Netherlands. Dordrecht: Foris, pp. 73–82. Hopper, P. J. and S. A. Thompson (1980) ‘Transitivity in Grammar and Discourse’, Language, vol. 56, pp. 251–99. McNally, L. (1994) ‘Adjunct Predicates and the Individual/Stage Distinction’, West Coast Conference on Formal Linguistics, vol. 12, pp. 561–76. Milsark, G. L. (1974) ‘Existential Sentences in English’, PhD dissertation: MIT. –––– (1977) ‘Toward an Explanation of Certain Peculiarities of the Existential Construction in English’, Linguistic Analysis, vol. 3, pp. 1–29.

Jiyoung Yoon 277 Rapoport, T. R. (1991) ‘Adjunct-predicate Licensing and D-structure’, in S. D. Rothstein (ed.), Perspectives on Phrase Structure: Heads and Licensing. San Diego: Academic Press, pp. 159–87. —— (1993a) ‘Stage and Adjunct Predicates: Licensing and Structure in Secondary Predication Constructions’, in E. Reuland and W. Abraham (eds), Knowledge and Language, vol. 2, Lexical and Conceptual Structure. Netherlands: Kluwer Academic Publishers, pp. 157–82. —— (1993b) ‘Verbs in Depictives and Resultatives’, in J. Pustejovsky (ed.), Semantics and the Lexicon. Netherlands: Kluwer Academic Publishers, pp. 163–84. —— (1999) ‘Structure, Aspect, and the Predicate’. Language, vol. 75, pp. 653–77. Thompson, S. A. and P. J. Hopper (2001) ‘Transitivity, Clause Structure, and Argument Structure: Evidence from Conversation’, in J. L. Bybee and P. J. Hopper (eds), Frequency and the Emergence of Linguistic Structure. Amsterdam: John Benjamins, pp. 27–60. Van Valin, R. D. and R. J. La Polla (1997) Syntax: Structure, Meaning and Function. Cambridge: Cambridge University Press. Vendler, Z. (1967) ‘Verbs and Times’, in Z. Vendler (ed.), Linguistics and Philosophy. Ithaca, NY: Cornell University Press, pp. 97–121.

13 Causative hacer and dejar Carmen Ruiz-Sánchez

Introduction In the past there has been some general discussion of Romance causatives, some of which have included Spanish samples, but little has been written specifically on Spanish causatives per se. To date, most studies dealing with Spanish causatives have concentrated on the hacer  infinitive construction, while some of them (for example Aranda, 1990) have focused on similarities and differences among lexical causative verbs (for example mostrar ‘show’ v. hacer ver ‘make see’) or the synonymy between the hacer  infinitive v. the dejar  past participle constructions (for example lo hice sonrojarse ‘I made him blush’ v. lo dejé sonrojado ‘I left him blushing’). Few studies, however, present a comparative analysis of causative hacer and dejar. The literature on Romance causatives has almost exclusively treated the syntax of causative constructions rather than their semantics. For example, many linguists have focused their attention on the question of argument structure (valency and actancy) (Tesnière, 1959), discovering that some verbs are subject to modifications of valency. Hacer is considered one of these verbs since it increases its valency by one actant when used in the causative hacer  infinitive construction. The goal of this chapter is to fill this void, unaddressed by previous studies, and explore the different semantic make up of causative hacer and dejar and the syntactic repercussions of these differences in causative infinitive constructions and small clauses. In arguing that hacer is the default causative verb in Spanish while causative dejar is a semantic extension of the main verb dejar, we claim that the difference between hacer and dejar is similar to that between the copula ser and estar. Before beginning, however, we will comment on the terminology to be used.

Terminology In the present study, we follow Curnow’s (1994: 167) definition of causative, as adapted from Shibatani’s (1976) definition: any construction with hacer or 278

Carmen Ruiz-Sánchez 279

dejar is interpreted as causative if the event is caused by the subject of the main clause and if it occurs only as a result of the intervention of the subject. In the case of dejar, the intervention of the subject is most often understood as a conscious nonintervention on the part of the subject, letting an event unfold by not impeding or stopping its trajectory. Regarding other terminology used in this chapter, we use the term implication to speak of the difference between the semantics of causative dejar and hacer. For example, in the sentence Juanita deja terminar su trabajo al niño ‘Juanita lets the boy finish his work’, there is an implication that the boy has already started the work and intends to finish it. In the analogous sentence Juanita hace terminar su trabajo al niño ‘Juanita makes the boy finish his work’, there is no such implication.

The semantics of hacer and dejar In this section we examine the non-causative meanings of hacer and dejar in order to isolate their respectively different semantic characteristics. The characterization of these verbs is based on the examination of some of their respective Transitivity features in the sense of Hopper and Thompson (1980): Transitivity of the verb, thematic role of the subject, lexical aspect and reference to the event structure. The following analysis will be used later to identify the semantic features retained by each verb in causative constructions. The semantics of non-causative hacer In its most standard denotation as ‘create, produce, fabricate, make, do, and so on’, hacer can be characterized as requiring a direct object and an animate subject, denoting a non-punctual dynamic situation (an accomplishment [nonpunctual, telic] or an activity [nonpunctual, atelic], but never an achievement [punctual, telic]), and having a scope of reference over the whole event: (13.1)

(13.2)

a. Los estudiantes hacen la tarea. ‘The students do their homework.’ b. *Los estudiantes hacen rápidamente. ‘The students do quickly.’ a. Los estudiantes/ los profesores/ Juan y María hacen su trabajo. ‘The students/ the teachers/ Juan and Marta do their work.’ b. *Los libros hacen su trabajo. ‘The books do their work.’

Thus, the (a) examples in (13.1)–(13.2) are well-formed because they have a direct object and an animate subject. On the other hand, (13.1b) and (13.2b) are ill-formed because the first has no direct object and the second is without

280 Causative hacer and dejar

an animate subject. As for the incompatibility of punctual events and the semantic feature(s) of hacer, it is compatible with object NPs denoting time intervals, but not with NPs denoting punctual events. Thus, (13.3a) is wellformed because una fiesta ‘a party’ is a durative event. By contrast, (13.3b) is ill-formed because the direct object NP llegada ‘arrival’ denotes a punctual event: (13.3) a. Marta hizo una fiesta el viernes. (durative) ‘Marta had a party on Friday.’ b. *Marta hizo la llegada el viernes. (punctual) ‘Marta had her arrival on Friday.’ With regard to the trait of referring to the whole event, the entailments in (13.4) can be said not to hold because hacer la sopa ‘make the soup’ makes reference to the whole event of creating the soup, not to a subset of its component parts: (13.4) a. Juan calentó la sopa -/- Juan hizo la sopa. ‘Juan heated up the soup’ -/- Juan made the soup.’ b. Juan cortó las verduras para la sopa -/- Juan hizo la sopa. ‘Juan cut up the vegetables for the soup’ -/- Juan made the soup.’ c. Juan hirvió los fideos para la sopa -/- Juan hizo la sopa. ‘Juan boiled the noodles for the soup’ -/- Juan made the soup.’ Having identified the most important semantic traits of hacer, let us turn to the semantic characterization of dejar. The semantics of non-causative dejar We group the different key meanings of dejar into two broad categories. The first is ‘separation, detachment or abandonment’, and the second ‘permission/ non-avoidance or non-impediment of an event’.1 Examples of the first set of entries are given in (13.5): (13.5)

a. Lo he dejado sobre la mesa. ‘I left it on the table’. b. He dejado la bebida/ los estudios/ de fumar. ‘I gave up drinking/ studying/smoking’. c. Ha dejado/abandonado a su familia. ‘He has left/abandoned his family’. d. Le dejé el coche a mi amiga. ‘I lent my car to my friend’.

Carmen Ruiz-Sánchez 281

Examples of the entries ‘permission/ non-avoidance or non-impediment of an event’ are shown in (13.6): (13.6) a. Juanita nos dejó salir. ‘Juanita let us leave.’ b. La mujer que estaba en el medio del pasillo me dejó pasar finalmente. ‘The woman who was in the middle of the hall let me finally go through.’ c. El policía nos dejó en paz. ‘The police officer left us alone.’ Non-causative dejar shares the following characteristics in its different entries in question: it is a (di)transitive achievement verb with an agent or experiencer subject, and implies either that the beginning of an event has already taken place or that a state already existed. If we compare the characterizations of hacer and dejar, their differences lie mainly in their respective lexical aspect, hacer is nonpunctual (telic or atelic) and dejar is punctual. Moreover, while hacer refers to a whole event and only to the event, dejar refers either to a point on the timeline of an event already started or to a point on the timeline of a state that already existed or was already happening. Based on the preceding semantic and syntactic characterization of these two verbs, we advance the following, uncontroversial hypothesis: the verbs hacer and dejar will exhibit different semantic features in their respective causative constructions, giving rise to different readings and interpretations. We also predict that these semantic differences can account for the different behaviours of hacer and dejar in small clauses such as those underlined in (13.7) and (13.8). In these examples, we see that causative hacer is compatible with an individual-level predicate like interesante ‘interesting’, but causative dejar is not. On the other hand, causative dejar is compatible with a stage-level predicate such as cansada ‘tired’ where hacer is not. We will discuss why this is so later in the chapter. (13.7) a. El profesor hizo la charla interesante. ‘The teacher made the lecture interesting.’ b. *El profesor dejó la charla interesante. ‘The teacher left the lecture interesting.’ (13.8) a. La abuela dejó cansada a Marta. ‘The grandmother left Marta tired.’ b. *La abuela hizo cansada a Marta. ‘The grandmother made Marta tired.’

282 Causative hacer and dejar

Causative dejar and hacer For our present purposes, this study is limited to dejar/hacer-plus infinitive constructions with agentive animate subjects.2 Regarding the methodology, in order to tease apart the semantic features of hacer and dejar, we use minimal pair sentences in which only one of the two verbs yields a well-formed sentence or in which one of the verbs is more acceptable than the other.3 In some cases, both verbs yield well-formed sentences, but each implies something different.4 Contexts contrasting causative dejar and hacer The examples to follow are classified into four categories according to the lexical aspect of the infinitive: states (STA), activities (ACT), accomplishments (ACC), and achievements (ACH) (Vendler, 1967, in Yoon, 2001).5 The motivation for this classification is to discern how the lexical aspect of the infinitive affects the behaviours of dejar and hacer. When contrasting the semantics of dejar/hacer-plus infinitive constructions, we focus on the following information: the possible avoidance/impediment of the action by the causer, intentionality of the causer, direct/indirect causation, the causee’s willingness for the event to happen, control over the event (coercion/ permission) on the part of the causer and the presence of semantic traces from the non-causative meanings of the verbs. In addition, some Transitivity features are considered, such as volitionality, involvement of the causer, and reference to the event structure. This information is compiled in two tables at the end of this section. A brief description of the different implications and readings follows each set of examples. In order to avoid repetition, the explanation for the first set of examples is more comprehensive and includes all of the implications, while the subsequent comments and example sets are limited to highlighting the most relevant difference/s that the constructed context reveals. States (13.9)

a. No quise justificarme y les dejé pensar lo que quisieran sobre mí. ‘I did not want to justify myself and I let them think whatever they wanted about me.’ b. No quise justificarme y *les hice pensar lo que quisieran sobre mí. ‘I did not want to justify myself and I made them think whatever they wanted about me.’

(13.10)

a. A veces mis compañeros de clase me hacen sentir un poco estúpido. ‘Sometimes my classmates make me feel a bit silly.’

Carmen Ruiz-Sánchez 283

b. A veces mis compañeros de clase *me dejan sentir un poco estúpido. ‘Sometimes my classmates let me feel a bit silly.’ Examples (13.9a) and (13.9b) show that when the sentence implies that the action could have been avoided, or was caused by lack of action on the part of the subject, only causative dejar is permitted. Therefore, the causer in (13.9a) is responsible for the action only in an indirect manner. By contrast, in the sentence with hacer (13.10a) there is a more direct involvement of the causer in the action. Another difference between these two verbs is that there is clear intentionality with dejar in (13.9a), but not necessarily in (13.10a). Moreover, dejar in (13.9a) refers to a state of mind already originated: the thinking and judging of his classmates had already started and the causer did not want to stop it, whereas in the sentence with hacer (13.10a) the implication is that the causer originates the emotion, caused in its entirety by his classmates’ actions or behaviour. This implication of hacer is traceable back to its original meaning ‘to create, to produce … from the beginning’. Examples (13.10a) and (13.10b) contrast hacer and dejar by illustrating the pure causative nature of hacer. In general, dejar adds more than causative meaning to the sentence, and it is not acceptable in those cases where the only intended meaning is causation. It appears as well that hacer implies that the causee did not want the event denoted by the verb to take place. Thus, causative hacer can express some sense of coercion, as it is more clearly demonstrated in following examples in (13.11) and (13.12): (13.11)

(13.12)

a. Mi padre me deja estar en su oficina si no lo molesto. Como lo moleste me echa. ‘My father lets me hang out in his office as long as I don’t bother him. If I do, he kicks me out.’ b. *Mi padre me hace estar en su oficina si no lo molesto. Como lo moleste me echa. ‘My father makes me hang out in his office if I don’t bother him. If I do, he kicks me out.’ a. Mi padre me hace estar en la tienda los fines de semana cuando yo prefiero jugar con mis amigos. ‘My father makes me be (work) in the store on the weekends when I would rather be with my friends.’ b. *Mi padre me deja estar en la tienda los fines de semana cuando yo prefiero jugar con mis amigos. ‘My father lets me be (work) in the store on the weekends when I would rather be with my friends.’

In (13.12a) hacer carries a strong sense of obligation as evidenced in the fact that it can be exchanged with the verb obligar. Besides, the causee has no

284 Causative hacer and dejar

desire to perform the action, as indicated by its incompatibility with the clause cuando yo prefiero jugar con mis amigos ‘when I would rather play with my friends’. Consequently, since the context in examples (13.11a) and (13.11b) implies the causee’s willingness to be in the office, causative dejar can be used but hacer is disallowed. In (13.11a) we can also see a cooperative effort between the two participants, and the causation is indirect. Accomplishments (13.13)

(13.14)

a. Me lesioné pero no tuve que salirme de la carrera, ¡menos mal! El juez fue bueno conmigo y me dejó correr las tres últimas millas a pesar de todo. ‘I injured myself but I didn’t have to quit the race. Thank God! The judge was very nice to me and he let me run the last three miles despite everything.’ b. Me lesioné pero no tuve que salirme de la carrera, ¡menos mal! El juez fue bueno conmigo y *?me hizo correr las tres últimas millas a pesar de todo. I had an injury but I didn’t have to quit the race. Thank God! The judge was very nice to me and he made me run the last three miles despite everything.’ a. Mi entrenador es muy exigente y me hace correr 10 millas todos los días como parte de mi entrenamiento. ‘My coach is very demanding and he makes me run 10 miles every day as part of my training.’ b. Mi entrenador es muy exigente y *me deja correr 10 millas todos los días como parte de mi entrenamiento. ‘My coach is very demanding and *he lets me run 10 miles every day as part of my training.’

All the implications considered for the stative verbs above apply in this set of examples as well. Hacer expresses some sense of coercion, direct causation, and reference to the whole event. Dejar contains the sense of permission for the continuation or non-impediment of the event, indirect causation, the causee’s volition in the carrying out of the event, and reference to a subpart or to a wider implications of the event. More examples of dejar and hacer with accomplishments follow in order to specifically illustrate dejar’s feature of letting some event continue its trajectory (the non-impediment of an ongoing activity or state): (13.15)

a. La señora estaba en medio del pasillo bloqueando el paso con el carrito, pero cuando nos vio acercarnos, se echó a un lado y nos dejó pasar.

Carmen Ruiz-Sánchez 285

‘The lady was in the middle of the hall blocking the way with her cart, but when she saw us approaching, she moved to one side and let us pass.’ b. La señora estaba en medio del pasillo bloqueando el paso con el carrito, pero cuando nos vio acercarnos, se echó a un lado y *nos hizo pasar. ‘That lady was in the middle of the hall blocking everyone’s way with her cart, but when she saw us approaching, she moved to one side and *made us pass.’ (13.16) a. En el libro se explica cómo los nazis hacían pasar a los judíos a las duchas de gas. ‘In that book they explain how the Nazis made the Jews go to the gas chambers.’ b. En el libro se explica cómo *?los nazis dejaban pasar a los judíos a las duchas de gas. ‘In that book they explain how *the Nazis let the Jews go to the gas chambers.’ Achievements (13.17)

(13.18)

a. El director estaba en la puerta recibiendo a los invitados. Por la expresión de su cara me di cuenta de que me había vestido muy informal para la ocasión, pero aún así me dejó entrar en su fiesta sin decir nada. ‘The director was at the door welcoming the guests. From the expression on his face, I could tell that I was dressed very informally for the occasion but, despite that, he let me enter his party without saying a word.’ b. El director estaba en la puerta recibiendo a los invitados. Por la expresión de su cara me di cuenta de que me había vestido muy informal para la ocasión y, aún así *?me hizo entrar en su fiesta sin decir nada. ‘The president was at the door welcoming the guests. From the expression on his face I could tell that I was dressed very informally for the occasion and, despite that, he made me enter his party without saying a word.’ a. Acusé a Manolo de haberse quedado con mi CD. Él se enfadó y la próxima vez que estuve en su casa, me hizo entrar en su habitación para que viera que no lo tenía. ‘I accused Manolo of taking my CD. He got angry and the next time I was at his house, he made me go in his room so I could see that he had not taken it.’

286 Causative hacer and dejar

b. Acusé a Manolo de haberse quedado con mi CD. Él se enfadó y la próxima vez que estuve en su casa, *?me dejó entrar en su habitación para que viera que no lo tenía. ‘I accused Manolo of taking my CD. He got angry and the next time I was at his house, he let me go in his room and see that he had not taken it.’ These sentences illustrate the expected implications of dejar and hacer. In particular, the sentences in (13.17a) and (13.17b) clearly display the difference between permission versus coercion readings of dejar and hacer respectively. The hacer causative in (13.17b) is ill-formed because there is not implication of direct causation on the part of the causer, but the use of hacer becomes acceptable in (13.18a), since the context is one in which Manolo is directly responsible for the causee’s action. In the same context, dejar is ill-formed because it wrongly implies that Manolo lets the event happen or gives his permission, counter to the meaning established by the context. This sentence would only be acceptable in a context in which it is understood that the person had asked Manolo to enter his room to check if he had the CD. Activities (13.19)

a. Cuando llegué a casa, mi hermana cantaba en la ducha. Aunque canta fatal, no le dije nada y la dejé cantar porque sé cuánto le gusta. ‘When I got home, my sister was singing. Even though she sings horribly, I didn’t say anything and let her sing since I know how much she likes it.’ b. Cuando llegué a casa, mi hermana cantaba en la ducha. Aunque canta fatal, no le dije nada y *la hice cantar porque sé cuánto le gusta. ‘When I got home, my sister was singing. Even though she sings horribly, I didn’t say anything and *made her sing since I know how much she likes it.’ (13.20) a. Antonio no quería cantar porque le daba vergüenza. Todos insistimos y al final lo hicimos cantar. ‘Antonio did not want to sing because he was shy. We all insisted and eventually made him sing.’ b. Antonio no quería cantar porque le daba vergüenza. Todos insistimos y al final *lo dejamos cantar. ‘Antonio did not want to sing because he was shy. We all insisted and eventually let him sing.’ These examples demonstrate the now-familiar ways in which dejar and hacer refer to an event structure. The context in (13.19a) conveys the idea that the

Carmen Ruiz-Sánchez 287

causer does not stop an ongoing activity but lets the causee continue with the activity. Due to its semantic make up, hacer in the same context is illformed, as seen in (13.19b). If the context is one in which the causer initiates the activity, as in (13.20a), only then is the use of hacer correct. In relation to this same idea, Aranda (1990: 69) states that when the causative verb is hacer, the impulse for the realization of the action originates in the subject/ causer, whose participation in the event is greater than when the verb dejar is used. According to him, this fact becomes evident in those cases when hacer can be interpreted as mandar and dejar as permitir. Finally, consider the following activity. (13.21) a. Dejaron arder los pastos toda la noche. ‘They let the pasture burn all night long.’ b. *?Hicieron arder los pastos toda la noche. ‘They made the pasture burn all night long.’ c. Hizo arder toda la casa con una sola cerilla. ‘He/she made the house burn with only one match.’ d. *Dejó arder toda la casa con una sola cerilla. ‘He/she *let the house burn with only one match.’ The phrase ‘toda la noche’ indicates duration, and can be used with dejar but not with hacer. Dejar arder in (13.21a) implies that the burning continued all night long while hacer arder, on the other hand, refers to the beginning of that action, that is, to the ignition of the fire, the whole phrase hacer arder being a synonym with quemar ‘burn’ or prender fuego ‘light a fire’. Even though arder is an activity, the phrase hacer arder becomes punctual and telic, an achievement. Sentence (13.21b) would only be acceptable if the action is interpreted as reiterative, meaning that they continued setting fire to the pasture all night. By contrast, if the context construed is one where the only possible interpretation is that of setting the fire as in (13.21c), then only hacer can be used but not dejar (cf. 13.21d). Having discussed causative hacer and dejar in various contexts involving states, accomplishments, achievements and activities, let us now summarize the respective properties that make up the two causative verbs. General characterization of causative dejar and hacer The previous discussion of causative dejar and hacer has identified consistent differences between the two verbs regarding their lexical semantic make up and in terms of their respective implications. Regarding the lexical aspect of the infinitive with which causative dejar and hacer appear, stative verbs are largely incompatible with causative hacer but not with causative dejar. With achievements and accomplishments (that is dynamic telic predicates), we also note that the reference of the verb complex can imply prior intention

288 Causative hacer and dejar

on the part of the causee. That is, while with stative and activity verbs, causative dejar refers to a subpart of an event, it encompasses more than the event proper with telic predicates. In Tables 13.1 and 13.2 we have attempted to summarize the difference between the causative verbs, trying to capture the different implications that each causative verb carries in the different contexts provided in the previous section. By comparing Tables 13.1 and 13.2, we arrive at the following characterization of causative dejar and hacer. Causative dejar carries the implication that the causer could have avoided or impeded the event or state being caused, but instead let it occur or continue, causing the action only indirectly, albeit intentionally. In most cases, causative dejar carries the implication that the causee wants the event or state to happen or continue and that s/he has some control over the action. The caused event or state is then the result of a cooperative effort between the two participants rather than the result of just the causer’s intervention. Causative dejar usually refers to an event that has already started, so its reference is partial in that sense. However, it can also include in its scope of reference the whole event/state as well as the intention on the part of one or plus the preceding intention. The reason for the unique semantic nature of causative dejar lies in that it retains traits from dejar as a main verb, namely ‘permission’ and/or ‘non-impediment’. In addition, the low involvement of the causer in bringing about the event Table 13.1 Characterization of causative dejar Characterization of dejar in the different verb categories

STA

ACC

ACH

ACT

Possible avoidance/impediment of the action by the causer









Intentionality (causer)









Direct/indirect causation

I

I

I

I

The causee’s willingness for the event to happen









Control over the event by the causee









Reference to the event structure

partial

partial

whole event  intention

partial

low

low

low

low









Involvement of the causer Semantic traces of noncausative meanings

Carmen Ruiz-Sánchez 289 Table 13.2 Characterization of causative hacer Characterization of hacer in the different verbs categories

STA

ACC

ACH

ACT

N/A

N/A

N/A

N/A

Intentionality (causer) Direct/Indirect causation

 D

 D

 D

 D

The causee’s willingness for the event to happen









Control over the event by the causee



 (some coercion)





Reference to the event structure

whole event

whole event

whole event

beginning

high

high

low

high

 (but  with verbs of feeling)







Possible avoidance/impediment of the action by the causer

Involvement of the causer Semantic traces of noncausative meanings

or its continuation or termination can be related to its traits ‘abandonment, separation’ as a main verb. Causative hacer displays characteristics of an inherent causative verb. In most cases, it adds no additional meaning to the sentence other than that of causation. The event happens only as a result of the intervention of the subject, who has direct responsibility for the event. Causative hacer carries the implication that the causee does not want the event or state to come about, and has no control over it, hence the sense of coercion in many instances. Causative hacer refers to the whole event. Generally, the idea of the initiation of the event or state is entirely clear. We propose that this is due to its semantics as main verb, namely ‘create, originate, produce.’

Dejar and hacer in small clauses6 In this section, we would like to support the developed characterization of dejar and hacer by briefly discussing the behaviour of these causative verbs in small clauses. Previous research on small clauses (Yoon, 2001) shows that hacer and dejar behave differently with stage-level predicates (SLPs) (for example sorprendido ‘surprised’, cansado ‘tired’) and individual-level predicates (ILPs) (such as inteligente ‘intelligent’, alto ‘tall’). After considering dejar and

290 Causative hacer and dejar

hacer in small clauses with SLPs and ILPs, we advance an account, based on the characterization of causative dejar and hacer above, as to why some SLPs can only take one of these causative verbs, and why some ILPs can be used only with hacer and under specific conditions. Given that so far our analysis of causative dejar/hacer-plus infinitive constructions has been limited to animate subjects, we will follow the same practice here. Regarding the analysis of dejar and hacer in small clauses, I will divide the examples in two subsections, one with animate objects and the other with inanimate objects to discover if animacy of the object is a variable in the choice of dejar or hacer. Small clauses with animate objects Consider the following sentences: (13.22)

a. Ese chico dejó a todo el mundo sorprendido/asustado/callado. (SLPs) ‘That guy left everyone amazed/scared/quiet/pleased.’ b. *Ese chico hizo a todo el mundo sorprendido/asustado/callado/ encantado. ‘That guy made everyone amazed/scared/quiet/pleased.’ (13.23) a. La peluquera dejó a mi mamá muy bonita/fea/atractiva. (SLPs) ‘The hairdresser left my mom very pretty/ugly/attractive.’ b. *La peluquera hizo a mi mamá muy bonita/fea/atractiva. ‘The hairdresser made my mom very pretty/ugly/attractive.’ (13.24) a. *Juan dejó a su hijo más inteligente/guapo/alto/tímido. (ILPs) ‘Juan left his son more intelligent/handsome/tall/shy.’ b. *Juan hizo a su hijo más inteligente/guapo/alto/tímido. ‘Juan made his son more intelligent/handsome/tall/shy.’ The acceptability judgments on these examples suggest that causative hacer cannot be used in this type of constructions with animate objects  SLP/ILP.7 By contrast, causative dejar is compatible with SLPs but not with ILPs. We can account for these facts by appealing to the semantics of causative hacer and dejar. As noted above, causative hacer refers to a whole event, an event which the causer initiates as an agent, and, therefore, it cannot be used with stage level predicates that refer to states which cannot be brought about at will by an agentive participant. In the (a) sentences in (13.22)–(13.23) the implication is that there is a situation previous to the event denoted in the sentence in which the subject participant was involved. Given its extended reference, causative dejar can accommodate a resultative interpretation, and the sentences are, therefore, acceptable. We argue that it is due to the lack of extended reference frame in the semantics of causative hacer that the corresponding sentences with hacer (the [b] sentences in 13.22–13.23) are ill-formed.

Carmen Ruiz-Sánchez 291

This account of the respective behaviour of causative dejar and hacer in small clauses is further supported by examining other structures. The sentences in (13.22) contain SLPs that are past participle forms of a corresponding middle reflexive verb (sorprendido ‘surprised’ from sorprenderse ‘become surprised’, asustado ‘scared’ from asustarse ‘become scared’, and callado ‘quiet’ from callarse ‘become quiet’). If instead of past participial SLPs we use the verbs, the sentences become well-formed, as shown in (13.25):8 (13.25)

Ese chico hizo a todo el mundo sorprenderse/asustarse/callarse. ‘That guy made everyone become surprised/scared/quiet.’

Here, ese chico ‘that guy’ is the initiator and agent of everyone becoming surprised, scared or quiet. There is a beginning and an end to the causative event. There is no implication of previous intention on the part of the subject, nor any implication that some event or state was already underway previous to the causative event. Aranda (1990) views this difference between causative dejar and hacer in the following way: dejar has a perfective value whereas hacer has a progressive value and represents the process from the perspective of its realization in which the subject is the starting point. Regarding the ILPs in the examples in (13.24), both (a) and (b) examples show that the two verbs in question cannot be used with ILPs (with animate object). This is arguably due to the fact that ILPs refer to permanent and unalterable qualities of the animate participants that cannot be brought about or be allowed to be brought about through human agency. As a result, they are incompatible with hacer, which must have an agentive subject. More particularly, given that in small clauses causative dejar refers to the resultant state of the event, it is incompatible with predicate adjectives denoting permanent features that do not change over time or can appear as a resultant state. However, if we assume that the subject is a creature with supernatural powers, the sentence with hacer becomes grammatical: Dios hizo al hijo de Luis inteligente/guapo/alto/tímido ‘God made Juan’s son intelligent/handsome/tall/shy’. Therefore, the previous statement should be redefined in the following way: hacer can take ILPs as long as the interpretation of the sentence does not conflict with our understanding of how the world typically runs. Small clauses with inanimate objects Consider the following sentences: (13.26)

a. La costurera dejó el vestido ancho/estrecho/grande/pequeño. (SLPs) ‘The seamstress left the dress loose/tight/big/small.’ b. La costurera hizo el vestido ancho/estrecho/grande/pequeño. (ILPs) ‘The seamstress made the dress loose/tight/big/small.’

292 Causative hacer and dejar

(13.27) a. Juan dejó la casa fea/bonita/limpia/desordenada. (SLPs) ‘Juan left the house ugly/pretty/clean/messy.’ b. Juan hizo la casa fea/bonita (ILP) /*limpia/*desordenada. (SLPs) ‘Juan made the house ugly/pretty/clean/messy.’ (13.28) a. Marta dejó la puerta abierta/cerrada/ manchada. (SLPs) ‘Marta left the door open/closed/stained.’ b. *Marta hizo la puerta abierta/cerrada/ manchada. (SLPs) ‘Marta made the door open/closed/stained.’ (13.29) a. *El profesor dejó su charla interesante/ entretenida/aburrida/original. (ILPs) ‘The teacher left his/her talk interesting/entertaining/ boring/original.’ b. El profesor hizo su charla interesante/ entretenida/aburrida/ original. (ILPs) ‘The teacher made his/her talk interesting/entertaining/ boring/original.’ The (a) examples in (13.26)–(13.28) reveal that causative dejar is also compatible with SLP small clauses with an inanimate object and incompatible with ILPs for the same reason as discussed above. As for causative hacer, it is incompatible with SLPs with inanimate objects because the use of these predicates (13.26–13.28) implies a previous state that was subject to some type of modification. The implication with hacer would be that the objects were created with those qualities, but these qualities are transitory, and do not admit such an interpretation. Causative hacer, however, is compatible with ILPs and inanimate object to the extent it can be created by the subject, as in the examples (13.26b), (13.27b) (with bonita ‘pretty’ and fea ‘ugly’), and (13.29b). To summarize, dejar takes SLPs in small clauses with both animate and inanimate objects, but cannot take ILPs. Hacer, on the other hand, cannot take SLPs regardless of the type of object, and occurs with ILPs mainly when the direct object is inanimate, as well as when the direct object is animate in those in cases in which the subject of the sentence can be interpreted as the creator of the object in question.

Similarities in the relation of causative hacer/dejar and ser/estar We suggest that the semantics of causative dejar and hacer in small clauses is analogous to the semantics of the copula verbs estar and ser, especially with regard to the type of adjectives they take and the implications involved. Clements (1988) argues that the basic distinction between the two Spanish copular verbs has to do with whether a connection to a locus or another

Carmen Ruiz-Sánchez 293

situation is implied or not. He notes that estar implies such a connection ([nexus]), while ser does not ([nexus]). The same distinction is applicable to causative dejar and hacer. Causative dejar would then be [nexus] while causative hacer implies no such thing and would thus carry the [nexus] feature. This distinction explains why causative dejar [nexus], like copular estar, can occur with SLPs adjectives (13.22a, 13.23a, 13.28a) but not with ILPs adjectives (13.24a, 13.29a) and why causative hacer [nexus], like copular ser, can occur with ILPs adjectives (13.29b, the ungrammaticality of 13.24b is explained at the end of this section) but not with SLPs adjectives (13.22b, 13.23b, 13.28b). More examples with a wider variety of each class of adjectives are given below in order to provide more evidence to this similarity between dejar/hacer and estar/ser: (13.30) a. Juan dejó el televisor apagado/encendido/roto/cubierto/sucio/ polvoriento. (SLPs) ‘Juan left the TV set off/on/broken/covered/dirty/dusty.’ b. El televisor está apagado/encendido/roto/cubierto/sucio/polvoriento. (SLPs) ‘The TV set is off/on/broken/covered/dirty/dusty.’ c. *Juan hizo el televisor apagado/encendido/roto/cubierto/sucio/ polvoriento. (SLPs) * ‘Juan made the TV set off/on/broken/covered/dirty/dusty.’ d. *El televisor es apagado/encendido/roto/cubierto/sucio/polvoriento. (SLPs) ‘The TV set is off/on/broken/covered/dirty/dusty.’ (13.31) a. Juan dejó la mesa arañada/coja/preparada/puesta/vacía/impecable. (SLPs) ‘Juan left the table scratched/wobbly/ready/set/empty/ impeccable.’ b. La mesa está arañada/coja/preparada/puesta/vacía/impecable. (SLPs) ‘The table is scratched/wobbly/ready/set/empty/impecable.’ c. * Juan hizo la mesa arañada/coja/preparada/puesta/vacía/impecable. (SLPs) ‘Juan made the table scratched/wobbly/ready/set/empty/ impeccable.’ d. *La mesa es arañada/coja/preparada/puesta/vacía/impecable. (SLPs) ‘The table is scratched/wobbly/ready/set/empty/impeccable.’ (13.32) a. *Juan dejó la mesa metálica/moderna/plegable/segura. (ILPs) ‘Juan left the table metallic/modern/folding/secure.’ b. *La mesa está metálica/moderna/plegable/segura. (ILPs)

294 Causative hacer and dejar

‘The table is metallic/modern/folding/secure.’ c. Juan hizo la mesa metálica/moderna/plegable/segura. ‘Juan made the table metallic/modern/folding/safe.’ d. La mesa es metálica/moderna/plegable/segura. (ILPs) ‘The table is metallic/modern/folding/secure.’ (13.33) a. *Juan dejó el pastel español/ enorme/nupcial/ligero ‘Juan left the cake Spanish/huge/wedding/light.’ b. *El pastel está español/ enorme/nupcial/ligero ‘The cake is Spanish/huge/wedding/light.’ c. Juan hizo el pastel español/ enorme/nupcial/ligero ‘Juan made the cake Spanish/huge/wedding/light.’ d. El pastel es español/ enorme/nupcial/ligero ‘The cake is Spanish/huge/wedding/light.’

(ILPs)

(ILPs) (ILPs) (ILPs) (ILPs)

In those cases in which estar or ser are acceptable, creating minimal pairs, Clements notes that ‘the speaker can choose to frame a given situation according to an individual norm, which corresponds to estar, or a class norm, corresponding to ser’. Some adjectives such as alto ‘tall’, ancho ‘wide’, fea ‘ugly’ and bonita ‘pretty’ permit this freedom of choice. In the feature system proposed by Clements, such adjectives carry a [␣resultative] feature, sensitive to the presence or absence of the [nexus] feature of the copula. In its presence [␣resultative] becomes [resultative]. That is, the adjective receives a resultative reading (an individual-norm reading) if it appears with the [nexus] copula (that is, estar). If it appears with the [nexus] copula (that is, ser), the absence of [nexus] supports a [resultative] reading (a classnorm reading). To illustrate, consider the examples in (13.34): (13.34) a. El edificio está bonito. ‘The building is (i.e. looks) pretty.’ (in reference to an existing assumption on the part of the speaker that in a previous state it did not look as pretty as it now does [an individual norm reading  stage-level interpretation of the adjective]) b. El edificio es bonito. ‘The building is pretty.’ (in reference to the class of buildings to which this particular building would be likely to be compared for beauty [a class-norm reading  individual-level interpretation of the adjective]) While sentence (13.34a) implies the prior existence of another state of the building known by the speaker, the speaker would say sentence (13.34b) if s/he were comparing the building in question to other buildings of its class. These two types of reading are, mutatis mutandis, also available with this same adjective class when they appear with causative dejar and hacer.

Carmen Ruiz-Sánchez 295

In examples (13.26a) and (13.28b) the adjectives ancho, estrecho, grande, pequeño are used both with dejar and hacer depending on whether they are interpreted as SLPs or ILPs. The sentence la costurera dejó el vestido ancho/ estrecho/grande/ pequeño ‘The seamstress left the dress loose/ tight/ big/ small’ implies that the dress was already made and the seamstress made some alterations to it, giving a resultative reading to the sentence, which corresponds to the semantics of SLPs. The sentence la costurera hizo el vestido ancho/estrecho/ grande/pequeño ‘The seamstress made the dress loose/ tight/ big/ small’ implies that the dress was made, from beginning to end, by the seamstress, and it does not allow a resultative reading, which corresponds to the semantics of ILPs and which is also true in the corresponding sentences with ser el vestido es ancho/estrecho/grande/pequeño. Adjectives that are hybrid SLPs/ILPs are arguably so because they carry the feature [␣Resultative]. The adjectives fea, bonita ‘ugly, pretty’ in (13.27) belong to this same class of adjectives that have two possible readings, and therefore they allow both dejar and hacer, as well as estar and ser.

Conclusion The results of this study confirm the initial hypothesis that causative hacer and dejar have different semantic implications. In addition, we have accounted for their different behaviours in small clauses and establish some interesting parallels between hacer and ser, and dejar and estar. As main verbs and as causatives in infinitive constructions as well as in small clauses, dejar and hacer refer to different extensions of a given event. Main verb and causative hacer refer to the whole event, as in hacer la comida ‘to make the food’, hacer entrar ‘to make enter’ or hacer la charla interesante ‘to make the lecture interesting’. This is captured in Figure 13.1. With activities, causative hacer refers to the beginning of the event, as in hacer arder ‘to set fire’, which is graphically represented in Figure 13.2. Main verb and causative dejar (Figure 13.3) refer not only to a whole event but also what immediately precedes it, as in dejar entrar ‘to let [someone] enter’ or dejarse el bolso ‘to forget one’s purse’.

Figure 13.1 Reference frame of hacer

296 Causative hacer and dejar

Figure 13.2 Reference frame for causative hacer with activities

Figure 13.3 Reference frame for main verb and causative dejar

However, the implications connected with causative dejar are more complex. The semantics of this verb can also imply an event or state that already exists, as in (13.9a), (13.11a) or (13.13a). This is graphically represented in Figure 13.4, where the arrow indicates the point of reference of causative dejar and the oval encompasses the event or state that is implied. In small clauses, causative dejar implies the existence of an entire event or state the result of which is referred to by the dejarsmall clause complex. This is shown graphically in Figure 13.5. The present study offers a deeper understanding of the different semantic implications that causative constructions with dejar ‘let’ and hacer ‘make’ have in Spanish. Since most of the discussion of Spanish causatives has been centred around hacer, this study is a significant contribution to the current knowledge about the semantics of dejar as a causative verb and about its contrast with the default causative hacer. A logical extension of the present study would be to examine in depth examples involving causative dejar and hacer with inanimate subjects. In addition, a similar analysis can be done for other Romance languages which have causative verbs comparable to Spanish hacer and dejar, such as French faire and laisser and Italian fare and lasciare. We may find that the causative verbs of these languages have the same or similar semantic implications as Spanish hacer and dejar.

Carmen Ruiz-Sánchez 297

Figure 13.4 Implied reference frame of causative dejar in certain cases

Figure 13.5 Reference to the result of an event in small clauses with an implied event or state

Notes 1 Hernanz (1982: 272–3) distinguishes three meanings of dejar: permitir ‘to allow’, no impedir ‘not impede’ and causar deliberadamente ‘cause deliberately’. He considers dejar a causative verb only when it cannot be exchanged by the other two or some other synonymous expressions. This third use of dejar as a causative is only exemplified in Lo dejé caer con todas mis fuerzas ‘I threw it with all my strength’. 2 For present purposes of accounting for the causative meaning of dejar, meanings of the verb with non-animate subjects were not included. Thus, we will not discuss uses of dejar with inanimate source or instrument subjects, while acknowledging that such a study would be highly interesting. As is apparent from the examples below, causative hacer and dejar with inanimate subjects have different syntactic behaviours that interact with negation, which would be a promising topic for future research. (i) a. El jarrón de la mesa no me deja ver la tele. ‘The table vase doesn’t let me see the television.’ b. *El jarrón de la mesa no me hace ver la tele. ‘The table vase doesn’t make me see the television.’ c. *?El jarrón de la mesa me deja ver la tele. ‘The table vase lets me see the television.’ d. *El jarrón de la mesa me hace ver la tele. ‘The table vase makes me see the television.’

298 Causative hacer and dejar (ii) a. El olor a huevo podrido me hace vomitar. ‘The smell of rotten egg makes me vomit.’ b. *El olor a huevo podrido me deja vomitar. ‘The smell of rotten egg lets me vomit.’ (iii) a. El ruido no me deja dormir. ‘The noise doesn’t let me sleep.’ b. *?El ruido me deja dormir. ‘The noise lets me sleep.’ c. *El ruido no me hace dormir. ‘The noise doesn’t make me sleep.’ d. ?El ruido me hace dormir. ‘The noise makes me sleep.’ 3 The examples analysed in this chapter are formulated by the author. They are all well-contextualized in an attempt to show that they are authentic and likely to appear in those situations. However, the analysis of a corpus of spoken Spanish would have contributed to yield a more authentic picture of causative constructions. 4 An interrogation mark together with an asterisk (*?) is used to express that the sentence could be regarded as grammatical and be given an interpretation, but that it would not be the natural one in the context in question. Three interrogation marks on a sentence indicate that we cannot know whether the sentence is well- or ill-formed, as the context does not provide enough information to determine that. 5 For further information about the criteria for these verb categories, see Vendler (1967), which is the main source for our classification of the infinitives and its arguments, Comrie (1976) and Yoon (2001). 6 See Yoon (2001) for a semantic-functional study of small clauses in Spanish. 7 Yoon (2001) considers the sentence (31) Juan hizo más educado a su hijo an example of the use of the causative verb hacer with an SLP, but I question the grammaticality of a sentence like this. 8 Encantado ‘charmed, pleased, enchanted’ has not been included because it is not the past participle of a reflexive verb, the infinitive being encantar ‘charm, please, enchant’ and not encantarse.

References Aranda, A. (1990) La expresión de la causatividad en español actual. Zaragoza: Libros Pórticos. Cano Aguilar, R. (1977) ‘Las construcciones causativas en español’, Boletín de la Real Academia Española, vol. 67, pp. 221–58, 323–51. Clements, J. C. (1988) ‘The Semantics and Pragmatics of the Spanish COPULA  ADJECTIVE Construction’, Linguistics, vol. 5, pp. 779–817. Comrie, B. (1976) Aspect. Cambridge: Cambridge University Press. Curnow, T. J. (1994) ‘Semantics of Spanish Causatives Involving “hacer” ’, Australian Journal of Linguistics, vol. 13, pp. 165–84. Hernanz Carbo, M. L. (1982) El infinitivo en español. Bellaterra: Universidad Autónoma de Barcelona, Departamento de Filología Hispánica. Hopper, P. J. and S. A. Thompson (1980) ‘Transitivity in Grammar and Discourse’, Language, vol. 56, pp. 251–99.

Carmen Ruiz-Sánchez 299 Shibatani, M. (1976) ‘The Grammar of Causative Constructions: A Conspectus’, in M. Shibatani (ed.), The Grammar of Causative Constructions. New York: Academic Press, pp. 1–40. Tesnière, L. (1959) Eleménts de syntaxe structurale. Paris: Klincksieck, pp. 260–82. Vendler, Z. (1967) Linguistics in Philosophy. Ithaca, NY: Cornell University Press. Yoon, J. (2001) ‘Small Clauses in Spanish: A Semantic-Functional Approach’, Indiana University dissertation. Bloomington, IN.

This page intentionally left blank

Index acquisition of language, lexicon/grammar and 1–2 adjectives copula choice in predicate adjective construction 161–2, 197–9; animate referent 174–83; directionality of underlying processes and events according to the animacy of the referent 171–4; inanimate referent 183–92; individual-level and stage-level predicates 166–8; role of subject referent and adjective 165–92; second-order entity as subject referent 192–7; semantic characterization of ser and estar 162–5; state types and time stability 168–71 position 203, 215–16; adjectives with professions 220–2; noun semantics and 219–20, 230–1; studies 208–14, 222–30; word length 203–8 adjunct position, semantic constraints on small clauses in 265–7, 274–5 definiteness of objects 270–1 stage-level predicates/verbs v. individual-level predicates/ verbs 267–70 transitivity hypothesis and adjunct predicates 271–4 affectedness of object 91–3 agency in a subject 9, 10–19 gustar-type verbs 88–91 two-argument constructions with postverbal subject 14–19 unaccusative constructions 10–14 Anacoluthon 35 antipassive constructions 246–8 aspect, se as aspectual marker 251–6 Austronesian languages, ergativeabsolutive marking in 116 case grammar 93 causative hacer and dejar 278, 295–6

contexts contrasting 282–7 general characterization 287–9 semantic aspects of non-causative hacer/dejar 279–81 similarities with ser/estar 287–95 small clauses 289–92; animate objects 290–1; inanimate objects 291–2 terminology of study 278–9 cohesion in discourse, discoursepragmatic viewpoint and 27–8 contrast, subject position and 38–9 copular verbs (ser/estar) causative hacer and dejar compared with 287–95 choice in predicate adjective construction 161–2, 197–9; animate referent 174–83; directionality of underlying processes and events according to the animacy of the referent 171–4; inanimate referent 183–92; individual-level and stage-level predicates 166–8; role of subject referent and adjective 165–92; second-order entity as subject referent 192–7; semantic characterization of ser and estar 162–5; state types and time stability 168–71 dative alternation 207 possessive 151–8 subjects of gustar-type verbs 93–7 definiteness 10–11, 19–23 non-anaphoric se co-varying with 256–8 null pronominalization and 135, 139–47 semantic constraints on small clauses in adjunct positions and 270–1 dependency grammar 82 discourse-pragmatic viewpoint subject order relative to verb 120

301

302 Index discourse-pragmatic viewpoint – continued word order in Spanish and 23–46; discourse cohesion 27–8; emphasis 28–9; focus and subject position 38–46; old/given information 25–7; thematic structure of Spanish 29–38, 47; theme and other discourse-pragmatic functions 24–9, 46 distance, functional approaches to noun referentiality and 56–7, 62–4 emphasis, discourse-pragmatic viewpoint and 28–9 ergative-absolutive marking 116, 124–8, 129–30 estar see copular verbs extraposition 37 focus subject position and 38–46 functional approaches functional sentence perspective 25 noun referentiality 55–8, 74–5; discussion 64–74; distance/recency approach 56–7, 62–4; episodic approach 57–8, 62–4; nominal devices for new and old referrals 60–2; results of study 58–60; study 58–60 generative grammar 94, 95, 119 gustar-type verbs 80–1, 108–9 dative subjects 93–7 direct construction v. inverse construction 97–101 indirect object v. direct object 101–8 Transitivity Hypothesis and 81–93; affectedness and individuation of object 91–3; kinesis, aspect and punctuality 86–8; participants of the verb 82–6; volition and agency 88–91 hanging-topic construction 33, 36–7 heavy noun phrase shift 207 impersonal constructions 248–9 inalienable possession, transitivity and 151–8

individuation of object intransitivity 3, 9

91–3

left-dislocation 33, 37–8 locative verbs 15, 16–17 negation, double 9 noun phrases 53–4 adjective placement and noun semantics 219–20, 230–1; adjectives with professions 220–2; study 222–30 functional approaches in the study of noun referentiality 55–8, 74–5; discussion 64–74; distance/recency approach 56–7, 62–4; episodic approach 57–8, 62–4; nominal devices for new and old referrals 60–2; results of study 58–60; study 58–60 heay noun phrase shift 207 Spanish syntax and 2 switch reference 55–6 null pronominalization 4, 53, 54, 134–6, 148–9 definiteness and 135, 139–47 marking the partitive 136–9 topicalization and 147–8 Transitivity Hypothesis and 135, 139–47 object(s) 115, 129–30 affectedness 91–3 definiteness, semantic constraints on small clauses in adjunct positions and 270–1 doubling 127–8, 147–8 evidence of nominative-accusative or ergative-absolutive marking in rule applications 127–8 indirect object v. direct object 101–8 individuation 91–3 null 4, 53, 54, 134–6, 148–9; definiteness and 135, 139–47; marking the partitive 136–9; topicalization and 147–8; Transitivity Hypothesis and 135, 139–47

Index 303 object(s) – continued primary–secondary object marking in mono- and di-transitive clauses 122–6; ergative–absolute marking patterns in pronominal system 124–6; full NP objects 123–4 split nominative–accusative v. ergative–absolutive marking 116 old/given information, discoursepragmatic viewpoint and 25–7 partitive 135, 136–9 passive clauses 85 evidence of nominative-accusative or ergative-absolutive marking in rule applications 127 possession dative possessive 151–8 inalienable, transitivity and 151–8 Prague School 25 predicate adjective construction, copula choice in 161–2, 197–9 animate referent 174–83 directionality of underlying processes and events according to the animacy of the referent 171–4 inanimate referent 183–92 individual-level and stage-level predicates 166–8 role of subject referent and adjective 165–92 second-order entity as subject referent 192–7 semantic characterization of ser and estar 162–5 state types and time stability 168–71 professions, adjectives with 220–2 pronouns 4, 53 ergative–absolute marking patterns in pronomial system 124–6 functional approaches in the study of noun referentiality 55–8, 74–5; discussion 64–74; distance/recency approach 56–7, 62–4; episodic approach 57–8, 62–4; nominal devices for new and old referrals 60–2; results of study 58–60; study 58–60

null pronominalization 4, 53, 54, 134–6, 148–9; definiteness and 135, 139–47; marking the partitive 136–9; topicalization and 147–8; Transitivity Hypothesis and 135, 139–47 reduplication 32 switch reference 55–6 transitivity and non-anaphoric se 236–40, 241, 258–9; decreasing transitivity 242–50; impersonal se 248–9; increasing transitivity 250–8; se as aspectual marker 251–6; se as valency-reducer 242–8; se co-varying with definiteness/referentiality 256–8 psychological reaction verbs 15, 17, 80–1, 96 see also gustar-type verbs recency, functional approaches to noun referentiality and 56–7, 62–4 reflexive pronouns 236–7 non-anaphoric se 236–40, 241, 258–9; decreasing transitivity 242–50; impersonal se 248–9; increasing transitivity 250–8; se as aspectual marker 251–6; se as valency-reducer 242–8; se co-varying with definiteness/referentiality 256–8 relational grammar 93, 95 relativization 85 semantics-based approaches to syntax 1 adjective placement and 219–20, 230–1; adjectives with professions 220–2; study 222–30 causative hacer and dejar 278, 295–6; contexts contrasting 282–7; general characterization 287–9; semantic aspects of non-causative hacer/dejar 279–81; similarities with ser/estar 287–95; small clauses 289–92; terminology of study 278–9 semantic characterization of ser and estar 162–5

304 Index semantics-based approaches to syntax – continued semantic constraints on small clauses in adjunct position 265–7, 274–5; definiteness of objects 270–1; stage-level predicates/verbs v. individual-level predicates/verbs 267–70; transitivity hypothesis and adjunct predicates 271–4 Spanish word order 9–10, 46; agency in a subject 9, 10–19, 46; definiteness of subject 10–11, 19–23, 46; discourse-pragmatic viewpoint 23–46 ser see copular verbs small clauses causative hacer and dejar in 289–92; animate objects 290–1; inanimate objects 291–2 semantic constraints on small clauses in adjunct position 265–7, 274–5; definiteness of objects 270–1; stage-level predicates/verbs v. individual-level predicates/verbs 267–70; transitivity hypothesis and adjunct predicates 271–4 stage-level restrictions on verbs 268–70 subject(s) agency in 9, 10–19; gustar-type verbs 88–91 dative subjects of gustar-type verbs 93–7 definiteness of 10–11, 19–23 focus and subject position 38–46 functional approaches in the study of noun referentiality 55–8, 74–5; discussion 64–74; distance/recency approach 56–7, 62–4; episodic approach 57–8, 62–4; nominal devices for new and old referrals 60–2; results of study 58–60; study 58–60 ordor relative to verb 116–22 role of subject referent and adjective in ‘copula  adjective’ construction 165–92; animate referent 174–83; directionality of underlying processes and events according to the animacy of the referent 171–4; inanimate referent 183–92; individual-level

and stage-level predicates 166–8; second-order entity as subject referent 192–7; state types and time stability 168–71 subject reference in Spanish 53–4 switch reference 55–6 syntax 1 Spanish see word order in Spanish themes discourse-pragmatic functions 24–9, 46 thematic hierarchies 21 thematic structure of Spanish 29–38, 47 Topic Continuity Theory 84 topicality hierarchies 21 topicalization 36 null pronominalization and 147–8 transitivity 240–2 cardinal 140 inalienable possession and 151–8 non-anaphoric se and 236–40, 241, 258–9; decreasing transitivity 242–50; impersonal se 248–9; increasing transitivity 250–8; se as aspectual marker 251–6; se as valency-reducer 242–8; se co-varying with definiteness/ referentiality 256–8 overt v. null pronominalization and 4 Transitivity Hypothesis 2–4 adjunct predicates and 271–4 gustar-type verbs 81–93, 104, 108; affectedness and individuation of object 91–3; kinesis, aspect and punctuality 86–8; participants of the verb 82–6; volition and agency 88–91 inalienable possession and 153, 157, 158 null pronominalization and 135, 139–47 unaccusative constructions 10–14, 246 definiteness of subject(s) and 20 Unaccusativity Hypothesis 9, 11, 14, 46; two-argument constructions with postverbal subject 14–19 urgency 35

Index 305 valency grammar 82 verbs causative hacer and dejar 278, 295–6; contexts contrasting 282–7; general characterization 287–9; semantic aspects of non-causative hacer/dejar 279–81; similarities with ser/estar 287–95; small clauses 289–92; terminology of study 278–9 gustar-type 80–1, 108–9; dative subjects 93–7; direct construction v. inverse construction 97–101; indirect object v. direct object 101–8; Transitivity Hypothesis and 81–93 order relative to subject 116–22 se as aspectual marker 251–6 semantic constraints on small clauses in adjunct position 265–7, 274–5; definiteness of objects 270–1; stage-level predicates/verbs v. individual-level predicates/verbs

267–70; transitivity hypothesis and adjunct predicates 271–4 Spanish syntax and 2 stage-level restrictions 268–70 syntactic propitious and meaning 1 unaccusative constructions 10–14 see also copular verbs; intransitivity; transitivity volition, gustar-type verbs 88–91 word length, adjective position and 203–8 word order in Spanish 2, 7–9, 46 adjective position 203, 215–16; adjectives with professions 220–2; noun semantics and 219–20, 230–1; Studies 208–14, 222–30; word length 203–8 semantic factors 9–10, 46; agency in a subject 9, 10–19, 46; discoursepragmatic viewpoint 23–46 subject order relative to verb 116–22 transitivity and inalienable possession 151–8