OUTSTANDING DISSERTATIONS IN LINGUISTICS
Edited by Laurence Horn Yale University
A ROUTLEDGE SERIES
OUTSTANDING DIS...
189 downloads
839 Views
1MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
OUTSTANDING DISSERTATIONS IN LINGUISTICS
Edited by Laurence Horn Yale University
A ROUTLEDGE SERIES
OUTSTANDING DISSERTATIONS IN LINGUISTICS LAURENCE HORN, General Editor THE EFFECTS OF DURATION AND SONORITY ON CONTOUR TONE DISTRIBUTION A Typological Survey and Formal Analysis Jie Zhang EXISTENTIAL FAITHFULNESS A Study of Reduplicative TETU, Feature Movement, and Dissimilation Caro Struijke PRONOUNS AND WORD ORDER IN OLD ENGLISH With Particular Reference to the Indefinite Pronoun Man Linda van Bergen ELLIPSIS AND WA-MARKING IN JAPANESE CONVERSATION John Fry WORKING MEMORY IN SENTENCE COMPREHENSION Processing Hindi Center Embeddings Shravan Vasishth INPUT-BASED PHONOLOGICAL ACQUISITION Tania S.Zamuner VIETNAMESE TONE A New Analysis Andrea Hoa Pham ORIGINS OF PREDICATES Evidence from Plains Cree Tomio Hirose CAUSES AND CONSEQUENCES OF WORD STRUCTURE Jennifer Hay THE PHONETICS AND PHONOLOGY OF GUTTURALS A Case Study from Ju’hoansi Amanda Miller-Ockhuizen TRUE TO FORM Rising and Falling Declaratives as Questions in English Christine Gunlogson
iii
PHONOLOGICAL AUGMENTATION IN PROMINENT POSITIONS Jennifer Smith CATEGORY NEUTRALITY A Type-Logical Investigation Neal Whitman THE UPS AND DOWNS OF CHILD LANGUAGE Experimental Studies on Children’s Knowledge of Entailment Relations Andrea Gualmini MARKEDNESS AND FAITHFULNESS IN VOWEL SYSTEMS Viola Miglio THE SYNTAX-INFORMATION STRUCTURE INTERFACE Evidence from Spanish and English Eugenia Casielles-Suarez
THE SYNTAXINFORMATION STRUCTURE INTERFACE Evidence from Spanish and English
Eugenia Casielles-Suarez
ROUTLEDGE New York & London
Sections 3. 3 . 1–3 . 3. 3 have been published with the title “Left-dislocated Structures in Spanish” in Hispania 86.2, 326–338 (2003). They are reprinted here by permission of Hispania. Published in 2004 by Routledge Taylor & Francis Group 270 Madison Avenue New York, NY 10016 www.routledge-ny.com Copyright © 2004 by Taylor & Francis Group, a Division of T&F Informa. Routledge is an imprint of the Taylor & Francis Group. This edition published in the Taylor & Francis e-Library, 2005. “To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.” All rights reserved. No part of this book may be reprinted or reproduced or utilized in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Catalog record is available from the Library of Congress
ISBN 0-203-50171-3 Master e-book ISBN
ISBN 0-203-58077-X (Adobe eReader Format) ISBN 0-415-97095-4 (Print Edition)
A mis padres, Cristina y Gonzalo
A Mark
A Alfonso
Contents
PREFACE
ix
ACKNOWLEDGEMENTS
xi
CHAPTER 1:
INTRODUCTION
3
CHAPTER 2:
TOPICS
15
2.1
Introduction
15
2.2
Brief Historical Survey
16
2.3
Analysis of Topic Characterizations
22
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS
53
3.1
Introduction
53
3.2
Topical Phrases
54
3.3
Topicalizing Constructions
71
3.4
Towards a Typology of Topics
88
CHAPTER 3:
CHAPTER 4:
BARE NOMINALS
101
4.1
Introduction
101
4.2
A Syntactic View
103
4.3
A TFAView
119
4.4
Towards a TFA-syntactic View
121
FOCUS
125
5.1
Introduction
125
5.2
Focus, Accent, New Information and Sentence-final Position
126
CHAPTER 5:
viii
5.3
Types of Focus
132
5.4
The Projection of Focus
162
5.5
Some Conclusions
164
FOCUS PREPOSING
171
6.1
Introduction
171
6.2
FP as a Topical Focus
172
6.3
FP as a Nontopical Focus
181
INFORMATION STRUCTURE AND SYNTACTIC STRUCTURE
193
7.1
Introduction
193
7.2
The Topic-Focus Articulation
194
7.3
The TFA and Syntactic Structure
209
7.4
Overall Conclusions
213
REFERENCES
219
INDEX
229
CHAPTER 6:
CHAPTER 7:
Preface
This book is a revised and updated version of my 1997 University of Massachusetts doctoral dissertation. I have maintained the original division into seven chapters, although each chapter has been revised and bibliographically updated. Thus, this study makes reference to important works published after 1997 (Chomsky 2001, Culicover and McNally 1998, Gómez-González 2001, Hajičová, Partee and Sgall 1998, Jacobs 1999, Kiss 1998, Lee 2000, Prince 1998, and Zubizarreta 1998, among others), as well as my own research on this topic since 1997 (Casielles 1998, 1999, 2000a, 2000b, 2001, 2003a and 2003b). I have modified the structure of the introductory chapter and rewritten parts to match changes introduced in later chapters. Chapters two, three and five contain the most substantial revisions, which include new sections not present in the original dissertation. In chapter 2, I have added a historical survey which includes a compilation and discussion of the original and exact definitions of topic offered by authors such as Weil, Mathesius, Firbas, Halliday, Contreras, Chafe, Hockett, Gundel, and Reinhart, among others (section 2.2). I have also modified the rest of the chapter and structured the analysis of topic characterizations into discourse (2.3.1) and syntactic characterizations (2.3.2). In chapter three, I have rewritten sections 3.2 and 3.3, and I have added a new section (3.4) where I suggest giving up on the traditional notion of topic and pursuing a feature-based typology. I have also added a new section in chapter 5 (5.3.6), and rewritten section 6.3.2 in chapter 6, and sections 7.3 and 7.4 in chapter seven. This work has benefited from presentations at the 1997 VII Coloquio de Gramática Generativa at the University of Oviedo, the 1997 First Hispanic Linguistics Colloquium at Miami University at Oxford, Ohio, the 1998 Second Hispanic Linguistics Symposium at the Ohio State University, and the 2000 XXVI Annual Meeting of the Berkeley Linguistics Society at the University of California, Berkeley.
x PREFACE
I would like to thank my colleagues at Wayne State University, and acknowledge financial support provided by this university: a Minority/Women Summer Grant in 1999, a Research and Inquiry Grant in 2000, and a Faculty Research Grant in 2001. My friends and family have also been a constant source of support. My deepest thanks go to my parents, Cristina and Gonzalo, for teaching me the most important things in life; to my dearest Dani, Cris, Nacho, Pedro, Miguel, Gon, Marta, and Gonzalo for their encouragement, understanding and affection; and very especially, to Mark, whose support, patience and love have made the completion of this work possible. For the last two years there has been somebody in my life who has not actually helped me with this project. However, from his birth on December 11, 2001 my son Alfonso has filled my life with happiness and thus deserves my warmest appreciation. Eugenia Casielles-Suarez February 22, 2004
Acknowledgments
In 1991, I came to Amherst as an exchange student to continue with my studies in linguistics and spend a year at the University of Massachusetts (which at the time seemed like a long time to me). Although I was quite lost at the beginning, I was extremely happy to have the opportunity of attending classes at one of the best linguistic departments in the world. I never thought that what started as a one year experience would change my life both professionally and personally. Yes. I decided to stay and do a Ph.D. here, at the University of Massachusetts, and after almost six years, now that I consider Amherst my home in the U.S. (which, being a graduate student, of course coincides with the time one must leave), it is time to thank all the wonderful people that made this possible, with whom I have had the pleasure of sharing this experience and without whose help, support, and friendship, I would not have been able to finish my graduate studies. First of all, I would like to thank the members of my dissertation committee, namely, (in alphabetical order) Harold Boudreau, Barbara Partee, Peggy Speas and Juan Zamora for their understanding, flexibility and willingness to read drafts in short periods of time. Although I was not very excited about being required to take a nonlinguistic class, I have to thank Harold Boudreau for showing me the beauty of literary criticism and reminding me about the existence of other interesting fields, besides linguistics. In addition to his superb teaching, he has given me invaluable and detailed comments on this manuscript for which I am very thankful. It is a especially difficult task to express how grateful I am to Barbara Partee. I know that no matter how hard I try, I am going to fall short of expressing the depth of my gratitude. Everybody knows Barbara is a superb researcher, excellent teacher and wonderful person. In addition, she is an outstanding mentor and advisor. Without Barbara’s always insightful suggestions, challenging questions, constant support and
xii ACKNOWLEDGMENTS
encouragement, I would not have finished this dissertation. Barbara always found time to meet with me, read my work and comment on it, in spite of her busy schedule. Not only has she advised me academically, but also professionally and personally. She believed in me, even when I did not, and reassured me in those times when I thought I would give up. Her support and understanding was crucial at a particularly difficult time in my fifth year. For everything she has done for me, I am deeply grateful. How can I thank the person that, when I thought I only had dispersed ideas and fragmented chapters, but not a dissertation, told me that I was wrong, that I did have a dissertation and showed me what it was? That is what Peggy Speas did for me and it was only the beginning. Since then, Peggy has helped me by giving me invaluable comments on several drafts of this dissertation. Although she is an expert in asking the toughest questions, she would always give me an idea of where to look for the answers. Peggy has been crucial not only for my academic work but also my mental health. She is one of the most supportive, helpful and thoughtful persons I know and I am very grateful for her encouragement and unconditional support through all these years. I have been very lucky to have had Juan Zamora as my advisor, teacher and friend from the very beginning. Juan is one of those professors that cares about all his students as a father would, and makes sure we have everything we need. Juan invited me to give my first talk and has not stopped helping me since then. He very patiently listened to all my problems at long lunches and always gave me excellent advice. He has always been available when I have needed him. Juan has taught me a lot about the history of linguistics, dialectology and bilingualism. Even more importantly, he has taught me a lot about life. I would also like to thank all faculty in the Linguistics and the Spanish and Portuguese departments for contributing to create a very friendly atmosphere and promoting true collaboration between the two departments. I have benefited greatly from attending classes taught by Emmon Bach, Hagit Borer, Harold Boudreau, Michael Fast, Lyn Frazier, Roger Higgins, Kyle Johnson, Angelika Kratzer, Barbara Partee, Lisa Selkirk, Peggy Speas, Tom Roeper and Juan Zamora. Besides the members of my committee, Emmon Bach, Kyle Johnson and Tom Roeper deserve special mention. I was very lucky to have had the opportunity of attending Emmon’s introduction to semantics class in my first semester, just before he retired. Since then, Emmon has been a constant source of wisdom, encouragement and support.
ACKNOWLEDGMENTS xiii
Kyle has always been extremely generous with his time. He has provided me with invaluable advise with my research. I remember being stuck with a paper (which now is part of chapter 4 of this dissertation) and he took the trouble to meet with me for lengthy sessions every week for several weeks. He has been very encouraging, supportive, and a lot of fun to learn with. I cannot thank Tom enough for helping solve what looked like an insurmountable problem that was going to affect the nature and even existence of this dissertation. He is a wonderful person and has given me invaluable advice. Before I came to the University of Massachusetts, I had read in several UMass dissertation acknowledgments about the wonderful and friendly atmosphere that both faculty and students had created at South College. Now, I know they were not exaggerating. The linguistics department at UMass is for sure one of the best places in the world to learn and do linguistics, and meet wonderful people. I would also like to thank two professors at the University of Oviedo who taught me about linguistics: Luis Escribano and Ana Ojea. Ana Ojea deserves special thanks. I was very fortunate to have her as my advisor. She helped me become more interested in linguistics and suggested that I should apply to the exchange program between the University of Oviedo and the University of Massachusetts. I will always be grateful to her for that and for being a wonderful advisor, teacher and friend. Also at the University of Oviedo I am thankful to a wonderful group of linguists in the Spanish department; specially to José Antonio Martínez and Serafina García with whom I have had great linguistic and non-linguistic conversations and shared so many wonderful moments in Oviedo and Amherst. Back at the University of Massachusetts, I would like to thank several wonderful women that have been responsible for the functioning of the department and the mental health of all its members; I am grateful to Narly Bedoya, Emily Kozodoy, Fran Taylor-Anderson and Jennifer Whiting in the Spanish and Portuguese department, and Kathy Adamczyk and Lynne Ballard in the Linguistics department for their invaluable work, help, patience and understanding all these years. Different parts of this dissertation have benefited from conversations with Emmon Bach, Hagit Borer, Heles Contreras, Lyn Frazier, Mercedes González, Roger Higgins, Bart Hollebrandse, Kyle Johnson, Winnie Lechner, Ana Ojea, Barbara Partee, Gemma Rigau, Tom Roeper, Hotze Rullmann, Jeff Runner, Peggy Speas and Satoshi
xiv ACKNOWLEDGMENTS
Tomioka. Also, I would like to thank the audiences at the XXIV Symposium on Romance Languages (UCLA-USC), the Workshop on Focus at the University of Massachusetts-Amherst, and the VII Coloquio de Gramática Generativa at the University of Oviedo for their comments and suggestions on several parts of this work. Special thanks to Mike W.Dickey for proofreading this thesis. Being a graduate student is not always easy and there are many people that have helped me greatly by creating a friendly atmosphere. First, I would like to thank all graduate students in both the Linguistics and Spanish and Portuguese departments, too numerous to name them all. Thanks to the cheerful crew from the Spanish department: César Alegre, Hilton Alers, Isabel Álvarez, Carolina Alzate, Francisco Borge, Ana Bugallo, Teresa Cabal, Sunny Cabrera, Molly Falsetti, Alina Franco, Javier Fruns, Janine Fuller, Gabriel García-Bajo, Michele Martin, Gloria Prosper, Gemma Reguera, Emma Rivera-Rábago, Iñaqui Rodeño, Carlos Rodríguez, Mercedes Valle, Michael Vrooman and Julia Wells. I remember many pleasant and fun moments at dinners, parties and barbecues, organized by graduate students in the Linguistics department, that I shared with John Alderete, Jill Beckman, José Benki, Laura Benua, Isadora Cohen, Pat Deevy, Mike and Laura W.Dickey, Mercedes González, David Holton, Bart Hollebrandse, Winnie Lechner, Janina Rado, Maribel Romero, Ed Rubin, Jeff Runner, Amy Schafer, Bernhard Schwarz, Rachel Thorburn, Satoshi Tomioka, Suzanne Urbanczyk, Elisabeth Villalta, Angeliek van Hout and Jochen Zeller. Special thanks go to the Spanish cooking team: Isabel Álvarez, Maribel Romero, Mercedes Valle and Elisabeth Villalta. Tom, Suzanne, and Michael Ready, and María, Juan Carlos and Paula Flores have also been great friends, with whom I have shared many wonderful moments. My life in Amherst would have not been the same without the wonderful times shared with my dearest girlfriends: Isabel Álvarez, Mercedes González, Raquel Medina, Dorrit Posel, Mercedes Valle, and Angeliek van Hout. Raquel deserves special mention. In addition to sharing with me her very special views on the notion of “topic,” she has cheered me up, put up with me, listened to all my problems (many times a day, sometimes) and given me all her support and wonderful advice. My neighbor and wonderful friend Ellen Landis has also been very supportive all these years and has always cheered me up. I am very grateful to her for her invaluable friendship.
ACKNOWLEDGMENTS xv
I remember many fun moments shared with a group of dearest friends we could refer to as “the Ale House team” (not a cooking team, exactly): Bart, Mark, Winnie, Raque, Dorrit, Ed, and Jochen. We have had wonderful times together. Special thanks to Bart and Winnie who were especially supportive and sweet to me when I needed it most. It has not been easy to be away for so long from Asturias and its people. The unforgettable moments I spent in Avilés, Salinas, Gijón and Oviedo gave me the energy to come back to my work and keep going. I would like to thank great friends that shared with me these wonderful summers in Asturias during these years: Olga & Julio, Ana, Sera, José Antonio, Cadín and Mateo. Very special thanks go to my dearest friends and daily companions in Avilés with whom I have spent so many intense days (and wild, long summer nights): Achim, José Luis, Paco and Tere. Five years ago, on May 17th, my friend Reyes convinced me to go with her to Mike’s Westview Café on a Sunday afternoon. I will be eternally grateful to her for that, since that is how I met Mark. How could I express my gratitude to Mark? He is the most wonderful person I have ever met and one of the reasons I decided to come back and stay in Amherst. Mark has supported and helped me in every possible way for five years. He has given me all his friendship, understanding, love, and affection. He has changed my life and I will always be grateful to him. I would also like to thank Rosalie Keller and Arthur Geetersloh, whose love and support have been so important to me through these years. Finalmente, me gustaría dar las gracias a las dos personas que más admiro y de quienes he aprendido las cosas más importantes: mis padres, Cristina y Gonzalo. Ellos han respetado siempre todas mis decisiones académicas y personales y me han ofrecido su apoyo incondicional todos estos años. Junto a ellos, están los que han compartido conmigo los momentos más importantes de mi vida y me han soportado, apoyado y mimado: Dani, Gon, Marta, Cris, Nacho, Miguel y Pedro. Sin su cariño, nada tendría sentido.
xvi ACKNOWLEDGMENTS
THE SYNTAX-INFORMATION STRUCTURE INTERFACE
OUTSTANDING DISSERTATIONS IN LINGUISTICS
2
CHAPTER 1 Introduction
This work investigates the syntax-information structure interface with particular reference to Spanish and English. There has been a renewed interest in recent linguistic studies within the framework of generative grammar in the role that notions such as topic and focus play in the syntax of various languages (i.e. Culicover and McNally 1998, King 1995, Reinhart 1995, Rizzi 1997, Soh 1998, Zubizarreta 1998, etc.).1 These proposals make use of topic/focus phrases, features, criteria or constraints, depending on the particular theoretical framework. In addition, several works have suggested the introduction of a whole new informational component into the model: Vallduví’s (1990) Informatics, which adds another branch to the T-model, Lambrecht’s (1994) Information Structure, which is proposed in the framework of Construction Grammar, and Erteschik-Shir’s (1997) F-structure, which replaces LF.2 However, this current investigation of the relationship between information structure, expressed in dichotomies such as Topic-Focus or Theme-Rheme, and syntactic structure encounters a great difficulty: the elusiveness of concepts such as topic and focus and the terminological chaos that surrounds them. Sometimes different terms are used to mean the same thing, and sometimes the same term is used to mean different things. Despite very useful discussions of these terms (particularly in Erteschik-Shir 1997, Lambrecht 1994, Prince 1981b, Reinhart 1981, 1995, and Vallduví 1990), linguists interested in investigating the Topic-Focus articulation (TFA)-syntax interface can only expect to find different and contradictory definitions of terms such as topic, theme, open proposition, background, comment, rheme, focus, etc., which appear scattered in a theoretically heterogeneous mixture of phonological, syntactic, semantic, pragmatic and discourse studies.3
4 SYNTAX-INFORMATION STRUCTURE INTERFACE
The following quote from Vallduví (1990) gives a picture of the terminological problem: An extreme case of misguiding terminology is found in the literature on Philippine languages, where, according to Herring (1990:fn1) “topic” designates the focus and “focus” designates the topic. Just in case the reader has lost count of the different types of topic and focus, here is a brief list: discourse-topic, Givón topic, sentence-initial topic (or topic - topic, if we may put one over the others) and Prague-topic (= topic - topic + part of the comment), and AI-focus (= Givón-topic), focus of new information (or focus - focus), and focus of contrast (= sometimes a focus-focus and sometimes a topic-topic). (Vallduví 1990:46) This study will start by offering a detailed analysis of the different notions of topic and focus and how they combine to form an information articulation. In addition, it will investigate the syntax of “topical” and “focal” elements with particular reference to Spanish and English, and more generally the relation between information structure and syntactic structure. Spanish contrasts sharply with English in the way these structures interact. First, in Spanish, there are two different positions available for subjects, which correlate with their informational status: while topical subjects appear in a preverbal position, focal subjects appear postverbally: (1)
(2)
¿Qué sabes de Juan? “What’s up with John?” ¿Quién ha llamado? Who has called?
Juan acaba de llamar “John has just called” Ha llamado Juan Has called John “JOHN has called”
That is, while in Spanish the informational difference is expressed syntactically, through two different positions (preverbal vs. postverbal), in English the difference is expressed intonationally. Since postverbal subjects in Spanish and other Romance languages are necessarily focal, any proposal about the licensing of these postverbal subjects should also account for their interpretation.4 Second, it is very common in Spanish to place a topical phrase in sentence-initial position, as in (3).5
INTRODUCTION 5
(3)
¿Quién comió las fresas? “Who ate the strawberries?”
Las fresas las comí yo the strawberries cl. ate I “I ate the strawberries”
As is well-known, English also has topicalizing devices such as Topicalization in (4)a, and Left-dislocation in (4)b: (4)
a.Julia I couldn’t reach b.John, I saw him yesterday
However, there are important differences between these structures and the construction in (3), known since Cinque (1990) as Clitic left-dislocation (CLLD). Notice the double achievement of (3). In addition to marking the direct object as topical (by placing it in sentence-initial position), it also marks the subject as focal, by placing it in sentence-final position. From this point of view, (3) looks like a direct reflection of the order Topic-Focus, which in the context given (where the direct object is topical and the subject focal) results in the order dislocated direct object-clitic-verb-subject. This is not possible in English. That is, topicalizing devices such as those in (4) cannot place a focal subject in sentence-final position, as the Spanish construction in (3) does. Furthermore, an additional interesting feature of CLLDs, not shared by English preposing mechanisms, is that they can place in sentence-initial position a variety of phrases: (5)
(6)
(7)
A Pedro los libros ya se los compré to Pedro the books already cl cl bought-1sg “I already bought the books for Pedro” A mí listo no me lo parece to me clever not cl cl seem “He doesn’t seem clever to me” Libros yo a Juan nunca se los dejaría books I to John never cl cl lend-1sg “I would never lend Juan books”
CLLDs pose very interesting questions both from a syntactic and pragmatic point of view. From a syntactic point of view, it brings up the well-studied, although not yet solved, issues having to do with the origin (base-generated vs. moved) and position (adjunction to CP, IP,
6 SYNTAX-INFORMATION STRUCTURE INTERFACE
etc.) of these topical phrases, the analysis of the clitic, and the position and licensing of the postverbal subject, among others. From a pragmatic point of view, we still do not know the differences regarding the contextual felicitousness between topicalizing strategies such as English Topicalization, English Left-dislocation, Romance Clitic Left-dislocation, Right-dislocations, and possibly many others. Third, in addition to this placement of topical elements in sentence-initial position through a Clitic Left-dislocation, it is also relatively productive in Spanish, as opposed to English, to place focused elements also in sentence-initial position in what is called Focus Preposing structures.6 (8)
EN PRIMAVERA visitó Juan Leningrado in spring visited-3sg John Leningrad “IN SPRING John visited Leningrad” (Hernanz and Brucart 1987)
This is the opposite of what we saw in (3), where the focal element appeared in sentence-final position. In this case the displacement reverses the unmarked Topic-Focus order. From a pragmatic point of view, we would like to know the significance of this reversed order, and from a syntactic point of view, the nature of this preposing, and more generally the relation between focus and certain syntactic positions. Notice that, as (9) shows, Focus Preposing in Spanish triggers subject-verb inversion, as opposed to its English counterpart in (10): (9)
*EN PRIMAVERA Juan visitó Leningrado
(10)
I think he was Japanese. No—KOREAN she was (Ward 1988:114)
Finally, topic and focus interact in very interesting ways in the distribution of Bare Nominals (BN) in Spanish. Thus, while postverbal BN subjects are allowed, preverbal BN subjects are ungrammatical: (11)
(12)
Jugaban niños en la calle played-3pl children in the street “Children were playing in the street” *Niños jugaban en la calle children played-3pl in the street “Children were playing in the street”
INTRODUCTION 7
Interestingly, preverbal dislocated topics (subjects and nonsubjects) are allowed: (13)
(14)
Estudiantes no creo que vengan students not think-lsg that come-3pl “Students, I don’t think they will come” Café no creo que tengan coffee not think-lsg that have-3pl “Coffee, I don’t think they have it” (Contreras 1986)
The first interesting question these examples pose concerns the nature of the preverbal subject position in Spanish. It has been proposed (Contreras 1991, among others) that the subject position in Spanish is not a specifier (Spec, IP), but an adjoined position, Although this proposal is appealing in many ways, the behavior of BNs does not favor that analysis. If both the dislocated and the subject positions are adjunctions, why is it that BNs are allowed in the former but not in the latter? The subject position seems to have certain requirements (probably having to do with agreement/case/specificity) that a Spanish BN cannot fulfill. Therefore, saying that both the dislocated and the subject positions are adjunctions does not help much, and it seems that, at least in order to account for the data above, the distinction between Spec, IP and an adjoined position is a reasonable one, In addition to the specific position of preverbal subjects, the contrast between preverbal BN subjects and BN dislocated topics brings up a second question which has to do with topics. Considering that preverbal subjects in Spanish are topical (as opposed to postverbal subjects, which are new to the discourse and focused), there seems to be a difference between this type of topics and the ones in (13)–(14): only the latter allow for BNs. Does the same notion of topic apply to both preverbal subjects and dislocated elements? The third question has to do with the fact that dislocated BN topics do not trigger the appearance of a clitic as in the case of nonbare nouns like the one in (3). (15)
Libros no traje books not brought-1sg “Books, I didn’t bring (them)”
8 SYNTAX-INFORMATION STRUCTURE INTERFACE
This brings up the question of the relationship between clitics and determiners (Uriagereka 1991, 1995) and the difference between these dislocated BNs and cases of English Topicalization, where no resumptive pronoun is present either. Let us now have a look at the interpretation of these nominal phrases. BNs in Spanish do not have the same possibilities regarding their interpretation that English BNs show. BNs in English can have an existential and a generic reading: (16)
Children were playing in the park
(existential)
(17)
Children (generally) play in the park
(generic)
From the point of view of Diesing’s (1992) mapping hypothesis, existentially interpreted BNs in English have to lower at LF inside VP in order to be mapped into the nuclear scope and be bound by existential closure, while generic BNs just stay in Spec, IP and from there they are mapped into the Restrictor and get bound by a generic operator. The existential interpretation of BNs is possible in Spanish. However, the subject has to be postverbal: (18)
*Niños jugaban en el parque
(19)
Jugaban niños en el parque
This shows us that, as opposed to English where the BN can move at LF and get the right interpretation, in Spanish it is syntactically placed at the right LF position and no LF lowering seems possible. If this lowering were possible, (18) should be able to get the existential interpretation. However, this cannot be the whole story. What prevents (20) from being grammatical with a generic reading? (20)
*Niños (generalmente) juegan en el parque
There is no LF lowering involved here. However, the sentence is ungrammatical. As is well known, the generic reading is expressed in Spanish with a definite DP. (21)
Los niños (generalmente) juegan en el parque
INTRODUCTION 9
However, to say that (20) is ungrammatical because only definite DPs can be generic in Spanish is not very satisfying. It does not capture any relationship between (18) and (20). Is the ungrammaticality of these two instances of preverbal BN subjects totally unrelated? We would like to offer a unified account for both and derive the fact that in Spanish the generic is expressed with a DP as a consequence. Finally, preverbal (nondislocated) subjects seem to improve considerably if they are prosodically focused: (22)
LANGOSTAS destruyeron las cosechas grasshoppers destroyed-3pl the crops “GRASSHOPPERS destroyed the crops”
(23)
ESCLAVOS construyeron las pirámides slaves built-3pl the pyramids “SLAVES built the pyramids” (Contreras 1986)
As opposed to (18) where no LF-lowering is possible, in (22)–(23) focus seems to license LF-lowering in Spanish. We will investigate the relation between these focused subjects and cases of Focus Preposing such as the one we saw in (8). In (24–(25) we have more examples: (24)
DINERO necesitan los estudiantes money need-3pl the students “MONEY students need”
(25)
DE DOS PARTES consta el examen of two parts consist-3sg the exam “TWO PARTS the exam consists of” (Hernanz and Brucart 1987)
These cases have been analyzed as cases of A’ movement, similar to wh-movement (Cinque 1990, Hernanz and Brucart 1987). Notice that in these cases there is no resumptive pronoun and, as mentioned above, inversion is triggered, just as with wh-movement. In order to account for all this data from Spanish, we will have to consider the differences between preverbal topics and preverbal subjects, the difference between preverbal and postverbal subjects, and the difference between focused and nonfocused phrases. This requires
10 SYNTAX-INFORMATION STRUCTURE INTERFACE
an investigation into the notions topic and focus, and an analysis of different topicalizing and focusing mechanisms which can place phrases in sentence-initial position, such as Clitic Left-dislocation, Left-dislocation, Topicalization and Focus Preposing. This first chapter has introduced the main goals of this work. The rest of the book is organized as follows. Chapter 2 investigates the notion of topic. After a brief historical survey, which offers some of the definitions available in the literature, section 2.3 considers in detail different discourse and syntactic characterizations of topics, which relate this concept to other notions such as aboutness, discourse-oldness, subject, sentence-initial position, etc. This detailed examination of the different conceptions shows that due to the variety of phonological, morphological, lexical and/or syntactic ways in which languages can mark, if at all, topical elements, any specific definition of topic based on a certain language will not be adequate to characterize topical elements in a different language. In order to understand the role of this concept in a particular language, we have to take into account the specific characteristics of the language, and examine the ways in which that language encodes nonfocal phrases. Chapter 3 examines topical phrases in Spanish and English, and considers several topicalizing devices available in these languages, We will start by considering the nature of topical phrases in section 3.2. As we will see, in Spanish, topical elements can be null or overt. If overt, they can appear in argument position or dislocated. From this perspective, the question is not anymore whether (all) topical phrases are in a certain position or whether they have certain discourse requirements, but rather how different types behave and what their nature is. Section 3.3 examines several topicalizing mechanisms: Clitic Left-dislocation, Pronoun Left-dislocation, Clitic Left-dislocation with no overt clitic, Topicalization, and Right-dislocation. Having considered in detail the features of topical phrases, I come back in 3.4 to the notion topic and tentatively propose how we could solve the problem of characterizing topics crosslinguistically. I suggest giving up on the traditional view of topic and I argue that particular combinations of certain phonological, syntactic, semantic and discourse features will give us different types of topical elements. From this perspective, the long-term goal is to examine in detail how different languages encode nonfocal elements, discover which are the relevant features, and how they combine, and ultimately establish a feature-based typology of topics.
INTRODUCTION 11
Two topical elements can be immediately distinguished. One, which we can refer to as the Sentence-Topic proper (STopic) shows the following features: a) it is restricted to a unique element; b) it correlates with a sentence-initial position (often a preverbal subject); c) it seems to be restricted to discourse referents; d) it is not necessarily discourse-old; and e) it is not necessarily unaccented. The other one, which we can refer to as the Background, shows exactly the opposite features: a) it is not restricted to a unique element; b) it does not correlate with a particular sentence position; c) it is not restricted to discourse referents; d) it is necessarily discourse-old and e) it is necessarily unaccented. Chapter 4 offers more evidence for the existence of these two elements by examining the (mis)behavior of BNs in Spanish. First, I consider the distribution of BNs from a syntactic point of view and suggest that in Spanish so-called NP-movement is restricted to DPs and that BNs, which are assumed not to be DPs, but NPs, cannot DP-move out of the VP. This explains the fact that they cannot appear in Spec, IP (which is argued to be a DP position in Spanish). It also accounts for their interpretative possibilities: the fact that they cannot have a generic reading like their English counterparts. I show that none of the grammatical cases of BNs (postverbal, dislocated and focused BNs) involve DP movement. Second, the data is considered from a TFA point of view. I suggest that the division of topical elements into STopic and Background explains the otherwise mysterious contrast between grammatical dislocated BNs and ungrammatical preverbal subjects. I propose that this division is syntactically expressed in Spanish and that dislocated phrases are Background (unfocused) elements rather than sentence topics. From this point of view, the contrast can be described as follows: BNs cannot be sentence topics. Third, I suggest how we could reconcile these two views if we assume that in Spanish DP movement is in fact a topic-feature checking mechanism. Further, I connect the position of dislocated elements with their Background (unfocused) nature. Following Reinhart (1995) and Zubizarreta (1998), I suggest that their dislocation has to do with having them escape VP: the focus domain which the focus feature can reach. Chapter 5 concentrates on the notion of focus. Focal phrases are also shown to be heterogeneous from the point of view of their syntactic and discourse features. We can have single (narrow) or multiple (wide) focal phrases; pronominal or lexical; discourse-old or discourse-new; and in a variety of sentence positions. Intonationally, however, a focal element (or a part of it, if it is wide) needs to be accented. Nevertheless, in the case of focus, we face a different problem, which has to do with
12 SYNTAX-INFORMATION STRUCTURE INTERFACE
the fact that too many different types of focus have already been suggested in the literature: a) presentational vs. contrastive (Rochemont 1986); b) psychological, semantic and contrastive (Gundel 1994); c) metalinguistic (Erteschik-Shir 1997); d) retrieve-add vs. retrieve-substitute (Vallduví 1990); e) sentence-, predicate- and argument-focus (Lambrecht 1994); and f) identificational vs. information focus (Kiss 1998). I consider these classifications and I suggest that, although a division such as the one in Vallduví (1990) and Kiss (1998) could be useful, it should not be interpreted as implying the existence of two kinds of focus. I argue that the different features that these two foci show are not due to their intrinsic different nature, but to their position in the sentence. Interestingly, focal phrases placed in a preverbal position share a number of syntactic and discourse features with the Sentence-Topic mentioned above. In chapter 6, a special focusing construction is considered: Focus Preposing. I examine different analyses of this construction and show that this type of preposing exemplifies a Focus-Background structure where the focus is marked as unambiguously narrow and the rest of the sentence as unambiguously nonfocal. The similarities between this structure and the case of focused subjects is then examined. Further, I consider some language-particular features of this construction, and I examine some similarities between focus preposing and wh in situ, on the one hand, and focus in situ and wh-movement, on the other hand. In Chapter 7, I examine the structure of the TFA and its relation with syntactic structure. First, I concentrate on the TFA structure (how topic and focus are put together). I consider the different proposals for the articulation of information structure, These include theories that propose that information is articulated as: a) one dichotomy (most authors); b) two different dichotomies (Dahl 1974a, Jacobs 1984 and Krifka 1991); c) a trichotomy (Vallduví 1990); and d) independent relations, such as the Topic relation or the Focus relation (Lambrecht 1994). I suggest that in the same way there is no unique syntactic structure to which all sentences conform, there is no unique articulation that can capture the information structure of different sentences. Sentences will have different information structures depending on the nature of their topical and focal elements. Although complex sentences probably involve embedded and/or superimposed complex information structures, we will see that for simple declarative sentences, two different information structures are shown to be relevant: Sentence Topic-Focus (STopic-Focus), and Background-Focus (or FocusBackground). While the STopic-Focus articulation is pragmatically,
INTRODUCTION 13
syntactically and phonologically unmarked, the Background-Focus is pragmatically marked, with languages varying in its syntactic or phonological special marking. Second, I examine the syntax of the TFA in Spanish. I show that in this language the two different information structures mentioned above (STopic-Focus and Background-Focus) can be read off the syntax. The STopic-Focus articulation, which is pragmatically unmarked (it can be uttered out of the blue), is also syntactically and phonologically unmarked both in English and Spanish. It involves a preverbal subject-predicate structure, and it is expressed through unmarked rightmost focus-related accent. However, an important difference between these two languages should be pointed out. Since Spanish has two other options for subjects that are not being introduced as sentence topics (null and postverbal subjects), a preverbal subject will be necessarily interpreted as the Sentence-Topic. English, however, which lacks null and postverbal subjects, requires subjects to be preverbal for independent reasons. Therefore, in English, preverbal subjects are informationally quite heterogeneous and cannot be automatically interpreted as sentence topics. The Background-Focus articulation, which has very specific contextual restrictions, is typically marked phonologically in English. The unique focal element is accented in situ (resulting in nonrightmost focus-related accent) and all the Background elements (the rest of the sentence) get de-accented. However, in Spanish this articulation also has its own syntax. Phrases considered to be part of the background can be base-generated outside the VP, appearing in left-or right-dislocated positions. This syntactic detachment not only marks these elements as Background elements, but equally important, by removing all these elements from the VP (the domain of focus projection), it also marks the element left in the VP as the unique focal element, without any special intonation. Thus, I argue that in Spanish there is a direct correlation between certain syntactic positions and certain interpretations as follows. While an overt DP in the preverbal specifier subject position is interpreted as the sentence topic in a STopic-Focus articulation, one or more left- or right-dislocated phrases in adjoined positions are interpreted as part of the Background in Background-Focus/Focus-Background structures. Finally, in the last section I offer some conclusions.
14 SYNTAX-INFORMATION STRUCTURE INTERFACE
NOTES 1. For Optimality theory works which consider the role of topic-focus see Choi (1999, 2001), Costa (2001), Grimshaw and Samek-Lodovici (1998), Lee (2000) and the references therein. 2. The other two connections—the syntax-semantics interface and the relationship between information structure and semantic representation have also been explored. For instance, a mapping has been proposed between the syntactic structure and the semantic representation (Diesing 1992). For the semantics-information structure interface see Fintel (1994), Krifka (1991), Partee (1991), and Rooth (1985), among others. 3. The elusiveness of concepts such as topic and focus may have contributed to their remaining quite peripheral in Chomsky’s work. Thus, although Topic-Comment structures are mentioned in Chomsky (1965) and Focus-Presupposition structures are considered in detail in Chomsky (1971), these notions have never occupied a central role in his theory. In Chomsky (1995), the significance of information structure is nevertheless acknowledged: “Notice that we are sweeping under the rug questions of considerable significance, notably questions about what in the earliest EST framework were called ‘surface effects’ on interpretation. These are manifold, involving topic-focus and theme-rheme structures, figure-ground properties, effects of adjacency and linearity, and many others.” (Chomsky 1995:220). Although not examined in detail, discourse-related properties are also mentioned in Chomsky (2001:11) in conjunction with scopal properties, as the kind of properties which motivate the operation internal Merge. 4. For instance, a purely syntactic analysis where these postverbal subjects get nominative case by government (as opposed to the usual spec-head relation that is assumed for preverbal subjects) does not account for their different interpretation. See Olarrea (1996), Ordoñez (2000) and Zubizarreta (1998) for some proposals. 5. Casielles (2000b) considers in detail the treatment of this type of word order variations by some of the most important grammars of the XIX and XX centuries. 6. For a pedagogical examination of the different ways syntactic structure can interact with information structure in Spanish, see Casielles (2003a).
CHAPTER 2 Topics
2.1 Introduction It is well-known that sentences show an informational articulation where there is a less informative part—the Topic or Theme—and a more informative part—the Focus or Rheme. The recognition of this split has taken different forms and the following dichotomies have been suggested: Theme-Rheme, Topic-Comment, Topic-Focus, Focus-Presupposition, Focus-Open Proposition or Focus-Background. As Vallduví (1990) has pointed out, while these proposals agree on the existence of a split inside sentences, they differ as to the point of the split and the kind of this division (continuum/dichotomy). Some of these dichotomies—Theme-Rheme and Topic-Comment—separate the topic or theme from the rest of the sentence. The others separate the focus from the rest of the sentence, which is referred to as presupposition, open proposition or background.1 Even inside the former proposals which separate the topic or theme from the rest of the sentence, there is no agreement as to how to identify the topic. Confronted with this picture, which includes a variety of articulations, terms, and an even greater variety of different conceptions of the same term, linguists, as Polinsky (1999) has recently put it, “appear to have given up on a definition of topic.” The paradoxical result is that the term topic is often taken for granted, as if there were a shared, agreed-upon notion already available. This chapter investigates the notion of topic in detail. I will start by offering a brief historical survey which will show some of the definitions of topic available in the literature. Section 2.3 will then examine the different characterizations and the motivation behind them. I will consider proposals which relate topics to discourse notions such as
16 SYNTAX-INFORMATION STRUCTURE INTERFACE
aboutness, discourse referent, discourse oldness and context-boundness as well as those which impose some syntactic restriction on topics such as being a subject, being in sentence-initial position or having pronominal form. We will see that there is no proposal which can account for all topical phrases. Due to the variety of phonological, morphological, lexical and/or syntactic ways in which languages can mark, if at all, topical elements, any specific definition of topic based on one language will not be adequate to characterize topical elements in a different language. I will suggest that in order to understand the role of this concept in a particular language, we have to take into account the specific characteristics of the language, and examine the ways in which that language encodes “topical” phrases. This will be done with particular attention to Spanish and English in chapter 3. In the meantime I will be using the terms “topical” and “topical phrase” rather vaguely and in a nontheoretical, intuitive sense to refer to nonfocal elements, which are assumed by the speaker to be salient and active in the discourse, and which, therefore, tend to be expressed by pronominal or unaccented lexical phrases. 2.2 Brief Historical Survey The division of the sentence into informational units, distinct from syntactic categories such as subject-predicate, goes back to Weil’s (1879 [1844]) thesis De I’ordre des mots dans les langues anciennes comparées aux langues modernes.2 In this work, Well points out that “the syntactic march is not the march of ideas” and refers to an informational split different from the syntactic division subject-predicate. He suggests that there is “a point of departure, an initial notion which is equally present to him who speaks and to him who hears, which forms, as it were, the ground upon which the two intelligences meet; and another part of discourse which forms the statement (l’énonciation), properly so called. This division is found in almost all we say.” (p. 29) He offers the following examples from Latin and notes that “the point of departure, the rallying point of the interlocutors, is Romulus the first time, Rome the second, and the third time the idea of founding”:
(1)
a. Idem Romulus Romam condidit b. Hanc urbem condidit Romulus
TOPICS 17
c. Condidit Romam Romulus “Romulus founded Roma” (Weil 1879 [1844]:29–30) Thus, he adds, the information, the goal of the discourse, is different in the three utterances. Weil’s pioneering study was soon followed by the crucial work of the Czech linguist Mathesius, who was familiar with Weil’s work and explicitly refers to it.3 Mathesius and other linguists of the Prague Circle focused on the differences between Czech and English and developed a theory of information structure referred to as Functional Sentence perspective (FSP).4 Mathesius (1975 [1961]) points out that theme-rheme should be distinguished from subject-predicate and that although in English the theme tends to be expressed by the grammatical subject and the rheme by the grammatical predicate, this is not the case in Czech.5 In this language, “the theme is placed in the initial section of the sentence, whatever sentence element may express it, the final section of the sentence being reserved for the rheme.” (p. 85). Thus, in addition to (2), which follows the order subject-predicate, we find (3), where the object is placed at the beginning of the sentence and the subject takes the sentence-final position: (2)
Tatínek napsal tenhle dopis “Father wrote this letter”
(3)
Tenhle dopis napsal tatínek this letter wrote father “FATHER wrote this letter” (Mathesius 1975 [1961]:85)
The utterance in (3) would occur in a situation in which the theme is “the letter” and the rheme “father.” In this work, Mathesius defines the theme as follows: “the element about which something is stated may be said to be the basis of the utterance or the theme.” (Mathesius 1975 [1961]:81). However, other definitions can also be found in his work. As pointed out by Daneš (1974a:106), in Mathesius (1939:234), he defines the starting point of the utterance (východisko) as “that which is known or at least obvious in the given situation and from which the speaker proceeds.” In Mathesius (1942) the foundation of the utterance (základ/téma) is defined as something “that is being spoken about in the sentence.”6
18 SYNTAX-INFORMATION STRUCTURE INTERFACE
In addition to this view of theme as a discrete element, we can also find views where thematic and rhematic elements are seen not so much as discrete elements but as gradient elements in a scale. Thus, one of Mathesius’ followers, Firbas (1964:272) introduces the notion of communicative dynamism and defines theme as “the sentence element(s) carrying the lowest degree(s) of CD [communicative dynamism] within the sentence.”7 He adds: “It follows from this definition that the theme need not necessarily convey known information or such as can be gathered from the verbal and situational context. It can convey even new, unknown information.”8 (p. 272). From this perspective, the articulation of information is not viewed so much as a dichotomy but as including elements such as theme, theme proper, transition, rheme and rheme proper. This nondiscrete vision of theme has also been explicitly rejected. Thus, Contreras (1976:16) argues that this notion of communicative dynamism is quite arbitrary and based on Chafe (1974) proposes that “the theme contains those elements which are assumed by the speaker to be present in the addressee’s consciousness.”9 An elaborate conception of theme can be found in Halliday (1967). This study points out that the information structure of the clause has to do with “the relation of what is being said to what has gone before in the discourse, and its internal organization into an act of communication.”10 (p. 199). Halliday offers the following examples, which show that even a language with quite a rigid word order such as English, allows for the thematization of nonsubject elements: (4)
a. John saw the play yesterday b. Yesterday John saw the play c. The play John saw yesterday (Halliday 1967:212)
For Halliday, the theme is what comes first in the clause. It is also what is being talked about; the point of departure for the clause as a message. Halliday distinguishes between theme and given: while given means “what you were talking about” (or “what I was talking about before”), theme means “what I am talking about” (or “what I am talking about now”).11 (p. 212) Thus, the term theme has been conceived both as a gradient or discrete element. As a discrete element, it has been defined in terms of aboutness, as the sentence-initial element and/or as given or salient in the discourse.
TOPICS 19
Regarding the term topic, we have to go back to Hockett (1958), who instead of using theme-rheme, introduces the terms topic and comment. Hockett notes that in English and other languages of Europe, topics are usually also subjects, but that this is not always the case, For instance, he points out that in (5) “that new book by Thomas Guernsey” is spoken first because “it specifies what the speaker is going to talk about: it is the topic of the sentence, though not its subject”12 (p. 201) (5)
That new book by Thomas Guernsey I haven’t read yet
Since then, numerous studies on both sides of the Atlantic have been dedicated to consider topics and their relationship to subject and/or old/given information in a variety of different languages.13 Thus, Gundel (1988:40) proposes that the topic is the expression which names what the sentence is about. It is always associated with the given (nonfocal) information in the sentence, and it never has primary stress.14 It should also be noted that present-day Prague school linguists also use the term topic. Thus, in Sgall et al. (1986:80), topic is defined as “the contextually bound part of the sentence; it corresponds to items the speaker supposes to be activated in the hearer’s memory at a given point of time.”15 Reinhart (1981) offers not only one of the most illuminating discussions about topic, but also a very appealing conception, which more recent proposals have adapted (i.e. Vallduví 1990, Erteschik-Shir 1997). She borrows from Stalnaker (1978) the notion of “context set” understood as “the set of the propositions that we accept to be true” and suggests that a discourse can be described as a joint procedure of constructing a context set.16 In this proposal, sentence topics have to do with the organization of the context set, that is, the propositions admitted into the context set are classified into subsets of propositions, which are stored under defining entries. From this perspective, NP sentence-topics are referential entries under which we classify propositions in the context set.17 In spite of the fact that Reinhart’s proposal and other recent research (i.e. Erteschik-Shir 1997, Lambrecht 1994, Reinhart 1995 and Vallduví 1990) have made available fairly explicit proposals, which make identifying sentence topics a less daunting task, there is still no agreement on the discourse and/or syntactic requirements for topics. Thus, while Reinhart argues against topics as old information, Erteschik-Shir (1997:11) has recently proposed that all topics are old or presupposed. While Vallduví (1990) restricts topics, which he calls
20 SYNTAX-INFORMATION STRUCTURE INTERFACE
links, to the sentence-initial position, Lambrecht (1994) argues against topics as sentence-initial elements.18 Lambrecht (1994) suggests that since topical elements are often encoded in pronominal forms, it makes no sense to restrict them to sentence-initial position. He restricts topics to discourse referents and proposes the following definition: “A referent is interpreted as the topic of a proposition if in a given situation the proposition is construed as being about this referent i.e. as expressing information which is relevant to and which increases the addressee’s knowledge of this referent.” (Lambrecht 1994:131) Thus, as we saw with the notion theme, in this type of articulations, which separate a topic-like element from the rest of the sentence, the term topic has been, and continues to be understood in different and sometimes incompatible ways.19 In (6) we have some of the definitions we have considered: (6) Some definitions of theme/topic • A point of departure, an initial notion which is equally present to him who speaks and to him who hears, which forms, as it were, the ground upon which the two intelligences meet. (initial notion, Weil (1879 [1844]:29) • That which is known or at least obvious in the given situation and from which the speaker proceeds. (východisko, Mathesius 1939:234) • The element about which something is stated may be said to be the basis of the utterance or the theme. (theme, Mathesius 1975 [1961]:81) • The sentence element(s) carrying the lowest degree(s) of CD [communicative dynamism] within the sentence. (theme, Firbas 1964:272) • The theme is what comes first in the clause. The theme is what is being talked about; the point of departure for the clause as a message. (theme, Halliday 1967:212) • The theme contains those elements which are assumed by the speaker to be present in the addressee’s consciousness. (theme, Contreras 1976:16) • That [information] which the speaker assumes to be already present in the addressee’s consciousness at the time of an utterance. (given information, Chafe 1974:111) • [The topic of the sentence] specifies what the speaker is going to talk about. (topic, Hockett 1958:201)
TOPICS 21
• [The topic is] the expression which names what the sentence is about. It is always associated with the given (nonfocal) information in the sentence. Hence it never has primary stress. (topic, Gundel 1988 [1974]:40) • NP sentence-topics, then, will be referential entries under which we classify propositions in the context set. (topic, Reinhart 1981:80) • Topic is the contextually bound part of the sentence; it corresponds to items the speaker supposes to be activated in the hearer’s memory at a given point of time. (topic, present-day Prague school, i.e. Sgall et al. 1986:29) • Within the ground there often is a “special” topic-like element, the link, which appears in sentence-initial position. (link, Vallduví 1990:57) • A referent is interpreted as the topic of a proposition if in a given situation the proposition is construed as being about this referent i.e. as expressing information which is relevant to and which increases the addressee’s knowledge of this referent. (topic, Lambrecht 1994:131) • All topics are old or presupposed. (topic, Erteschik-Shir 1997:11) To complicate matters further, as we mentioned above, some proposals for the articulation of information separate the focus from the rest of the sentence, which is referred to as open proposition, presupposition or background. Despite this terminological variation, the definitions for this “topical” element are less heterogeneous than those we have seen for theme or topic. Here are some of them: (7) Some definitions of open proposition/background, etc. • The presupposition [is] an expression derived by replacing the focus by a variable. (presupposition, Chomsky 1971:100) • The information in the sentence that is assumed by the speaker to be shared by him and the hearer. (presupposition, Jackendoff 1972:16) • The nonfocused part of the sentence. (background, Dahl 1974a:1)20 • A string P is c-construable in a discourse δ if P has a semantic antecedent in δ. (c-construable, Rochemont, 1986:47)21
22 SYNTAX-INFORMATION STRUCTURE INTERFACE
• An open proposition…represents what is assumed by the speaker to be salient in the discourse. (open proposition, Ward 1988:4–5) • The ground is the complement of the focus. (ground, Vallduví 1990:58) As opposed to proposals about theme/topic, in these definitions we do not find a reference to a particular place in the sentence or to the notion of aboutness. However, they do share with definitions of theme/topic references to concepts such as shared knowledge, discourse-old, salient, etc. Regarding the type of data that these proposals are meant to account for, they often investigate sentences involving certain prosodic variations, such as those in (8), rather than sentences involving certain word order variations. (8)
a. JOHN saw Bill b. John SAW Bill c. John saw BILL (Jackendoff 1972:16)
In these type of sentences what is considered to be topical (shared information, given, discourse-old or salient) is the nonfocal part of the sentence. Notice that this topical part can correspond not only to subject +verb chunks, as in (8)c, but also to discontinuous elements as in (8)b. This brief survey has shown that each definition of what counts as a topical element tends to concentrate on certain requirements, which often depend on the language or structure under examination. The problems that these conceptions face will be now considered in detail. 2.3 Analysis of Topic Characterizations There probably would not be much disagreement about the following general characterization: a topic tends to be what the sentence is about; what the speaker takes as the point of departure for the sentence; that which is not part of the focus; placed towards the beginning of the sentence; often a subject; often active or salient in the discourse; and often expressed by pronominal or unaccented lexical phrases.22 The problems arise when we have to decide which of these tendencies
TOPICS 23
are sufficient or necessary to identify the topic of a sentence in a particular language. Many characterizations end up combining some phonological, syntactic, semantic and/or pragmatic features. However, the defining requirement in most cases tends to be either syntactic or discourse-related.23 Therefore, I have divided the different notions to be considered here into discourse and syntactic characterizations. Let us consider the motivation behind the different proposals and the problems they face. 2.3.1 Discourse Characterizations Four different discourse characterizations of topical elements will be considered. They have to do with the following notions: aboutness, discourse referent, discourse-oldness and context boundness. 2.3.1.1 Aboutness The most intuitive notion of sentence topic is probably “what the sentence is about.” As we have seen, this is one of the notions proposed by Mathesius, who views the theme or basis of the utterance as the element about which something is stated.24 It is also present in more recent studies such as Lambrecht (1994). However, it has proved very hard to make this definition into a useful tool in identifying the topic of a sentence, since it is not always easy to find a constituent in a sentence which expresses what the sentence is about.25 Several tests have been devised in order to identify the topic (understood as what the sentence is about), by authors such as Gundel (1988) [1974] and Reinhart (1981). These include the following: (9)
The “As-for” test: an element can be considered a topic, if it can be dislocated and preceded by “as for.” (10) The “What about” test: an element can be considered a topic, if the sentence can answer the question “What about x?,” where x is the topic. (11) The “Said-about” test: an element can be considered a topic, if it can be inserted as x in the frame “S/he said about x that ‘comment’.”
24 SYNTAX-INFORMATION STRUCTURE INTERFACE
Several problems have been pointed out in the literature concerning the validity of these tests, both by their own proponents, such as Gundel, and by other authors such as Prince (1984), Ward (1988) and Vallduví (1988). These tests have been found to be too strong in some instances, since some topicalized phrases fail to pass the tests, and also too weak in other instances, since they can identify as topics too many elements. Regarding the first problem, Ward points out that 69% of the Topicalizations in his corpus fail the “As-for” test and 53% fail the “What-about” test (Ward 1988:19–20). Here is an example that fails the three tests: (12) To illustrate with a simple analogy, consider a person who knows arithmetic, who has mastered the concept of numbers. In principle, he is now capable of carrying out or determining the accuracy of any computation. Some computation he may not be able to carry out in his head. Paper and pencil are required to extend his memory. But the person does not have to learn something new to carry out a more complex computation, using paper and pencil. (Ward 1988:19, taken from Chomsky 1980:221) According to Ward, all the topic tests applied to this example produce infelicitous results: (13) #As for some computation, he may not be able to carry it out in his head (14) Q: What about some computation? A: #Some computation he may not be able to carry out in his head (15) #He says about some computation that he may not be able to carry it out in his head However, this may not be proving that these are not valid topic tests but rather that Topicalizations may be dealing with special topics (perhaps only new topics) or even that this construction is not the preposing of a topic, but of something else. In fact, that is what Ward proposes. He says that the function of Topicalizations (and other preposing mechanisms) in English is not to mark sentence topics but to mark both the preposed element as a Backward Looking Center (BLC), and the
TOPICS 25
Open Proposition (OP) as salient in the discourse.26 If he is correct, and Topicalizations should not be analyzed in terms of the notion of topic, the fact that the tests are not valid for that construction can be used to support his theory, but not to question the validity of the tests. In fact, if he is right, the tests do pretty well in that they do not accept those preposed elements as topics. Regarding the second problem with these tests (their weakness), Vallduví (1990) considers the following example, which does not involve a Topicalization: (16) She told me I needed a change in my life, like getting a new job. It was to no avail. Linguistics fascinated me. Wall Street would have to wait. He says that the tests would identify as a topic more than one element, since both “linguistics” and “me”seem to pass the tests: (17)
(18)
a) I said about linguistics that it fascinated me b) (?) As for linguistics, it fascinated me c) What about linguistics? It fascinated me a) I said about myself that linguistics fascinated me b) As for myself, linguistics fascinated me c) What about myself? Linguistics fascinated me
However, there might not be anything wrong with this, since, as we will see, it has been suggested that sentences can have multiple topics. Although I do not want to imply that these tests are sufficient to identify any topic, they do not seem to fail as badly as has sometimes been suggested. In fact, the infelicitous results shown in (13)–(15) above that these tests have when applied to examples such as (12), seem to be related to the type of NP we have in this example.27 That is, if we change “some computation” into a definite or generic NP, the tests do fine. Take the “As-for” test, for instance: “As for computations…, as for the computations…, as for some of the computations….” So, it looks like these tests may not be good enough for indefinite topics, such as that in (12). However, this does not mean they are not good, valid tests for other topics. Thus, the fact that we may still be lacking the tests that can identify any type of topic, does not mean that we have to forget about the relation between the notion of topic and the concept of aboutness.28
26 SYNTAX-INFORMATION STRUCTURE INTERFACE
2.3.1.2 Discourse Referent Another notion of topic relates the topical element to a discourse referent. Although there are different notions of discourse referent, in the context of information structure, it is used in general to refer to discourse entities.29 Lambrecht (1994), for instance, characterizes the topic as follows: The topic relation is the relation of aboutness between a proposition and a discourse entity. A proposition is interpreted as being about an entity if it is understood as conveying relevant information with respect to this entity, i.e. as increasing the hearer’s knowledge of it. A topic entity must exist in the universe of discourse independently of what is being predicated of it in a given proposition, i.e. it must be a discourse referent (Lambrecht 1994:335) Thus, according to Lambrecht (1994), only referring expressions can be topics. That is why the subjects of (19)–(20) cannot be topic expressions: (19) (20)
It is raining There’s nobody in the room
In addition, he claims that this restriction against nonreferring expressions applies also to indefinites and other quantified expressions. That is the reason, he says, why (21) is bad in German: (21)
*Kein Arzt ist er no doctor is he
However, he does mention that, as noted by Reinhart (1981), universally quantified NPs can sometimes be topics, provided that their referents are coextensive with the entire class designated by the NP. One can imagine a sentence beginning with As for all his friends, they … but hardly *As for some people, they …is explains, he says, why a generic indefinite such as “ein Arzt” is a possible topic expression: (22)
Ein Arzt kann dir nicht helfen A doctor can’t help you
TOPICS 27
One problem for this restricted notion o topic is the fact that some languages like Spanish allow cases such as those in (23)–(27): (23) (24) (25) (26) (27)
De la conferencia no he oído nada of the lecture not have-lsg heard nothing Listo no lo parece clever not cl seems Estudiando nunca està studying never is A todos no los he visto todavía prep all not cl have-1sg seen yet Que fumas lo sabemos todos that smoke-2sg cl know-1pl all
That is, in Spanish almost any category can be placed in sentence-initial position if it is considered to be topical: PP, AdjP, VP, QP, CP, etc.30 If these sentence-initial phrases are topics, they do not seem to refer to any discourse entity. 2.3.1.3 Discourse-oldness One of the most popular characterization of topics equates them to old or given information.31 As we mentioned above, the notion of topic in terms of old information can also be traced back to Mathesius. Thus, in Mathesius (1975) [1961], we can find both a characterization in terms of aboutness that we considered in section 2.3.1.1, and a characterization in terms of givenness in (28). (28) Theme of the utterance: What is given by the context or what naturally presents itself, in short what is being commented upon. (Mathesius 1975:156) However, it is not clear that “given by the context” should be interpreted as meaning context-old. It might be closer to the notion of context-boundness to be examined next. A recent proposal, however, is very clear about requiring topics to be discourse-old. As mentioned above, Erteschik-Shir (1997) has suggested that being old or presupposed is an absolute requirement of topics. This characterization of topic as a discourse-old element faces, as pointed out in the literature (Reinhart 1981, Lambrecht 1994 and
28 SYNTAX-INFORMATION STRUCTURE INTERFACE
Vallduví 1990, among others), two immediate problems. First, being discourse-old does not seem to be sufficient, since, as (29) and (30) show, discourse-old elements can constitute the sole focus of the utterance. (29) Who did they call? (30) Who did you see, David or Tom?
Pat said they called HER I saw TOM
Therefore, the discourse-old vs. discourse-new status of the referent of an expression should be distinguished from its information status, as topical or focal. Second, being discourse old may not be necessary, either. It has been noted that topics can be discourse-new elements. (31) That was a student of mine. Her HUSBAND had a HEART attack (Lambrecht 1994:326) That is, new topics can be introduced into the discourse. These new topics are typically marked in English with Jackendoff’s B-accent, and, as the following examples show, these can also appear in the construction known as Topicalization.32 (32) Do you watch much sports? Not much football. BASEBALL I like a lot BETTER (Ward 1988) (33)
The PLAY John saw YESTERDAY (Chafe 1976:49)
It should be noticed here that these examples not only show evidence of the nonidentity between topical and discourse-old, they also point to a distinction between focused and accented. As Lambrecht has suggested, accent, at least in English, can either mark a topic relation or a focus relation.33 Despite these arguments against equating topical with discourse-old, topics do have certain requirements concerning their discourse-oldness. Since the terms “old” and “new” have been found to be misleading, other terms have been proposed to talk about the context features of referents.34 Prince (1981b) has proposed that entities may be discourse-old/discourse-new (may or may not be already introduced in
TOPICS 29
the current discourse) and hearer-old/hearer-new (may or may not be already present in the hearer’s knowledgestore). Due to the impossibility of something being discourse-old and hearer-new, this classification gives us three different referential status: discourse-new/hearer-new (brand-new), discourse-new/hearer-old (unused) and discourse-old/hearer-old (evoked). Lambrecht (1994) distinguishes the following discourse states of referents: Unanchored (34)
Unidentifiable
Identifiable Anchored
Inactive Accessible Active
Unidentifiable referents are those for which a representation exists only in the speaker’s mind. An unidentifiable referent can be anchored if it is linked by means of another NP, properly contained in it, to some other discourse entity. For example, “a bus” would be unanchored and “a guy I work with” would be anchored.35 On the other hand, identifiable referents are those for which a shared representation already exists in the speaker’s and the hearer’s mind at the time of utterance. Now, having a representation of a referent is different from thinking about it at the time of the utterance and this is where the concept of activation comes in. That is, a hearer is not thinking about all the referents for which he has a representation. Only some of these are active, salient. Lambrecht here follows Chafe (1987:22ff) who distinguishes three activation states: active, semi-active (or accessible) and inactive. Active referents are those which are currently “lit up.” Unaccented referential expressions have active referents, although not all active referents appear as unaccented expressions. Similarly, all pronominal expressions have active referents, but not all active referents are expressed pronominally.36 Semiactive or Accessible referents are in the peripheral consciousness. According to Lambrecht, there are three ways a referent can be made accessible. A referent may become accessible through deactivation from a previous state, that is, by having been active at an earlier point in the discourse. This referent is textually accessible. Alternatively, a referent can be inferentially accessible if it can be inferred from some other active or accessible referent in the universe of discourse.37 Finally, Lambrecht claims that a referent can also be situationally accessible, that is, its salient presence in the text-external
30 SYNTAX-INFORMATION STRUCTURE INTERFACE
world can make it accessible too.38 Accessible referents can be coded either as inactive or as active depending on various factors. They do not have a direct phonological or morphological correlate although they can have an indirect correlate in the syntax. Inactive or unused referents are in a person’s long term memory. They are necessarily relatively prominent prosodically and typically coded in English as definite lexical NPs.39 Now, what is the relationship between these states and the notion of topic according to Lambrecht? For a referent to be interpretable as the topic of a proposition, the topic referent must have certain activation properties. Lambrecht points out: In selecting a topic for a sentence, a speaker makes a communicative decision as to the “point of departure” for the new information, i.e. as to the entity that she wishes to convey information about. But before making this communicative decision, the speaker must make certain hypotheses concerning the status of the referent of the topic in the mind of the addressee at the time of the utterance. On the basis of these hypotheses, the speaker then decides upon the form of the sentence in which the topic is to be coded. (Lambrecht 1994:162–3). That is, for a statement to be relevant with respect to a topic, the topic itself must be of current interest. Therefore, it must either be already established in the discourse or easily relatable to one that is already established. This means that insufficiently accessible topics will tend to be interpreted as ill-formed. Lambrecht establishes a correlation between the states of topic referents and the acceptability of sentences expressed in the form of the following topic acceptability scale (4.34 in his work): (35)
The Topic Acceptability Scale Active Accessible Inactive Brand-new anchored Brand-new unanchored
most acceptable
least acceptable
The preferred topics are active in the discourse since the mental effort necessary to process sentences containing them is not increased by the additional task of accessing the topic referent by retrieving it from long-term memory or by drawing inferences leading to its assessment.
TOPICS 31
Therefore, the preferred topic expression is going to be an unaccented pronominal (or inflectional or zero) morpheme. Thus, even though, as we have seen, topical cannot be equated with discourse-old, topics seem to have certain discourse requirements. This brings us to our next category. 2.3.1.4 Context Boundness According to current Praguian work, topic is the contextually-bound part of the sentence.40 An element is contextually bound “if it is ‘given’ or chosen by the speaker among the items s/he assumes to be easily accessible in the hearer’s memory, i.e., salient, activated above the upper threshold in the stock of shared knowledge.” (Sgall et al. 1986:179).41 In this definition, “given” does not mean discourse-old necessarily. As Hajičová et al, (1998) have recently pointed out, it is not precise to equate the linguistic opposition contextually bound (CB) and nonbound (NB) with the cognitive opposition between “given” and “new” information. They state: The term “contextual boundness” is not to be understood in its etymological sense. First, an item previously mentioned may occur as NB, in the (contrastive) focus, as e.g. in (Did you talk to Jane and Jim?) I only talked to JIM. Second, not only items mentioned in the preceding verbal co-text can occur as CB, but also those referring to entities activated by the situation of the discourse.42 (1998:59) Thus, this notion of context-bound differs from the discourse-old conception just examined and takes into account that not all discourse-old elements are topical and that not all topical elements are discourse-old. Notice that in this theory, topics are not restricted to discourse entities. In fact, they are not restricted to discrete elements. The topic can encompass all the elements that are context bound, including verbs. For instance, Sgall et al. (1986:57) point out that in one of the three readings of (36), as an answer to a question such as “What do beavers build?,” “beavers build” would be considered to be topic, (36)
Beavers build DAMS (Sgall et al. 1986:57)
32 SYNTAX-INFORMATION STRUCTURE INTERFACE
Thus, in that context “the verb has also been activated at the given time-point so that one only states that it is dams that beavers build.”43 Note that this notion is closer to “all the elements in a sentence that are not part of the focus,” than to a particular sentence-initial element or elements. In this sense, this conception of topic is very similar to definitions of Open Proposition/Presupposition/ Background that have been proposed to account for discourse-old, multiple topical components. Even though context-bound cannot be equated with discourse-old, a definition such as the one in Sgall et al. (1986), does not seem to allow for discourse-new topics. As we have seen, although it is not the preferred option for topics, it is possible for a discourse-new element to be introduced as a topic. Thus, in this examination of discourse characterizations of topics we have seen that even though not all topics are discourse-old, topics are most successful when active in the discourse. In addition, sentences can often present topical discourse-old chunks of elements as we saw with “beavers build.” However, this type of nondiscrete unfocal part of the sentence is quite different from the prototypically discrete, sentence-initial topic. One way to resolve this duality is to distinguish two different elements. Dahl (1974a), for instance, suggests that an utterance such as “John drinks beer” would have the following structure, as an answer to the question “What does John drink?”:
(37)
What does John drink?
background|focus John drinks beer topic|comment
That is, on the one hand, “John drinks” is considered the background (discourse-old) and, on the other hand, “John” is considered the topic. This analysis takes into account the fact that neither “John” nor “drink” are part of the focus and at the same time preserves a difference between verbs (which can be part of the Background, but not topics) and NPs (which can be considered to be topics). That is, while “John drinks” would be considered a topic by Prague School scholars, Dahl suggests that even if these two elements share some discourse status, it is useful to distinguish them. From this perspective, “John” can be considered to be a topic (by being in sentence-initial position, the point of departure for the sentence, an NP, even a discourse referent in this case), while it is much harder to view “drink” as a topic in this sense.
TOPICS 33
In chapter 7 we will come back to Dahl’s proposal which consists of two superimposed articulations -Topic-Comment and Focus-Background-, and, therefore, allows for two topical elements.44 A different strategy which also allows for different topical elements is Vallduví’s (1990) hierarchical trichotomy. Vallduví notes a certain redundancy in Dahl’s system in that background and comment overlap. That is, “drink” in (37) is both part of the background and part of the comment. He suggests a tripartite hierarchical structure that separates first the focus from the rest of the sentence, the ground, and then separates the topic, which he refers to as link, from the rest of the ground: the tail. From this perspective one topical element (his link) is restricted to sentence-initial position (see 2.3.2.1 below) while another topical element, his tail, can occur in any position in the sentence. It should be noted that this articulation is not meant to account for the discourse status of verbs, but for non-sentence-initial unfocused elements, such as the right-dislocated element in (38):
(38)
ground focus ground L’amo l’ODIA, el bròquil link tail
“The boss HATES broccoli”
Vallduví’s system will be considered in detail in chapter 7.45 For now, we will just keep in mind the possibility of two topical primitives: one which tends to be discourse-old and might consist of nondiscrete chunks (i.e. “beavers build”) and another discrete element, possibly restricted to a particular place in the sentence. 2.3.2 Syntactic Characterizations As the discussion on discourse characterizations has made clear, some proposals in the literature impose some syntactic restrictions on topics. Thus, topics have been equated with subjects; have been required to appear in the sentence-initial position, and have been correlated with pronominal form. I will now consider each of these correlations.
34 SYNTAX-INFORMATION STRUCTURE INTERFACE
2.3.2.1 Sentence-initial Position Although there probably would not be much disagreement regarding the claim that there is a tendency towards the order topic-focus and, therefore, a tendency for topics to appear sentence-initially, there is no consensus as to how far we want to push this tendency into a necessary and sufficient condition for phrases to be considered topics.46 At one extreme, we find proposals such as Halliday (1967) for whom sentence-initial position is not only necessary but sufficient for a phrase to be considered the sentence topic or theme. That is, for him, all sentence-initial phrases are themes, regardless of their discourse-status. He claims that themes are “given” in the unmarked case, but that they do not necessarily need to be “given,” where “given” is understood as what the speaker or addressee were talking about before and “theme” what the speaker is talking about now. According to him, all the sentence-initial elements in the following examples, taken from Gundel (1988:27) are themes: (39)
a) Bill/sent the package back b) A large cloud/appeared c) The one who called/was Alice d) A girl/broke the vase e) Did/he call? f) Who/called? g) Close/the door h) You/close the door
The most striking consequences of this theory are that not only are the subjects of presentational verbs such as “appear” in (39)b considered themes, but also, wh-words as in (39)f and the auxiliary “do” in (39)e. In addition, as Gundel (1988) points out, (39)g above is claimed to be about “close” and (39)h about the hearer.47 Thus, we see that the notion theme for Halliday includes what other authors would refer to as focus material and newly-introduced topics, as well as given topics. The usefulness of such a broad concept can be called into question. Apart from appearing in first position, do these elements have anything in common regarding their informational status?48 In view of the fact that the first position in a sentence can be occupied by many different things, other authors restrict themselves to the claim that sentence-initial position is necessary but not sufficient for topic
TOPICS 35
identification. That is, although topics do occupy this position, not all sentence-initial phrases are topics. But even inside this weaker version of the relationship between topics and sentence-initial position, there are different views. As we have seen, for present-day Prague School linguists, for instance, whose broad notion of topic allows for topical material to extend beyond one element and cover all contextually bound material, the sentence-initial requirement is quite loose. As mentioned in the previous section, Vallduví (1990) takes a different position, and divides the contextually bound material into two different primitives: links and tails. In this way, he can capture the fact that topical material can occur in more than one position. He can also account for right-dislocations like the one in (40) where topic-like material occurs sentence-finally.49 (40)
L’amo l’ODIA, el bròquil “The boss HATES broccoli”
Thus, considering that it seems to be true that although topics tend to appear in sentence-initial position, they can also occur somewhere else in the sentence, two positions have been taken: widen the concept of topic so that it can encompass all kinds of topical material (Prague School) or divide it up into two different elements (Vallduví’s theory). Finally, there is another view which suggests that although some topics do occur in sentence-initial position, there is no relation whatsoever between being a topic and that position in the sense that it cannot be used to define or recognize a topic. This is Lambrecht’s (1994) position. He claims that topics are not restricted to sentence-initial position. Very typically, he says, topics are grammatically expressed by pronouns (full or empty). Considering that pronouns usually have a fixed position in the sentence (very often very near the verb), it does not make sense to require topics to be sentence-initial. For example, imagine a situation in which the speaker and the hearer are talking about John. The speaker wants to add that in fact he saw John yesterday. In this situation, considering that John is salient, the speaker could use a pronoun to refer to John, as in (41): (41)
I saw him yesterday
Should “I” be considered the only topic just because it is in sentence-initial position, or is “him” a topic also, even if it is not in that
36 SYNTAX-INFORMATION STRUCTURE INTERFACE
position?50 In the context described above, “him” can be a very plausible topic. It can be viewed as what the sentence is about (in fact it passes the three tests) and it also qualifies as a Prague-topic: it is contextually bound. However, it does not fulfill the sentence-initial requirement. Notice, however, that the fact that object pronouns cannot appear sentence-initially is just a language-particular feature. The Spanish counterpart of (41) in (42) would indeed place the direct object pronoun in sentence-initial position.51 (42)
Lo vi ayer
Nevertheless, the existence of syntactic mechanisms such as Left-dislocation or Topicalization also needs to be considered. Topicalizing constructions such as those in (43) and (44) are assumed to place topical material sentence-initially, and if we are not willing to accept a direct relation between being a topic and being in that position, we have to explain why these mechanisms place these elements in exactly that position. (43) (44)
John I haven’t seen A Juan no lo he visto52
We could say that these elements are not placed in that position due to the fact that they are topics, but due to the fact that they are being introduced as new topics. This is what Lambrecht (1994) does. According to Lambrecht, these elements will appear in sentence-initial position when they are being promoted to topic status, very often because there is the necessity of a topic-shift or because the topical status of the element is not clear enough, so it cannot be taken for granted. From this perspective, only not-yet-clearly-established topics will appear sentence-initially. Lambrecht points out that what constructions such as those in (43)–(44) do is separate the reference of the topic (expressed by the lexical NP) from its role in the predication (DO argument, expressed by the clitic or pronoun) due to the fact that, when the topical character of a phrase cannot be taken for granted and the element needs to be introduced, it is better to introduce it first and then predicate something of it.53 Finally, when considering the relation between being a topic and being in sentence-initial position, we also have to take into account the fact that this position is very often occupied in languages such as
TOPICS 37
English by the subject, which cannot be placed postverbally in the unmarked case.54 However, in a language like Spanish, there are preverbal and postverbal subjects. Interestingly, postverbal subjects cannot be topical. Therefore, regarding the correlation between sentence-initial position and topic, we also need to explain the fact that when subjects are allowed to appear in different positions, only sentence-initial subjects are topical. This is important because here we have a choice, while in the case of the position of clitics in Spanish or pronouns in English we do not. 2.3.2.2 Subject In order to consider the relation between subjects and topics, we have to take into account the types of subjects we are dealing with. That is, the features of the subject may matter in the relation between these and topics. Therefore, the behavior that a particular language allows for its subjects may affect the nature of the relation we want to examine. I am going to compare the situation in Spanish with that in English. The fact that Spanish is a null-subject language is going to be crucial. The two important features that this parameter has, for our purposes, are the existence of null and postverbal subjects. It has been said that in English the subject is the unmarked topic. That is, although certain predicates have a preference for nontopic subjects, with most predicates, the subject is understood as the topic, unless there are prosodic, morphosyntactic or semantic clues to the contrary.55 However, subjects in English have a fixed (preverbal) position, let us say Spec, IP, and it may be that it is not so much being a subject but being in that position (without a special intonation) that is important for being a default topic. In fact, there is an exception to the obligatorily preverbal nature of subjects in English: so-called stylistic inversion. As illustrated by the following examples taken from Rochemont (1986), postverbal subjects are also possible in English. (45) a. Into the forest ran ROBIN HOOD b. Next to the fireplace stood a large old SOFA c. Less fortunate are the people without JOBS d. Sitting on the bed was his long lost BROTHER e. There was heading towards them, an ominous cloud with a long FUNNEL
38 SYNTAX-INFORMATION STRUCTURE INTERFACE
In this case, however, the subject is not interpreted as topical, but as focal.56 Let us see what the situation is in a language that allows different positions for subjects such as Spanish. Consider the following examples where we have a preverbal (46), postverbal (47) and sentence-final (48) subject: (46) (47) (48)
José trabaja en el campo “Jose works in the country” En el colegio trabaja Paco at the school works Paco En mi casa friega los platos mi padre in my house does the dishes my father
Of these subjects only the one in (46), the preverbal one, could be understood as topical. In (47) and (48), it is the preverbal PP that is understood as topical. Therefore, these sentences seem to show that in Spanish not all subjects are topical (“Paco” and “mi padre” in (47) and (48) respectively cannot be interpreted as topics) and that not all topical phrases are subjects (“En el colegio” and “En mi casa” are topical and they are not subjects). However, one could still try to maintain the generalization that subjects are the default topics also for Spanish, by saying that (47) and (48) are marked in some way, and, therefore, similar to those cases in English in (45). I do not think this is the case. The possibility of producing structures like those in (45) in English is very restricted; in this language postverbal subjects are not productive. However, constructions such as those in (47) and (48) are very productive in Spanish. For instance, the most natural, unmarked way to answer a question such as “Who has brought these books?” in Spanish is the following: (49)
(Estos libros) los ha traído el cartero
That is, to place a subject postverbally is totally productive in Spanish.57 Notice that this would be ungrammatical in English, which shows the restrictions that postverbal subjects such as those in (45) have , as shown by (50).58 (50)
*These books has brought the postman
Thus, I would like to suggest that there is nothing marked in Spanish about either nontopical subjects or nonsubject topical phrases.
TOPICS 39
However, what all the examples we have considered seem to point to is to the correlation between being topical and being in a certain sentence-initial position.59 It could be that the reason that in English subjects correlate with topics is that they correlate with that unmarked topic position. Two points should be made before we proceed. First, the motivation to place nonsubject topical phrases in sentence-initial position may be different in these languages. That is, although we are going to consider these topicalizing mechanisms in more detail below, it should be pointed out here that (49) and the grammatical version in English (51) may have not only different syntactic structures but also different pragmatic motivation : (51)
These books, the postman has brought them60
We are going to see that one of the motivations to place topical elements in sentence-initial position in Spanish is to remove them from the focus domain so that only phrases that do constitute the focus show up postverbally. That is, by putting the object in sentence-initial position in (49) we leave in focus position only the subject (the answer to the question). This, as (51) shows, cannot be the motivation to prepose the object in English. Second, is it true that all preverbal subjects are interpreted as topical? Consider the following; (52)
¿Qué pasó? “What happened?”
María ha dejado el programa “María has left the program”
An answer to a question such as that in (52) is often taken to be an all-focus sentence and show no topic. If it is true that a sentence used as an answer to a “What happened” question shows no topic—the whole sentence is focal—we may have to say in view of (52) that preverbal subjects in Spanish are not always topical. It could be the case that while a postverbal subject has to be part of the focus and cannot ever be topical, a preverbal subject can go either way, This would mean that there would be no way to know if a preverbal subject was a topic or not in an out-of-context sentence. However, I think that it is not clear that we have to consider the preverbal subject in (52) part of the focus just because it happens to be the answer to such a question. Lambrecht (1994) has pointed out that that kind of question can be answered with a Topic-Comment structure with no problem. He suggests that if the event to be reported involves a
40 SYNTAX-INFORMATION STRUCTURE INTERFACE
Topic, the corresponding element is very easily understood as a topic by the hearer. In (52) we have a proper noun. However, this is possible with different types of NPs other than proper nouns. In fact any type would do: a definite (53), a specific indefinite (54), a potentially nonspecific indefinite, (55) and even a quantifier (56) could also be interpreted as topical as long as they are preverbal subjects. (53) (54) (55) (56)
La estudiante española ha dejado el programa Una estudiante de mi clase ha dejado el programa Una estudiante ha dejado el programa Alguien ha dejado el programa
That is, it looks like a preverbal subject is immediately “accommodated” as a topic even in the context of the question above.61 Thus, there seems to be more relation between the sentence-initial position and topics than between subjects and topics in the sense that only subjects in that position are interpreted as topical, Of course, in a language where almost all subjects are in that position, as in English, the relation between subjects and topics is hard to deny. However, as mentioned in the previous section, although in many cases topics are placed in sentence-initial position, topical material can appear somewhere else in the sentence. In addition, we will need to be more explicit as to what we mean by sentence-initial position. First, we have to separate that topic position from a preverbal lower position, which can host wh phrases and focused phrases. Second, we will also need to separate the preverbal subject position from the position where dislocated topical elements are placed. 2.3.2.3 Pronominal Form The final syntactic characterization of topics that I will consider here establishes a correlation between pronominal elements and topics. As pointed out by Lambrecht (1994), pronouns are typically topical. In the exchange we considered in (41) above, repeated here as (57), is hard to deny the topical nature of the pronominal direct object form. (57)
I saw him yesterday
It seems to be what the sentence is about, it is contextually bound, it points to a discourse referent, and, in the context mentioned above, it is
TOPICS 41
the only topical element in the sentence. From this perspective, Erteschik-Shir (1997) has suggested that pronouns must be interpreted as topics. Again, we have to be careful here, because in the right context, a pronoun can in fact be the focus of the utterance: (58)
Who did they call?
Pat said they called HER
Furthermore, in regard to the informational status of pronominal elements, we have to take into account not only the position they occupy, but also the difference between full pronouns, null pronouns and clitics. As we have seen above, the position of pronouns can vary from language to language. Thus, in the Spanish counterpart the pronoun would appear before the verb, as in (59). (59)
Lo vi ayer
In addition, notice that sentence-initial pronouns: (60)
we
can
also
have
dislocated,
Him I saw yesterday
This example shows that the dislocated topic does not have to be discourse-new. However, it does look like the structure is making clear which topic (of several that may be salient in the discourse) is being chosen. I think a good continuation for this example would be the following: (61)
Him I saw yesterday, her I haven’t seen
This is also possible in Spanish and in this case the full pronoun appears sentence-initially followed by the clitic: (62)
A él lo vi ayer, a ella no la he visto
Finally, regarding the different discourse features of topics, Lambrecht points out the following correlations going from grammatical coding to cognitive states: a) pronominal coding and absence of pitch prominence are sufficient, but not necessary, conditions for activeness of a referent; b) presence of an accent and lexical coding are necessary, but not sufficient, conditions for inactiveness of a referent; c) definite vs. indefinite is neither a necessary nor a sufficient condition for either
42 SYNTAX-INFORMATION STRUCTURE INTERFACE
identifiability or activation state, at least in English, even though the tendency is strong for unidentifiable referents to be coded as indefinite noun phrases.62 Therefore, the only one-to-one correlation between a formal category and a cognitive state is the one between lack of prosodic prominence and/or pronominal coding and activeness. In English activeness is the only cognitive state that can be unambiguously expressed by grammatical means. Lambrecht points out that there is an apparent exception to this correlation. Consider (63): (63)
I’m looking for a BOOK. I FOUND one /I FOUND a book
If the indefinite NP “a book” is nonspecific, the anaphor must be an indefinite expression (one/a book).63 Lambrecht points out that the referents of the anaphor “one” and “a book” are unidentifiable in the sense that the addressee is not assumed to be able to identify the particular book that the speaker has in mind. However, as indicated by the lack of accentuation on the anaphor and, in the case of “one,” by the use of a pronominal form, the referent is treated as active, hence, identifiable. He explains this phenomenon as follows. When an NP is uttered, whether definite or indefinite, the CATEGORY denoted by the lexical NP is necessarily activated, in addition to the individual. Independently of whether the addressee can identify the particular referent the speaker has in mind or not, the active status of this category may be reflected in an anaphoric expression. That is, the TYPE may be active, even if the TOKEN is not.64 2.3.3 Some Conclusions We have examined some of the important issues surrounding the concept of topic. We have seen that although the notion of aboutness is quite intuitive it does not seem enough to identify the topical elements in a sentence. Regarding the topichood tests available in the literature, a distinction between nonspecific vs. specific topical elements might be helpful. Instead of concluding, as has sometimes happened, that these tests are not reliable and should be discarded, it might be that they are good tests for specific topics only and that we need to come up with a good test for nonspecific topics. Thus, the problem with (64), seems to be the nonspecificity of the topic, as opposed to (65):
TOPICS 43
(64) What about some students? (65) What about the students/first-year of linguistics?
students/students
We also considered the notions of discourse referent, discourse old and context-bound. We saw that not all topical phrases refer to discourse referents and not all topics are discourse-old. Although the notion of context-bound does allow for elements to be bound even if not discourse-old, it cannot account for discourse-new topics either. Regarding the syntactic characterizations, there seems to be a relationship between the sentence-initial position and topics in the sense that a) preverbal subjects are interpreted as topical (as opposed to postverbal) and b) other nonsubject phrases are placed in a sentence-initial position in so-called topicalizing structures, However, it should be taken into account that some devices, namely, right-dislocations, place these phrases sentence-finally. In addition, pronominals such as “him” in (66), can also be interpreted as topical in situ. (66)
I saw him yesterday
As the discussion above has shown, in order to investigate the relation between topic, sentence-initial position and subject, we need to take into account the type of language we are dealing with. That is, we need to take into consideration whether subjects are obligatorily placed in a preverbal position (as in English) or whether there is a choice (as in Spanish). Among positions, we should also distinguish between dislocated positions, which we can have at either sentence-edge (either left or right) and nondislocated positions. From this perspective, topics cannot be characterized syntactically without taking into account the syntactic features of the language, since languages will vary in the way they mark topics syntactically, if at all. Thus, before we offer a characterization of topics which does not run into the problems we have just considered, we will examine in detail how topical elements can be encoded in Spanish as compared to English, and the topicalizing mechanisms available in both languages.
44 SYNTAX-INFORMATION STRUCTURE INTERFACE
NOTES 1 For a summary of the main problems that the notions such as topic and focus pose, see Casielles (1999). 2. Weil’s thesis (1879 [1844]) was translated into English by Charles W.Super (1842–1939) in 1887. Quotes from this work are taken from the 1978 re-edition of the English version. 3. In addition to Weil’s work, Mathesius also mentions von der Gabelentz (1891), Paul (1886) and Wegener (1885). 4. See Mathesius (1915, 1939, 1942, 1975 [1961]), Beneš (1968), Daneš (1964, 1974a) and Firbas (1964, 1966, 1971), among others. For current Prague School approaches see Sgall, Hajičová and Panevová (1986) and Hajičová, Partee and Sgall (1998). 5. It should be pointed out that in the original work in Czech, Mathesius used the term východisko/východiště, which was translated by Vachek as “basis” and later identified by Firbas as “theme.” Nineteenth century authors such as Paul and von der Gabelentz had used “psychological subject” and “psychological predicate.” According to Daneš et al. (1974) the terms thema and rhema were introduced by H.Ammann. 6. Firbas (1964:277) points out that Mathesius (1939) seems to be doing a distinction between východiště and thema. However, in that work he offers no definition of the phenomenon referred to as thema and does not distinguish between those two elements in any subsequent work. The terminological problem is also considered in Daneš et al. (1974:217–222) 7. The degree of communicative dynamism carried by a sentence element is defined in Firbas (1966:240) as “the extent to which the sentence element contributes to the development of the communication, to which, as it were, it ‘pushes’ the communication forward.” 8. Firbas also points out that his concept of theme should be distinguished from what Beneš (1959:216) refers to as “basis,” which “as the opening element of the sentence links up the utterance with the context and the situation, selecting from several possible connections one that becomes the starting point, from which the entire further utterance unfolds and in regard to which it is orientated.” 9. It should be noted that in addition to the term theme Contreras also uses the term topic to refer to dislocated elements, which in Spanish co-occur with clitics. 10. See Gómez-González (2001) for a recent evaluation of three functional theories: the Praguian Functional Sentence Perspective, Functional Systemic Grammar and Functional Grammar. 11. Halliday’s notion of theme will be examined in detail in 2.3.2.1. As pointed out by Prince (1981b), the term given has also been understood in many different ways. See the discussion in 2.3.1.3.
TOPICS 45
12. Other early studies considering related issues are Bolinger (1954, 1965) Crabb (1955) and Hatcher(1956). 13. See Adjémian (1978), Allerton (1978), Chafe (1976), Dahl (1974a), Givón (1983), Gundel (1988) [1974], Kuno (1972, 1976), Li (1975, 1976), Prince (1981a, 1981b), and Reinhart (1981), to cite just a few important works from the seventies and early eighties. 14. Gundel is particularly concerned with Russian. See King (1995) for a recent proposal for this language. 15. This is a gradient notion, similar, as we will see below, to conceptions of open proposition/presupposition/background. 16. She clarifies that in Stalnaker (1978) the context set is the set of worlds compatible with these propositions. 17. Although Reinhart (1981) restricts herself to NP topics, she suggests that her analysis could be extended to other types of topics. 18. As we will later consider, Vallduví (1990) also allows for another topical element, the tail, which does not have a particular place in the sentence. See Portner and Katsuhiko (1998) for a semantic and pragmatic analysis of topic phrases based on Vallduví’s notion of link. 19. The problem of coming up with a definition that allows to identify the sentence topic is absent in languages that have a morphological marker for topics, such as Japanese “wa.” See Kuroda (1965) and Kuno (1972, 1976) for early studies and Tateishi (1991) for a more recent proposal. 20. Dahl defines topic as “that part of the sentence that refers to what the sentence is about.” (Dahl 1974a:7). See 2.3.1.4 for his proposal, which can divide a sentence into both topic-comment and focus-background. 21. A string P has a semantic antecedent in a discourse δ, δ={φ1, …, φn}, if, and only if, there is a prior and readily available string P′ in δ, such that the uttering of P′ either formally or informally entails the mention of P. (Rochemont 1986:47) 22. The term topic in this work will always be used in the sense of sentence topic rather than discourse topic. The notion of “discourse topic,” referring to “what a text is understood to be about,” is suprasentential, and while it may be indirectly connected with the syntactic structure of sentences, the connections undoubtedly involve interaction among many other factors as well, and, therefore, will not be considered here. Daneš (1974a:109) refers to this higher notion as hypertheme. See Reinhart (1981) for some discussion. 23. One problem we face when we try to classify some of the proposals is the vagueness of some of the concepts included in the definitions. For example, when a topic is characterized as the point of departure, this expression can sometimes mean a) placed towards the beginning of the sentence, b) constituting shared knowledge or c) being discourse-old. 24. In the philosophical tradition, Strawson (1971:97) also introduces the notion of topic in terms of aboutness. He refers to “the idea of the topic
46 SYNTAX-INFORMATION STRUCTURE INTERFACE
or centre of interest of a statement, the idea of what a sentence could be said, in this sense, to be about.” 25. In Partee (1989) the following semantic definition of aboutness in possible-worlds semantics is offered: “An empirical proposition can be said to be about a particular individual (or property, etc.) if the truth of that proposition varies systematically with the properties of that individual (or property, etc.).” She points out that this semantic notion may not correspond to any constituent in a sentence. For instance, she exemplifies this with the sentence in (i) which can be understood as being about “Jones’s age” (in addition to being about “Jones”). However, there is no constituent in the sentence corresponding to the concept “Jones’s age.”
(i)
Jones is over 40
26. A BLC is a discourse entity which is related to the set of previously evoked discourse entities, via a salient scalar relation, where a scale is defined as any partially ordered set. The OP is a sentence which contains one or more variables and it represents what is assumed by the speaker to be salient in the discourse at the time. We will consider Ward’s theory in more detail in chapter 6. 27. Thanks to Peggy Speas for pointing this out to me. 28. See Reinhart (1981) for the concept of pragmatic aboutness. 29. As Vallduví (1990) points out, there is some confusion in the literature on topics regarding the relation between the notion of topic and its referent. While for some authors topic refers to the phrase, for others it refers to the referent. 30. This is also true of Italian. Cinque (1990) gives the following examples:
(i)
(ii) (iii)
(iv) (v)
Al mare, ci siamo già stati. to the seaside there have-1pl already been Bella, non lo è mai stata. beautiful not it ever was-3sg Messo da parte, non lo e mai stato. got out of the way not it ever was-3sg Tutti non li ho visti ancora. All not them have-lsg seen yet. Che bevi, lo dicono tutti. That drink-2sg it says everybody
(PP)
(AP) (VP)
(QP) (CP)
31. On the relation between topic, givenness and other notions see Allerton (1978) and Chafe (1976), among others.
TOPICS 47
32. Topicalizations such as (32)-(33) should not be confused with so called Focus-Movement or Focus-Preposing as in (i):
(i)
FIFTY SIX HUNDRED DOLLARS we raised yesterday
Topicalizations have two accents, one on the topic and a focus accent, which are different (the topic accent is L+H* in Pierrehumbert’s system). In Focus-Preposing, there is only one accent, a focus accent, which is placed in sentence-initial position, We will come back to these constructions in the next chapters. 33. For the relation between accent and focus see 5.2. 34. Prince (1981b) points out three different interpretations of old/given: a) information which is predictable from previous discourse, b) salient (in the immediate awareness of speaker and addressee) and c) information which constitutes shared knowledge, 35. Unidentifiable referents do not coincide completely with indefinites. See Lambrecht (1994:79–87) for discussion and Ward and Prince (1991) for indefinite topics. 36. Generic and deictic uses of pronouns are exceptions in that they do not need to be activated. See Lambrecht (1994:95). Besides, there are several situations which might cause an active referent to be expressed as a lexical NP instead of a pronoun (Lambrecht 1994:95–6). 37. In Chafe’s words, this type of accessible referent belongs to the set of expectations associated with a schema. This concept of schema is related to Fillmore’s semantic frame. See Lambrecht (1994:99) where he considers these notions. He also points out that Clark (1977) and Prince (1981b) consider various types of inferences that may cause a referent to be accessible. 38. See Lambrecht (1994:36–43) for his division of the universe of discourse into a text-external world and a text-internal world. He also points out (Lambrecht 1994:100) that this type of accessibility would not fit well with the general notion of accessible as in the peripheral consciousness. If reference is made to an object in the external world, which can be assumed to be easily accessible, it is not necessary to assume that this object is in the addressee’s consciousness. He says that the two categories “textually accessible” and “situationally accessible” correspond to the text-internal and the text-external world respectively, while the category “inferentially accessible” is neutral to this distinction: a referent can be inferred from an element in the linguistic as well as in the extra-linguistic context.
48 SYNTAX-INFORMATION STRUCTURE INTERFACE
39. Lambrecht (1994:100) points out that the distinction between active, accessible and inactive referents is not made grammatically. From a grammatical (phonological and/or morphological) point of view, the only distinction that is made is between referents that are marked as being active (attenuated pronunciation and/or pronominal coding) and those which are not so marked. 40. This is so in the primary case. However Sgall et al. (1986:58) point out that “due to the complexity of a sentence involving embedding, the difference between the topic and the focus of the sentence is not always identical with that between CB [contextually bound] and NB [nonbound] expressions.” 41. This notion is connected with Firbas’s notion of communicative dynamism. See Hajičová, Partee and Sgall (1998) for some discussion. 42. They add that associative links can also make an element CB. For instance, because of the relationship between the verbs “to like” and “to hate,” “liked” in (i) is CB:
(i)
John hated EVERYBODY and Mary only liked the DANCER (Hajičová et al. 1998:59)
43. In another reading, “beavers” is the topic and the rest of the sentence belongs to its focus. The sentence also has a third reading, as a general statement without any topic. 44. As will be considered in later chapters, there are also proposals in the literature which suggest having one dichotomy (the Focus-Background) embedded both inside the Topic and inside the Comment, as in (i), in order to account for cases where focus seems to occur inside topics. See the discussion in 6.2.2 and 7.2.1.
(i)
Topic (Focus-Background)—Comment (Focus-Background)
45. As we will see, Vallduví’s primitive tail has some definitional problems, and sometimes it is not clear what the difference is between link, tail and ground. 46. The identification of topic/theme with the beginning of the sentence can be traced back to Praguian works such as Trávniček (1962). See Firbas (1964) and Beneš (1959) for arguments against this conception. 47. Gundel also points out the following surprising consequences of Halliday’s theory: the sentence in (i) would be about probably (or probability??); (ii) would be about nobody and (iii) about somebody (neither is used to make a statement about Bill).
(i)
Probably he’ll conference again
call
off
the
news
TOPICS 49
(ii) (iii)
Nobody saw Bill Somebody saw Bill
48. In this regard it is surprising to find recent works such as Gómez González (2001), which considers under theme/topic not only topicalizing devices such as left- and right dislocations but also focalizing devices such as clefts and pseudo-clefts. 49. See chapter 3 for a detailed examination of topicalizing mechanisms such as right-dislocations, left-dislocations and topicalizations. 50. In the case of (41) and due to the inherent topicality of a first person pronoun, both could be considered topical. We will later consider the case of multiple topics. However, if the speaker’s contribution in (41) would have been something like (i) or even
(ii), (i) (ii)
I think we would want to say that only “him” could be considered a topic. A friend of mine saw him yesterday Peter saw him yesterday
Notice that the subjects in (i)-(ii) require an accent in this context. 51. At least considering only the overt material. The null subject pronoun pro may be occupying a position to the left of the clitic. 52. It is very common for these structures in Spanish to have either negation or the emphatic affirmative “sí” right after the topical element:
(i)
A Juan sí lo he visto
53. It should be pointed out here that for Lambrecht, there are two ways an NP can be promoted to topic status, depending on its activation state in the discourse. If the discourse referent is accessible, it can be promoted through a dislocated structure. However, if it is brand-new, it will be first introduced into the discourse in a presentational clause of its own. Therefore, for Lambrecht, presentational sentences such as (i) are also promoting a referent.
(i)
Once there was a wizard. He was very wise, rich.,. (Lambrecht 1994:177)
This is necessary when the status of this referent is not accessible enough to both introduce it and predicate of it in the same sentence, which is what a dislocated structure does. From this viewpoint, we do not expect brand new elements in detachment structures nor active (specific) elements in presentational sentences, which tends to be the case.
50 SYNTAX-INFORMATION STRUCTURE INTERFACE
54. We will see some instances of postverbal subjects in English in the next section. 55. The predicates that have a preference for nontopic subjects are mostly unaccusatives. They constitute an exception in that in the unmarked case, their subjects are accented, as in the following examples: (i) (ii) (iii)
My CAR broke down The CHILDREN have arrived Your BROTHER has called
(Lambrecht 1994)
Lambrecht (1994) points out that in examples such as those in (i)-(iii), either the whole sentence is understood as the focus or just the subject. 56. See Rochemont and Culicover (1990) for the restrictions and focus-effects of this construction. Also Levin and Rappaport (1995) for a detailed study of the types of verbs that can appear in this construction and for the claim that it cannot be considered an unaccusative diagnostic. 57. In fact, later I am going to assume that the postverbal position is the unmarked, base position of subjects in Spanish. They are only placed preverbally if they are sentence topics. 58. As discussed as early as Mathesius (1915), the main way for English to place “the postman” in final position is through a passive: (i)
These books have been brought by the postman
In fact, this is not the only way. Although with different implications, the following structure is also possible: (ii)
The person who brought the books was the postman
59. To avoid the above-mentioned problems that we saw with Halliday’s system we would have to be more specific about this sentence-initial position (A vs. A′, for instance). 60. Although I have used a Left-dislocated structure in (51), as the English counterpart of the Spanish example in (49), it could be that a Topicalization is a closer counterpart. In chapter 3 I will pursue this idea and provide examples of the construction I think is closer to English Left-dislocation. 61. The term “accommodation” used here is borrowed from Lewis’ (1979) notion of presupposition accommodation.
TOPICS 51
62. Regarding the relationship between the activation states of referents and their grammatical coding (presence vs. absence of an accent; pronominal vs. lexical coding; and definite vs. indefinite phrase), Lambrecht points out that going from the cognitive state to the formal expression, we can suggest the following: -an active referent may be coded as an unaccented or accented, pronominal or lexical, definite or indefinite. That is, all formal types are compatible with the cognitive state active. --a nonactive (identifiable or unidentifiable) referent necessarily appears as an accented, lexical noun phrase, which may be definite or indefinite. 63. If, on the other hand, the indefinite is specific, the anaphor has to be a definite pronoun or NP: (i) I’m looking for a book. I found it/I found the book 64. Lambrecht points out that the process of category activation can apply going from one individual to another, as in the case of (63); from an individual to a category (i); and from a category to an individual (ii): (i) (ii)
I’m looking for a book. I LOVE books I love BOOKS; in fact I’m reading one right NOW
52 SYNTAX-INFORMATION STRUCTURE INTERFACE
CHAPTER 3 Topical Phrases and Topicalizing Constructions
3.1 Introduction We have considered different notions of topic in the literature and suggested that in order to understand the role of this concept in a particular language, we have to take into account the syntactic characteristics of the language, which may be crucial, for instance, for the potential correlation between topic and subject. We have seen that in the absence of particular phonological or syntactic special features, subjects, and more specifically, preverbal subjects can be considered to be unmarked topical phrases and can be very easily accommodated as such in out-of-the-blue utterances.1 However, we have pointed out that the differences between English and Spanish, such as the (im)possibility of having postverbal and null subjects, are crucial for this correlation and should be taken into account. As we are going to see below in more detail, a consequence of these differences is, for instance, that while a) an active topical element, b) an inactive topical element and c) a nontopical element can be all syntactically expressed by a preverbal subject in English, only b) would be expressed by a preverbal subject in Spanish. An active topical element would be expressed by a null subject and a nontopical by a postverbal subject.2 In regard to the correlation between topic and the sentence-initial position, we suggested that we have to be more specific about which sentence-initial position we are talking about and that we should at least differentiate between the sentence-initial position occupied by wh-words and other focused phrases and the sentence-initial position occupied by topical phrases. In addition, we also claimed that although both preverbal subjects and dislocated phrases can be considered to be topical phrases, a distinction should be made between them, since while
54 SYNTAX-INFORMATION STRUCTURE INTERFACE
the former occupy a specific argument position, the latter are in different (right or left) detached positions. This chapter examines topical phrases in Spanish and English, and considers several topicalizing devices available in these languages. We start by considering the nature of topical phrases in 3.2. As we will see, in Spanish, topical elements can be null or overt. If overt, they can appear in argument position or dislocated. From this perspective, the question is not anymore whether (all) topical phrases are in a certain position or whether they have certain discourse requirements, but rather how different types behave and what their nature is. Section 3.3 examines several topicalizing mechanisms: Clitic Left-dislocation, Pronoun Left-dislocation, Clitic Left-dislocation with no overt clitic, Topicalization, and Right-dislocation. Having considered in detail the behavior of topical phrases and topicalizing constructions in these languages, I come back in 3.4 to the notion topic and tentatively propose how we could solve the problem of characterizing topics crosslinguistically. I suggest giving up on the traditional view of topic and I argue that particular combinations of certain phonological, syntactic, semantic and discourse features will give us different types of topical elements. From this perspective, the long-term goal is to examine in detail how different languages encode nonfocal elements, discover which are the relevant features, and how they combine, and ultimately establish a feature-based typology of topics. Two topical elements can be immediately distinguished. One, which we can refer to as the Sentence-Topic proper (STopic) shows the following features: a) it is restricted to a unique element; b) it correlates with a sentence-initial position (often a preverbal subject); c) it seems to be restricted to discourse referents; d) it is not necessarily discourse-old; and e) it is not necessarily unaccented. The other one, which we can refer to as the Background, shows exactly the opposite features: a) it is not restricted to a unique element; b) it does not correlate with a particular sentence position; c) it is not restricted to discourse referents; d) it is necessarily discourse-old and e) it is necessarily unaccented. 3.2 Topical Phrases This section will examine different types of topical phrases in Spanish and English.3 First we will see that topical phrases in Spanish can be
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 55
null or overt. That is, a topical element can be expressed by a null category or by different overt phrases (pronouns or lexical phrases). In addition, they can be internal (in argument position) or external (in dislocated positions). Regarding null phrases, we will start with null subjects and consider Lambrecht’s (1994) view, which allows the verbal morphology to qualify as a topic. I will suggest that this is not necessary and that in these cases we have an implicit topical element: pro . The case of direct object (DO) topical phrases will be considered next. I will examine pairs such as those in (1) and suggest that maybe they are not as similar as they seem. (1)
a. I saw him yesterday b. Lo vi ayer
Regarding overt topical phrases, we will consider: a) full pronouns, b) lexical phrases, and c) dislocated topical phrases. First, we will see that in Spanish full pronouns are interpreted either as part of the focus or as unused or contrastive topical elements. They are never interpreted as active topical phrases, as opposed to English pronouns, which can be. Active topical elements are expressed by null pro. Second, we will consider lexically full topical phrases, I will point out a contrast between subjects and objects: while a lexically full preverbal subject can be interpreted as topical, a lexically full object cannot, unless it is dislocated or de-stressed. Finally, I will examine both left- and right-dislocated topical phrases, such as those in (2) and (3), (2) (3)
a. John, I haven’t seen (him) b. A Juan no lo he visto a. He is a nice guy, your brother b. Es un buen tío, tu hermano
As we will see, there are important differences between these dislocated topical phrases and the nondislocated ones, which point to a distinction between external vs. internal topical phrases. 3.2.1 Null Phrases It is well-known that in languages such as Spanish, which belong to the so called null-subject parameter, subjects can be dropped:
56 SYNTAX-INFORMATION STRUCTURE INTERFACE
(4)
__Llamé a Pedro __called-lsg part. Pedro
However, it should be emphasized that only topical subjects can be dropped. That is, the answer in (5), with a null subject is not felicitous. A full pronoun must be used: (5)
¿Quién llamó a Pedro? who called part. Pedro
# Lo llamé cl. called
Lo llamé YO cl. called I
Thus, rather than allowing for null subjects in general, languages like Spanish allow more specifically for null topical subjects. These topical null subjects contrast, at least apparently, with the possibility of having empty DOs, as (6) shows: (6)
* A Juan vi__ayer part. Juan saw-1sg__ yesterday
Notice that in this respect, English behaves just the opposite: while subject pronouns have to be overt (7), the English counterpart of (6) in (8) is grammatical: (7) (8)
* __Called Peter That movie Peter has not seen__yet
However, when we look at cases in which the DO has a certain activation state and can, therefore, be expressed through a pronoun, the situation looks very similar in English and Spanish. The DO pronoun has to be overt: (9) (10)
Peter has not seen it (*Peter has not seen__) Pedro no lo ha visto (*Pedro no ha visto__)
Nevertheless, the correspondence between DO pronouns in English and Romance DO clitics can be called into question. If the analysis of clitics as object agreement markers is on the right track, similarly to (11), where we have a null topical subject, a sentence such as that in (12) might also involve a pro in the object position, and, therefore, another type of null topic in Spanish and other Romance languages.
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 57
(11) (12)
pro Llamé a Pedro Pedro no lo ha visto pro 3.2.1.1 Subject pro
As mentioned above, a subject in Spanish can be left unexpressed. When this happens, it is currently assumed, in G&B and later developments (Chomsky 1982, Rizzi 1982, etc.), that there is an empty pronominal: pro. Since subjects can only be dropped when topical, we could say that these empty subjects are actually null topics, Lambrecht (1994) also considers these cases involving empty subjects in null-subject languages. However, he suggests that the topical element in cases like (11) is the verbal suffix itself. The set of unaccented pronominal expressions in Lambrecht’s (1994) theory, which according to him are the cognitively preferred topic expressions, includes free and bound pronouns, inflectional morphemes and null arguments. From this perspective, the inflectional morpheme on the verb in (11), -é, would also belong to this set of unaccented pronominal expressions. However, I think this is problematic. First, it is difficult to view the verbal suffix as an unaccented pronominal when it carries the stress, not only in that word, but it is also capable of being the only accented syllable in a complete sentence such as (13): (13)
Lo llam-é “I called him”
Second, considering the verbal inflection to be a topical element causes a problem that Lambrecht points out. Although for Lambrecht, unaccented pronominals are the cognitively preferred topic expression, not all unaccented pronominals are topic expressions. Certain pronominals cannot be topical phrases because (i) they have no referents, (ii) their referents are not pragmatically salient enough, or (iii) the clauses in which they occur cannot be interpreted as topic-comment clauses for independent reasons. In (14), we have examples of these three cases respectively; (14)
a. It is raining b. Has Pat been called yet?
58 SYNTAX-INFORMATION STRUCTURE INTERFACE
Pat said they called her twice c. Lleg-ó Juan arrived-3sg Juan The most relevant case for us now is (14)c. Lambrecht suggests that due to the fact that the NP “Juan” is not a topic constituent, the third person suffix -ó in (14)c, which counts as an unaccented pronominal in his analysis, has no topic referent to refer to and, therefore, cannot be a topic expression. Lambrecht’s suggestion in order to account for these apparent counterexamples to his claim that unaccented pronominals are the preferred topic expressions is to say that in these three cases the subject pronominals are interpreted as default morphemes, which are required by the grammatical system of the language. That is, he says that the weather pronoun “it” in (14)a, for example, is not required for semantic or pragmatic but for structural reasons, and he suggests that whatever explanation is used to account for the dual function of English pronouns such as expletive “it” or “there,” will account for the dual function of the person-number morphemes in inflectional languages. However, I do not think we need to consider the verbal suffix in (11) to be a topical element, since there is a perfect candidate for it: the empty subject pro If we consider the empty subject pronoun pro to be the implicit topical element, we can account for the difference between (11) and (14)c, repeated here as (15) and (16) for convenience: (15)
(16)
pro Llamé a Juan called-1sg John “I called John” Llegó Juan arrived-3sg John “JOHN arrived”
In both the verbal suffix is part of the agreement system. In (15) we have a null topical element (pro), while in (16) we do not. The subject in this case is explicit and it is not topical, as its postverbal position shows. We do not need to talk about a dual function of the inflectional morphemes, In both cases, they are agreement markers. Thus, in cases like (15), I will consider pro to be the null topical element, rather than the verbal morphology.
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 59
3.2.1.2 Direct Object pro? As mentioned above, the pair of sentences in (17) and (18) look pretty similar: (17) (18)
I saw him yesterday Lo vi ayer
In the right context, following something like “I haven’t seen John recently,” for instance, the DO pronoun “him” in (17) can be interpreted as topical. Now, the question is: are (17) and (18) as similar as they look? Can we consider “lo” topical, too? Lambrecht (1994) would say “yes,” since clitics also belong to the set of unaccented pronominals in his theory. However, (15) and (18) might be more similar than they seem at first sight.4 As recent analysis of clitics have suggested, clitics can also be considered to be part of the agreement system. From this perspective, there could also be a pro in object argument position.5 This empty DO pro could also be viewed as a null topic in cases like (18).6 From this perspective, Spanish might also allow for null objects. If so, the ungrammaticality of (6) might not be due to the fact that no null DOs are allowed in Spanish, but to the fact that the clitic, the agreement marker, is missing.7 Thus, if we look at clitics as agreement markers, we can see the parallelism between subject and DO agreement in (19): in the same way subject pro is licensed by the subject agreement suffix on the verb, DO pro could be licensed by the clitic: the DO agreement marker. (19)
a. pro Llamé a Juan b. Pedro no lo ha visto pro
If this is on the right track, and Spanish allows for both null subjects and null objects, as long as they are topical, the null-subject parameter might be viewed as a topic-drop parameter. Nevertheless, even if clitics can be considered to be agreement markers, there are some differences between subject and object agreement, which should be mentioned here. In Spanish, subject agreement on the verb is always overt, regardless of the null or full nature of the subject:
60 SYNTAX-INFORMATION STRUCTURE INTERFACE
(20) (21)
Tu hermana acab-a de salir “Your sister has just left” pro acab-a de salir pro has just left
However, object agreement seems to be only overt when the object is null. (22) (23)
Vi a tu hermana La vi
(full DO—no object agreement) (empty DO—object agreement)
If the object is an overt NP, the presence of the clitic is ungrammatical. (24)
*La vi a tu hermana
Notice, however, that there is some dialectal variation here. Porteño Spanish shows what has been called clitic doubling for direct objects.8 (25)
La vi a tu hermana
(full DO—object agreement)
Thus, in Porteño Spanish, object agreement seems to work similarly to subject agreement. The agreement is overt regardless of the null vs. full nature of the object. Even with an overt direct object as in (25), the object agreement shows up. On the other hand, clitic doubling for indirect objects is a general phenomenon in all dialects of Spanish. IO agreement seems to be similar to subject agreement. It also appears with overt IOs. (26)
a. Le dimos las flores a tu hermana ayer cl. gave-lpl the flowers to your sister yesterday b. Le dimos las flores ayer
Interestingly, subject agreement is overt not only with topical subjects (either overt or null), as in (20) and (21), but also with nontopical subjects: (27)
¿Quién le dio las flores a mi hermana? who cl. gave the flowers to my sister
This is also true of IO agreement.10
Se las dimos NOSOTROS9 cl. cl. gave-1pl we
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 61
(28)
¿A quién le disteis las flores? to whom cl. gave-2pl the flowers
Se las dimos a tu HERMANA cl. cl. gave-1pl to your sister
What about DO clitic doubling? As mentioned above, there has been a lot of debate about the Porteño Spanish data, and its specificity restrictions. However, I have not found anything about the topical/nontopical character of the object. That is, I do not know if the doubling is allowed both with topical, and nontopical objects, as long as they are specific, or if (25) is only possible if the object is topical. That could be the case, if the verb in (25) is stressed and the object de-stressed. However, I do not have any intonational information about the data. This is important because specificity does not seem to explain the situation in Standard Spanish. That is, although much effort has been put into explaining why (25) is good in Porteño Spanish, I do not know of an explanation of why (24) is bad in the rest of the dialects, as opposed to (29). (24) (29)
*La vi a tu hermana cl saw-1sg part your sister La vi a ella cl saw-1sg part she
The full NP and the strong pronoun do not differ in specificity, but they may differ in their topical status. That is, it may be that (24) is bad not because the clitic co-occurs with a full NP, but because the object, being in argument position (not dislocated) and with a normal intonation, will not be interpreted as topical. Before we turn to overt topical phrases, let us explore what the relation might be between the appearance of the clitic in Spanish and the topical character of the object. 3.2.1.3 Object Clitics and Topical Objects We have seen that DO clitics in the standard dialects only show up with null (therefore, topical) DOs, while subject agreement shows up with both null and full (both topical and nontopical) subjects. A question we might ask is what triggers the appearance of the DO clitic. Is it the null feature or the topical feature? That is, does the clitic mark a topical object (overt or covert) or the existence of a pro?
62 SYNTAX-INFORMATION STRUCTURE INTERFACE
First, DO clitics do show up in cases like (30), where there is an overt topical (although dislocated) DO: (30)
A tu hermana la vi ayer part your sister cl saw-1sg yesterday
However, the fact that these overt DOs are dislocated may point to the existence of a pro in the DO argument position; (31)
A tu hermana la vi pro ayer
Therefore, these cases cannot help us answer the question above, since they may involve both an empty pro and an overt topical DO. Second, it is also possible in Spanish, as in English, to shift focus to the left and mark a DO topical by de-stressing it, as in (32): (32)
Mi novia ODIA a los republicanos “My girlfriend HATES republicans”
In this case, we have an overt DO which is marked topical intonationally (by de-stressing it and stressing the verb), and the clitic does not show up. If these unstressed lexical NPs count as topical phrases, it looks like what triggers the appearance of the clitic is not the topical character of the DO by itself, but the null character of the DO. In this case, there is no pro, and the clitic does not show up. In fact, in the right-dislocated counterpart of (32), which presumably involves an empty pro in argument position, the clitic does show up. (33)
Mi novia los odia pro, a los republicanos
Thus, up to here, it looks like the clitic only appears if the postverbal object position is empty, as in (23), which can co-occur with a dislocated overt topical phrase, as in (31) and (33). In (24), this position is full, and, therefore, the clitic does not appear, This cannot be the whole story, however. As mentioned above, a clitic can, and in fact has to show up with an overt pronominal form, as opposed to a lexical form: (34) (35)
*La vi a tu hermana La vi a ella11
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 63
We could try to relate this contrast with the fact that pronouns are often the syntactic form used to encode topical elements. That is, given that topical phrases are very often expressed through pronominal (full or empty) forms, the fact that, as shown in (34)–(35), clitics can co-occur with pronouns but not with fully-lexical phrases could be considered evidence supporting the correlation clitic-topical element rather than clitic-empty object position. However, that does not work either because the clitic shows up, even with a clearly focal pronoun: (36)
¿A quién invitaste? “Whom did you invite?”
La invite a ELLA *Invité a ELLA cl invited-1sg part she invited-1sg part she
Thus, it appears that a pronoun, even if it is focused, triggers the appearance of a clitic. So, unless there is a pro also in cases such as (35) and (36), it seems that object clitics co-occur only with pronominals (empty or full). Why should this be? Is this related with the correlations we saw in chapter 2 (2.3.2.3) between pronominal coding and activeness, and lexical coding and inactiveness? Or is it just a purely syntactic correlation? Can it be that the NPs occupy a different position from the pronouns in (35) and (36)? All these questions will have to be left for further research.12 3.2.2 Overt Phrases In addition to null phrases, topical elements can also be expressed by three types of overt phrases: full pronouns, lexical phrases and dislocated phrases. 3.2.2.1 Full Pronouns We have seen that Spanish might allow null pronouns in both subject and object position, as long as these are topical. That is, while English has a full pronoun in a context in which the subject or the object are active topical elements, as in (37) and (38) respectively, Spanish shows an empty subject pronoun, as in (39), and a clitic in (40), which might also involve an empty object, as we have discussed above. (37)
I saw John yesterday. He was very happy
64 SYNTAX-INFORMATION STRUCTURE INTERFACE
(38) (39) (40)
I saw John yesterday. I invited him to the party Vi a Juan ayer. pro Estaba muy contento Vi a Juan ayer. pro Lo invité pro a la fiesta
What about Spanish full pronouns? When are they used? Let us consider the case of full subject pronouns first. In a case such as (39), a full pronoun is not felicitous. A full pronoun, as opposed to the available null pro, indicates disjoint reference and there is no other referent available in the discourse to which the pronoun could refer. (41)
Vi a Juan ayer. #Él estaba muy contento
A full pronoun can be used, however, when there is more than one active topical element, when we have so-called contrastive topical elements, as in (42): (42) Vi a Juan y a Raquel ayer. Él estaba muy contento, pero ella no. “I saw Juan and Raquel yesterday. He was very happy, but she wasn’t” The case of object full pronouns is similar. If we fill in the object position in (40) with a full pronoun, the result is not felicitous in this context, in which there is only one topical element. (43)
Vi a Juan ayer. # Lo invite a él a la fiesta
However, similarly to subject full pronouns, if there is more than one active topical element available, the full form (marker “a”+subject tonic pronoun) can be used: (44) Vi a Juan y a Raquel ayer. Lo vi a él por la mañana y a ella por la noche “I saw J. and R. yesterday. I saw him in the morning and her at night” In addition, full pronouns can be used when they constitute the focus, as in English: (45)
¿A quién invitaste? ¿a él o a ella? part who invited-2sg part he or part she
Lo invite a ÉL cl invited part he
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 65
So, it looks like while in English full pronouns can be used when they are focused, contrastive topical phrases or just single active topical elements, in Spanish they can only be used when they are focused or contrastive topical elements, but not for active topical elements. This is probably related to the fact that Spanish full pronouns cannot be de-stressed.13 3.2.2.2 Lexical Phrases Putting aside dislocated phrases, there is an important difference between fully-lexical subjects and nonsubjects regarding their possibilities of being interpreted as topical elements. Let us consider lexical DOs. When DOs are lexical phrases in argument position (when they are not either right-, or left-dislocated), they cannot be interpreted as topical.14 In contrast, subjects have their preverbal argument position when they are topical, and their postverbal argument position when they are not. That is, while a lexical preverbal subject is interpreted as topical, and a postverbal one as nontopical, a lexically-full (nondislocated) DO is always interpreted as nontopical. As opposed to subjects, therefore, lexical DOs do not seem to have an argument position that they can occupy when they are topical. Compare the following sentences: (46) (47)
Los niños fueron al cine “The children went to the movies” Los padres llevaron a los niños al cine ‘The parents took the children to the movies”
The nonpronominal subject “los niños” in (46) can be interpreted as topical, even in the absence of any context, and as an answer to a question such as “What happened?”15 However, the nonpronominal object “a los niños” in (47) will not be understood as a topical element, unless it is empty or dislocated. (48) (49)
Los padres los llevaron al cine A los niños los padres los llevaron al cine
So, it looks like lexical material inside VP cannot be interpreted as topical. From this perspective, the difference between a preverbal lexical subject and a lexical DO, regarding their possibility of being interpreted
66 SYNTAX-INFORMATION STRUCTURE INTERFACE
as topical, might be a consequence of their different positions. There is an exception, which we have already mentioned: a de-stressed lexical DO in argument position can be interpreted as a topical element.16 (50)
Los padres LLEVARON a los niños al cine “The parents TOOK the children to the movies”
(no los recogieron) (they didn’t pick them up)
This is only felicitous in a context in which the stressed phrase is the only focus in the sentence. That is, (50) can only be uttered in a situation in which the action is the only focus, meaning something like: “concerning the parents, the children and the cinema, they brought them there (they didn’t pick them up).” That is, this mechanism not only marks the DO topical, it marks everything except the stressed phrase as topical. Both “the parents” and “to the movies” are necessarily topical. This brings up again the problem of distinguishing between topic and background. Notice that the same results can be obtained by focusing any other phrase: (51)
Los PADRES llevaron a los niños al cine
(no los abuelos) (not the grandparents)
The sentence in (51) is interpreted like “concerning taking the children to the cinema, the PARENTS did that.” In these cases, in which the verb is not focused, are all unstressed phrases (including the verb) topical phrases? We would also have to consider a case like (52), where the focused element is sentence-final. (52)
Los padres llevaron a los niños al CINE
As we will see in more detail in chapter 5, where the notion of focus is examined, this case is different, since sentence-final focus can project up, while the focus in (50)–(51) cannot. In fact, we will see that in Spanish (52) would have a VP focus (concerning the parents, they took the children to the cinema) rather than a PP focus (concerning the parents taking the children somewhere, they took them to the cinema, not to a play), which means that sentence-final focus obligatorily projects up.17 In order to have PP focus, we would have to get “the
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 67
children” out of the VP, through a dislocation: “A los niños los padres los llevaron al cine.” Going back to the de-stressed object in (50), is not clear then if “the children’ is a topic, or if it is just part of the background in a Focus-Background structure. It is interesting to compare these de-stressed objects with right-dislocated objects, Is “republicans” part of the Background in (53), but a topic in (54)? (53) (54)
My sister HATES republicans My sister hates them, republicans
For Vallduví (1990) both DOs in (53) and (54) above would be “tails.” However, while he would consider a right-dislocated object as in (55) a tail, he would consider a left-dislocated object a topic (his link). (55) (56)
I saw him yesterday, your brother Your brother, I saw him yesterday
Do (55) and (56) involve different primitives? We come back to this question below. 3.2.2.3 Dislocated Phrases As we have seen, dislocated phrases can be dislocated to the left or to the right. In (57) and (58) we have familiar examples of a left-dislocated direct object and a right-dislocated one: (57) (58)
A tu hermana la vi ayer part your sister cl saw-1sg yesterday La vi ayer, a tu hermana
The dislocated phrase does not have to be lexical, it can be pronominal: (59) (60)
A ella la vi ayer La vi ayer, a ella
In addition, subjects can also be dislocated: (61)
Tu hermano__es un buen tío “Your brother, he is a nice guy”
68 SYNTAX-INFORMATION STRUCTURE INTERFACE
(62)
__Es un buen tío, tu hermano “He is a nice guy, your brother”
The left-dislocation of a subject in (61) is hard to see. However, we can make it visually more obvious by dislocating the subject of an embedded clause, as in (63): (63)
Tu hermano, no creo que pro venga a la fiesta your brother not think-lsg that come-3sg to the party “Your brother, I don’t think he will come to the party”
Also, an instance of multiple topical phrases, where the subject occurs to the left of other dislocated phrases, can help us see that the subject is dislocated: (64) Tu hermano, la opera pro la odia your brother det opera cl. hates “Your brother, opera, he hates it / Your brother, he hates it, opera” Thus, we have seen that both lexical and pronominal subjects and nonsubjects can be left- or right-dislocated. Several questions remain. First, are dislocated topical phrases inherently different from nondislocated topical elements? Second, is there a crucial difference between right- and left-dislocations as Vallduví (1990) suggests? In the next section I suggest that there is a difference between nondislocated or what we could refer to as internal topical phrases and dislocated or external topical phrases. However, I will claim that both left-and right-dislocation involve the same primitive. 3.2.3 Internal vs. External Topical Phrases We have seen that topical phrases in Spanish can be either null or overt. Now, we are going to consider another distinction that cuts across that one. There are topical phrases in argument position (either null or overt), which could be referred to as internal, and dislocated topical phrases in nonargument position (necessarily explicit), which appear detached to the left or to the right edge of the sentence, which we can call external, I believe that this is an important distinction in Spanish, since
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 69
there are constraints on internal topical phrases that do not affect external ones. When we considered the relation between topic and subject in section 2.3.2.2, we saw that in Spanish preverbal subjects, as opposed to postverbal subjects, are interpreted as topical, unless intonationally focused. We also saw that this was the case regardless of the type of phrase. As mentioned in chapter 2, we can have a definite, specific indefinite, a potentially nonspecific indefinite and a quantifier, which are very easily accommodated as topical elements.18 (65) (66) (67) (68)
La estudiante española ha dejado el programa “The Spanish student has left the program” Una estudiante de mi clase ha dejado el programa “A student in my class has left the program” Una estudiante ha dejado el programa “A student has left the program” Alguien ha dejado el programa “Somebody has left the program”
What we cannot have, however, is a BN: (69)
*Estudiantes han dejado el programa “Students have left the program”
Does this mean that a BN in Spanish cannot be accommodated as a topic or that it cannot be a subject? Examples such as (70)–(71), where BNs do appear as dislocated topical phrases seem to favor the second view: (70) (71)
Estudiantes no he visto students, not have-1sg seen Café no quiero coffee, not want-1sg
However, if we are going to pursue the idea that BNs cannot be subjects in Spanish and assume the lack of the determiner to be the reason, we will have to specify that it is not so much that BNs cannot be subjects, but rather that they cannot be preverbal subjects, as the contrast between (72)–(73) shows.
70 SYNTAX-INFORMATION STRUCTURE INTERFACE
(72) (73)
Jugaban niños en el parque were-playing children in the park *Niños jugaban en el parque “Children were playing in the park”
This again brings us close to the notion of topical element, since one important difference between postverbal and preverbal subjects in Spanish, as we have seen, is the fact that only the latter can be interpreted as topical. Thus, there seems to be a difference between topical dislocated phrases, such as those in (70)–(71), which do allow for a BN, and topical subjects in (73), which do not. An important difference between (70)–(71), on the one hand, and (73), on the other hand, is that the topical phrases in the former are dislocated, and therefore not in argument position. So, it may be that BNs cannot be internal topical phrases, topical phrases in argument position. Notice that although the dislocated topical phrases in (70) and (71) are not subjects, the same difference can be found between a nondislocated subject, as in (74), and a dislocated subject, as in (75). (74) *Niños no vendrán después de las siete children not come-3pl after prep det seven “Children will not come after seven” (75) Niños, me imagino que no vendrán después de las siete children cl imagine-1sg that not come-3pl after prep det seven “Children, I imagine they will not come after seven” Thus, rather than a subject-object asymmetry, the contrast between (74) and (75) shows that there is something else going on. It looks like a BN cannot be a nondislocated topical subject, although it is fine as an external topical phrase, both as a dislocated subject as in (75), and as a dislocated object, as in (70)–(71). The ungrammaticality of (74) shows that a topical element in subject position cannot be expressed by a BN. What about a topical element in object position? If the same situation obtains, we can generalize this phenomenon by using the contrast internal/external and saying that a BN cannot be an internal topical phrase in Spanish, but only an external, dislocated topical element. I think this is the case. In (76), the BN in object position cannot be interpreted as topical.19
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 71
(76)
Los vaqueros mascan tabaco the cowboys chew-3pl tobacco
(that’s what they do)
However, this can change if we make it nonbare. The sentence in (77) is interpreted with “tobacco” as topical. (77) Los vaqueros mascan el tabaco (that’s what they do with it: they chew it) the cowboys chew-3pl the tobacco The fact that a BN cannot be interpreted as a topical element in (76) is taken as evidence that in Spanish BNs cannot be internal topical phrases. In object position, they are interpreted as part of the focus, and they are ungrammatical in the preverbal subject topical position. In chapter 4, we will see why this should be the case, and we will also consider the interpretation of BNs in Spanish. We will see that in Spanish, as opposed to English, BNs can only have an existential interpretation, and it will be proposed that this is due to the fact that only DPs can get out of the VP. To summarize this section, we have seen that Spanish has null as well as overt topical phrases. While null topical elements are active in Lambrecht’s sense, overt pronominal (as opposed to their English counterparts) cannot. Regarding lexically full topical phrases, these can appear in the preverbal subject position or in dislocated positions, but not in the in situ DO position, unless the DO and everything except the focal element is de-accented. Finally, I have provided some evidence which points to a distinction between external and internal topical phrases, by showing that while BNs can be external topical phrases, they cannot be internal ones. Next I will examine different topicalizing mechanisms in English and Spanish. 3.3 Topicalizing Constructions20 Topicalizing constructions such as the structures referred to by Ross (1967) as Left-dislocation in (78) and Topicalization in (79) have attracted considerable attention.21 (78) (79)
John, I saw him yesterday Julia I couldn’t reach__
72 SYNTAX-INFORMATION STRUCTURE INTERFACE
Similar structures have been attested in many different languages and there has been extensive research both on their syntactic as well as their pragmatic nature.22 From a pragmatic point of view, the English structures in (78) and (79) have both been considered to be topicalizing mechanisms in the sense that they seem to prepose an element which is taken to be the topic of the sentence. However, they are syntactically quite different; while topicalizations have a gap, where the topicalized phrase would have appeared in the nonpreposed version, left-dislocations include a resumptive pronoun. Based on these two English constructions, the labels Topicalization and Left-dislocation have been used to refer to apparently parallel preposing mechanisms in other languages, which also place certain phrases towards the beginning of the sentence. With particular reference to Spanish, the relevant constructions are those in (80) and (81), taken from Rivero (1980). (80) El dinero acepto que pretendan que lo tienen ya23 the money accept-lsg that pretend-3pl that it have-3pl already “The money I accept the claim that they have it already” (81) Dinero dicen que no tiene___ money say-3pl that not have-3sg “Money they say she/he doesn’t have” (Rivero 1980:365–366) Since both structures prepose a topical element (and, in a way that is similar to the English constructions, one structure shows a resumptive pronoun while the other seems to have a gap), Rivero (1980) applies the English labels to these constructions and refers to them as Left-dislocation and Topicalization respectively. However, a detailed examination of these two constructions in Spanish will show that they differ from the English constructions syntactically, pragmatically and phonologically. In 3.3.1, we will examine the constructions involving a resumptive pronoun in (78) and (80). I will propose that, despite apparent similarities, these are two different constructions. In 3.3.2, I will show that in fact Spanish has a second left-dislocation (also present in other Romance languages), which does behave like English left-dislocation. Next, I will focus on the structures involving a gap and suggest that Topicalization does not exist in Spanish (3.3.3). It will be proposed that the structures involving a gap in Spanish should be analyzed as cases of clitic left-dislocations with no overt clitic. Finally, section 3.3.4 will consider Right-dislocation.
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 73
3.3.1 Clitic Left-dislocation vs. English Left-dislocation Prima facie, the English construction in (78), repeated below as (82), and the Spanish structure in (83) appear to be the same: (82) (83)
John, I saw him yesterday A Juan lo vi ayer
Syntactically, they both involve the preposing of the direct object and the co-occurrence of a resumptive pronoun. From a pragmatic point of view, both structures are considered to be topicalizing mechanisms in the sense that they are used to introduce, establish or comment on a discourse topic, John in (82) and Juan in (83).24 Nevertheless, we can immediately perceive two differences, which have to do with particular syntactic features of Spanish: (i) the appearance of the particle a before the human direct object and (ii) the different position and nature of the pronoun, which in Spanish is a weak pronoun or clitic and has to appear preverbally. These are general characteristics that would also occur in nondislocated structures and should thus not be the only ones that we use to tease these two constructions apart. However, there are additional syntactic differences that do point to the existence of two different types of left-dislocation. First, the Spanish construction(s), which will be referred to as Clitic left-dislocation (CLLD) following Cinque’s (1990) proposal for Italian, allow(s) not only for the dislocation of noun phrases (NPs), but also for any kind of phrase. Thus, as the following examples show, we can dislocate an adjective, a quantifier or even a clause: (84) Listo no lo parece25 clever not cl seem-3sg “He doesn’t seem clever” / “Clever he doesn’t seem” (85) A todos no los he visto todavía prep all not cl have-1sg seen yet “I haven’t seen everybody yet” / “Everybody I’ve yet to see” (86) Que fumas lo sabemos todos that smoke-2sg cl know-1pl all “We all know that you smoke”/“That you smoke we all know” As is well-known, English Left-dislocation (LD), on the other hand, only allows for the dislocation of referential NPs, such as the one in
74 SYNTAX-INFORMATION STRUCTURE INTERFACE
(82). Thus, the be ungrammatical: (87)
English
counterparts
of
(84)–(86)
would
a. *Clever he doesn’t seem it b. *Everybody I haven’t seen them yet c. *That you smoke we all know it
Second, there is no limit to the number of phrases that a Spanish CLLD can prepose. Thus, we can have two (88)–(89), three (90) or even four dislocated phrases, as in (91). (88) A Pedro los libros ya se los compré to Pedro the books already cl cl bought-1sg “I already bought the books for Pedro”/ “The books for Pedro I already bought” (89) A mí listo no me lo parece to me clever not cl cl seem-3sg “He doesn’t seem clever to me”/?“Clever to me he doesn’t seem” (90) Estos libros yo a Juan nunca se los dejaría these books I to Juan never cl cl lend-1sg “I would never lend Juan these books”/ ?“To Juan these books I would never lend” (91) Un libro a mí Juan en Navidad nunca me lo ha regalado a book to me john for Christmas never cl cl have-3sg given (as a gift) “John has never given me a book for Christmas”/ “A book for Christmas John has never given me” Although, as the second glosses in (88)–(91) show, English possibly allows for the topicalization of two elements with no resumptive pronoun, English LD does not allow for multiple left-dislocations. Thus, the following multiple left-dislocations are ungrammatical: (92)
a. *These books to John I would never lend them to him b. *John a gift he has never bought it
Third, the Spanish construction can accomplish two tasks simultaneously: it can place a direct object in sentence-initial position and at the same time place a focal subject in sentence-final position, as in (93);
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 75
(93)
La casa la limpié yo the house cl cleaned-1sg I “I cleaned the house”
This is particularly appropriate in a situation where the direct object is topical and the reference of the subject unknown, which would be the case if (93) occurred as an answer to a question such as ¿Quién limpió la casa? “Who cleaned the house?” As is well-known, the English counterpart of (93) would be quite ungrammatical since English left-dislocations do not allow for the order object-verb-subject. Thus, we have seen three important syntactic differences between English LD and Spanish CLLD: while the Spanish structure can dislocate any type and number of elements and even result in an object-clitic-verb-subject order, English LD only allows for the dislocation of a single NP in the fixed structure NP-subject-verbpronoun. Such differences suggest that these are two different constructions—two different types of dislocation. Some of these differences have also been pointed out by Cinque (1990) for Italian CLLD. Thus, Cinque also notes that Italian CLLD can dislocate any type and number of phrases.26 In addition, he provides additional syntactic evidence that further supports the heterogeneous nature of English LD and Italian CLLD. First, Cinque points out that in the Italian CLLD structure, the resumptive pronoun can only be a clitic, never a strong pronoun: (94) a. In quella città, non ci sono mai stato27 in that town not-there-(I)-have ever been “I’ve never been to that town”/“To that town I’ve never been” b.
*In quella città, non sono mai stato là in that town not (I) have ever been there *”To that town I’ve never been there”
As the ungrammaticality of (95)b shows, the same restriction holds in Spanish: (95)
a. A Julia no la pude encontrar prep Julia not cl could-lsg find “I couldn’t find Julia’/‘Julia I couldn’t find” b. *A Julia no pude encontrar a ella
76 SYNTAX-INFORMATION STRUCTURE INTERFACE
prep Julia not could-lsg find prep pr *”Julia I couldn’t find her” Second, Cinque points out that while English LD can only place a constituent at the head of main clauses, the left-dislocated phrase in Italian CLLDs can occur at the front of virtually any subordinate clause type: (96) L’unica persona che a Gianni, non gli ha mai fatto un favore,… the only person who to Gianni not-to-him-has ever done a favor *”The only person who to Gianni never to him has done a favor…” Again, this is also the case in Spanish. It is possible to have dislocated elements occurring in a subordinate clause: (97)
La única persona que a Juan nunca le ha hecho un favor…
Third, in English, the relation between the left-dislocated phrase and the resumptive pronoun is not sensitive to island constraints, Thus, the example in (98), taken from Chomsky (1977) is fine: (98) As far as John is concerned, I will never believe the claims that have been made about him However, Italian CLLD is sensitive to island constraints. In (99) we have an example of the complex NP constraint: (99) *[A Carlo], ti parlerò solo del[le persone [che gli piacciono]] to carlo I will talk to you only about the people that to him appeal As (100) shows, Spanish CLLD is sensitive to island constraints, too: (100)
*A Carlos conozco sólo a las personas que le gustan28 to Carlos I know only the people that to him appeal
Thus, Spanish CLLD behaves like its Italian counterpart. These Romance dislocations involving a clitic sharply contrast with
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 77
English LD, and should not be considered instances of the same construction. We have seen that the similarities between the examples in (82) and (83) are only apparent and that in fact they exemplify two completely different types of dislocation: Romance CLLD and English LD. While the Romance structure can dislocate any type and number of elements and even result in an object-clitic-verb-subject order, English LD only allows for the dislocation of a single NP in the fixed structure NP-subject-verb-pronoun. However, that does not mean that English LD has no counterpart in Romance languages. As we are going to see in the following section, Italian and Spanish each have a different dislocation that does resemble English LD in certain ways. 3.3.2 Spanish Strong-Pronoun Left-dislocation Cinque (1990) has noted that in addition to the CLLD structure, Italian also has a nonclitic left-dislocation with a strong resumptive pronoun similar to English LD, as in (101): (101)
Quella città, non sono mai stato là that town I’ve never been there
Note that as opposed to (94)b, which is ungrammatical, in (101) we do not dislocate the full prepositional phrase (PP), but just the NP “that town.” That is, similarly to English LD, the Italian LD in (101) involves a full pronoun, and it seems to be only possible with NPs. Does Spanish manifest a Left-dislocation which does not involve a clitic but a strong pronoun, similar to the Italian construction in (101)? Although it is harder to see in Spanish due to its idiosyncratic pronominal system, I will now try to show that Spanish also has a left-dislocation that is different from the CLLD we considered in the previous section. In Italian, the difference between (101) and (94)a, repeated here as (102), can be clearly seen due to the existence of the locative clitic ci and the locative pronoun là. (102)
In quella città non ci sono mai stato in that town not cl have-1sg ever been
However, Spanish lacks both the clitic and the full pronoun for this type of locative prepositional phrase. When a PP such as that in (102) is
78 SYNTAX-INFORMATION STRUCTURE INTERFACE
left-dislocated in Spanish, no clitic appears, since this type of clitic does not exist. Thus, the Spanish equivalent of (102) would be (103), a CLLD with no overt clitic.29 (103)
En esa ciudad nunca he estado
The closest we can get in Spanish to the Italian construction in (101) is with the structure in (104), where we could add allí “there” in lieu of a resumptive pronoun. (104)
Esa ciudad, nunca he estado allí
Thus, due to the absence of locative pronouns, the contrast between the two structures is not as sharp as in Italian. However, I think this type of left-dislocation, which is restricted to NPs and involves a strong pronoun, nevertheless does exist in Spanish. Consider the following contrast: (105)
(106)
Con Juan hablé ayer with John spoke-1sg yesterday “I spoke with John yesterday” Juan, hablé con él ayer30 John, spoke-lsg with him yesterday “John, I spoke with him yesterday”
As opposed to (105), where we have the straightforward left-dislocation of the whole PP, in (106) we prepose only the NP, without the preposition, and we include a resumptive strong pronoun. I would like to suggest that it is this structure, and not the Romance CLLD we considered above, which is akin to English LD. In order to distinguish this dislocation from CLLD, I will refer to it as the strong-pronoun left-dislocation (SPLD). We have already seen that Spanish and Italian SPLDs share with English LD two syntactic features: the appearance of a strong pronoun, and the restriction to the dislocation of NPs. In addition, both English LD and Romance SPLD have special phonological and pragmatic characteristics, which have also been pointed out by Cinque for the Italian construction. Typically, examples of English LDs include a comma, which separates the dislocated element from the rest of the sentence and points to the existence of a pause. This seems to be also the case for Italian and
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 79
Spanish SPLDs. Thus, as opposed to CLLDs, the preposed element in (101), (104) and (106) is separated from the rest of the clause by a pause. As Cinque points out, in terms of its intonation the dislocated element has a contour which is somewhat similar to that of questioned NPs. Furthermore, according to Cinque (1977) and (1990), Italian SPLD also has special pragmatic features. He suggests that this construction mainly serves to promote an NP to topical status at a point in the discourse when it was not topical, This would contrast with CLLDs, which, according to Cinque, dislocate an element that is already a topic in the discourse. Although the pragmatic features of English LD have not been clearly established, the same distinction has been noted by Rodman (1974) with respect to English LD and English Topicalization. Thus, Rodman suggests that in English, Left-dislocations are used to establish a topic while Topicalizations are used to prepose topical phrases which have already been established. He offers the following examples: (107) (108)
What can you tell me about John? Nothing. But Bill, Mary kissed him/*But Bill, Mary kissed What can you tell me about John? John, Mary kissed/*John, Mary kissed him
If this is on the right track, English LD and Italian SPLD might have another feature in common: they serve to establish a topic. However, we should be cautious here, since this pragmatic feature of English LD is not as clear as the syntactic features we considered above. Thus, there seem to be some problems with Rodman’s data. Ward (1988), for instance, does not share Rodman’s grammaticality judgments in (107) and (108) above, and points out that for him neither structure is felicitous in that context. However, while it is true that the exchanges in (107)–(108) do sound quite unnatural, as Ward points out, this might be related to the particular words chosen, rather than to the structures. The following examples do sound more natural; (109) (110)
What can you tell me about John? Nothing. But Bill, I really like him What can you tell me about John? John, I really like
80 SYNTAX-INFORMATION STRUCTURE INTERFACE
Nevertheless, Prince (1984) offers the following naturally occurring example, which shows that, at least in some cases, English Topicalization and LD might be interchangeable: (111) She had an idea for a project. She is going to use three groups of mice. One, she’ll feed them on mouse chow, just the regular stuff they make for mice. Another, she’ll feed them veggies. And the third she’ll feed junk food. She notes that in this example, the LDs could be replaced by Topicalizations and the Topicalization could be exchanged for an LD. Thus, more research seems to be needed to clearly establish the pragmatic differences that might exist between English Topicalization and English LD. I think the same can be said of the possible pragmatic difference between Romance CLLD and SPLD. Although the difference pointed out by Cinque in terms of old vs. new topic seems to be on the right track, more naturally occurring data needs to be gathered and analyzed. Nevertheless, the syntactic and phonological differences between these two constructions are clear. We have seen that, in addition to CLLD, Spanish and Italian have a left-dislocated structure which shows a strong pronoun, only allows for the dislocation of NPs, includes a pause (which separates the dislocated element form the rest of the sentence), and gives the preposed element a rising intonation similar to that of questioned NPs. All these features are also present in English LD. 3.3.3. CLLD with no overt clitic vs. English Topicalization Let us now examine the constructions involving a gap in English and Spanish. In (112) and (113) we have the examples introduced in section one, repeated here for convenience: (112) (113)
Julia I couldn’t reach Dinero dicen que no tiene——
The first interesting feature of the Spanish construction in (113) is that the dislocated element is not a full NP, but what has been referred to as a bare or naked nominal: a noun without a determiner.31 This is a crucial feature of (113), because, as we will shortly see, this structure
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 81
has a gap not because it is an instance of Topicalization, but because the dislocated phrase is a bare noun. In addition to a proper name, as in (112), English topicalization can dislocate any type of noun. In (114) we have a full NP. (114)
That lesson Mary knows——very well Contreras (1991)
As pointed out by Contreras (1991), this is not possible in Spanish. The only way to express (114) in Spanish is by using a CLLD, as in (115)a. Without the clitic the sentence is ungrammatical. (115)
a. Esa lección María la sabe muy bien b. *Esta lección María sabe muy bien
Thus, the ungrammaticality of (115)b shows that Topicalization is not possible in Spanish. From this perspective, the apparent cases of Topicalization such as the one in (113) need to be accounted for. I will suggest that (113) is a case of CLLD, similar to those we saw in 3.3.1, but with no overt clitic.32 The reason it does not show a clitic is that it involves the dislocation of a bare noun (BN). As opposed to other Romance languages such as Italian and Catalan, Spanish lost the type of partitive clitic that a dislocated bare noun triggers in languages that actually have them. In (116) we have an example from Catalan.33 (116)
(De) diners diuen que no en té (of) money say-3pl that not cl have-3sg “She says she doesn’t have any money” (Vallduví 1988)
The same applies to the dislocation of PPs. In languages such as Catalan that have retained this type of clitic, an overt clitic appears, while in languages such as Spanish, which lacks this clitic, the CLLD of a PP shows no clitic, Compare the Catalan examples in (117)–(118) with their Spanish counterparts in (119)–(120): (117)
(118)
De l’examen ningú no n’ha parlat encara of the exam nobody not cl have-3sg talked yet “About the exam, nobody has said anything yet” Al jardí els nens s’hi diverteixen molt in the garden the children refl cl enjoy-3pl much
82 SYNTAX-INFORMATION STRUCTURE INTERFACE
(119) (120)
“In the garden the children enjoy themselves a lot” (Hernanz and Brucart 1987) Del examen nadie ha hablado——todavía of-the exam nobody have-3sg talked yet En el jardín los niños se divierten mucho in the garden the children refl enjoy-3pl much——
Thus, the absence of the clitic in examples such as (113), (119) and (120) is not due to the existence of Topicalization in Spanish, but rather to the incomplete clitic system of modern Spanish, which has not preserved the type of clitics that are triggered in other languages when a BN or a PP are dislocated. From this perspective, despite the apparent similarities between (112) and (113), (113) should not be considered an instance of Topicalization but an instance of CLLD with no overt clitic. Rivero (1980) makes a similar argument about cases such as (121), where we have a preposed subject and no resumptive pronoun. (121)
El niño parece que està protestando siempre the child seems that is complaining always “It seems that the child is always complaining” (Rivero 1980:363)
She suggests that these are not raising structures or cases of Topicalization but dislocated structures where the pronoun has been deleted by a transformation. However, she does not extend this kind of argument to the preposing of BNs, such as that in (113). She notices that these tend to be indefinites, but concludes that Topicalization in Spanish must be restricted to indefinite NPs.34 Further evidence that cases such as (113) should be considered instances of CLLD with no overt clitic and not cases of Topicalization comes from multiple CLLDs. As exemplified above, in Spanish we can have several clitic-left-dislocated elements in the same sentence, as in (122): (122)
EI regalo a mi madre no se lo he dado todavía the present to my mother not cl cl have-1sg given yet “I have not given my mother the present yet”
If cases such as (113), (119) and (120) were real instances of Topicalization, we would not be able to have in the same sentence a
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 83
clitic left-dislocated element and a dislocated BN or PP, since we cannot have a Topicalization and a LD in the same sentence. That is, a sentence such as (123), which includes a left-dislocated subject and a topicalized object, is ungrammatical: (123)
*John, the books, he didn’t bring——
However, if it is true that the dislocation of a PP or a BN in Spanish is just another instance of the construction referred to as CLLD, then we expect to be able to mix these elements in the same sentence. As (124) shows, it is indeed possible to have both a dislocated element with an overt clitic and a dislocated BN with no overt clitic in the same sentence: (124)
A mí dinero Juan nunca me deja to me money john never me lend-3sg
The fact that we can have these dislocated bare nouns co-occurring in the same sentence with clear cases of CLLD means that these are not Topicalizations, but rather instances of the same construction: CLLD. This argument faces a potential problem, however. In Spanish we can have both a clitic left-dislocated element and a wh-moved element in the same sentence, as in (125), and even a clitic left-dislocated element and a Focus-moved element, as in (126). (125)
(126)
El regalo ¿quién lo tiene? the present who cl has “The present, who has it?” A sus padres MENTIRAS les cuenta siempre to his parents lies cl tell-3sg always “He always tells his parents lies”
The fact that this is possible, one could argue, does not mean that wh-movement, focus-movement, and CLLDs are instances of the same construction. However, there is an important difference between the structures in (125)–(126) and cases such as (122) and (124). When it is possible to have in the same sentence such different processes as CLLD and wh-movement, as in (125), or CLLD and focus-movement, as in (126), a strict order is observed. Dislocated elements must precede wh-words or focused-moved phrases. The opposite order is not grammatical.35
84 SYNTAX-INFORMATION STRUCTURE INTERFACE
(127) (128)
*¿Quién el regalo lo tiene? *MENTIRAS a sus padres les cuenta siempre
However, in cases of multiple CLLDs, the dislocated elements can appear in any order whatsoever: (129) (130)
El regalo a mi madre no se lo he dado todavía A mi madre el regalo no se lo he dado todavía
This is the case not only when we have two clitic left-dislocated elements with overt clitics, as in (129)–(130), but also when we have a clitic left-dislocated element and a dislocated bare noun with no overt clitic as in (131): (131)
a. A mí dinero Juan nunca me deja b. Dinero a mí Juan nunca me deja
In fact, we can have the bare noun intermingled in whatever order both with the clitic left-dislocated indirect object and with the subject: (132)
a. Dinero Juan a mí nunca me deja b. Juan a mí dinero nunca me deja c. Juan dinero a mí nunca me deja d. Dinero a mí Juan nunca me deja e. A mí Juan dinero nunca me deja f. A mí dinero Juan nunca me deja
The fact that these dislocated bare nouns with no overt clitic can appear in any order, intermingled with clear cases of clitic left-dislocated elements constitutes proof that we are dealing with the same structure: CLLD.36 Furthermore, although a detailed discussion of the different proposals for the syntactic representation of CLLDs is beyond the scope of this section, it should be mentioned that the fact that we can have as many phrases as we want and in any order, as (132) shows, also constitutes evidence against analyses of this construction which suggest a movement of left-dislocated elements to recursive topic positions such as the analysis that appears in Rizzi (1997). As opposed to the behavior shown in (132), movement processes are quite constrained as to the
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 85
number and order of elements moved. As proposed in Casielles (2001), CLLDs are better analyzed as base-generated adjunctions. Thus, in this section I have claimed that despite the apparent similarities between structures in Spanish which do not trigger the appearance of a clitic and the syntactic structure known as Topicalization in English, these are two different constructions. I have argued that Topicalization does not exist in Spanish and that these cases are CLLDs with no explicit clitic. The absence of the clitic has to do with the nature of the preposed element (a BN or a PP) and not with the existence of Topicalization in Spanish. I will consider now the topicalizing mechanism referred to as Right-dislocation. 3.3.4 Right-dislocation In addition to the topicalizing constructions we have considered—Left-dislocation, Clitic left-dislocation (with and without an overt clitic), and Topicalization—there is another structure which differs from the others in that the dislocation is to the right, as in (133): (133)
He is a nice guy, your brother37 (Lambrecht 1994:203)
Although most of the examples in the literature of right-dislocations in English involve the dislocation of a subject, Gundel (1988) offers the following cases, which involve the right-dislocation of direct objects: (134) a. I’ll never be able to figure them out, women b. You better get it out before the party on Saturday, this spot in the rug c. I promise I’ll write him tomorrow, your father Hernanz and Brucart (1987) offer some examples of this construction in Spanish: (135)
a. Hoy no lo ha visto nadie, a Juan today not cl has seen nobody, part. John b. La policía las recuperó ayer, las joyas the police cl recovered yesterday, the jewels
86 SYNTAX-INFORMATION STRUCTURE INTERFACE
These examples show that similarly to the dislocation of elements to the left, when English right-dislocates an element, there is an announcing pronoun, while in Spanish a clitic appears. Considering that we have claimed that CLLD and Left-dislocation are two different structures, the first question that we could address here is the following: Is Romance Clitic Right-dislocation (CLRD) different from English Rightdislocation? Notice that in the case of left-dislocation we had two structures in both Romance and English: Pronoun left-dislocation and Clitic left-dislocation in Romance and Pronoun left-dislocation and Topicalization in English: (136) (137)
a. Pedro, no he hablado con él todavía b. A Pedro no lo he visto a. Peter, I haven’t talk to him yet b. Peter I haven’t seen
First, I am not sure that the Nonclitic Romance dislocation is possible to the right: (138)
?No he hablado con él todavía, Pedro
It may improve if we repeat the preposition: (139)
No he hablado con él todavía, con Pedro
Second, I do not think that we can have a right Topicalization: (140)
?I haven’t seen——, Peter
So, it looks as if in the case of right-dislocation , there is only one structure available in both languages. We could see if the differences between CLLD and LI) that Cinque has noted apply also to CLRD and RD. The following Catalan example shows that similarly to CLLD, CLRD can dislocate phrases other than NPs and more than one at a time in whatever order. Also, the resumptive pronoun has to be a clitic. This is also true of Spanish RD. (141)
a. l’hi fiquem, el ganivet, al calaix b. l’hi fiquem, al calaix, el ganivet
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 87
A second question we can ask is what is the difference between a left-dislocated and a right-dislocated element? Does the notion of topical element apply to right-dislocated elements, too? Lambrecht (1994) has pointed out that one of the differences between right and left-dislocation is that while left-dislocations can achieve a topic-shift or introduce fairly inactive elements, right-dislocation cannot do this. That is, the right-dislocated element has to have an active status in the discourse. As opposed to left-dislocated elements, right-dislocated phrases are always unaccented. He points out that this can be clearly shown in German where the distinction has grammatical reflexes. German has two sets of personal pronouns in the third person, the set er, sie, es, sie and the set der, die, das, die. Both of these sets mean “he, she, it, they.” He states that, as a general rule, pronouns of the er-series are used when a referent is active and already topical, and pronouns of the der - series are used when a referent is active but not yet an established topic. Lambrecht points out that in most anaphoric contexts, only the er -series can be used: (142)
Wenn er/der isst, macht er/*der so komische Gerausche “When he eats, he makes kind of funny noises”
He gives the following examples of left- and right-dislocations: (143)
(144)
a. Die Mullers, die wohnen im dritten Stock b. *?Die Mullers, sie wohnen im dritten Stock “The Mullers (they) live on the third floor” a. Die wohnen im dritten Stock, die Mullers b. Sie wohnen im dritten Stock, die Mullers
While a left-dislocation only allows for the der series pronouns, pronouns marking the referent as active but not yet established as a topic, no such constraint is imposed in the case of right-dislocation. Lambrecht points out that these data confirm the existence of a general information structure principle according to which right-dislocation requires greater pragmatic salience of the topic referent than left-dislocation. So, there seems to be a difference between left- and right-dislocated elements in that while left-dislocated elements may be already established topical phrases or not, right-dislocated elements have to be already established topical phrases.
88 SYNTAX-INFORMATION STRUCTURE INTERFACE
However, as we have already seen, other authors like Vallduví claim that it is better to restrict topical phrases (which he refers to as links) to sentence-initial position and consider these right-dislocated elements as nonlink, nonfocus elements (his tails). That is, for Vallduví, the most important difference between a left- and a right-dislocated element is its position. He does not consider the difference between them in terms of the discourse status of the dislocated element, which can be an already established topical element or not in the case of left-dislocation, but only the former in the case of right-dislocation. Notice that from the point of view of the notions of aboutness and discourse-oldness, right-dislocated elements are even more “topical” than left-dislocated elements, which can sometimes be discourse-new. It seems to me that we do not need to account for the differences between right- and left-dislocated elements by saying that only left-dislocated are topical phrases. We have seen that topical phrases, even if we restrict them to a fixed position, can have different discourse features and, therefore, nothing extra is needed to characterize these right-dislocated elements. As suggested in 3.2.3, rather than a division between left-and right-dislocated topical elements, we see an important difference between dislocated or external topical phrases, on the one hand, and nondislocated or internal topical phrases, on the other hand.38 In this section I have considered several topicalizing devices. I have argued that, as suggested by Cinque for Italian, Spanish CLLD is different from English Left-dislocation and that a similar Pronoun Left-dislocation exists also in Spanish. In addition, I have shown that there is no syntactic structure in Spanish similar to English Topicalization and that structures in Spanish apparently similar to English Topicalizations are in fact cases of CLLD with no overt clitic. Finally, I have considered right-dislocated structures and have shown that in this case, there seems to be only one structure both in English and Spanish. 3.4 Towards a Typology of Topics Having considered in detail the nature of topical phrases and topicalizing devices in Spanish and English, we can come back to the problem of characterizing topics crosslinguistically. I would like to suggest that the reason why no appropriate definition of topic has been found is because this unitary entity we have been trying to define does not exist. That is, due to the variety of phonological, morphological,
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 89
lexical and/or syntactic ways in which languages can mark, if at all, topical elements, any specific definition of topic based on a particular language will not be adequate to characterize topical elements in a different language. From this perspective, I suggest giving up on the traditional view of topic and working towards establishing a feature-based typology of topics. I believe that particular combinations of certain phonological, syntactic, semantic and discourse features will give us different types of topical elements. From this perspective, the long-term goal is to examine in detail how different languages encode nonfocal elements, isolate the relevant features, and find out how they combine. Obviously, this type of comprehensive crosslinguistic study is beyond the scope of this work. However, taking into account the different characterizations of topic discussed in chapter 2, and the features of topical phrases in Spanish and English that we have seen in this chapter, we can tentatively offer a collection of features which can start us off. Thus, instead of trying to find a single requirement that all topical phrases must comply with, we can organize the characteristics that nonfocal elements may have in terms of optional features. For instance, from a phonological point of view, topical elements can be unaccented or accented. That is, although there is a tendency for topical elements to be unaccented, some topics, especially nonpronominal, discourse new elements can get an accent.39 This is especially true of English Topicalizations such as that in (145): (145) Do you watch much sports? Not much football. BASEBALL I like a lot BETTER (Ward 1988) In these cases, the nonfocal element gets Jackendoff s (1972) B accent. From this perspective, nonfocal elements can be ±unaccented. From a syntactic point of view, topical elements can be lexical or pronominal. In fact, in languages like Spanish they can also be null. Thus, as (146) shows, a topical subject can be dropped: (146)
a. ¿Qué sabes de Juan? “What do you know about John?” b. ___Está bien. ___Acaba de comprar una casa “He is well” “He just bought a house”
90 SYNTAX-INFORMATION STRUCTURE INTERFACE
Thus, we could also talk of ±overt topics. Another important syntactic feature would be ±sentence-initial. In many different languages, topical elements appear in this position. In (147), we have some examples from Latin, Czech, Russian and Japanese: (147) a. Hanc urbem condidit Romulus this city founded Romulus b. Tenhle dopis napsal tatínek this letter wrote father “FATHER wrote this letter” c. Ivan, ja ego ne ljublju (Gundel 1988 [1974]: 185) Ivan I him not like “Ivan, I don’t like him” (Gundel 1988 [1974]: 17) d. Ano-hon-wa John-ga kat-ta “That book, John bought it” However, not all topical elements appear in this position. We have mentioned pronominal topical elements which can appear in situ, as “him” in (148), and right-dislocated elements, which appear sentence-finally, as in (149): (148) (149)
I saw him yesterday Raquel los odia, los tomates Raquel cl hates-3sg the tomatoes
In addition, as we have seen, topical elements can be ±preverbal subjects and ±dislocated. In fact, in languages like Spanish which allow for multiple dislocations, topics can be ±single. Thus, in (150), we have multiple topical phrases.40 (150)
a. A Pedro los libros ya se los compré to Pedro the books already cl cl bought-1sg “I already bought the books for Pedro” b. A mí listo no me lo parece to me clever not cl cl seem “He doesn’t seem clever to me” c. Libros yo a Juan nunca se los dejaría books I to Juan never cl cl lend-1sg “I would never lend Juan books”
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 91
We also have specific or nonspecific topical elements such as that in (151): (151) Some computation he may not be able to carry out in his head This is related to the semantic-pragmatic referential feature. Although many topics refer to a discourse entity, some, like those in (152), do not: (152)
a. Estudiando nunca està studying never is b. A todos no los he visto todavía prep all not cl have-1sg seen yet c. Que fumas lo sabemos todos that smoke-2sg cl know-1pl all
In addition, many topics are discourse-old. However, we have also seen discourse-new topics: (153) That was a student of mine. Her HUSBAND had a HEART attack (Lambrecht, 1994:326) (154) Do you watch much sports? Not much football. BASEBALL I like a lot BETTER (Ward 1988) We could then say that topics can be ±discourse-old. In fact, as we have seen, we would need more specific features such as ±unused or ±active along the lines suggested by Prince (1981b) and Lambrecht (1994). Finally, Left-dislocation in English and SPLD in Spanish show a preposed topical NP, which is separated from the rest of the sentence and has a rising intonation. (155) (156)
John, I haven’t talked to him yet Juan, no he hablado con él todavía
Taking into account the fact that in this case there is a pause and a rising intonation, this type could be characterized as +prosodically independent. These are then some of the features we could start with:
92 SYNTAX-INFORMATION STRUCTURE INTERFACE
(157) Prosodic ±unaccented ±pros independent
Tentative Features Morphosyntactic ±pronominal ±overt ±dislocated ±preverbal subject ±discrete ±single ±sentence-initial ±clitic-doubled ±pronoun-doubled
Semantic-pragmatic ±referential ±discourse-old ±unused ±active (±specific) (±identifiable)
Obviously, these are very basic, broad features, which will need to be refined. In order to discover the precise features that would be most appropriate to classify topical elements crosslinguistically, we would need to examine carefully what are considered to be topical elements in many different languages. Further, recent work has suggested more specific semantic and pragmatic characteristics of topical elements. Thus, Prince (1998) has shown that what had been considered to be homogeneous topic-marking constructions seem to involve different discourse functions. She points out that Left-dislocation in English involves at least three different functions: a) simplifying discourse processing by introducing a discourse-new entity so that something can be predicated of it, b) triggering an inference on the part of the hearer that the entity represented by the initial NP stands in a salient partially ordered set relation to some entity or entities already evoked in the discourse, and c) amnestying an island violation. In a similar spirit, Jacobs (1999) has pointed out that there are different semantic relations between the topic and the comment in Topic-Comment structures. He identifies four dimensions (informational separation; predication; addressation and frame-setting) and analyzes three Topic-Comment structures in German (left-dislocation, I-topicalization and free topic) in terms of these dimensions. Whatever the most appropriate features turn out to be, they should help classify different types of topics in different languages. One of the advantages of a feature-based typology is that it can also capture the differences we saw between the element which dichotomies such as topic-comment, theme-rheme want to capture, which is typically
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 93
discrete and sentence-initial, and the element that dichotomies such as Background/Open proposition want to capture, which is typically nondiscrete. The topical element in examples such as those in (158) is sentence-initial, often a subject, and refers to a discourse entity: (158)
a. John drinks beer b. The children went to school and the parents went to bed
On the other hand, the nondiscrete topical component in sentences like (159) tends to be discourse-old and does not have any particular place in the sentence. (159)
What did Mark buy for the children? Mark bought a BOOK for the children
An additional feature that we could add to this topical element is that it necessarily co-occurs with narrow focus, focus restricted to one element. From this point of view, topical element #1 seems to have the following characteristics: a) it is restricted to a unique element; b) it correlates with the sentence-initial position, often a preverbal subject; c) it seems to be restricted to discourse entities d) it is not necessarily discourse-old and e) it is not necessarily unaccented. Topical element #2 seems to have exactly the opposite features: a) it is not restricted to a unique element; b) it does not correlate with a particular sentence position; c) it is not restricted to discourse referents; d) it is necessarily discourse-old and e) it is necessarily unaccented. Let us refer to these two types of topical elements as sentence topic (STopic) and Background respectively. We could tentatively express the differences between them with the following phonological, syntactic and discourse features.41 (160)
Sentence Topic +single +sentence-initial +referential ±discourse-old ±unaccented
Background ±single ±sentence-initial ±referential +discourse-old +unaccented
Some of these features might be the consequence of the different syntactic positions of these elements, That is, the syntactic restrictions
94 SYNTAX-INFORMATION STRUCTURE INTERFACE
that the topic seems to have might be due to its occupying a very specific syntactic position such as a specifier position. Background elements, on the other hand, do not have a special syntactic position and, therefore, are not restricted to a single, sentence-initial referential element. Thus, it seems that the topic is syntactically restricted and phonologically and discourse free while the background is discourse-restricted as well as phonologically restricted and syntactically free. Thus, in this section I have argued for a nonunitary characterization of the notion topic. I have suggested that particular combinations of certain phonological, syntactic, semantic and discourse features will give us different types of topical elements. I have pointed out some of the phonological, syntactic and semantic-discourse features we might want to start with. These features have been shown to distinguish two topical elements, which seem to have almost opposite features: the sentence topic and the background. In the next chapter we offer more evidence which supports this distinction, and which comes from the distribution of BNs in Spanish. NOTES 1. The phonological and syntactic special mechanisms that I am referring to here are cases of accented or postverbal subjects of unaccusatives in English and Spanish respectively, or cases of syntactic constructions such as passives or clefts: (i) The CHILDREN have arrived (ii) Han llegado los niños (iii) He hasn’t been seen by his family recently (iv) It was HIM that I saw 2. Consider the following examples: (i) (ii) (iii)
She has just left She has just left but he is still here SHE has just left
Acaba de salir Ella acaba de salir pero él todavía està aquí Acaba de salir ella
In English, the same preverbal pronominal subject (although with different intonations) is used in the three sentences, when we have an active (i), an inactive, new or contrastive topical element
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 95
(ii) and a focal subject (iii). However, in Spanish we have a null, a preverbal and a postverbal subject respectively. 3. There has been much research done on topical phrases and the relation between topic and subject in many different languages, at least since Li (1976). I have already mentioned the crucial work of Prague scholars for Czech. The work on Hungarian (Kiss 1981 and Horvath 1986) is also well-known. The notion of topic is particularly important in Asian languages. See Tateishi (1991) and Dwivedi (1994) for recent works addressing the status of topical phrases in Japanese and Hindi respectively. 4. There is a considerable amount of literature on the nature of Romance clitics, and on their phonological, morphological and syntactic behavior, From a syntactic point of view, there have been two main analyses: an analysis involving movement of the clitic, and a base-generation analysis. See Sportiche (1996) for a review of the two most important analyses on clitics and for the proposal that what moves is the XP associated with the clitic. 5. For the view of clitics as agreement markers see Borer (1984) and Suñer (1988) among others. See Uriagereka (1995) for an analysis of the different placement of clitics in several Romance languages, and the view of clitics as determiner heads. 6. As is well-known, these clitics were enclitisized to the verb in Old Spanish. What is interesting is that the change in the position of the clitics and the loss of verb-second from Old to Modern Spanish (Fontana, 1993) may be related to the disappearance of Topicalization. Kroch (1989) suggests that the loss of verb-second word order in
French took place via the replacement of Topicalization by Left-dislocation. 7. When a Bare Noun is detached (i) or simply dropped in a short answer to a question (ii), there is no need of a clitic: (i) (ii)
Cintas no traje tapes I didn’t bring ¿Trajiste cintas? “Did you bring tapes?”
No, no traje *”No, I didn’t bring”
The short answer in (ii) contrasts with (iii) where a nonbare noun is being dropped: (iii)
¿Trajiste las cintas?
*No, no traje
No, no las traje
96 SYNTAX-INFORMATION STRUCTURE INTERFACE
We will consider Bare Nouns in detail in chapter 4. 8. The fact that this doubling is only possible with specific objects has contributed to supporting the relation between these clitics and specificity (Suñer 1988). There have been different proposals to account for the Porteño data, and for the analysis of the doubled NP. These include a right-dislocation analysis, an apposition analysis and the proposal that doubles are adjunctal in some languages. See Uriagereka (1995) for references. 9. The IO pronoun “le” is realized “se” when it precedes another clitic. See Bonet (1991) for a morphosyntactic analysis of Romance clitics. 10. Notice the co-occurrence of the clitic with the wh-word in (28), as opposed to cases such as (i):
(i)
*¿Qué lo compraste? what it bought-2sg *“What did you buy it?”
11. Although Spanish does not have special nonsubject tonic pronouns for all persons, just for first and second person singular (mí and ti), subject pronouns preceded by the marker “a” can be used for both direct and indirect tonic forms. 12. Something that should be taken into account in investigating this correlation is the changes that occurred from Old Spanish to Modern Spanish, since in Old Spanish clitics did not co-occur with pronouns. In addition, a dislocated phrase did not trigger the appearance of a clitic. See Fontana 1993 for the proposal that in Old Spanish clitics were XPs in argument position while in Modern Spanish they are Xs. 13. Thanks to Barbara Partee (p.c.) for bringing this to my attention. 14. Unless a special intonation is given to the sentence, as we saw in (32) above. 15. See the discussion in 2.3.2.2, example (52). 16. The de-stressing of the DO and this shifting of stress to the left could be seen as a mechanism that marks the end of the VP. That is, it may be that the end of the VP can be marked syntactically (any right-dislocated phrase is outside VP) or phonologically (any postverbal de-stressed phrase is outside VP). Later we will see that while the phonological option is available in both Spanish and English, it is not available in Catalan. 17. While the verb can be replaced with a pro-form in the case of (51) where the whole predicate is part of the Background (‘the PARENTS did that’), we cannot do that in (52): *they did that to the MOVIES) 18. It has been suggested that preverbal subjects in Spanish are always specific and that quantifiers are not allowed for that reason. However, it is not clear that in (i) the subject is specific. Regarding quantifiers, I think they are fine as long as they can be accommodated as topics, which makes me
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 97
think that the requirement is the topichood, which may have as a consequence that in most cases the subject will be specific too.
(i)
Un estudiante ha dejado el programa pero no sé quién es. En realidad, nadie lo sabe.
19. Krifka (1992) uses these examples taken from Laca (1990) to point out that Spanish can use the determiner to map an element to the restrictor in tripartite structures. See Partee (1991) for the correlation between material in the restrictor and topical material. 20. Sections 3.3.1–3.3.3 have been published with the title “Left-dislocated Structures in Spanish” in Hispania 86.2, 326–338 (2003). They are reprinted here by permission of Hispania. 21 The term Topicalization has been used in a variety of different ways in the literature. Here it is always used to refer exclusively to the English syntactic structure which involves a gap and a preposed topical element. It is not used to refer to a particular syntactic position, or to the general preposing of topics or to the preposing of focused elements. 22 See Adjémian (1978), Allerton (1978), Bolinger (1954), Chafe (1976), Chomsky (1977), Contreras (1976), Gundel (1988) [1974], Halliday (1967), Hatcher (1956), Kuno (1972), Li (1976), Prince (1981a), (1984), Rodman (1974) and Vallduví (1988), among others. 23 I have modified slightly Rivero’s English glosses of examples (80) and (81). 24 In this sense, they contrast with cases of Focus Preposing. See chapter 6. 25 In the English glosses of examples (84)-(86), in addition to the nonpreposed version, a gloss containing a preposed element is also given, since it more accurately mirrors the Spanish counterparts. Note, however, that these glosses do not have a resumptive pronoun, and, therefore, are not instances of English LD, but rather English Topicalizations. As (87) shows, the Left-dislocated version with the resumptive pronoun is ungrammatical in English.” 26 In (i)-(iv) we have some of the examples that Cinque (1990:58) offers of different types of dislocated phrases and in (v) a case of multiple left-dislocation:
(i) (ii) (iii) (iv)
Al mare, ci siamo già stati to the seaside there-(we)-have already been Bella, non lo è mai stata beautiful not-it-(she)-ever was Tutti, non li ho visti ancora All not-them (I)have seen yet Che bevi, lo dicono tutti That (you)drink it says everybody
98 SYNTAX-INFORMATION STRUCTURE INTERFACE
(v)
Di vestiti, a me, Gianni, in quel negozio, non mi ce ne ha mai comprati clothes to me Gianni in that shop (he) not-tome-there-of-them ever bought
27 In all the examples taken from Cinque (1990), I have maintained his English glosses of the Italian examples whenever possible or helpful. 28 However, see 4.2.2 for other examples of CLLD which do not seem to be sensitive to island constraints. 29 See the next section for a more detailed analysis of this type of dislocation. 30 Notice the similarity between this structure and English Left-dislocation:
(i) (ii)
John, I haven’t talked to him yet Juan, no he hablado con èl todavía
31 See Bosque (1996), Casielles (1996), Contreras (1986), Suñer (1982), and chapter 4. 32 The phrase “CLLD with no overt clitic” is used here to express the fact that in Spanish, due to its incomplete pronominal system, some of these CLLDs—in particular those involving BNs, PPs or subjects—do not show a clitic. Thus, it is not intended to imply the existence of an empty category necessarily, although it does not preclude an analysis where there is a null clitic. Evidence for the existence of a null partitive clitic in cases of dislocated BNs can be found in Clements (1994). 33 Also, as Vallduví (1988) points out, further proof that the absence of the clitic in cases such as (113) is due to the BN and not to their not being CLLD is that BNs do not manifest a clitic when they are dropped either, as has been shown in Campos (1986), where no preposing mechanism is even taking place, as in (i).
34 It should be mentioned that Rivero (1980) does not take intonation into account and, therefore, uses the term Topicalization for what look like cases of Focus Preposing. Thus, she notes that there seem to be cases of Topicalization where the preposed element is not indefinite such as (i) from Hatcher (1956).
(i)
La reina de las Españas te mereces, hermoso the queen of the Spains yourself deserve, beautiful “The Queen of the Spains you deserve, beautiful one”
TOPICAL PHRASES AND TOPICALIZING CONSTRUCTIONS 99
I think, however, that this example will only sound natural when pronounced with an emphatic stress on the preposed element. That is, as an instance of Focus Preposing. 35 This restriction has to do with the different nature of these movements and the different syntactic positions involved. Both Wh-movement and focus-movement have been analyzed as A-bar movements, which involve an operator and which move elements to a unique specifier position. However, CLLDs are better seen as instances of base-generated recursive adjunctions. See Casielles (2001), Cinque (1990), Contreras (1991), Rizzi (1997) and Zubizarreta (1998) for different proposals. 36 See Rivero (1980) for a very interesting case where we seem to have a double preposing of the phrase mucho dinero, which ends up split in two:
(i)
Dinero dicen que María cree que mucho sí que tiene_____ money say-3pl that María believes that much yes that has
37. Lambrecht refers to this structure as the Antitopic construction and Vallduví as a Right-detachment. It should be pointed out that the difference is not only terminological. While Lambrecht considers the right-dislocated element to be topical, and, therefore, the existence of this construction as a proof that topics do not have to be sentence-initial, Vallduví considers these right-dislocated element to be “tails,” as opposed to “links” (his topics), which are restricted to the sentence-initial position. 38 See 4.2.2 and chapter 7 for further discussion and evidence. 39. Null topics and big chunk topics are necessarily unaccented 40. Japanese wa-topics can also be multiple. This language, however, also allows for multiple ga-subjects. See Tateishi (1991). 41. More specifically, the STopic could probably be referred to as a preverbal subject topic (Spec, TP Topic, for instance). In addition, these could be just two types at both ends of a continuum. That is, we expect to find many other types in addition to the STopic (or Spec, TP Topic) and the Background (or Wide Topic). For instance, consider the topical element which occurs in English Left-dislocation (i), and Spanish SPLD (ii), which we could refer to as a Free Topic:
(i) (ii)
John, I haven’t talked to him yet Juan, no he hablado con él todavía
English Leftdislocation Spanish SPLD
100 SYNTAX-INFORMATION STRUCTURE INTERFACE
This Free topic could tentatively be described as:
On the other hand, the Null Topic which is allowed in Spanish sujectless sentences (iv)could be described as in (v)
We could try something similar for the topical element in English Topicalization, which we can refer to as the Preposed Topic:
CHAPTER 4 Bare Nominals
4.1 Introduction We have seen that in Spanish we can have both preverbal and postverbal subjects and that this positional difference correlates with a different interpretation in terms of their information status. That is, while postverbal subjects are interpreted as part of the informative part of the sentence or focus, preverbal subjects are interpreted as topical. Thus, the sentence in (1) would be an appropriate answer to a question such as “What has John done” and (2) would be an appropriate answer to “What has happened?”/“Who has called?” (1)
Juan ha llamado john have-3sg called “John has called”
(2)
Ha llamado Juan have-3sg called john “JOHN has called”
However, this is not possible with BNs. Although they can appear postverbally (3), BNs are ungrammatical as preverbal subjects (4). (3)
Jugaban niños en la calle played-3pl children in the street “Children were playing in the street”
102 SYNTAX-INFORMATION STRUCTURE INTERFACE
(4)
*Niños jugaban en la calle children played-3pl in the street “Children were playing in the street”
In addition, preverbal dislocated topics (subjects and nonsubjects) are allowed: (5)
(6)
Estudiantes no creo que vengan students not think-lsg that come-3pl “Students, I don’t think they will come” Café no creo que tengan coffee not think-lsg that have-3pl “Coffee, I don’t think they have it” (Contreras 1986)
Finally, preverbal (nondislocated) subjects seem to improve considerably if they are focused (the informative part and intonationally prominent): (7)
LANGOSTAS destruyeron las cosechas grasshoppers destroyed-3pl the crops “GRASSHOPPERS destroyed the crops” (Contreras 1986)
So, this is the distribution of BNs that we need to explain:
From the data in (8), we can extract the following generalization: (9) BNs are not allowed as preverbal (nondislocated) subjects unless focused. In order to explain this mysterious generalization we need to consider the difference between a preverbal and a postverbal subject, the differences between a preverbal subject and dislocated topical elements, and the role of focus. This chapter is organized as follows.
BARE NOMINALS 103
In section 4.2, I consider these data from a syntactic point of view. First, I examine the differences between preverbal and postverbal BN subjects, taking into account their interpretation. I suggest that the ungrammaticality of preverbal BN subjects has to do with the fact that BNs cannot leave the VP in Spanish. Second, I consider preverbal BN subjects and dislocated BNs and suggest that, in order to understand the differences between them, we have to take into account the different positions they occupy. Furthermore, in order to reconcile the VP-external position of these topics with the generalization proposed above (BNs cannot leave the VP), I adopt a base-generation analysis of these dislocated topics. Third, I consider the case of focused preverbal subjects and I suggest that similarly to postverbal subjects, they are inside VP at LF. In section 4.3, I examine preverbal subjects and dislocated BNs, taking into account that both are interpreted as topical. I will address the following question: taking into account that preverbal subjects (as opposed to postverbal subjects) are interpreted as topical, why do subjects ever appear dislocated? That is, what is the difference in topicality between a nondislocated preverbal subject and a dislocated subject? I will suggest that while the former is indeed a sentence topic, the latter should be viewed as part of the Background. In 4.4, I relate the informational differences between preverbal subjects and dislocated elements to their syntax.
4.2 A Syntactic View In this section, I am going to look at the distribution of BNs in (8) from the point of view of the syntactic position that they occupy. I will take into account if the BN is inside or outside VP, if movement is involved and, if it is, what type of movement it is. As we will see, this will allow us to describe the distribution of BNs as follows:
104 SYNTAX-INFORMATION STRUCTURE INTERFACE
We will then be able to change (9) to (11):
(11)
BNs cannot be DP-moved
An explanation of why something like (11) would hold in Spanish (as opposed to English, for instance) will have to wait until section 4.4. I will now consider the syntactic differences between the different BNs, starting with the contrast between preverbal and postverbal subjects. 4.2.1 Preverbal vs. Postverbal Subjects 1 In order to explain the ungrammaticality of preverbal BN subjects in Spanish, I would like to start by examining BNs in a language where this is not the case, a language which allows for preverbal BN subjects, such as English, to see if this difference in grammaticality correlates with other differences regarding the nature of BNs in the two languages, which could give us a clue as to what causes (4) to be ungrammatical in Spanish.2 As is well-known, English allows for preverbal BN subjects and, therefore, the English counterpart of (4) in (12) is fine: (12)
Children were playing in the street
The first potentially crucial difference between English and Spanish is that the English counterpart of (3), the postverbal BN subject, is ungrammatical. That is, the grammaticality of (12) correlates with the ungrammaticality of (13): (13)
*Were playing children in the street3
Is this a coincidence or is it the case that a language will only allow for preverbal BN subjects if postverbal subjects are out? Before we answer this question, let us look at another important difference between English and Spanish BNs. It is also well-known that English BNs can have two different interpretations: an existential and a generic interpretation.4 Thus, while (14) has an existential interpretation, (15) has a generic interpretation: (14)
Children were playing in the street
(existential)
BARE NOMINALS 105
(15)
Children (generally) play in the street
(generic)
What about Spanish? The sentence in (4) is ungrammatical both with an existential reading (16) and a generic reading (17): (16) (17)
*Niños jugaban en la calle *Niños (generalmente) juegan en la calle
(existential) (generic)
On the other hand, the postverbal BN subject can only have the existential reading: (18) (19)
Jugaban niños en la calle Juegan (generalmente) niños en la calle5
existential *generic
So, it looks like the generic reading of a BN is impossible in Spanish (regardless of the position), while the existential reading is possible for the postverbal subject. Let us look at another language. Diesing (1992) points out that in German the two interpretations are possible, as in English. However, they correlate with two different positions. If the BN occurs following particles such as “ja doch” (which are supposed to mark the VP-boundary), the BN gets an existential interpretation (20), and if it occurs before these particles (21), it gets a generic interpretation. (20) …weil ja doch Kinder auf der Straße spielen. (existential) …since prt prt children on the street play “…since there are children playing in the street” (21) …weil Kinder ja doch auf the Straße spielen. (generic) …since children prt prt on the street play “…since (in general) children play in the (Diesing 1992) street” So, German is similar to Spanish in allowing two different positions for subjects (one outside the VP and one inside VP). However, as opposed to Spanish, it allows BNs to occupy the VP external position, where they get the generic interpretation. In order to account for the differences between English and German, Diesing (1992) adopts the Kamp-Heim theory of NP interpretation
106 SYNTAX-INFORMATION STRUCTURE INTERFACE
(Kamp 1981, Heim 1982), and assumes Wilkinson’s (1991) analysis of the generic reading of BNs. She then connects the two interpretations of BNs with the two subject positions that the VP-internal subject hypothesis (Koopman and Sportiche 1991) has made available (Spec, IP and Spec, VP) and proposes the mapping hypothesis in (22): (22)
Mapping Hypothesis: Material from VP is mapped into the Nuclear Scope Material from IP is mapped into the Restrictor
From this point of view, BNs that have an existential interpretation have to be inside VP before they are mapped into the Nuclear Scope (NS) and get bound by Existential Closure while generic BNs have to be in Spec, IP, and from this position, they are mapped into the Restrictor and bound by a Generic Operator. For a language such as English where all subjects have to be in Spec, IP at S-structure, this means that an existential BN has to lower to the position inside VP at LF, while a generic BN stays in Spec, IP. However, in German, as (20) and (21) show, there are two different positions for subjects at S-structure (which, as we have seen, can be distinguished by their relative position with respect to sentential particles such as ja doch, which are assumed to mark the VP boundary). The subject in (20) is inside VP and gets an existential interpretation and the subject in (21) is in IP and gets a generic interpretation. From this perspective, Spanish shares with German the fact that there are two different subject positions at S-structure: a preverbal and a postverbal subject position. Thus, the grammatical sentence in (18) is similar to the German (20) in that the subject can be inside VP at
BARE NOMINALS 107
S-structure. If it is the case that LF lowering is only allowed in languages which do not have a VP-internal position available for subjects, like English, it is not surprising that neither Spanish nor German allow their VP-external BNs to get the existential reading. This would explain the ungrammaticality of (16) with an existential reading. However, we still have to explain why (16) is not grammatical with a generic reading. That is, in which way is Spanish different from German? The Spanish counterpart of the English (15) and German (21) generic BNs is a nonbare definite phrase: (23)
Los niños (generalmente) juegan en el parque
If we assume that the only way to get a generic reading is by being mapped into the restrictor and bound by a generic operator, and that Diesing’s mapping hypothesis is on the right track and elements in the IP are mapped into the restrictor, what seems to be going on in Spanish is that BNs, as opposed to non-BNs such as that in (23) cannot reach the position inside IP. Let us assume that this is the case and suggest that in Spanish, the following generalization holds: (24)
The Bare Noun Movement Constraint (BNMC) I BNs cannot move out of the VP
If this is right, BNs are stuck inside VP. Thus, the only reading they can get is the existential interpretation. Why would this be? I am going to adopt Abney’s (1987) DP hypothesis and also assume that BNs are not DPs but NPs.6 From this point of view, I am going to suggest that (24) is a consequence of the NP-nature of BNs and that in Spanish so-called NP-movement is restricted to DPs, If this is on the right track we expect passives and raising structures to be ungrammatical with BN subjects. As (25)–(26) show, this is confirmed: (25) (26)
*Niños son maltratados “Children are mistreated” *Niños parecen estar contentos “Children seem to be happy”
Furthermore, the generalization in (24) not only predicts BNs to be unable to reach the preverbal subject position. As it is stated, it predicts that they are unable to move to any position outside VP, not only to the subject position. I’ll show evidence which confirms that prediction.
108 SYNTAX-INFORMATION STRUCTURE INTERFACE
Diesing (1992) points out that an object can be scrambled at S-structure in German and at LF in English and get a generic interpretation. That is, that in a sentence such as (27), the BN object can be mapped into the restrictive clause and receive a generic interpretation in a context in which whenever Esther comes upon a novel, she reads it. This interpretation is easier to get if we focus the verb. (27) Esther reads novels (that’s what she does with novels: she reads them) If the generalization in (24) is correct, and if the only way to get mapped into the restrictor in Spanish is to move out of the VP at S-structure, the Spanish counterpart of (27) should not be able to have the generic reading intended. This seems to be true. The sentence in (28) cannot have the interpretation that whenever Esther comes upon a novel, she reads it. (28)
Esther lee novelas (no generic reading)
This generic interpretation, which is also easier to get if the verb is focused, can be expressed by a definite DP object . Thus, (29) can have the interpretation intended, which could also be expressed by left-dislocation of the object, as in (30). (29)
Esther lee las novelas
(30) Esther las novelas las lee/Las novelas Esther las lee esther the novels cl read-3sg/the novels esther cl read-3sg “Esther, the novels, she reads them/The novels, Esther, she reads them” Krifka (1992) notes this fact about Spanish and other Romance languages and offers the following examples, taken from Laca (1990), which show how a DP can get out from the Nuclear Scope and be mapped into the Restrictor. Thus, the sentence in (31) is interpreted as “what cowboys do is chew tobacco” while (32), on the other hand, is interpreted with “tobacco” in the restrictor, that is: “what cowboys do with tobacco is chew it.” (31) Los vaqueros mascan tabaco the cowboys chew -3pl tobacco
BARE NOMINALS 109
“Cowboys chew tobacco” (that’s what they do: they chew tobacco) (32) Los vaqueros mascan el tabaco the cowboys chew-3pl the tobacco “Cowboys CHEW tobacco” (that’s what they do with tobacco: they chew it) (Krifka 1992, taken from Laca 1990) Thus, I have shown evidence supporting the claim that BNs, which are assumed to be NPs, cannot be moved out of the VP in Spanish, and, therefore, cannot reach the preverbal subject position. As we have seen, this analysis can also account for their interpretative possibilities. BNs inside VP get the existential interpretation, as in German. However, since they cannot leave the VP, they cannot get the generic interpretation of their English and German counterparts, which explains why it is that the generic reading is expressed with a full DP in Spanish, If this is on the right track, neither the case of dislocated BN topics (5)–(6) nor the case of preverbal focused subjects (7), introduced in section 4.1, can be instances of movement (at least not the same type of movement), since, as we have seen, they contrast with preverbal BN subjects in being grammatical. I will consider these two cases in turn. 4.2.2 Preverbal Subjects vs. Dislocated Phrases It has been claimed (Suñer 1982) that cases such as (33)–(34) differ from preverbal subjects in being somewhat contrastive. (33)
(34)
Estudiantes no creo que vengan students not think-1sg that come-3pl “Students, I don’t think they will come” Café no creo que tengan coffee not think-1sg that have-3pl “Coffee, I don’t think they have it” (Contreras 1986)
Let us examine Suñer’s arguments. Given the correlation that we know exists in Spanish between being a preverbal subject and being a topic,
110 SYNTAX-INFORMATION STRUCTURE INTERFACE
one might think that the ungrammaticality of preverbal BN subjects such as that in (4), repeated here, is due to the fact that BNs are intrinsically nontopical (or unsuitable topics) and, therefore, they cannot be preverbal subjects. Suñer (1982) explores this possibility in her first version of the Naked Noun Constraint (NNC) which says that BNs cannot be thematic subjects.7 (4)
*Niños jugaban en el parque
However, she notices the existence of examples such as (35) where similarly to our examples in (33)–(34) we have a thematic BN. (35)
Petróleo no surgió pero sí agua oil not bubbled-up-3sg but yes water “Oil did not bubble up but water did” (Suñer l982)
In (35), as in other examples she examines, there is an element (water) which contrasts with the thematic subject. She suggests that these thematic subjects are contrastive and that this makes them fine. She proposes, then, that BNs cannot be noncontrastive themes. However, even if we wanted to maintain that dislocated thematic BN subjects are allowed because they are contrastive, we would have to explain why this possibility is not available for preverbal, nondislocated subjects. That is, what would prevent (4) from being grammatical with a contrastive reading? Also, are these dislocated topics really contrastive? It seems to me that from the fact that these structures with preverbal topics may have a contrastive feeling we do not have to conclude that the topic itself is contrastive. Dislocated BN topics are also fine in examples like (36)–(37), where the topic is discourse-old and does not contrast with any other topic. (36)
(37)
-No tengo dinero. not have-lsg money “I don’t have money”
-No te preocupes. Dinero tengo yo not pron, worry money, have-1sg I “Don’t worry. Money, I have”
¿Vinieron niños a la función de la noche? came-3pl children to the show of the night “Did children come to the night show?
BARE NOMINALS 111
-Niños, vinieron a la función de la tarde children came-3pl to the show of the afternoon “Children, they came to the afternoon show” There seems to be a contrast in the informative (underlined) part. That is, between the person that has the money in (36) and between the different show the children attended in (37). However, the dislocated BNs “money” and “children” are clearly topical and they do not have to be considered contrastive.8 Therefore, I am not going to assume that the dislocated BNs in (33)–(35) are grammatical because they are contrastive, and, therefore, somehow rhematic, as Suñer does. I am going to accept that BNs can be topical, as shown by these examples. I would like to suggest that the contrast in grammaticality between (4), on the one hand, and (33)–(35), on the other hand, has to do with the position of the BN, rather than with contrastiveness. In (4), it is in the preverbal nondislocated subject position and in (33)–(35) it is dislocated; it is in a preverbal topic position. As we will see, in this position, we can have not only BNs, but basically any type of phrase. Further, I will consider these dislocated topics from the point of view of the generalization in (24), and explain how these BNs managed to get out of the VP. Suñer (1982) does not take into account the fact that her example in (35) is a dislocated BN preverbal subject, as opposed to the ungrammatical (4), which is not dislocated. Contreras (1986) notes this difference and points out that preverbal subjects should not be confused with topics. In the preverbal topic position we can have subjects as in (33) and (35) or nonsubjects as in (34) where we have a dislocated object. In chapter 3 (section 3.3.3) I have provided evidence suggesting that BN dislocated topics are cases of the same construction known as Clitic Left-dislocation (CLLD) since Cinque (1990). One of the most compelling bits of evidence, I think, is the fact that these BNs can co-occur with other clear cases of CLLD and intermingle with them in any order, as (38) shows: (38)
a. A mí dinero Juan nunca me deja b. Dinero a mí Juan nunca me dejac. c. Juan a mí dinero nunca me deja etc. “John never lends me money”
112 SYNTAX-INFORMATION STRUCTURE INTERFACE
Now, I am going to argue that, if this is true, and BN topics are instances of CLLDs, this can help us understand the differences between these topics and preverbal subjects. However, an obvious difference between dislocated BNs and clear cases of CLLD is the fact that no overt clitic appears, so before I proceed, I will address this issue. As we saw in chapter 3, the lack of a clitic when a BN or a PP is dislocated in Spanish, is related to the absence of this clitic in this language.9 That is, to the nature of what is being dislocated, rather than to these cases being instances of a different construction from CLLD (such as English Topicalization). In fact, as Vallduví (1988) points out, more evidence supporting the claim that the absence of the clitic is due to the nature of BNs, and not to their not being a CLLD, is that BNs do not show a clitic either when they are dropped, as shown in Campos (1986), where no preposing mechanism is even taking place, as in (39). (39)
¿Trajiste tenedores? brought-2sg forks “Did you bring forks?” (Vallduví 1988)
Sí, sí traje yes yes brought-1sg *Yes, I brought
I would like to connect the non-DP nature of these BNs, which is assumed here, with the fact that Spanish, in the absence of a special BN clitic does not make use of the DO clitics available, That is, with the fact that only DPs obligatorily trigger the appearance of these DO clitics in Spanish, while non-DPs do not. The correlation between the appearance of a clitic as an object agreement morpheme and specificity has already been noted in the literature (Suñer (1988) and Runner (1993) and (1994), among others. Here, I would like to relate the fact that BNs, which are assumed to be non-DPs, do not trigger the appearance of a clitic with Uriagereka’s (1991) and (1995) recent proposal that clitics can be considered determiner heads, which Roca (1992) also assumes. In this light, when a DP is dislocated or dropped as in (40) and (41), it obligatorily triggers the appearance of the clitic and when a BN (an NP) is dislocated or dropped, as in (42) and (43), no clitic has to appear. (40)
(41)
a.*Los libros no traje b. Los libros no los traje “The books, I didn’t bring them” -¿Trajiste los libros?
-No, no los traje/*No, no traje
BARE NOMINALS 113
“Did you bring the books?” (42) (43)
“No, I didn’t”
Libros no traje -¿Trajiste libros?
-No, no traje
Now, how can the analysis of BN Topics as instances of CLLD help us to explain the differences between preverbal BN subjects and dislocated BN topics? Although, as far as I know, there is no standard syntactic analysis of CLLD in the sense of a widely accepted or assumed structure, it is reasonable to suggest, as shown in (44), that these dislocated elements are adjoined to a maximal projection (let us vaguely say IP) rather than occupying a specifier position, since we can have any number of them (in any order). (44)
Subjects, on the other hand, of which we can only have one (at least in Spanish and English) are assumed to raise from the VP-internal subject position to occupy the VP-external specifier position: (45)
One of the most important differences between these two positions (XP, IP and Spec, IP) for our purposes, is the fact that we can have as many adjoined phrases as we want. In addition, while the subject has to be a nominal/determiner phrase, in these adjoined positions, we can have any type of phrase whatsoever. Recall our by now familiar cases of PP,
114 SYNTAX-INFORMATION STRUCTURE INTERFACE
AdjP, VP, QP and CP topics mentioned in chapters 2 and 3, repeated here for convenience:
(46) (47) (48) (49) (50)
De la conferencia no he oído nada of the lecture not have-1sg heard nothing Listo no lo parece clever not cl seems Estudiando nunca està studying never is A todos no los he visto todavía prep all not cl have-1sg seen yet Que fumas lo sabemos todos that smoke-2sg cl know-lpl all
From this point of view, it is not surprising that BNs are allowed in the adjoined left-dislocation position, since this is a very liberal position where any type and number of phrases are allowed. On the other hand, they are not allowed in the specifier position, where nominal agreement features have to be checked in a spec-head configuration, which, may only be possible in the presence of a (full) D. However, even if it sounds reasonable that BNs be allowed in adjoined positions where any type of phrase is fine, we are still left with the question of the origin of these dislocated BN topics. Looking at the position of these BN topics, it appears that these BNs managed to escape the VP, which would mean that they are violating our BNMC in (24). There is a syntactic analysis of CLLD (Cinque 1990) which can give us a way out of this problem. According to Cinque, CLLDs are base-generated structures, which involve no movement. As Vallduví (1988) points out, these dislocated BNs do violate the wh-island constraint as in (51), and can also violate the CNPC, as in (52).10 (51) Dinero te pregunta (que) por qué no tiene money cl ask-3sg (that) why not have-3sg “Money, she asks you why she doesn’t have” (52) Dinero dicen que cree que tiene ganas de ahorrar money say-3pl that believe-3sg that have-3sg desires of save “Money, they say that he believes that he has the desire to save” (Rivero 1980)
BARE NOMINALS 115
Thus, if this is right, and CLLDs involve no movement, there has been no violation of (24). These topics have been base-generated up there and, therefore, they never escaped the VP. Further, even if it turns out that these dislocated phrases involve some type of movement (or whatever the right metaphor for the relation between two positions is: be it real movement, copying, coindexing, etc.), it would certainly be a different type of movement than subject-raising. From this perspective, (24) might not be so much a restriction on movement but a restriction on a specific type of movement. Thus, let us substitute “move” in (24) with “DP-move.” (53)
THE BNMC II BNs cannot be DP-moved out of the VP
From this point of view, even if the topics in (38) have been moved, this is obviously not a DP-movement like the one involved in subject-raising, as the different non-DP phrases in (46)–(50) show. Now, (53) is a language-particular stipulation and although it looks simpler than the one we started with in (9) (BNs are not allowed as preverbal (nondislocated) subjects unless focused), it is not less mysterious. In which way are Spanish BNs different from English BNs, for example? We would like to derive such a generalization from some general non ad hoc principles. That is, is there any known difference between English and Spanish of which (53) is a consequence? One possibility is that the nature of BNs may be different in these two languages (e.g. it may be a difference between being an NP and being a DP (with a null D)). Another possibility is that DP-movement may be different. In 4.4, I am going to explore the second possibility and suggest that (53) is a consequence of the different nature of DP-movement in English and Spanish. I will now consider the case of preverbal focused subjects. 4.2.3 Focused Subjects As has been noticed in the literature, BN preverbal subjects seem to be possible, if focused intonationally: (54)
LANGOSTAS destruyeron las cosechas grasshoppers destroyed-3pl the crops “GRASSHOPPERS destroyed the crops”
116 SYNTAX-INFORMATION STRUCTURE INTERFACE
(55)
ESCLAVOS construyeron las pirámides slaves built-3pl the pyramids “SLAVES built the pyramids” (Contreras 1986)
Prima facie, one of the striking things about these focused subjects is their position. As we have seen, focused subjects appear postverbally in Spanish and it is only topic subjects that are placed in the preverbal subject position. So, the first question we can ask regarding the position of these focused subjects is if they really are in the preverbal subject position. In fact, not only subjects but also objects, as in (56), and Prepositional Phrases as in (57) and (58) can be focused and placed in a preverbal position. (56)
DINERO necesitan los estudiantes money need-3pl the students “MONEY students need”
(57)
DE DOS PARTES consta el examen of two parts consist-3sg the exam “TWO PARTS the exam consists of”
(58) EN PRIMAVERA visitó Juan Leningrado in spring visited-3sg John Leningrad “IN SPRING John visited (Hernanz and Brucart 1987) Leningrad” So, we could think that the subject in (54) and (55) is not really in the preverbal subject position but has been moved to that preverbal position where all kinds of focused-moved elements can appear. From this viewpoint, all the constructions in (54)–(58), could be cases of Focus-movement, which, as (59) shows, also exists in English.11 (59)
FIDO they named it (Prince 1981a)
The position of these preverbal focused elements could be Spec, ΣP (Laka 1990) or a focus projection (Uriagereka 1992). However, saying that they are not really in the preverbal subject position would not solve our problem, since we still have to explain how
BARE NOMINALS 117
they got there. In fact, in the light of (60)–(63), which show that a preverbal subject is incompatible with a focused preverbal element or with a wh-word, it has been argued (Diesing 1990, Bonet 1990, Goodall 1992, and also Vallduví 1991) that in the languages in which this happens, they are all occupying the same position: Spec, IP. (60) (61)
*DE DOS PARTES el examen consta *EN PRIMAVERA Juan visitó Leningrado
(62)
*¿De cuántas partes el examen consta? of how-many parts the exam consists “How many parts does the exam have?”
(63)
*¿Qué Juan visitó? what John visited-3sg “What did John visit?”
Therefore, it may be the case that they are in fact in the preverbal subject position at S-structure. Now, why is it that preverbal nonfocused BN subjects are not allowed in this position and preverbal focused BN subjects are? First, the type of movement involved may be different. Focus-movement has been claimed to be similar to wh-movement (Cinque 1990). It involves an operator. Therefore, even if preverbal focused subjects occupy the same position as unfocused subjects we could still explain the difference in terms of the type of movement. That is, we could still maintain our generalization, since it only refers to DP movement, not to Focus-movement, Second, it may be the case that these focused subjects are only in the same position as unfocused subjects at S-structure. Partee (1991) and Diesing (1992) have pointed out that there is a correlation between Focus and the Nuclear Scope; in Diesing’s mapping theory, between Focus and the VP. From this point of view, the focused subjects in (54) and (55) may be inside the VP at LF. That is, through Focus movement, BNs move out of the VP at S-structure, but they lower at LF. If this is correct, all cases of focused preverbal subjects are predicted to be cases in which the BN gets an existential reading. That is, the generic reading of a preverbal focused BN is predicted to be excluded (in order to have a generic reading the BN would have to stay in IP). As (64) shows, this is the case. Focus does not allow a BN to stay in IP. A generic phrase can be focused but it has to be a definite DP, as in (65).
118 SYNTAX-INFORMATION STRUCTURE INTERFACE
(64)
(65)
*LINGUISTAS son inteligentes linguists are-3pl intelligent “LINGUISTS are intelligent” LOS LINGUISTAS son inteligentes
Thus, this type of movement, which involves an operator, allows BNs to come back inside VP at LF and get an existential interpretation. However, BNs cannot be DP-moved (so-called “NP movement”). This has two consequences: BNs cannot have a generic interpretation (BNs cannot move to Spec, IP and stay there) and preverbal BNs cannot get an existential interpretation (BNs cannot DP-move to Spec, IP and lower at LF). This explains the difference between the grammatical (66) and the ungrammatical (67)-(69). (66) (67) (68) (69)
NIÑOS jugaban en el parque *Niños (generalmente) juegan en el parque *Niños son maltratados *Niños jugaban en el parque
In sum, in this section we have considered the distribution of BNs in Spanish from a syntactic point of view, taking into account the position and origin of the BN and the type of movement involved. I have suggested that an important syntactic difference between the ungrammatical case of preverbal BN subjects and the rest of the grammatical cases is that only the former involves DP-movement We have argued that neither postverbal subjects nor preverbal topics involve movement. Finally, we have seen that focused subjects may be instances of focus-movement which is different from DP-movement. From this point of view, the distribution of BNs is analyzed as follows:
Thus, from a syntactic point of view, we can describe the distribution of BNs as a consequence of our BNMC in (53); BNs cannot be DP-moved out of the VP. Now, what does that mean? In order to answer that question, I am going to examine the data from an information point of view.
BARE NOMINALS 119
4.3. A TFA View We have just examined the distribution of BNs from a syntactic point of view, taking into account the different syntactic positions and type of movements involved. However, if we look at this distribution from a TFA point of view, in terms of their topical vs. focal status, as in (71), it is even harder to describe what is going on, since the grammatical cases are the focused BNs+one type of topic (our external topical phrase) and the ungrammatical, the other type of topic (our internal topic phrase). (71) a. Jugaban niños en el parque b. LANGOSTAS destruyeron las cosechas c. Dinero tengo yo d. *Niños jugaban en el parque
Focal Focal Topical (external) Topical (internal)
So, now the challenge is to explain why it is that BNs cannot be internal topical elements. In fact, we should ask not only why should one type of topical element be restricted to DPs, but also, why does Spanish have two different positions for topical phrases? More specifically, considering that subjects have a particular position in the sentence that they occupy when they are topical, why do subjects ever get dislocated? That is, what is the difference (besides the position) between a dislocated subject and a nondislocated one? Furthermore, do phrases get dislocated because they are topical or because they are not part of the focus? Is there a difference between being topical and being nonfocused?12 I will argue that the syntactic difference that Spanish makes between dislocated and nondislocated phrases correlates with an informational difference, From this point of view, although I do not think the notion of contrast is involved, Suñer’s intuition that dislocated topical elements differ from preverbal subjects is on the right track. I will suggest that dislocated phrases are not so much topics as unfocused elements, that is, Background elements. From this perspective, only preverbal subjects are topics in Spanish.13 Therefore, a sentence has only one topic, if any. As suggested in chapter 3, we can refer to this strict notion of topic as the sentence topic (STopic). Dislocated phrases, on the other hand, are Background elements, which means that they are not part of the focus.14 If this is right, phrases get dislocated in Spanish not because they are topics, but because they are not part of the focus. Consider the following contrast between an English Topicalization and a CLLD:
120 SYNTAX-INFORMATION STRUCTURE INTERFACE
(72) (73)
These articles I haven’t read Estos libros los ha traído el cartero these books cl has brought the postman
As mentioned in previous chapters, one of the differences between the CLLD in (73) and topicalizing mechanisms in English such as that in (72) is that by removing the DO from its VP-internal position in (73) we leave the subject in sentence-final focus position. That is, (73) is an adequate answer to a question such as “Who has brought these books?” and by displacing the DO (which is not part of the informative part), we maintain the correlation focus-sentence-final position (see 5.2). Now, is this dislocation of uninformative elements obligatory? That is, what would be wrong with answering that question with the sentence in (74) where the DO is left in situ preceding the subject? In this case, the correlation between the informative part and the sentence-final position would be maintained also. (74)
?Ha traído estos libros el cartero
The problem with (74) has to do with the fact that, as we will see in more detail in the next chapter, sentence-final focus projects. Therefore, the DO in (74) is interpreted as informative, which clashes with the question. From this point of view, CLLDs can be viewed as the dislocation of uninformative elements, which is caused by the projective nature of focus. Uninformative elements, which we can refer to as Background elements, are elements that are nonfocal, rather than elements that are STopics. Thus, CLLDs differ from structures such as English Topicalizations. Notice, that in addition to the difference noted above with respect to the placement of the subject in the two structures in (72)–(73), another difference between these two structures is the fact that while only one element can be preposed in the English construction, we can clitic left-dislocate several phrases, as we have seen, This is not surprising, if the English construction involves the preposing of a topic (only one per sentence) while the CLLD involves the dislocation of Background elements (we can have as many nonfocal elements as we want). From this perspective, the reason why dislocated preverbal BNs are fine is because these are not topics but Background (nonfocused). Now, if we look at the distribution of BNs in Spanish again, we can say that what the three grammatical cases have in common is the fact that
BARE NOMINALS 121
the BN is not a sentence topic. It is either focused, by its position as in (75), or intonationally as in (76), or part of the Background as in (77); (75) (76) (77) (78)
Jugaban niños en el parque LANGOSTAS destruyeron las cosechas Dinero tengo yo *Niños jugaban en el parque
Focus Focus Background *STopic
Therefore, from a TFA point of view, we can describe the distribution as follows: (79)
The BN Topic Constraint BNs cannot be STopics
Now, what is the relationship between this constraint and our syntactic constraint in (53)? I am going to suggest that they are equivalent. I am going to argue that DP-movement in Spanish is Topic movement. From this perspective, the different behavior of BNs in English and in Spanish will be argued to be the result of the different nature of DP movement in the two languages. In addition, we need to explain the connection between the two types of topical elements and their syntax. That is, we would like to know what drives topics to the preverbal subject position and why are Background elements dislocated. 4.4 Towards a TFA-syntactic view In this section, I am going to tentatively suggest that we can put our two generalizations together by viewing DP-movement in Spanish as a topic-driven movement. At least since Chomsky (1995), movement is viewed as a last resort feature-checking motivated mechanism. Movement of subjects to Spec, IP, in particular, is claimed to be related to the checking of agreement features in a spec-head configuration. If this is on the right track, the restriction to DP-movement can be related to spec-head agreement requirements, as follows: the spec-head agreement/checking procedure between the verb and the subject in Spec, IP can only take place through the Determiner head. However, this would not account for the fact that only topical subjects DP-move in Spanish. That is, in English, VP-internal subjects are not allowed and it is assumed that they cannot get nominative case. In Spanish, this cannot be the reason for subject raising, since
122 SYNTAX-INFORMATION STRUCTURE INTERFACE
VP-internal subjects are fine. That is, as we have seen, subjects only raise to a preverbal position if they are topics. Therefore, subject raising in Spanish is not related to agreement or checking (or nominative case). I propose that subject raising in Spanish is related to the notion of topic. Nevertheless, Chomsky’s general conception of movement as a feature-checking mechanism can be kept, if we suggest that subject raising in Spanish is motivated by the checking of one feature: a topic feature. This is an optional DP feature and, therefore, it is only available for DPs, not BNs. That is, a DP with the feature [+topic] has to DP-move to the preverbal subject position (the topic position) where the feature gets checked. From this perspective, BNs in Spanish could never be DP-moved since they cannot have the topic feature.15 What about Background elements, why are they dislocated? As mentioned in the previous section, I think this has to do with being out of the focus domain, so that the focus feature that projects upward does not reach them. As we will see in detail in the next chapter, sentence-final focus can project up to the whole VP, but not beyond. Therefore, elements which are not part of the focus are placed outside the VP. In terms of features, I am not sure if they should be marked [−focus] or if any element which is not marked [+focus] gets automatically base-generated outside the VP. Although sentence-final focus also projects in English, English chooses another mechanism to prevent focus from reaching these Background elements. Instead of placing them outside the focus domain through dislocations, it prevents focus from projecting by placing it in a non-sentence-final position. For instance, notice the contrast between the following sentences: (80)
What did Mark buy for the children? Mark bought a BOOK for the children
(81)
¿Qué les compró Mark a los niños? A los niños Mark les compró un LIBRO
That is, in a context like this, in which everything but the DO is Background information, we have two options: we can either place the focus in a position where it cannot project (non sentence-final), as in English, or we can place the Background elements outside the focus projection domain, as in Spanish.16 In sum, in this chapter I have considered the distribution of BNs in Spanish. First, I have suggested that BNs are syntactically constrained in that they cannot DP-move. Second, I have argued that from a TFA
BARE NOMINALS 123
point of view, the situation can be described as BNs not being able to be sentence topics. Third, I have suggested how we could reconcile these two views, by assuming that in Spanish DP movement is in fact a topic-feature checking mechanism and that the topic feature is a feature of D, so BNs can never bear this feature. Finally, I have suggested that dislocated phrases are Background elements (unfocused) rather than topics and I have connected the position of these elements with their unfocused nature by suggesting that their dislocation has to do with escaping the focus domain (the VP) which the focus feature can reach.17 Above, we asked ourselves why Spanish would have two positions for topics and we have now suggested that the answer is that it does not, that what we thought were topics are not, and that there are at least two different types of topical elements. Is that the case of focus, too? Nothing I have said so far explains why in some instances focused subjects, such as that in (76), can appear in a preverbal position, the position normally reserved for topical subjects, when they could have occupied their unmarked postverbal position. Now, we have to examine the relation between focus and different syntactic positions and see if there are different types of foci. I turn to these questions in chapter 5, NOTES 1. Different parts of 4.2.1 and 4.2.2 have appeared in Casielles (2001), which also contains some material from 7.3. 2. See Contreras (1986) for an analysis of the ungrammaticality of preverbal BN subjects in Spanish in terms of an ECP violation caused by an empty QP, which is postulated inside the structure of the BN. 3. This contrasts with (i), which is also fine in Spanish (ii).
(i) (ii)
There were children playing in the street Había niños jugando en la calle
4. For analyses on the interpretation of BNs in English see Carlson (1977), Heim (1982), Kratzer (1989) Wilkinson (1991) and Carlson and Pelletier (1995). 5. It should be pointed out that (19) is ungrammatical with the intended meaning (children in general play in the street). That is, with a generic interpretation of the BN itself, which can only be expressed in Spanish by a definite: “Los niños generalmente juegan en la calle.” However, (19) is fine with an existential reading of the BN and a generic time (Generally, there are children playing in the street).
124 SYNTAX-INFORMATION STRUCTURE INTERFACE
6. See Longobardi (1994) for the assumption that BNs are DPs with an empty D and the analysis of ungrammatical preverbal BNs in Italian as an ECP effect. It should be noted that this purely syntactic proposal cannot account for the unavailability of a topical interpretation for postverbal BN subjects. 7. Suñer uses the terms “thematic” and “rhematic.” Although as we have seen in chapter 2, there are differences between the notions of topic and theme (and even between different notions of “theme”), none of these differences is crucial for the discussion here. Therefore, we can equate, for the purposes of this section, thematic with topical and rhematic with focused. 8. Vallduví (1990) points out that the feeling of contrast that is obtained in most of these constructions is just an artifact of the set understanding that licenses the construction. See Prince (1984), Ward (1988) and Ward and Prince (1991) for the view that contrastiveness is derivable as a side-effect. 9. See Fernández-Soriano (1993) for a very interesting collection of articles on the analysis of clitics in Spanish. 10. These examples contrast with the examples in 3.3.1 where it seemed that both Italian and Spanish CLLDs were sensitive to islands. So, it looks like a careful analysis of the type of examples and type of islands is needed. I leave this for further research. 11. This construction is considered in detail in chapter 6. 12. Thanks to Winnie Lechner for bringing this to my attention. 13. As we will see in more detail in chapter 7, this is similar to Vallduví’s suggestion of restricting topics (his links) to the sentence-initial position. I restrict them even further to a particular sentence-initial position: the preverbal subject position. 14. I consider the question of how topic, background and focus form an information articulation in chapter 7, where I examine Vallduví’s trichotomy. 15. The relation between this and the unavailability of the generic reading needs to be investigated, since an instance of an informative generic DO (which a priori has nothing to do with Spec, IP or with being a topic) has to be a DP anyway:
16. From this perspective, Spanish syntax is mandated not only by the feature +topic but also by the nature of focus. In English, on the other hand, the syntax is mandated by case requirements which overrule any topic/focus needs, which are then expressed phonologically. 17. I am using VP here vaguely, just in contrast with IP. Zubizarreta (1998) suggests more specifically that the focus domain is formed by all the categories dominated by AspP.
CHAPTER 5 Focus
5.1 Introduction As already mentioned in the introduction to chapter 2, although it is generally agreed that sentences have a less informative or more topical part, and a more informative or less topical part, there has been no agreement as to the point and nature of this split. We have already considered the problems of identifying a topic and the different notions available in the literature, and we have proposed that two different topical elements should be distinguished. Now, we turn to the notion of focus. Prima facie, the task of identifying the focus of a sentence may seem easier than that of identifying a topic, since, as opposed to topics, foci are explicitly marked phonologically: they are prosodically prominent. In addition, since this part of the dichotomy constitutes the informative part, we have what looks like the perfect test for identifying the focus of a sentence. In an appropriate response to a wh-question, the answer to the question, the informative part constitutes the focus. So, as an answer to the question in (1), we can identify in (2) the focus part of the sentence as that which constitutes the new information and is intonationally prominent, namely, “John.” (1)
Who did Peter see?
(2)
Peter saw JOHN
The answer in (2) also exemplifies another common characteristic of focused elements: their sentence-final position.1
126 SYNTAX-INFORMATION STRUCTURE INTERFACE
However, the relationship between focus, new information, stress and sentence-final position is not so simple as what the exchange in (1)-(2) might suggest. We will see below that not all focused elements get stress, that some stressed elements are not necessarily focused, that focus may fall on not-so-new information and that some focused phrases prefer sentence-initial position. This chapter is organized as follows. I will consider in 5.2 some of the mismatches between focus, stress, new information and sentence-final position. Section 5.3 will be devoted to examining some of the types of focus proposed in the literature. We will consider Rochemont’s (1986) division into Presentational and Contrastive Focus, Gundel’s (1994) Psychological, Semantic and Contrastive Focus, Erteschik-Shir’s (1997) notion of Metalinguistic Focus, Vallduví’s (1990) Retrieve-Add vs. Retrieve-Substitute Focus, Lambrecht’s (1994) Sentence-, Predicate-, and Argument-Focus, and Kiss’s (1998) Identificational vs. Information Focus. I will then consider the position of focus in the syntax and how it can project in section 5.4. Finally, I will offer some concluding remarks in 5.5. Focus Preposing constructions will be examined in detail in the next chapter. 5.2 Focus, Accent, New Information and Sentencefinal Position As mentioned in the introduction to this chapter, focus can in many occasions be identified by being stressed, being the informative part of the utterance and occurring towards the end of the sentence. However, this is not always the case. Let us consider first the relation between focus and accent.2 First, there are at least two cases which suggest that stress may not be a necessary feature of focused elements. The first case has to do with the fact that not all the elements of what constitutes the informative part of a question-answer pair are always stressed. As pointed out by Selkirk (1984 and 1995), verbs do not have to be accented even when they are part of the focus of a sentence. Consider the exchange in (3): (3)
What did Mary do?
She cooked LASAGNA
Although the answer to the question, the informative part, includes the verb and the direct object in this context, only the direct object gets accented. According to Selkirk, cases like these show that focus can
FOCUS 127
project from an internal argument to its head. Later we will come back to these cases and consider in more detail how focus projects. The second case pointed out by Rochemont (1986) involves wh-questions in English such as that in (4): (4)
What did you BUY?
The wh-word in (4) need not be accented in order to receive its interpretation as focus, which again suggests that accenting may not be a necessary condition for interpretation as focus. Second, the example in (4) also shows that accenting may not be sufficient for interpretation as focus. As pointed out by Rochemont (1986), while the verb “buy” in (4) gets an accent, it need not be interpreted as focused. Another case suggesting that accenting may not be sufficient for interpretation as focus has to do with the existence of accented topics. As we have seen, in a situation in which there are several topics in the context, or in which a new topic is being introduced, we can have an accent on these topics (in addition to the focus-related accent). Thus, in (5) in addition to the focus-related accents on the informative part (school and bed), we also have accents on two different topics (children and parents): (5) What did the children and the parents do? The CHlLDREN went to SCHOOL and the PARENTS went to BED Another instance in which we can have accented topics is in English Topicalization, as in (6): (6)
THOSE Articles I haven’t READ
There are several proposals concerning these cases of stressed topics, which will be considered later. Some maintain a strong correlation between stress and focus, consider this accent on topical elements to be also a focus-related accent, and, propose that focus can occur both inside the topic and inside the comment in the Topic-Comment dichotomy. Other analyses, however, will propose that accent can mark either a topic or a focus relation, that there are no such things as focused topics, and furthermore that the accent that topics can
128 SYNTAX-INFORMATION STRUCTURE INTERFACE
get is different from the focus-related accent (Jackendoff B and A accents respectively). Thus, despite the relatively direct relationship between accentand focus, there are both cases of unstressed focused elements (such as verbs which inherit focus from an argument and wh-phrases, which seem to be inherently focused) and cases of stressed and not necessarily focused elements (such as verbs in the sentence-final position of a wh-question and contrastive subjects or topics). Regarding the relation between focus and new information, Rochemont (1986) has pointed out that there are cases where focus does not correlate with new information. He suggests that the existence of focused pronouns exemplifies this. That is, for him pronouns are old information and since these can be focused, the correlation between focus and new information does not hold in these cases.3 However, we will see that it is important to separate the notion of discourse-old and that of new information, meaning the informative part of an utterance. As mentioned already in chapter 2, the fact that a pronoun may refer to a prominent, discourse-old referent (“active” in Lambrecht’s terminology) does not mean that it cannot constitute new information and be the focus of the sentence. Thus, in the following exchange, the answer to the question, the informative part or focus is the direct object, which happens to be active enough that it can be encoded in a pronominal form: (7)
Who did they call?
Pat said they called HER
I will consider these cases in detail in the next section in the context of Rochemont’s Contrastive Focus. Another case which indirectly may call into question the relationship between focus and new information is the structure referred to as Focus Preposing or Focus Movement, as in (8): (8)
—I think she was Japanese —No. KOREAN she was
Although it is clear that “Korean” constitutes new information, and in fact the only new information in the sentence in which it takes place, since “she was” is clearly part of the background, this structure looks prima facie more similar to Topicalization than to wh-movement, which can point to the existence of stressed or focused topics.4 Furthermore, it has been suggested that Focus Preposing (FP) in English (Ward 1988)
FOCUS 129
shares with other preposing mechanisms such as Topicalization the fact that the preposed element is related to a previously evoked set. From this perspective, there may be something topical or discourse-old about these focused preposed elements. However, even if FP in English happens to be analyzed as a case of topical focused elements, that does not mean that in these cases the focused element constitutes old information. As we have seen, topics do not have to be discourse-old necessarily. Discourse-new elements can be introduced as topics. Thus, if we agree on a definition of new information in terms of the informative part of a proposition regardless of its previous appearance in the context, the relation between focus and new information seems pretty strong. We will come back to the relation between focus, accent and new information, introduced here, at different points in this chapter and the following. Let us finally present some preliminary comments on the relationship between focus and sentence-final position. Similarly to the relation we considered between topics and the sentence-initial position, there is a relation between focus and the sentence-final position. As pointed out by Rochemont (1986), unmarked sentence accent is sentence-final in English. That is, putting aside for the moment the discrepancies between stress and focus, which arise in wh-questions, this means that focused elements in declarative sentences tend to be sentence-final. In fact, the unmarked character of sentence-final accents is one of the reasons for which nonfinal sentence accents have been referred to as “Contrastive,” as we will see in the next section. The correlation between focus and sentence-final position is enforced in Romance languages, which can place any focused element (including subjects) in sentence-final position. Thus, the examples in (9) and (10) from Spanish and Catalan respectively are perfectly adequate answers to a question such as “Who washed the clothes yesterday?” (9)
Ayer lavó la ropa Pedro yesterday washed the clothes Pedro “PEDRO washed the clothes yesterday”
(10)
Ahir va rentar la roba el Pedro
In English, however, and with the exception of a few cases mentioned in example (45) in chapter 2 (section 2.3.2.2), this is not possible. In fact, in Catalan, this tendency of correlating focus with sentence-final
130 SYNTAX-INFORMATION STRUCTURE INTERFACE
position does not admit exceptions. That is, focus has to be rightmost and anything following the focus has to be syntactically detached and phonologically separated by using a right-dislocation structure: (11)a.
*L’amo ODIA el bròquil The boss HATES broccoli
b.
L’amo l’ODIA, el bròquil The boss cl-HATES, broccoli (Vallduví 1990)
In English, focus can occur further to the left without affecting the syntax. That is, although it is true that an out-of-the-blue statement is probably going to follow the structure Topic-Focus, and, therefore, place the accented focus towards the end of the utterance in sentence-final position, it is possible in many contexts to focus a phrase further to the left, including the leftmost element in a sentence. In (12)–(14), we have some examples: (12)
—I heard he has given his house to his wife —He has given EVERYTHING to his wife
(13)
Has he cooked the chicken? No, he has BURNT the chicken
(14)
Who called you this morning? MARK called me
Thus, in English we can basically focus in situ any element whatsoever, if the right context is provided. That is, we can allow phonology to do the job of marking the focus and not worry about a nonfocused element occurring to the right of a focused one in the syntax. Spanish tends to maintain sentence-final focus and, therefore, the English sentences in (12)–(14) can be rendered as in (15)–(17): (15)
(16)
—He oído que le ha dado su casa a su mujer —A su mujer le ha dado TODO ¿Ha cocinado el pollo? No, el pollo lo ha QUEMADO
FOCUS 131
(17)
¿Quién te llamó esta mañana? Me llamó MARK
However, I think that it is also possible in Spanish to have a non dislocated unfocused element occurring to the right of a focused element, as in English: (18)
—He oído que le ha dado su casa a su mujer —Le ha dado TODO a su mujer
(19)
¿Ha cocinado el pollo? No, ha QUEMADO el pollo
(20)
¿Quién te llamó esta mañana? MARK me llamó
Thus, it looks like some languages, such as Catalan, admit no exceptions to the correlation focus/sentence-final position. Other languages such as Spanish can go either way in that even when this correlation can be enforced as in (15)–(17), where focused elements are placed in sentence-final position, the option of phonologically shifting focus to the left is also allowed. Finally, in English this correlation cannot always be enforced syntactically and, therefore, focus can occur further to the left freely.5 I will finish these introductory comments about the relation between focus and the sentence-final position by noting that cases of FP mentioned in regard to the relation between focus and new information above, also constitute another instance of a mismatch in the correlation between focus and the sentence-final position. Interestingly, the three languages mentioned above, although with different restrictions that we will see later, allow for this structure: (21)
(22) (23)
I can’t believe this! The boss is going crazy! BROCCOLI he wants now BRÉCOL quiere ahora BRÒQUIL vol ara
It is worth pointing out that this structure is very peculiar in that the syntactic displacement of the element involved (the direct object in the
132 SYNTAX-INFORMATION STRUCTURE INTERFACE
case of (21)–(23) above) cannot be said to be motivated by an effort to maintain the TFA, as in the case of all the topicalizing devices we have seen, which place a nonsubject topic in sentence-initial position. In this case, the syntactic displacement does not help maintaining the unmarked order Topic-Focus. On the contrary, it involves a reversal of this order. This construction will be considered in detail in the next chapter. 5.3 Types of Focus In addition to the not-so-straightforward relation between focus, accent, new information and sentence-final position that we have just considered, when approaching the notion of focus we also face the problem of extracting a unified notion from all the different types of focus that we find in the literature. That is, we face the opposite problem to the one we encountered when examining the notion of topic. In the case of topic, we found too many incompatible notions trying to characterize a supposedly unique and homogeneous topic object. Confronted with that situation, we pointed out that the reason that no definition could account for all observed instances of topical elements was that there was no such thing as a unique object. In the case of focus we face the opposite problem: there are already too many divisions, and we need to consider if all those types are necessary or if they can be unified and reduced. I am going to consider several proposals in the literature concerning different types of focus and point out some of the problems they face. I will start with Rochemont’s (1986) division between Presentational and Contrastive Focus, 5.3.1 Rochemont’s Presentational vs. Contrastive Focus 6 Before he considers different types of focus, Rochemont (1986) examines the relation between focus and accent and points out some cases where we cannot identify these two notions, as we have mentioned in the previous section. He also argues against the view that every nonrightmost accent is a case of a contrastive accent.7 For instance, he points out that there is nothing contrastive about the accent in a context such as that in (24):
FOCUS 133
(24)
Who followed Ralph into the bedroom? LAURIE followed Ralph into the bedroom
That is, he points out that under appropriate contextual conditions, new information can be signaled by nonrightmost accents, such as that in LAURIE, without any contrast being involved. From this point of view, sentential stress contrasts with lexical stress, which, he points out, is predictable, regardless of the context.8 Sentential stress, on the other hand, contributes to the wellformedness of discourse information structure. Thus, even though sentential stress tends to be rightmost, it can be moved to the left, as in the answer to the question in (24), without any contrastive interpretation.9 Another case of nonrightmost noncontrastive stress that Rochemont considers-and that is going to be important later—is the case of presentational verbs such as “appear,” which have in the unmarked case an accented subject He offers the contrast in (25)–(26), and suggests that verbs like “appear” are special in that they do not function as a focus of new information, but just serve to draw attention to the new information status of the subject, which is accordingly accented: (25)
The case was judged. Then a LAWYER appeared
(26)
The case was judged. Then a lawyer APPEALED
It should perhaps be clarified here that Rochemont is not arguing against the existence of such a thing as a Contrastive Focus, which as we are going to see below he acknowledges, but against considering a nonrightmost accent such as those in (24) and (25) to be contrastive. This is important since I will argue below that a problem with his definition of Contrastive Focus is that the focus in (24), which he claims involves no contrast, comes out as Contrastive in his theory. This said, let us consider his Presentational Focus. Rochemont begins the discussion of Presentational Focus with the term “old information.” He offers a definition of what it means for something to be “under discussion,” using the term c-construable.10 His preliminary definition of c-construable is as follows: (27) A string P is c-construable in a discourse δ if P has a semantic antecedent in δ. (Rochemont 1986:47)
134 SYNTAX-INFORMATION STRUCTURE INTERFACE
Now, what does it mean to have a semantic antecedent? This is the definition that he offers: (28) A string P has a semantic antecedent in a discourse δ, δ -{ φ1,…, φn}, if, and only if, there is a prior and readily available string P′ in δ, such that the uttering of P′ either formally or informally entails the mention of P.11 He then presents a preliminary definition of focus (understood as new information) as an expression which is not c-construable: (29)
A string P is Focus if, and only if, P is not c-construable
He points out that this proposal guarantees that all that is c-construable is not focused and that only and all that is “new” to the discourse is focused. He notes an immediate problem for this theory, which is the case of focused pronouns mentioned above. That is, when we have focused pronouns such as those in (30), we seem to be facing something that is c-construable and focused: (30)
John hit Mary, and then SHE hit HIM
Confronted with these cases, Rochemont points out that this problem only arises if we assume a uniform semantic interpretation of focus. He suggests abandoning this assumption and proposes that there are distinctly characterized types of focus. The first is focus interpreted as new information, which he calls Presentational Focus and the second, exemplified by the focused pronouns in (30) is Contrastive Focus. Before we consider the definitions of these two types of focus, let me offer a different solution for the problem in (30) hinted at in the previous section. The pronouns in (30) are only problematic if we define new information as that not previously mentioned in the context. However, this problem does not arise under the assumption that new information is constituted by the informative part of an utterance, regardless of the previous mentioning of any of its elements in the discourse. That is, the informative or focus part of both (31) and (32) in that context is “TOM,” regardless of the fact that it has been mentioned in (32), but not in (31).
FOCUS 135
(31)
Who did you see?
I saw TOM
(32)
Who did you see, David or Tom?
I saw TOM
From this point of view, which I am going to adopt, and which is also assumed in recent proposals such as Vallduví (1990) and Lambrecht (1994), we do not need to divide focus into different types to solve the problematic example in (30), since it may not be a problem. That is, faced with those types of examples, we can still maintain a uniform notion of focus, as long as we do not define focus in terms of “not previously mentioned in the discourse,” which, as will be argued for later, is neither necessary nor adequate. However, we will see that there are other problems for a uniform notion of focus and that both Lambrecht and Vallduví do have different types of focus. Since there are cases of c-construable focus, such as that in (30), which are claimed to be cases of Contrastive Focus, the definition in (29) can only apply to Presentational Focus, which is therefore defined as in (33): (33) An expression P is a Presentational Focus in a discourse δ, δ={φ1,…, φn}, if, and only if, (i) P is an expression in φi, and (ii) at the time of utterance of φi in δ, P is not c-construable Since Rochemont assumes that all new information must be focused, a problem arises with cases such as (25) involving verbs of appearance. These are not c-construable and should therefore be focused, but they are not. In order to solve this, he suggests that in addition to having a semantic antecedent, other situations can make a string c-construable. For instance, verbs of appearance and other lexical items such as indexical expressions like the personal pronouns I, you, we, locative and temporal adverbs such as here, there, now, then, today, tomorrow, last night, etc., “which are apparently c-construable by virtue of the speaker’s wishing to establish a suitable setting for the discourse scene being introduced” are going to be considered c-construable. In fact, these are going to be indirectly c-construable, as opposed to items under discussion (those which have a semantic antecedent), which are directly c-construable. From this point of view, (25) is not a problem anymore,
136 SYNTAX-INFORMATION STRUCTURE INTERFACE
since the verb is indeed c-construable (indirectly) by virtue of being a verb of appearance.12 Thus, we have directly c-construable elements (those that have a semantic antecedent) and indirectly c-construable elements, such as verbs of appearance or indexical expression which do not need a semantic antecedent in the discourse. In addition, Rochemont further subdivides indirectly c-construable elements or acceptable scenesetters into those which are conventionally so (verbs of appearance and indexical expressions belong to this type, and are necessarily c-construable) and those which can be conversationally so, and are only optionally c-construable. This happens when certain expressions are taken by the speaker to be unremarkable as scenesetters when they refer to recurring scenarios familiar to the audience. For instance, Rochemont suggests that ‘into my office’ does not have to be focused in (34), because it can be conversationally c-construable in this fashion, that is, taken to be an unremarkable setting for the speaker’s discourse to revolve around. (34)
A strange MAN came into my office today
I see two problems with these types of c-construable elements. First, the notion of conversationally c-construable may be too permissive, since it allows us to make anything that is not focused-and should be according to the definition—a conversationally c-construable item. Second, we may not need to say that verbs of appearance are intrinsically c-construable. This is only necessary if we assume that everything that is considered to be new information has to be stressed. However, we know that not all elements which are considered to be part of the focus get an accent. That is, as we have mentioned in the previous section, and, as Rochemont himself notes, citing Chomsky (1971), one of the features of the “unmarked” rightmost sentence accent is that its scope is ambiguous or, in other words, that it can project upward. For instance, a sentence such as (35) can be an appropriate answer to any of the questions in (36): (35) (36)
Laurie followed Ralph into the BEDROOM13 a. Where did Laurie follow Ralph? b. What did Laurie do? c. What happened?
FOCUS 137
In order to explain why “followed” and “Ralph” are not focused in (35)), as an answer to (36)b, we do not have to say that they must be c-construable. We can say, as has been suggested (Selkirk 1984, 1995), that they are indeed part of the focus, though not accented. Now, is it possible to suggest that this is what happens in cases such as those in (37)? (37)
a. A MAN appeared b. A LETTER arrived
Prima facie, we face a problem, since it has been noted that while rightmost focus such as that in (35) can project up to the whole sentence (Selkirk 1984, 1995), focus on a preverbal subject cannot. That is, the sentence in (38) cannot be an answer to any of the questions in (36).14 (38)
LAURIE followed Ralph into the bedroom
In view of this, we might think that, since the examples in (37) that we are interested in involve focused subjects, this focus could not have projected up to the verb. However, the cases in (37) differ from other cases of focused subjects in that the types of verbs they involve overlap with unaccusatives.15 From this perspective, and if current syntactic theories of unaccusativity are on the right track, it is very possible that focus can project in the case of (37) and not in the case of (38), since only in the former case is the focused subject in fact an internal direct object.16 Selkirk (1995) suggests that this is indeed the case, that focus can project from the subject of unaccusative verbs, and suggests two ways of accounting for this in a theory of focus projection, which we will consider later. Thus, if the cases of verbs of appearance that Rochemont considers are indeed cases of unaccusative verbs, we can suggest that in cases such as (37) focus projects from the subject up to the verb and, therefore, there is no need to stress the verb. From this perspective, we do not have to stipulate that these verbs are special in that they can be indirectly c-construable without the need of a semantic antecedent in the discourse. We can derive their behavior from the current syntactic analysis of unaccusativity and of focus projection, both of which are available and needed for independent reasons. We have seen some arguments against the necessity of different types of c-construable elements. The notion of Contrastive Focus is also
138 SYNTAX-INFORMATION STRUCTURE INTERFACE
problematic. Recall that this type of focus is proposed to account for cases in which the focused part is also c-construable. From this perspective (39) would be considered to be a case of Contrastive Focus. (39)
Who did John hit?
He hit HIMSELF
On the one hand, notice that, as opposed to the example in (30) that Rochemont considers, where we have two potentially contrasting elements, there does not seem to be anything contrastive about (39), at least not more contrastive than if the focus had been the non c-construable “HIS MOTHER.” However, “himself” is c-construable in his theory in that it has a semantic antecedent in the discourse, and should, therefore, be considered as contrastive as (30). On the other hand, I think the definition of Contrastive Focus is itself problematic. First, even though it was cases such as (30), where we have c-construable elements that are also focused, that suggested the proposal of a Contrastive Focus, when Rochemont defines this type of focus, he defines it in terms of the nature of the rest of the sentence (the nonfocused part, which has to be c-construable) and not in terms of the focused element itself. Consider the definition in (40); (40) An expression P is a Contrastive Focus in a discourse δ, δ—{φ1,…, φn}, if, and only if, (i) P is an expression in φi, and (ii) if P/φi is the result of extracting P from φi, then P/φi is c-construable, and φi is not c-construable17 According to this definition, the focus in both the examples in (41)–(42) should be considered Contrastive, since the nonfocused part is c-construable in both. It does not seem to matter that only the focus in (41) is c-construable while the one in (42) is not. (41) A: Bill’s financial situation is a source of constant concern to Mary B: Bill’s financial situation is a source of constant concern to BILL (42) John hit Mary, and then he KICKED her Rochemont suggests that it is a desirable consequence of how Contrastive Focus is defined that a Focus can be both Contrastive and
FOCUS 139
Presentational. This would be the case of (42). That is, KICKED in (42) qualifies both as a Contrastive Focus (the nonfocused part is c-construable and the whole thing is not) and as a Presentational Focus (the focus is not c-construable). The focus in (41), on the other hand, would be solely Contrastive.18 However, it is hard to see how these cases of solely Contrastive Focus, such as that in (41) follow the general definition in (40). This definition does not distinguish between only Contrastive and both Contrastive and Presentational (this depends on the c-construability of the focus itself, which is not mentioned in the definition). However, it does require that the whole thing (φi) not be c-construable. That is, according to the definition in (40), it cannot be the case that both the whole discourse and the unfocused part are c-construable. But this is what happens when we have a solely Contrastive Focus such as that in (41). On the one hand, the unfocused part is c-construable, and, on the other hand, the focused part is c-construable, which makes φi c-construable.19 Thus, the definition in (40) seems to be good only for cases such as that in (42), which are both Presentational and Contrastive, according to Rochemont. That is, it accounts for cases where the focused part is not c-construable and the unfocused part is, but it is not a definition about the case of a c-construable focused element that prompted this type of focus, such as the one in (30). A second problem with the notion of Contrastive Focus is that, according to the definition in (40), the focused element in an answer to a subject-wh question such as the one in (21), repeated here in (43), is a case of Contrastive Focus: (43)
Who followed Ralph into the bedroom? LAURIE followed Ralph into the bedroom
That is, the unfocused part is directly c-construable and the whole utterance is not. Recall that this was one of the cases which Rochemont had argued had nothing contrastive about it, in the context of considering the relation between contrastive accent and nonrightmost accent, mentioned in the previous section. There, he suggested that there was nothing contrastive or marked about a focused subject. Rochemont points out that it is a problem that cases like (43) come out as Contrastive and suggests that we may be facing a different type of focus, perhaps what Culicover and Rochemont (1983) refer to as Informational Focus.20
140 SYNTAX-INFORMATION STRUCTURE INTERFACE
Although the definitions of only-Contrastive, only-Presentational and both Contrastive and Presentational are not offered, considering a) the examples he gives of each, b) what is pointed out about the differences between them, and c) the change mentioned in footnote 104, which adds “directly” to the definition in (40), the following seems to hold in this theory: (44) An expression P is: A. Only Contrastive if the unfocused part is directly c-construable and the focus is c-construable B. Both Contrastive and Presentational if the unfocused part is directly c-construable and the focus is not c-construable C. Only Presentational if the focus is not c-construable and the unfocused part is not directly c-construable This raises several questions: Why is it the case that the difference between direct and indirect c-construability is only important for nonfocused material? Is it the case that the fourth logical possibility (the focus is c-construable and the unfocused part is not directly c-construable) does not exist? Consider the following example: (45) A: Who went to the meeting, his lawyer or his doctor? B: When the meeting was about to end, his LAWYER appeared This example may come close. The focus “his LAWYER” is c-construable by virtue of having a semantic antecedent in the discourse, and the unfocused part, “appear,” is not directly c-construable. If this is possible, we would have to say that it is neither Presentational nor Contrastive. Another question we might ask is: what is the value of the division between Contrastive and Presentational if most Presentational foci are at the same time Contrastive? Notice that the only non-Contrastive Presentational is the very limited case of presentational verbs.21 The implied consequences of the definitions of Presentational and Contrastive Focus in (44) seem far from a non ad hoc, general and simple typology of foci. Thus, we have seen that Rochemont’s notion of Contrastive Focus has problems. On the one hand, it identifies as Contrastive, instances that do not seem to be contrastive such as (39) and (43), repeated here in (46), and, on the other hand, only applies to cases which are both Contrastive and Presentational like (42),
FOCUS 141
(46)
a. Who did John hit? He hit HIMSELF b. Who followed Ralph into the bedroom? LAURIE followed Ralph into the bedroom
In addition, according to the definition in (40), both c-construable focused elements and not c-construable focused elements come out as contrastive, which loses the initial intuition that equated contrastively focused elements which c-construable focused elements such as those in (30). All these problems might be avoided if, as hinted at before, we do not restrict focused elements to non c-construable elements. As will be defended below, focus can be conceived as the informative part in an utterance be it already mentioned in the discourse as in (41) or unmentioned as in (42). From this perspective, there is not going to be any difference between (41) and (42). Both will be considered cases of narrow focus, cases where focus does not project, which form so-called Focus-Background structures. Thus, there will be nothing contrastive about (30), (39) or (41)–(43) and no Contrastive Focus different from Presentational Focus. As we will see later, focus in general has been claimed to make a set of alternatives salient (Rooth’s 1985 p-set) in all cases. From this point of view, the fact that some of these alternatives may be in some cases more obvious or even totally spelled out does not change the nature of focus. The only division that we will have is that between narrow and wide focus. This division does not affect the nature of focus; that is, it does not divide focus into different types, but only depends on the syntactic position of the focused element. That is, it is a consequence of the position and not of a double nature of focus. 5.3.2 Gundel’s Psychological, Semantic and Contrastive Focus Gundel (1994) points out that the term “focus” has been used in the literature in at least three different senses, which she refers to as Psychological, Semantic and Contrastive. I will now consider these three different senses and suggest that we restrict the term “focus” to her Semantic Focus to avoid confusion. Psychological Focus corresponds to the psychological notion of focus of attention, which Hajičová (1987) calls AI focus. This is used in much work in artificial intelligence to refer to active, salient elements in the
142 SYNTAX-INFORMATION STRUCTURE INTERFACE
discourse, which are encoded by unstressed personal pronouns, zero anaphors and weakly stressed constituents. In (47) we have an example of an entity assumed to be in psychological focus, which can, therefore, be represented by the use of an unstressed personal pronoun (in italics): (47) Emily hasn’t changed much. She still looks like her mother, doesn’t she? (Gundel 1994:457) Since all these unstressed elements would be considered to be topical in the linguistic literature, I will not use the term “focus” to refer to this notion. I think it would be less confusing to refer to it with a term mentioned by Gundel: “current center of attention.” Putting aside Psychological Focus, Gundel notes that the term “linguistic focus” has been used in the literature with two senses. The first and more narrow sense was introduced in Halliday (1967) and developed later in several generative works.22 Gundel quotes the following definition taken from Hajičová (1987): “that part of the (meaning) of the sentence that conveys some (irrecoverable) information predicating something about the ‘given’, recoverable, contextually bound part (i.e. of the topic of the sentence), distinguishing thus a dichotomy of topic and focus.” Gundel is going to refer to this notion as Semantic Focus. In (48) we have an example adapted from Gundel, where the Semantic focus is capitalized and underlined: (48)
Do you know who called the meeting? BILL called the meeting
In addition, the term focus, Gundel points out, has also been used in a broader sense, since Halliday, to refer to any constituent which is linguistically prominent (prosodically and sometimes also syntactically). Crucially, constituents may be given prominence for reasons other than their newness with respect to a topic. As Gundel points out: Constituents may also be made prominent because the speaker/writer does not think that the addressee’s attention is focused on a particular entity and for one reason or another would like it to be, because a new topic is being introduced or reintroduced (topic shift) or because one constituent (topic or semantic focus) is being contrasted explicitly or implicitly, with something else. (Gundel 1994:459)
FOCUS 143
This is her Contrastive Focus, which can be exemplified by the capitalized (without underlining) element in (49): (49) That COAT you’re wearing I don’t think will be WARM enough Notice that in (49) there is also another prominent element, a semantic focus, marked with capital letters and underlining. Thus, in Gundel’s proposal, Semantic Focus refers to the new information that is being asserted. It is “the part of the sentence that answers the relevant wh-question (implicit or explicit) in the particular context in which the sentence is used.” As Gundel points out, it is, therefore, an obligatory part of any sentence and it can be linguistically marked by pitch accent, by word order and other aspects of syntactic structure, by special focus position, by focus marking particles or by some combination of one or more of these features. This is the notion of Focus I will adopt. Note that this notion would include both Rochemont’s Presentational and Contrastive types. As Gundel points out, context-active, discourse-old elements (in her terminology, current centers of attention or entities in psychological focus) can be semantically focused, as in (50): (50)
Mary said that it was SHE (=Mary) who called (Gundel 1994:461)
That is, the fact that Semantic Focus falls on a previously mentioned entity does not alter its nature. As Gundel points out, Psychological Focus (a discourse-old, background element, or in Rochemont’s terms a c-construable element) is not the antithesis of Semantic Focus. The two notions are logically and empirically independent of one another. Regarding Gundel’s notion of Contrastive Focus, it should be noted that this notion is totally different from Rochemont’s Contrastive Focus. Recall that the latter has to do with the unstressed part of the utterance being c-construable, as in (41)–(43), repeated here for convenience.23 (41) A: Bill’s financial situation is a source of constant concern to Mary B: Bill’s financial situation is a source of constant concern to BILL
144 SYNTAX-INFORMATION STRUCTURE INTERFACE
(42)
John hit Mary, and then he KICKED her
(43)
Who followed Ralph into the bedroom? LAURIE followed Ralph into the bedroom
That is, all these cases involve the occurrence of just one focus, which would be considered to be a case of Semantic Focus by Gundel, since it follows her definition in terms of being the new information that it is being asserted, the part of the sentence that answers the relevant wh-question. Gundel’s notion of Contrastive Focus is very different. It has to do with making an element prominent, regardless of whether or not it is the informative part of the sentence. In fact, even though it is possible to put Contrastive Focus on a Semantic focus, as we will see below, due to the fact that Semantic Focus is already inherently contrastive, as Gundel points out, Contrastive Focus primarily falls on topics. That is, it involves making a constituent prominent in order to focus the addressee’s attention on a particular entity, very often because a new topic is being introduced or reintroduced or because it is being contrasted with something else. This prominence on a topic can be marked phonologically only, as in (51) or also syntactically by using a Topicalization as in (49) above, repeated here as (52): (51)
What did Bill’s sisters do? Bill’s YOUNGEST sister kissed JOHN
(52) That COAT you’re wearing I don’t think will be WARM enough Furthermore, as (51) and (52) show, this Contrastive Focus, as opposed to Rochemont’s, co-occurs with a Semantic Focus (marked with capitals and underlining, in these two examples). That is, it is not possible to have a Contrastive Focus only; it always co-occurs with a semantic focus. It is also worth pointing out that the stress on this type of Contrastive Focus is not obligatory. In fact, as we will see later, none of the focused topics in (51)-(52) would get an accent in a language like Spanish. They would only be marked syntactically. In addition, as Gundel points out, the type of accent by which Contrastive Focus is expressed is different from Semantic Focus accent. The former is Jackendoff’s
FOCUS 145
(1972) B accent (Pierrehumbert’s complex L+H*) while the latter is Jackendoff’s A accent or Pierrehumbert’s simplex H* tone. I think that, in this case also, we would avoid some confusion if we did not use the term focus, but rather, something like “Emphatic Stress.” That is, I think that there are enough differences between the notion of Semantic Focus and this type of prominence to favor the use of a different term. Thus, I will use the term “focus” only to refer to Gundel’s Semantic Focus; that is, to the informative and irrecoverable (from the context) part of the sentence that answers the relevant wh-question (implicit or explicit) in the particular context in which the sentence is used. Focus in this restricted sense is marked only with an A accent, with an exception that will be mentioned below that involves putting Emphatic Stress on a Semantic Focus. In fact, in a footnote, Gundel points out that, although she has kept the term “Contrastive Focus” due to the fact that her goal is to distinguish different ways in which the term focus has been used (rather than to propose terminology), she partly agrees with Hajičová’s suggestion (p.c.) that the term “Contrastive topic” may be more appropriate for what she is referring to as Contrastive Focus. I prefer the term Emphatic Stress to Contrastive topic. First, because of the “contrastive” part. I would like to avoid using contrastive alone for these cases, since focus itself is inherently contrastive. That is, as pointed out by Gundel, “whenever information is predicated about some topic, it is inherently in contrast with other things that might be predicated of that topic.” Second, because of the “topic” part. As mentioned before, it is also possible for this type of Emphatic Stress to fall on a Semantic Focus rather than on a topic. As an example of this, Gundel mentions Rooth’s (1992) example in (53), where, if we put special contrastive stress on “passed,” the answer strongly implies that the speaker could have done better. (53)
How did the exam go? Well, I PASSED
That is, a Semantic Focus can also be given this type of Emphatic Stress, which in my view makes this latter tem preferable to “Contrastive topic.” Cases such as (53) are also interesting from another perspective. In a case like this, we have a Semantic Focus which is also emphatically stressed. Recall that, as Gundel points out, Emphatic Stress on topics such as the one on coat in (52) involves Jackendoff’s B accent, while Semantic Focus involves Jackendoff’s A accent. Does Emphatic Stress
146 SYNTAX-INFORMATION STRUCTURE INTERFACE
on a Semantic Focus also involve a B accent or a different one? Neither Gundel nor Rooth clearly says that the accent in (53) (with the intended meaning) is indeed a B accent. However, whatever this accent may be, it is not an A accent, since although this accent is possible in (53), it would not have the intended meaning. So, it seems that in this case the accent used to mark Emphatic Stress on foci overrules Semantic Focus A accent. There is another interesting case, which may shed some light on this issue, where we can also have a Semantic Focus emphatically stressed (in Gundel’s terms an element that is both a Semantic and a Contrastive Focus). Gundel considers some cases where Semantic Focus can also be syntactically marked as in (54): (54)
What do you think I should take on the camping trip? That COAT you’re wearing, I think you should take
As Gundel points out, the fact that this is possible as well as the apparent similarities with cases such as (52) have caused a lot of terminological confusion in the literature, where sometimes the term “Topicalization” has been used for both cases.24 Notice that in (54) as opposed to (52), there is only one focus. That is, in these cases Semantic Focus is placed in a sentence-initial position (otherwise occupied by topical elements). One of Gundel’s arguments in proposing that cases like (54) involve Semantic Focus while cases like (52) involve her Contrastive Focus (Emphatic Stress on a topic), is the fact that the accent in (54) is an A accent. However, she points out that these instances of syntactically marked semantic foci such as the one in (54) can also be emphatically stressed. In her terminology, these cases of semantic focus can also be contrastively focused. That is, an example such as the one in (54) could also be considered a case where “coat” gets both Semantic and Contrastive Focus, if the utterance in (54) is followed for instance by something like: “but I wouldn’t take the boots.” In this case, Gundel says, “coat” gets the complex L+H* tone or B accent characteristic of Contrastive Focus (our Emphatic Stress). That is, according to Gundel, in this case the B accent overrules the A accent, even when we are dealing with a case of Semantic Focus. If this is the case, that is, if the preposed element in cases like (54) can also get a B accent, as in the case in (52), the potential confusion between the two structures is even greater, since, if we put a B accent on the preposed element in (54), the
FOCUS 147
only difference between (52), and (54) would be the existence of an A accent towards the end of the utterance in the case of (52). Further, this possibility of giving the preposed element in (54) a B accent has two important consequences. First, it constitutes an exception to equating Semantic Focus with an A accent, referred to above, since the preposed element in (54) can get a B accent, even though it is the Semantic Focus. Second, it also exemplifies a case where a B accent does not co-occur with another accent, which we had not seen up to now. I will delay until chapter 6 any further analysis of this construction, simply noting here these two special features that this construction may have. To sum up, I have considered in this section Gundel’s differentiation between three senses of the term “focus” in the literature, and I have suggested keeping the term focus just in the sense of Semantic Focus only and referring to her Psychological Focus and her Contrastive Focus as “Current center of attention” and “Emphatic Stress” respectively. From this point of view, stress is not equated with focus. That is, the fact that there are cases of stressed topics does not mean these should be considered focused topics. This means that from this perspective, focus will not occur inside topics, as other proposals suggest. It should be noted that the decision to keep stress and focus apart is not new. As we saw in 5.2, there are cases where focused elements do not get an accent and also cases where an accent is not focus-related. I am just extending this situation to the case of accented topics. Recall that, as we saw in chapter 2, this is also Lambrecht’s (1994) position when he suggests that accent can mark different relations: it can mark elements as topics or as foci. Although there are several recent analyses which involve focused topics (topics with a focus feature) such as Krifka (1991) and Fintel (1994), among others, I will pursue an analysis where, due to the definition of topic and focus, there cannot be such a thing as a focused topic, but only accented topics. As we will see, one of the important differences between focus and the accent on topics is that while the former affects truth conditions, the latter does not, as Gundel also points out. It could be argued that by trying to keep a unified and homogeneous notion of focus understood only as Semantic Focus we are just avoiding the problem and shifting the burden onto the notion of stress. That is, from this perspective there is only one focus but stress can fall on a semantic focus or on a topic. I think this may indeed be the right move. The fact that both topics and foci may share a particular feature (in this
148 SYNTAX-INFORMATION STRUCTURE INTERFACE
case, the fact that both can be marked prosodically) is not an isolated phenomenon. They are known to share another feature: the fact that both can be syntactically marked as subjects. That is, we can have both topical subjects as in (55) and focused subjects as in (56): (55)
What did the children do? The children went to SCHOOL
(56)
Who went to school? THE CHILDREN went to school
The fact that there is a relatively strong relation between topics and subjects and that both the topical element in (55) and the focus in (56) are syntactic subjects has not been taken to suggest that perhaps there is a topic feature inside the focused subject. It has been taken to suggest that the informative division between topic and focus does not always coincide with the syntactic division subject-predicate, and, therefore, we can have both topical subjects and focused subjects. I think the same argument can be extended to the notion of stress. The fact that there is a relatively strong relation between stress and focus and that we can have both accented topics and accented foci doesn’t have to be taken to mean that there may be a focus feature inside the topic. It can mean that the semantic notion of focus does not coincide with the phonological notion of accent, and, therefore, we can have both accented topics and accented foci. From this perspective, in the same way that foci can be marked phonologically and/or syntactically depending on the language, topics can also be marked phonologically and/or syntactically. In fact, some contrasts between English and Spanish point to this conclusion. In Spanish the difference between a topical and a focused subject can be marked syntactically instead of phonologically, as in English. That is, instead of accenting the focused subject as in (56), we can mark it syntactically by placing it postverbally as in (57): (57)
¿Quiénes se fueron? “Who left”
Se fueron los niños left the children
Similarly, while the “new topics” in (51) and (52) are accented in English, they would not be accented in their Spanish counterparts:
FOCUS 149
(58)
¿Qué hicieron las hermanas de Bill? La (hermana) más pequeña besó a JUAN
(59) Ese abrigo que llevas no creo que sea lo bastante CALIENTE That is, the accent in the topics in (51)-(52) can be seen as an optional way of marking them as topics (which some languages like Spanish do not use) rather than as the phonological realization of a focus feature. Notice that while the notion of Semantic Focus is valid for both Spanish and English (and potentially universal with languages varying only in the way it is marked: phonologically, syntactically, morphologically, etc.), it would cease to be valid if we were to associate the accent in (51)–(52) with focus. Thus, I will maintain that while these cases of new topics can be marked with a pitch accent in English, they should not be considered cases of focus. It may be worth pointing out that while the focus-related accent marks a syntagmatic relation, that is, a relation between a topic and what is predicated of it, what I have referred to as Emphatic Stress seems to mark a paradigmatic relation, That is, either a relation between a topic and other possible salient (explicit or implicit) topics as in (51)–(52) or a relation between some focus and other (implicit or explicit) foci as in (60) where (with a B accent on COAT) a relation is being marked between the Semantic Focus “COAT” and the Semantic Focus “BOOTS” (in addition to the syntagmatic relations that are established between “I think you should take” and “COAT,” on the one hand, and “I wouldn’t take” and “BOOTS,” on the other hand.) (60) What do you think I should take on the camping trip? That COAT you’re wearing, I think you should take but I wouldn’t take the BOOTS 5.3.3 Erteschik-Shir’s Metalinguistic Focus Erteschik-Shir’s (1997) theory of focus is very complex and it involves regular focus and restrictive focus, main and subordinate predication, and also main and subordinate topics.25 I cannot examine here her whole proposal, so I am only going to consider her notion of “metalinguistic focus.” This concept is based on Horn’s (1989) notion
150 SYNTAX-INFORMATION STRUCTURE INTERFACE
of metalinguistic negation; it is a type of restrictive focus and it has to do with contrast. She specifically states that contrast is the metalinguistic equivalent of restrictive focus. However, even though the notion of Metalinguistic Focus (MF) has to do with contrast, it is very different from Gundel’s Contrastive Focus. First, Erteschik-Shir’s notion is more constrained. She points out that metalinguistic foci are a subcategory of restrictive foci and that they are more restricted in that the metalinguistic restrictive sets have exactly two members. Second, and more importantly, Gundel’s Contrastive Focus always co-occurs with a “real” Semantic Focus and this is not the case for Erteschik-Shir’s MF. Consider one of the examples of MF that she offers. (61)
A MAN is intelligent (Erteschik-Shir 1997:121)
She says that this sentence shows an instance of a contrastively interpreted indefinite interpreted generically, meaning something like “a man, not a woman.” The MF on “a man” is the only focus in the sentence and contrasts with (62) which shows Gundel’s Contrastive Focus co-occurring with Semantic Focus: (62)
A MAN is INTELLIGENT, a WOMAN is CLEVER
Let us have a closer look at (61), which is the only example of MF that Erteschik-Shir offers. For Erteschik-Shir, “a man” in (61) is the topic of the main predication. At the same time, it constitutes the focus in the subordinate predication, with respect to the other member of the contrast (a woman), which is not overt, but is present as part of the subordinate predication, and necessary for the interpretation of the sentence. Thus, having two levels of predication, allows for an element to be a topic in one level and a focus in another. However, it is important to notice that in (61) there is only one focus, and, according to her, it is inside the main topic. That is, in the main predication, “is intelligent” is the Comment and should have a Focus, but it does not. Erteschik-Shir’s solution for this problem in sentences such as (61) is to say that MF overrules regular focus with respect to stress. I guess this would mean that the focus on “a man” in (62) would not be considered metalinguistic, as it does not overrule regular focus. She does not consider these cases. It seems to me that to explain the absence of focus in the main predication by saying that MF overrules regular focus is more
FOCUS 151
stipulative than explanatory. Although Erteschik-Shir’s theory of focus deserves much more attention than I can give it here, I am not sure that the analysis of (61) in terms of MF and in terms of a main and subordinate predication does anything more for us than a simpler analysis in terms of a regular focus and an Open Proposition or Background. Although there seems to be something special about focused generics such as that in (61), I will consider cases like these to be just cases of narrow focus.26 We can derive their special flavor from features they may share with other instances of narrow focus; from specific characteristics that they possess due to the position of this narrow focus (Spec, IP), and from the nature of the focused phrase: a subject of an individual level predicate, which is interpreted generically. From this perspective, the focus involved in (61) is not a different kind of focus but it interacts with the nature and position of the phrase that receives the focus. If this is the case, the differences between (63), an example similar to (61), and the example in (64), will be due to the nature of the NP that is being focused rather than to (63) being an instance of a MF distinct from the regular focus in (64): (63) Do you think there are animals that are intelligent?/I don’t believe animals are really intelligent/Which animal do you think is intelligent? DOLPHINS are intelligent (64) Who cleaned the house? PETER cleaned the house We could try to separate all factors bearing on the interpretation of (63) and try to form a set of examples that gradually go from an uncontroversially nonmetalinguistic narrow focus to a case like (63): (65) For which colloq dinner did Rightmost Narrow Focus you cook paella? I cooked paella for SAITO’s colloq dinner (66) What did you cook for that Nonrightmost Narrow dinner? Focus I cooked PAELLA for that dinner
152 SYNTAX-INFORMATION STRUCTURE INTERFACE
(67) Did you burn the paella? I COOKED the paella
Narrow Focus on the verb
(68) Who cooked the paella? EUGENIA cooked the paella
Narrow Focus on a subject
(69) Who is cooking next Friday? Narrow Focus on a STUDENTS are cooking next nongeneric BN subject Friday (70) Do you know anybody that Narrow Focus on a generic BN eats paella once a week? subject SPANIARDS eat paella once a week From this perspective, then, the case of focused generics will not be considered a different type of focus in its nature, Its “special” flavor will be derived from the position and nature of the focused element involved.27 I will now consider Vallduví’s (1990) division of focus. 5.3.4 Vallduví’s Retrieve-Add vs. Retrieve-Substitute Focus Vallduví (1990) is a very interesting and complete proposal about information structure and its relation to the syntactic structure and interpretation of sentences. As mentioned in chapter 2, he proposes a trichotomy, where the focus is separated from the rest, which he calls the ground. This nonfocused part, the ground, is further divided into the topic (called link) and the rest, which is therefore nontopic, nonfocus: the tail. (71)
S={FOCUS, GROUND} GROUND={LINK, TAIL}
I cannot consider here Vallduví’s whole proposal. However, since I will follow him in distinguishing between two topical elements in chapter 7, I will introduce his system now. I will consider here his notion of focus and the four structures that he proposes.28 The focus in this trichotomy corresponds to the focus in articulations such as Topic-Focus,
FOCUS 153
Focus-Open proposition or Focus-Background, and it constitutes the only informative part of the sentence. It is invariably intonationally prominent and it is the only nonelidable part of the sentence. Considering that both the link and the tail are optional, his articulation predicts four types of informational structures for a sentence: link-focus, all-focus, link-focus-tail and focus-tail. In (72)–(75), we have an example of each, both in English and Catalan: (72) a. The boss [F visited a broccoli plantation in Link-focus COLOMBIA] b. L’amo [F va visitar una plantació de bròquil a COLOMBIA] (73) a. Waiter! [F There’s a fly in my cream of All-focus broccoli soup]! b. Cambrer! [F Tinc una mosca a la crema de bròquil]! (74) a. The boss HATES broccoli Link-focus-tail b. L’amo l’ODIA, el bròquil (75) a. I can’t believe this! The boss is going Focus-tail crazy! BROCCOLI, he wants now. b. No m’ho hagués cregut mai! L’amo està ben boig! BROQUIL, vol ara These examples show that what the two tailful structures have in common—which is not shared by the other two structures without tails—is the fact that in both (74)–(75) the focus is narrow. In fact, Vallduví (1990:65) points out that the two tailful structures correspond to Prince’s (1981a, 1986) focus/open proposition sentences and that these tailful types have also been referred to as sentences that have “narrow focus,” “constituent focus,” or “contrastive focus.” Thus, in Vallduví’s theory focus is going to come in two brands, depending on the existence of a tail in the structure. Vallduví’s proposal is based on the notion of Information Packaging (adapted from Chafe 1976 and Prince 1986), which is described as the structuring of the sentence into instructions with which the speaker directs the hearer to enter the information carried by the sentence into his/her knowledge-store, which is described in terms of Heim’s (1982)
154 SYNTAX-INFORMATION STRUCTURE INTERFACE
file metaphor. Packaging instructions are a speaker-designed information-retrieval mechanism for the hearer. From this perspective, focus, which is understood to be an operator, can be read as “retrieve information.” The ground acts as a vehicular frame for the informative focus: it guarantees an appropriate entry of the information into the hearer’s knowledge- store, indicating to the hearer where and how the information must be entered. The link denotes an address in the hearer’s knowledge-store under which s/he is instructed to enter the information, and the tail provides further directions on how the information must be entered under a given address. That is, the tail, which is the complement of the link within the ground can be viewed as an element that acts as a signaling flag to indicate exactly how the information carried by the sentence must be entered under a given address. If there is a link, it instructs hearers to “go to” the address it denotes in their knowledge-store and then enter the information provided by the sentence under that address. A link, therefore, is informationally interpreted as a bipartite element: the instruction “to go” and the address hearers are instructed to go to. From a speaker’s point of view, it may be procedurally read as follows: “I instruct you to go to address a in your knowledge-store and then retrieve the information of the sentence by adding focus under a.” For instance, the Link-focus structured sentence in (76) can be read as “I instruct you to go the address ‘the boss’ in your knowledge-store and retrieve the information of the sentence by adding under ‘the boss’ that he called.” (76)
The boss CALLED
Vallduví represents these instructions formally. An all-focus instruction is represented as in (77), where the symbol Φ is the “focus operator” and the focus is within the scope of Φ in boldface. (77)
Φ [focus]
According to him, this structure, where an entire sentence is information, is the simplest of information packaging instructions. Sentences with a ground have a more complex structure. A link is represented with a quantifier-like element: Λx is read “go to x”and the address denoted by the link constitutes the range of x. For an address a the representation of the link is in (78) and the abstract representation of a Link-Focus sentence in (79):
FOCUS 155
(78) (79)
Λx, x=a Λx, x=α [Φ [x focus]]
For instance, the representation of the instruction for the sentence in (76) would be as follows: (80)
Λx, x=the boss [Φ [x called]]
Let us now consider tailful structures, which are the ones where focus is not read as “retrieve information.” If the ground contains a tail, it means that the information of the sentence cannot be simply added under the address denoted by the link. Instead, it indicates that part of the proposition communicated is knowledge already contained under that address and that the information of the sentence must be construed in some way with that knowledge instead of merely added. Therefore, the presence of the tail affects the nature of the packaging instruction. The operator Φ comes in two brands. This operator instructs the hearer to retrieve information contained in its scope. In the case of link-focus sentences, the operator is read as “retrieve information by adding focus” and in the case of a sentence containing a tail, the operator is read “retrieve information by substituting focus for the blank in the ground ‘link—tail’ (which is already under the address denoted by the link).” He abbreviates these two brands as “retrieve-add” and “retrieve-substitute.” So, the presence of a tail in a packaging instruction alters the nature of focus, turning it from a “retrieve-add” to a “retrieve-substitute.” He gives the following examples: (81)
The boss hates BROCCOLI
(82)
The boss HATES broccoli
These sentences have the same link, so the representation of the link is identical in both sentences and can be read as “I instruct you to go to the address ‘the boss’ and then….” In the case of the link-focus sentence, once the address denoted by the link is “gone to,” the hearer is in a position to retrieve the information provided by the sentence. In the case of the link-focus-tail sentence, the hearer needs to know more about the way in which the information must be entered. In the former, the instruction continues “…and then retrieve the information of the
156 SYNTAX-INFORMATION STRUCTURE INTERFACE
sentence by adding under ‘the boss’ that he hates broccoli.” In the tailful sentence, the instruction continues “…and then retrieve the information of the sentence by substituting hates for the blank in he—broccoli, which is already under ‘the boss’”. He considers the case of a tailful sentence with no link, like the one in (83) and suggests that in this case the hearer is instructed to substitute broccoli for the blank in he—wants under the current address. (83)
I can’t believe this! The boss is going crazy! BROCCOLI, he wants now.
Now, how are tails represented? The tail, being part of the ground, must escape the scope of Φ. Vallduví represents this by abstracting the tail away from the clause, leaving a variable behind within the scope of Φ using a lambda-like construct. He borrows λ from the lambda-calculus used in logical semantics and states that all he is borrowing is the symbol and none of the semantics of the lambda-calculus. So, a tail, which specifies how the information must be entered under a given address will be represented as follows for a given tail β: (84)
λx [Φ [focus x]] (β)
For instance, a focus-tail sentence such as (83) would be represented as in (85) and a link-focus-tail sentence as in (82) would be represented as in (86): (85) (86)
λx [Φ [he x broccoli]] (wants) Λx 1 , x 1=the boss [ λx 2 [Φ [x 1 hates x 2]] (broccoli)
So, these are the packaging instructions associated with the four informational articulations: (87)
Packaging Instructions: All-Focus: Φ [focus] Link-Focus: Λx 1 , x 1=α[ Φ [x 1 focus]] Link-Focus-Tail: Λx 1 , x 1=α[λx2 [Φ [x 1 focus x 2]] (β) Focus-Tail: λx 2 [Φ [focus x 2]] (β)
Vallduví points out that the interpretation of the instructions can be expressed in shorthand notation, as follows:
FOCUS 157
(88) Interpretation All-Focus (ΦA): RETRIEVE-ADD (focus) Link-Focus (Λ, ΦA): GO-TO (link), RETRIEVE-ADD (focus) Link-Focus-Tail (Λ, ΦS): GO-TO (link), RETRIEVESUBSTITUTE (focus) Focus-Tail (Φs): RETRIEVE-SUBSTITUTE (focus) Therefore, Vallduví distinguishes two ways of interpreting focus. In information structures without a tail the information is retrieved by adding focus and in tailful structures information is retrieved by substituting focus in a specific slot or blank in the structure. I think that the difference between his retrieve-add focus and retrieve-substitute focus, which is basically the difference between narrow and wide or broad focus has to do with the position of the focus rather than with two types of focus. Thus, although as we will see later I will accept Cinque’s (1993) difference between unmarked or default focus vs. marked focus, this results from the position of the focus and not from a double nature of focus. Further, I think Vallduví’s tailful structures are problematic and I will depart from him in the structure that results when we have a retrieve-substitute focus or narrow focus. The position I will take (which will not be defended in detail until chapter 7) is that his tailful structures are all instances of the Focus-Background structure that results whenever we have an instance of narrow focus. I think that both (82) and (83) are instances of the same information structure: a Focus-Ground structure where the difference between links and tails is no longer relevant. That is, these sentences are only adequate in a context in which everything except the focused element is part of the background. From this perspective, “the boss” in (82) is as much part of the background as is “he” in (83) or in its nonpreposed counterpart “He wants BROCCOLI now.” From this point of view, one of the structures with narrow focus (retrieve-substitute), namely, his tailful structure with a link, does not capture the fact that whenever we have narrow focus the rest of the sentence is necessarily part of the background (ground in his terminology), which a simpler structure for these cases in terms of Focus-Ground would. Another change that I will also suggest in chapter 7 has to do with left-dislocated phrases, which I will consider to be also part of the Ground in the same way right-dislocated phrases are.
158 SYNTAX-INFORMATION STRUCTURE INTERFACE
5.3.5 Lambrecht’s Sentence-, Predicate- and ArgumentFocus In Lambrecht’s (1994) theory, topic and focus do not form a dichotomy. They are two different pragmatic relations.29 Therefore, there is a topic relation and a focus relation and, from this point of view, focus cannot be defined as the complement of topic. Both topic and focus are relational categories. They have to do with the pragmatic relations between referents with given identifiability and activation properties and propositions in given contexts. He adopts a definition of topic as that which the proposition expressed by the sentence is about. Focus has to do with the conveying of new information and since all declarative sentences convey information, all such sentences must have a focus. However, not all sentences have a topic. Focus is not just “new information.” The loose equivalent of “new information” in his system is his Pragmatic Assertion, which is a proposition that is superimposed on and includes the Pragmatic Presupposition. The focus is an element of information that is added to, rather than superimposed on the pragmatic presupposition. That is, just as the topic is included in the Presupposition without being identical to it, a focus is part of the Assertion without coinciding with it. The focus is, therefore, the element of information whereby the presupposition and the assertion differ from each other. It is that portion of a proposition that cannot be taken for granted at the time of speech. It is the unpredictable element in an utterance. Lambrecht proposes that the focus articulation of sentences can be divided into a number of distinct TYPES which correspond to different types of pragmatically structured propositions. There are three types of sentences: Topic-Comment, Identificational, and Event-reporting Sentences. Each of these shows a different focus structure, which correspond to three basic communicative functions: Predicate-Focus, Argument-Focus and Sentence-Focus. The communicative functions are: that of predicating a property of a given topic (Predicate-Focus, Topic-Comment function); that of identifying an argument for a given proposition (Argument-Focus, Identificational function); and that of introducing a new discourse referent or of reporting an event (Sentence-Focus, Presentational or Event-reporting function). Here are some examples:
FOCUS 159
(89) (What did the children do Topic-Comment next?) Focus The children went to SCHOOL
Predicate-
(90) (Who went to school?) Identificational The CHILDREN went to Focus school
Argument-
(91) (What happened?) My CAR broke down
Event-reporting Focus
Sentence-
In the Predicate-Focus structure, the predicate is the focus and the subject (plus any other topical elements) is in the presupposition, In the Argument-Focus structure, the focus identifies the missing argument in a presupposed open proposition. Finally, in the Sentence-Focus structure, the focus extends over both the subject and the predicate (minus any topical nonsubject elements). The Argument-Focus and the Sentence-Focus structures are often homophonous or near-homophonous. He also points out that sentences can express several of these functions at once, via formal combinations of different focus structures.30 I think that Lambrecht’s types of focus structure account very nicely for the fact that some sentences (the event reporting or Sentence-focus) do not seem to have a topic (at least not an explicit one), some sentences have wide focus (the Topic-Comment or Predicate-focus) and some sentences have narrow focus (the Identificational or Argument-focus). I think the Predicate-focus type poses no problem, but I would like to add a comment about the two other types. Regarding the event-reporting type, where the whole sentence is the focus, I am not sure that the fact that there is no explicit topic means there is no topic in this structure. In fact, it has been claimed (Erteschik-Shir 1997) that this type of event-reporting sentences have an stage topic, understood as the “here-and-now” of the discourse. That is, this stage topic expresses the time/place at which the event expressed by the sentence takes place. If that is the case, the Sentence-focus type can be reduced to the Predicate-focus type. Regarding the Argument-focus type, it seems to me that this type coincides with cases of narrow focus and maybe “argument” is not the best term for it. All the examples that he gives for this structure involve a focused subject, as in (92):
160 SYNTAX-INFORMATION STRUCTURE INTERFACE
(92)
(Who went to school?)
The CHILDREN went to school
However, I am not sure that cases of narrow focus on objects (93), PPs (94) or even on verbs (95) should not be considered to be instances of this Argument-Focus. (93) (What did the children eat?)
The children ate PROSCIUTTO
(94) (Where did the children go?) The children MOVIES
went
to
the
(95) (What is it with the children The children LOVE prosciutto and prosciutto?) It seems to me that all these sentences could have Lambrecht’s identificational function, which is prompted by the question. For instance, in (93) instead of considering it a Topic-Comment structure, we could think that the verb is as topical as “the children” in the same way that “went” and “to school” are topical in the case of (92). If (92) is a case of Argument-Focus, then perhaps all of them are. The case of (95), in particular, where a predicate is focused, suggests considering whether Argument-Focus is the best name for this type of focus. I think all these cases are identificational in Lambrecht’s sense and all show an instance of narrow focus. Thus, I would like to suggest that at least two of Lambrecht’s types of focus structures do exist and coincide with what we have been referring to as wide and narrow focus or Vallduví’s retrieve-add and retrieve-substitute. 5.3.6 Kiss’s Identificational vs. Information Focus Kiss (1998) suggests that Identificational Focus should be distinguished from Information Focus, since it has additional syntactic and semantic properties that information focus does not share. This division is similar to Rochemont’s Presentational vs. Contrastive or Vallduví’s Retrieve-add vs. Retrieve-substitute, that is, that is, a division between wide vs. narrow focus, although there are important differences. Kiss proposes that Information Focus differs from Identificational Focus as follows. Information Focus has the features listed in (96):
FOCUS 161
(96)
Information Focus a) merely marks the nonpresupposed nature the information. b) allows for any type of phrase c) does not take scope d) does not involve any movement e) can be either smaller or larger, i.e. it can project
Identificational Focus, following characteristics:
on
the
other
hand,
presents
of
the
(97) Identificational Focus a) expresses exhaustive identification b) does not allow for all kinds of phrases (universal quantifiers, also—phrases, and even-phrases cannot function as identificational phrases) c) takes scope d) moves to the specifier of a functional projection e) is always coextensive with an XP available for operator movement (does not project), although it can be iterated. First, it should be noted that some of these features which Kiss uses to characterize identificational focus are specific to Hungarian. As she notes, crosslinguistically narrow focus can be ±exhaustive.31 Second, narrow focus is not always restricted as in Hungarian. Thus, as (98) shows, Spanish allows for universal quantifiers, and even-phrases.32 (98)
(99)
TODOS LOS SOMBREROS quería llevarse la niña ALL THE HATS wanted-3sg to take with her the girl “ALL THE HATS the girl wanted to take with her” HASTA UN SOMBRERO quería llevarse EVEN A HAT wanted-3sg to take with her “EVEN A HAT she wanted to take with her”
Third, even though iteration is listed as one of the features of Identificational Focus, languages such as Spanish, English, Catalan, Italian and possibly many others do not allow for multiple narrow foci. This iterative feature seems to be a Hungarian-specific characteristic.
162 SYNTAX-INFORMATION STRUCTURE INTERFACE
Fourth, narrow focus does not always involve movement. Thus, in English in situ narrow focus is quite common: (100)
Mary picked A HAT for herself
Furthermore, the fact that this type of focus takes scope in Hungarian might be a consequence of its position and not an inherent feature of narrow focus. From this perspective, it is not clear that Identificational Focus has such distinct features that it should be considered a different type of focus. In fact, despite some of the differences in the characterization of Kiss’s Identificational focus when compared to contrastive or narrow focus, the same arguments already put forward against such a division can be applied to her proposal: the fact that focus can be wide or narrow does not necessarily mean that these are two different kinds of focus.33 The different features that wide vs. narrow focus seem to have might be due to their different syntactic position, and not to a double nature of focus. In sum, we have seen different subdivisions of focus in the literature. I have suggested a homogeneous characterization of focus, where there are no different types of focus. However, as we will see in more detail below, focus can project, i.e., it can encompass just one phrase (narrow focus) or several up to the verb (wide focus). Therefore, we can talk about wide and narrow focus, bearing in mind that we are not referring to two different types of focus in terms of its nature (as, for instance, in Rochemont’s division between presentational and contrastive focus or Kiss’s division between information and identificational), but to the size of the focused phrase. Furthermore, I have suggested keeping the term “focus” only in the sense of Gundel’s semantic focus and using other terms to refer both to her psychological focus (current center of attention here) and her contrastive focus (emphatic stress for us).34 Next, I will consider how focus projects in the syntax. 5.4 The Projection of Focus As mentioned before, the phenomenon known as focus projection goes back to Chomsky (1971) where it is observed that in the following sentence, any of the constituents in brackets can be part of the focus:
FOCUS 163
(101) He was (warned (to look out for (an ex-convict (with (a red SHIRT))))) In a very important work, Selkirk (1984) correctly observes that focus can project from an argument to its head: (102)
What did John do? John cooked LASAGNA
However, as pointed out in previous sections, this is only possible with rightmost focus such as the one in (101) and (102).35 That is, in a sentence such as (103), which has nonrightmost focus, focus cannot project: (103)
Laurie followed RALPH into the bedroom
That is, in this case the focus is necessarily narrow. Although I cannot consider here all the theories of focus projection, it should be mentioned that the relation between focus projection and the position of the focus has been noted at least since Contrera’s (1976) pioneering work on word order in Spanish, and Prague School scholar’s work on systemic order in Czech. Von Stechow and Uhman (1986) consider in detail the insights and shortcomings of three different theories of focus projection: Culicover and Rochemont (1983), Selkirk (1984) and Contreras (1976). I am going to follow recent proposals based on Cinque’s (1993) theory of stress such as Reinhart (1995), Zubizarreta (1998), and Nash (1995) and assume a distinction between unmarked and marked accent. Unmarked accent is given by the grammar and it falls on the most deeply embedded constituent. Unmarked accent identifies the unmarked focus of the sentence. Only unmarked focus projects. For instance, Zubizarreta offers the following rule of focus propagation: (104) Rule of focus propagation The focus may propagate upward from the constituent that bears the unmarked accent along a continuous path that includes the nodes of the recursive side of the tree and the nodes that are projections of the head. The fact that focus projects is a very important feature, which, as we mentioned in chapter 4 forces some uninformative elements escape the
164 SYNTAX-INFORMATION STRUCTURE INTERFACE
focus domain. As we will see later, languages seem to vary in the way they cope with this feature of focus. While English makes use of nonrightmost (marked) focus-related accent to avoid the propagation of focus into uninformative elements, other languages like Spanish remove uninformative elements from the focus domain and maintain unmarked focus-related rightmost accent. I will consider these differences in chapter 7. 5.5 Some Conclusions In this chapter I have considered the relation between focus and other notions (accent, new information, and sentence-final position), and different types of focus proposed in the literature, I suggested distinguishing between accent and focus. I adopted the view that accent can mark a topic or a focus relation and proposed using the term “emphatic stress” to refer to the use of B accent in English to mark contrastive topics or contrastive foci. I pointed out that a notion of focus which includes these cases of accented topics would only apply to English and suggested instead keeping the term focus to refer to the informative part of the sentence marked with A accent. I considered different types of foci suggested in the literature and argued for keeping a unified notion. I have adopted Gundel’s definition of Semantic Focus and rejected any other subdivision of this notion. This notion of focus, is valid for English, Spanish and many other languages. Although a division such as the one in Vallduví (1990) and Kiss (1998) could be useful, it should not be interpreted as implying the existence of two kinds of focus. I argue that the different features that these two foci show are not due to their intrinsic different nature, but to their position in the sentence. Among them is the fact that unmarked rightmost focus can project while nonrightmost focus cannot. NOTES 1. As we will see later, this does not mean that focus has to occur in that position, First, in a language like English, focus can be easily shifted to the left intonationally. Second, there are syntactic constructions such as Focus Preposing which move focused elements to sentence-initial position. Third, some languages like Hungarian have a preverbal focus position. 2. I will be using “accent” and “stress” interchangeably.
FOCUS 165
3. In fact, as we are going to see below, he considers “new information” to be just one of the interpretations that focus can get, another being what he calls “focus of contrast.” 4. Note that in this case, as opposed to the case of a Topicalization such as that in (6), the accent on “Korean” is the only accent in the sentence. It is focus-related in the sense of being the only informative part and it is not a B accent but just the same A accent typical of uncontroversial sentence-final focus accent. 5. It should be noted that one way of enforcing this correlation syntactically that English does allow is the use of an agentive passive. That is, by transforming the subject in (i) into a by-phrase as in (ii), we have succeeded in placing that subject in sentence-final position:
(i) (ii)
Cervantes wrote el Quijote El Quijote was written by Cervantes
6. Rochemont also considers what he refers to as Cleft Focus and Constructional Focus. Since these are subtypes of the main two types—Cleft Focus is Contrastive and Constructional Focus is Presentational—I will not consider them here. 7. He is concerned with the inadequacies of the Nuclear Stress Rule in (i) from Chomsky and Halle 1968:
(i)
NSR: Assign nuclear stress (=accent) to the rightmost lexical category in S (Rochemont 1986:6)
8. Rochemont points out that lexical stress is predictable not only regardless of the context, but even despite the context. That is, he suggests that even in a situation in which the context would favor a shift of accent to maintain the information structure, this is nevertheless impossible. He offers the following example:
(i)
What else do you have that’s black?
I have BLACKboard
a
9. From the fact that the answer in (24) does not seem to have any contrastive interpretation, Rochemont concludes that there is nothing special about an accented subject and that it should not be considered in any way more marked or nonnuclear than any other accent. However, this may not be a justified conclusion. In fact, if we take into account the relatively strong correlation that we examined in chapter 2 between preverbal subjects and being topical, it may be that there is indeed something “special” about a focused preverbal subject. Note that putting aside the case of presentational verbs, which we are about to consider, an utterance such as that in (24) cannot start a discourse, and
166 SYNTAX-INFORMATION STRUCTURE INTERFACE
10. 11.
12.
13.
14.
there is only one context in which it would not be contrastive, namely, as an answer to a who (subject)-question. This term is taken from Culicover and Rochemont (1983) and it stands for “construable from the context.” Rochemont explains what he means by formal and informal entailment. I will not comment on his definitions here since it would take us too far away from our present goal. We are not so much concerned here with the appropriate technical definition, but with what he means by c-construable, which is basically discourse-old. That is, the term “c-construable” in the definition of Presentational Focus in (33) includes both directly and indirectly c-construable. As we will see, this is not going to be the case in the definition of Contrastive Focus. Although the projection of this type of rightmost focus up to the verb is uncontroversial, I think that the suggestion that it can also include the subject and, therefore, encompass the whole sentence is not straightforward. While it seems possible to answer a “what happened” question with a sentence with focus only on the rightmost element, like the one in (35), this may not mean that focus has extended all the way up. This would only be the case if we assume that the answer to a “what happened” question is necessarily an all-focus structure. However, as we have mentioned in chapter 3, it has been proposed (Lambrecht 1994) that a sentence with a topic-focus structure can be an answer to a “what happened” question. If this is the case, the fact that (35) can be an appropriate answer to the question in (36)c does not mean focus has extended to the whole clause. In fact, I think that the fact that focus cannot project from the subject in (38) is only a subcase of a more general phenomenon which has to do with narrow focus not being able to project in any position and not only in preverbal subject position. That is, none of the following sentences would be appropriate answers to the questions in (36):
(i) (ii)
Laurie FOLLOWED Ralph into the bedroom Laurie followed RALPH into the bedroom
If this is right, we will have to review the way focus projects. As we will consider later, it is assumed that focus can project from an internal argument to the verb, as in (iii), where focus projects from the only stressed element “lasagna” to its head “cook,” so that the whole VP is focused as the question requires: (iii)
What did Laurie do?
She cooked LASAGNA
FOCUS 167
However, if that were the case, (ii) should be an appropriate answer to the same question, since in Selkirk’s theory we would expect the focus on “Ralph” to be able to project up to the verb. Notice that one of the differences between the two focused objects is the fact that the focus which projects is the rightmost. The focus in (i) and (ii) has been shifted to the left, which has as a consequence the fact that it is narrow, it cannot project and marks the rest of the sentence as background information or in Rochemont’s terms as c-construable. 15. Lambrecht (1994) suggests that even when unaccusative and presentational verbs do overlap, they do not totally coincide. For a detailed consideration of unaccusativity see Levin and Rappaport (1995) and the works cited there. 16. For further discussion see chapter 6, section 6.3.1. 17. In fact, later Rochemont will change the two instances of “c-construable” in (40) into directly c-construable. This is due to the fact that as the definition stands, the focus in (i) would come out as Contrastive:
(i)
A LETTER arrived for you today
By changing the requirement that the nonfocused part be directly c-construable, he makes cases like that in (i) not eligible for Contrastive Focus, since the nonfocused part is only indirectly c-construable. It is interesting to note that the reason behind trying to prevent (i) from being considered contrastive is the fact that it can be discourse-initial. 18. Rochemont says that presentational sentences such as (i) exemplify the case of strictly Presentational Focus:
(i)
A LETTER arrived
In this case, the focus cannot be also Contrastive since the nonfocused part is not directly c-construable. 19. Notice that the change from c-construable to directly c-construable that Rochemont suggests, mentioned in footnote 104, does not get rid of this problem. (φi in (41) does not involve indirectly c-construable elements such as presentational verbs. Both the unfocused part and the focused part are directly c-construable. 20. Culicover and Rochemont (1983) have three different types of focus: “presentational,” “contrastive” and “informational.” Their Presentational Focus is different from Rochemont’s (1986) and it only involves Rochemont’s verbs of appearance. However, their contrastive includes
168 SYNTAX-INFORMATION STRUCTURE INTERFACE
21.
22. 23.
24. 25.
the informational, which is equivalent to Rochemont’s “presentational.” Therefore, given that Culicover and Rochemont’s (1983) “informational” type is included in the “contrastive” type, it is not clear how suggesting that cases like (43) could be instances of Culicover and Rochemont’s (1983) “informational” could rescue (43) from being considered a Contrastive Focus. Notice that these cases of presentational sentences would still be only presentational if, as suggested above, we consider that in presentational sentences the focus projects from the internal argument to the verb. They would be seen as all-focus structures and without an unfocused part, they would be presentational only. Among these, we can mention Chomsky (1971), Jackendoff (1972), Rooth (1985), Rochemont (1986), Sgall et al. (1986), etc. As we noted above, (41) involves a solely-Contrastive Focus, which, in fact, does not follow the definition (the whole utterance—φi—should not be c-construable, but it is); the focus in (42) is both Contrastive and Presentational and the one in (43) should not come out as Contrastive, but it does. The construction in (54) will be referred to here as “Focus Preposing” and will be considered in detail in chapter 6. Restrictive focus is defined in the context of Heim’s (1982) Filing system as involving a set of cards. Erteschik-Shir also points out (1997:26) that restrictive focus is somewhat androgynous: “On the one hand it is stressed and triggers a search for a card as a proper focus should, on the other it functions as a topic with respect to predication.” These remarks refer to the following example:
(i)
Which of your friends is bald? THE KING OF FRANCE is bald
For Erteschik-Shir, “is bald” is always going to be considered the predicate and its subject is always going to be considered the topic, even if it is focused. However, if we equate topic with subject of a predicate (not in a syntactic, but only in a semantic sense) and focus with what is predicated of a topic, we could consider that in this context “being bald” is part of the topic and what is predicated of it, is that it can apply to the king of France. 26. One of the interesting features of cases like (61) is the fact that this focus does not correlate with the nuclear scope. That is, while it has been noticed (Partee 1991) that topics are among the things that can be mapped into the restrictor in a tripartite semantic structure while focused elements fall into the nuclear scope of operators, these cases involving focused generics may call into question this correlation between focus
FOCUS 169
and nuclear scope, if the assumption that generically interpreted NPs are in the restrictor part is right (Diesing 1992, Wilkinson 1991). On the other hand, it may be the case that the correlation between focus and the nuclear scope is true only of VP-internal focus. 27. In fact, the special flavor of the focused phrase in Erteschik-Shir’s example in (61) seems to arise more from being the subject of an individual level predicate rather than just being generic. Focus on a generic direct object as in (i) seems perfectly “unflavored”:
(i)
Which pet do you like best? I like DOGS
But, even (ii) is also fine, which may mean that provided with the right context there is nothing particular about focused generic subjects either. (ii)
Who do you think are more intelligent, men or women? WOMEN are more intelligent
28. In chapter 7 I will consider his trichotomy in more detail. 29. In fact, Lambrecht’s Information Structure involves two other sets of categories, in addition to the topic relation and the focus relation. The first has to do with the speaker’s assumptions about the hearer’s state of knowledge and awareness at the time of an utterance. It involves the notion of Propositional Information with its two components: Pragmatic Presupposition and Pragmatic Assertion which have to do with the structuring of propositions into portions which a speaker assumes an addressee already knows or does not yet know. The second has to do with the speaker’s assumptions about the nature of the representations of the referents of linguistic expressions in the hearer’s mind at the time of the utterance and with the constant changes which these representations undergo in the course of a conversation. It involves the notions of Identifiability and Activation, which we already considered in chapter 2. 30. A fourth communicative function expressed via focus structure that he mentions but that he does not consider is that of marking the polarity of a proposition. 31. Although it is not used in the general characterization of identificational focus, later she also suggests that identificational focus can also be ±contrastive. 32. In the next chapter we will consider this construction in great detail. 33. It is worth pointing out that Kiss seems to allow for information focus to occur inside identificational focus. See Kiss (1998:261) for details.
170 SYNTAX-INFORMATION STRUCTURE INTERFACE
34. I cannot consider here the different proposals in the literature which examine the interpretation of focus. However, I should mention that although most studies on the semantics of focus emphasize the fact that focus introduces a set of alternatives into the discourse (Rooth’s 1985 p-set, for instance), as understood here, that is not the only or even most important feature of focus. The most important function of focus, from the perspective advocated here is its predication semantics, which works such as Peregrin (1996) have pointed out. 35. I cannot consider here all the details of Selkirk’s theory. She allows focus to project even in cases like (103). See Von Stechow and Uhman (1986) for a critique of her theory.
CHAPTER 6 Focus Preposing
6.1 Introduction1 In the previous chapter we have considered the notion of focus in detail. We have seen that focus and accent do not always correlate, and we have pointed out that cases of accented topics in English should not be considered focused topics. We suggested that accent can also mark a topic relation and .showed that the type of accent that topics get is different from focus-related accent. In addition, there is a preposing mechanism, which has been claimed to involve the preposing of a topical focus: Focus Preposing (FP). In this chapter I am going to consider this construction in detail, explore different possibilities regarding its informational articulation, and suggest that FP should not be considered the preposing of a topical focus. As the example in (1) shows, the preposed element in FP differs from other preposing mechanisms (Topicalization, Left-dislocation, Clitic Left-dislocation) in that it contains the focus of the utterance and bears, therefore, the single accented syllable. (1)
SIX DOLLARS it costs
(Ward 1988)
However, as Ward (1988) has pointed out, in English, the preposed focused element in FP is also a Backward Looking Center (BLC); that is, it is related to a set of previously evoked discourse entities via a salient scalar relation. Does this mean that this preposed focused element is also topical in English? Is this a language particular restriction or is FP in general precisely the preposing of a topical focus? These are some of the questions that will be addressed in this chapter.
172 SYNTAX-INFORMATION STRUCTURE INTERFACE
This chapter is organized as follows. In 6.2, the possibilities available in the literature that could allow us to have topical foci are explored. Obviously, a Topical focus, that is, an element that is focused and is considered to be topical at the same time, cannot be explained only by appealing to a Topic-Focus articulation. I consider three different possibilities to account for topical foci. These comprise a Reconstruction analysis as proposed by Vallduví (1990), the interaction of two dichotomies, which can allow focus to occur inside a topic, and the existence of a Metalinguistic Focus in the sense of Erteschik-Shir (1997). After examining these proposals, I consider other languages and other dialects of English in section 6.3, and I suggest that the preposed element in FP does not have to be considered topical and that it can be viewed as a Focus-Background structure where the focus is unambiguously marked and placed in a preverbal focus position, different from the topic position. Crucially, as pointed out by Vallduví (1990), the verb is unambiguously excluded from the focus. Further, I consider some similarities between FP and focused preverbal subjects (6.3.1), I examine some language-particular features (6.3.2), and I finally offer some tentative comments about the relationship between FP and wh-movement, which suggest that there may be a correlation between narrowness and marked constructions, which is shown by both FP and wh in situ, as opposed to their unmarked counterparts: focus in situ and wh-movement. 6.2 FP as a Topical Focus In his extensive analysis of preposing mechanisms in English, Ward states that both FP and Topicalization perform two simultaneous discourse functions: they mark the preposed element as a Backward Looking Center (BLC) and they also mark the Open Proposition as salient in the discourse. The Open proposition is the structure that results from putting the preposed element back to its base position and substituting a variable (“It costs x,” in the case of the example in (1). So, from this point of view, both FP and Topicalization involve the preposing of a BLC, which is the notion that Ward thinks should replace sentence topic. The only difference between these two constructions, according to Ward, is where the focus is: in FP, the BLC and the focus coincide, while in Topicalization they do not. According to Ward, the relations that the preposed element can have with the evoked set include, among others, a quantity scale (2), an
FOCUS PREPOSING 173
entity/attribute relation (3)–(4), and a temporal relation, as in (5). These are naturally occurring examples, (203), (204), and (205) respectively in Ward’s work. (2)
I made a lot of sweetbreads. A COUPLE of POUNDS I think I made for her.
(3)
Are there black kids in that school now? Not many. I had two really good friends. DAMON and JIMMY their names were
(4)
The Contras worked with a new strategy REVOLUTIONARY they called it…
(5) I promised my father—on CHRISTMAS EVE it was—to kill… In addition, there are also cases where the focused BLC constitutes an alternate value on the scale. Consider, for instance the examples in (6) and (7): (6)
(7)
Waitress: Did you want tea? Customer: COFFEE I ordered, I think I think she was Japanese. No—KOREAN she was (Ward 1988:114, examples (206) and (207))
In these examples, a value is mentioned in the immediately preceding discourse (“tea” in the scale “hot beverages” in (6) and “Japanese” in the scale “Asian nationalities” in (7)), and the focused BLC can be seen as an alternate value on the relevant scale (“coffee” and “Korean” respectively), which is being substituted for the evoked value.2 In the case of Topicalization, however, the BLC is not focused: (8) You see every Woody Allen film as soon as it comes out No-Stardust Memories I saw (only) yesterday (Ward 1988:42) If FP only differs from Topicalization in that the BLC is focused in the former, but not in the latter, we need to explore how to account for these focused BLCs or topical foci, as I have referred to them above. Let us see how this could be done. In order to account for the apparent existence of such things as topical foci, I am going to consider three
174 SYNTAX-INFORMATION STRUCTURE INTERFACE
different possibilities: an analysis based on reconstruction, a two-dichotomy approach and a two-types-of-focus analysis. 6.2.1 The Reconstruction Analysis Vallduví (1990) briefly considers Ward’s analysis and points out the problem of having an element that is both a focus and a BLC—a link in his theory. This is not possible either in a Topic-Focus articulation or in his hierarchical trichotomy, which divides a sentence into focus and ground and further subdivides the ground into the link and the tail: (9)
S—{FOCUS, GROUND} GROUND={LINK, TAIL}
Vallduví suggests that the position of the preposed element may reflect its force as a link and points out that we can get the focus value by reconstructing the preposed phrase. Thus, he assumes that the preposed element is topical in English and concentrates on the focus force of this element. However, he also points out that the scale or set itself can be interpreted as the link or BLC and the actual lexical content of the phrase as the focus. If it is the scale or set that constitutes the BLC, and the focused preposed phrase constitutes the focus, then it may not be exactly the case that in FP the BLC and the focus coincide, that the preposed phrase is both topical and focal. In fact, when Ward talks about Topicalization (Ward 1988:58), he explicitly suggests that “perhaps it is not so much the referent of the preposed constituent of a preposing [construction] which corresponds to the topic, or what the sentence is about, but rather it is the scale on which that referent can be said to lie.” For instance, he says that in the example in (10), instead of labeling baseball as the topic we may want to say instead that the topic is the scale “sports” and that “baseball” represents a value on that scale. From this point of view, the topic of a sentence need not be explicitly mentioned. (10)
Do you watch much sports? Not much football. Baseball I like a lot better
Thus, not only may it be the case that the preposed element in FM is not in itself topical but this may apply to Topicalizations, too. That is, the much more accepted fact that English Topicalizations involve the
FOCUS PREPOSING 175
preposing of a topic may need to be revised, too. I will not consider this problem here. Putting aside for the moment the possibility that the scale may be the BLC rather than the preposed element itself, it is worth pointing out that even if the preposed element itself is understood to be the BLO—to be related to an evoked set as, for instance, Korean in the FP construction in (7), which is related to Asian nationalities—this would also be the case for its nonpreposed counterpart in (11): (11)
I think she was Japanese. No—she was KOREAN
That is, Korean in (11) is also related to a set of previously evoked entities or scale. This case, however, would not be claimed to be a case of a topical focus, but just regular focus, in spite of the fact that the same relation holds. Therefore, being related to a set of previously evoked entities should not be considered sufficient evidence of a topical nature. The preposing may be felicitous partly due to the existence of that relationship but that may not affect the nature of the focused element. Regarding the position of the preposed element, it is not clear that it shows its force as a link, as Vallduví suggests. Vallduví (1990:42) himself points out that although links do occupy the sentence-initial position, this position can also be occupied by other nontopical elements such as wh-phrases. The preposed focused phrase in FP may well pattern with wh-phrases rather than with links or topics. For instance, in Spanish, it behaves just like wh-movement: only one focused phrase is allowed, the focused phrase cannot co-occur with a wh-word, and it triggers the verb-subject order exactly as wh-movement does: (12)
En PRIMAVERA visitó Juan Leningrado in spring visit John Leningrad “In the SPRING John visited Leningrad”
(13)
*En PRIMAVERA Juan visitó Leningrado (Hernanz and Brucart 1987)
Topics do not have any of these features. Also, a topic and a focus preposing can co-occur, and if they do, the focused element has to be in a position lower than the topic, just like wh-phrases:
176 SYNTAX-INFORMATION STRUCTURE INTERFACE
(14)
a. Al jefe, BRÉCOL le compraron sus empleados to the boss, BROCCOLI cl bought his employees b. *BRÉCOL al jefe le compraron sus empleados
(15)
a. Al jefe, ¿qué le compraron sus empleados? b. *¿Qué al jefe le compraron sus empleados?
Therefore, it is not clear that FP is related to the preposing of topics rather than to other operations such as wh-movement. While Ward’s analysis in terms of what all preposing mechanisms seem to have in common in English is very insightful and may be telling us a lot about the discourse requirements for the displacement of elements to be felicitous in English, it may not capture the nature of FP in particular, as opposed to other preposing mechanisms. For him, Focus Preposing and Topicalization are assumed to perform exactly the same functions. This does not tell us in what way, if at all, the fact that the preposed element is focused in Focus Preposing affects these two functions; it should be mentioned that he only considers Focus Preposing briefly. I think that a certain saliency of the Open Proposition does seem to be a hallmark of Focus Preposing and that in many cases, there is also a relationship between the focused preposed element and other elements in the discourse. However, I am going to claim that we do not have to consider the preposed element to be topical, and that the two characteristics just mentioned do not have to be viewed as the functions of the construction. I have been suggesting that we may not need to consider the preposed element to be topical and, therefore, we may not run into the problem of accounting for topical foci. However, one might object to this and point out that there are proposals that can account for topical foci with no problem and, therefore, since topical foci may not be a problem, avoiding it should not be mentioned as a good consequence of my proposal. So, before I pursue an analysis of Focus Preposing which does not consider the preposed element to be topical, I will consider two analyses for which the existence of topical focus would not be a problem: a two-dichotomy approach and a two-types-of-focus approach.
FOCUS PREPOSING 177
6.2.2 The Interaction of Two Dichotomies The first analysis that would allow us to account for the existence of topical foci has to do with the interaction of two dichotomies: a Focus-Background and a Topic-Comment dichotomy. This interaction has been suggested to account for the possibility of an accented element inside a topic, as in (16). (16)
What did Bill’s sisters do? Bill’s YOUNGEST sister kissed JOHN
(Krifka 1991)
This occurrence of an accented element inside a topic could be accounted for, if we allowed the Focus-Background articulation to occur both inside the Topic and inside the Comment. From this perspective, in (16) “Bill’s youngest sister” is the Topic and “kissed John,” the Comment, In addition, it could be said inside the Topic we have a Focus-Background articulation, where “youngest” is the focus and the rest of the NP the background. Thus, the existence of focused topics or focused elements inside topics has been acknowledged. Now, can this help us with the case of FP? Notice that there is an important difference between the case in (16) and our cases of FP. The preposed focus in FP is the only focus allowed in the sentence. What the interaction of two articulations allows for in cases like (16), is to have two foci (one inside the Topic and one inside the Comment). This is not what we need in order to account for FP. There is only one focus and if it is inside the topic we have to explain the fact that the Comment does not and cannot have another focus, which given the usual understanding of Comment, seems unusual. In fact, it is not clear that we actually need two different articulations even to account for cases like (16), where we do seem to have two foci. As pointed out in chapter 5, we should be careful with equating accent with focus. The fact that a topic or part of a topic can get an accent does not necessarily mean it is focused. As we have seen, accent does not always seem to correlate with focus. Thus, in (17), the wh-word need not be accented in order to receive its interpretation as focus, while the verb “buy,” which does get an accent, need not be interpreted as focused. (17)
What did you BUY?
(Rochemont 1986)
178 SYNTAX-INFORMATION STRUCTURE INTERFACE
As Lambrecht (1994) has suggested, accent can mark either a Topic relation or a Focus relation. Lambrecht suggests that when the role of a referent as a topic is not expected or taken for granted or has not yet been established, it can be accented, as in the following example: (18) That was a STUDENT of mine. Her HUSBAND had a HEART attack From Lambrecht’s perspective, after uttering the first sentence, the student is a possible topic and could be taken for granted. If the following sentence were about her (“That was a student of mine. She had a heart attack”), we would not need an accent on the pronoun “she.” However, the topic of the following utterance is not the student but her husband and not being yet established or expected as a topic, it carries an accent which marks its role as the new topic. The same could be said about (16). The expected topic would be “Bill’s sisters” but the speaker wants the topic to be “Bill’s youngest sister.” Therefore, in order to mark this unexpected topic relation, an accent is used. From this perspective, accent does not correlate with focus and can mark either a topic or a focus relation. Furthermore, the accent that can occur inside a topic as in (16) and (18) has been argued to be different from the focus-accent. For instance, this is mentioned by Gundel (1994), who points out that the accent that these topics show is Pierrehumbert L+H* or Jackendoff B accent, which is different from that of the focus of the utterance. This applies to cases such as those in (16) and (18), and also to cases of Topicalization, as in (19): (19) That COAT you’re wearing I don’t think will be WARM enough (Gundel 1994) In these cases the type of accent in the topic is different from the accent in the focus. In addition, this accent in the topic cannot be the only accent in the utterance. I would like to add another argument from Gundel supporting the difference between the accent inside the topic and the accent inside the focus, or, in Gundel’s terminology, the difference between Contrastive Focus and Semantic Focus. Gundel notes that only the accent inside the focus affects truth conditions. She gives the examples in (20) and (21) taken from Partee (1991), which show the truth-conditional effects of
FOCUS PREPOSING 179
Semantic Focus, where different choices of focus determine different truth conditions: (20) The largest demonstrations took place in PRAGUE in November (in) 1989 (21) The largest demonstrations took place in Prague in NOVEMBER (in) 1989 (Partee 1991) In contrast, she says, the sentence in (22), with a Contrastive Focus on “Prague,” and a Semantic Focus on “November” has the same truth conditions as (21), not (20). The Semantic focus is underlined in (22): (22) The largest demonstrations took place in PRAGUE in NOVEMBER . (Gundel 1994) Therefore, in view of all these differences, if we want to maintain the two articulations approach for these cases, we would have to admit that the focus that can occur inside the topic has different features from the focus inside the Comment. Gundel’s notion of Contrastive Focus is related to the other approach I am going to consider next, which has to do with Erteschik-Shir’s (1997) notion of Metalinguistic Focus. 6.2.3 Metalinguistic Focus In order to account for topical foci in FP, we could explore the idea that the type of focus that this construction shows may be different from the focus in the Topic-Focus articulation. The Focus in FP could be a case of what Erteschik-Shir (1997) calls Metalinguistic Focus (MF). We have briefly considered this notion in chapter 5. As pointed out there, although this notion has to do with contrast, it is very different from Gundel’s Contrastive Focus. First, Erteschik-Shir’s notion involves restrictive sets which have exactly two members, Second, Gundel’s Contrastive Focus always co-occurs with a “real” Semantic Focus and this is not the case for Erteschik-Shir’s MF. As her example of MF in (23) shows, the focus on “a man” is the only focus in the sentence. (23)
A MAN is intelligent (Erteschik-Shir 1997)
180 SYNTAX-INFORMATION STRUCTURE INTERFACE
This contrasts with (24) which shows Gundel’s Contrastive Focus co-occurring with the Semantic Focus: (24)
A MAN is INTELLIGENT, a WOMAN is CLEVER
Note that the focus in Erteschik-Shir’s example of MF in (23) is much more similar to our examples of FP than to Gundel’s examples of Contrastive Focus. For Erteschik-Shir, a man in (23) is the topic of the main predication. At the same time, it constitutes the focus in the subordinate predication, with respect to the other member of the contrast (a woman), which is not overt, but is present as part of the subordinate predication, and necessary for the interpretation of the sentence. This is reminiscent of the mechanism of having a focus inside a topic, as we saw the interaction of two articulations also tried to do. Instead of two articulations, she has two levels of predication, which allows for an element to be a topic in one level and a focus in another. As we pointed out in chapter 5 (section 5.3.3), it is important to notice that in (23) there is only one focus, and, according to her, it is inside the main topic. That is, in the main predication, “is intelligent” is the Comment and should have a focus, but it does not. This is exactly the situation we face in the analysis of FP if we say that the preposed element is both topical and focus. Erteschik-Shir’s solution for this problem in sentences such as (23) is to say that Metalinguistic Focus overrules regular focus with respect to stress. As suggested before, I am not sure that the analysis of (23) in terms of MF and in terms of a main and subordinate predication does anything more for us than a simpler analysis in terms of a regular focus and an OP. In addition, there are two differences between her examples of MF and our instances of FP that could be important. In FP, the set is not restricted to two members. Further, the preposed element in FP does not tend to be generic as in (23). So, although Erteschik-Shir’s concept of MF deserves much more attention that I can give it here, this may not be the most useful notion to account for the characteristics of FP. Up to this point, we have seen that to assume that the preposed focused element is topical and to try to account for this feature of FPs in terms of the interaction of two dichotomies or the existence of a different type of focus has not been very satisfactory. In the next section I provide more evidence which suggests that the preposed element in FP does not have to be consider topical. This construction will then be viewed as a simple, one-dichotomy, Focus-Background articulation where the focus is narrow, and disambiguated, in that it excludes the
FOCUS PREPOSING 181
verb. I will suggest that the function of FP, and the motivation for placing the focus in sentence-initial position may be to unambiguously mark narrow focus on the preposed element. 6.3 FP as a Nontopical Focus If we look at other languages and other dialects of English, we can see that there are instances of FP which do not require the preposed element to be related to a previously evoked set. This could be taken to mean that the feature noted by Ward may be the result of English particular restrictions regarding the displacement of elements and not a hallmark of FP. In fact, even when Ward claims that for all types of preposings, it is the case that the preposed element is a BLC, when he considers so-called Yiddish Movement (YM) by Prince (1981a), he points out that the relationship between the preposed constituent and other entities in the discourse is less direct. In (25)–(27), we have some examples of YM taken from Ward. (25) Then came bald Uncle Hymie, one fist shaking violently in the air—like LENIN he looked!… (26)
Butter! She’s dreaming about butter. RECIPES she dreams while the world zips…
(27) You’ve got clean underwear? I’m washing it at night. I’m OK, Aunt Gladys. By hand you can’t get it clean. It’s clean enough. Look, Aunt Gladys, I’m having a wonderful time SHMUTZ (filth) he lives in and I shouldn’t worry! Ward points out, for instance, that the preposed focused element (SHMUTZ) in (27), is only loosely related to the discussion of clean underwear. Therefore, it looks like the requirement that the preposed focused element be a BLC (be related to a set of previously evoked entities) does not have to be met by the subgroup of FPs referred to as YM. As Prince (1981a) had already noted* one of the features of YM is the relative newness of the preposed constituent. I think this is also true of FP in languages such as Catalan and Spanish. Unfortunately, as far as I know, no corpus such as that
182 SYNTAX-INFORMATION STRUCTURE INTERFACE
collected by Ward for English, where we have all the contexts of the FPs, is available for these languages. Vallduví (1990) offers the following example in Catalan, where the preposed element can be new and need not be related to a previously mentioned set of entities: (28) No m’ho hagués cregut mai! L’amo està ben boig! BROQUIL, vol ara “I can’t believe this. The boss is going crazy! BROCCOLI, he wants now” This is also possible in Spanish: (29) No me lo puedo creer. El jefe se està volviendo loco. BRÉCOL quiere ahora Thus, it seems that the required relation between the preposed element and the evoked set may be a feature having to do with felicity conditions on displacement of elements in general in English (since, as noted by Ward, also applies to Topicalizations), rather than being an inherent feature of FP. What seems much more topical in all instances of this construction, regardless of the language, is the Open Proposition, which is assumed by the speaker to be salient in the discourse at the time of the utterance: “She was x,” in the case of (7), repeated in (30) for convenience. Certain saliency or topicality of the OP does look like a hallmark of FP. (30)
I think she was Japanese. No—KOREAN she was
One interesting feature of this construction is that the verb is unambiguously excluded from the focus. For instance, in English, the sentence in (31) is ambiguous as to the extent of the focus material, The focus can comprise any of the material between brackets. (31)
[I [put [the knife [in the DRAWER]]]]
That is, it can be an appropriate answer to any of the following questions: Where did you put the knife?/What did you put where?/What did you do?/What happened? However, if we prepose the locative as in (32), the focus is unambiguously restricted to the preposed element: (32)
In the DRAWER I put the knife/it
FOCUS PREPOSING 183
According to Vallduví, the exclusion of the verb from the focus is not just a feature of this construction in Catalan, but rather its raison d’être. That is, that is the function of FP in Catalan: it is used to express cases in which the focal material excludes the verbal head. From this perspective, the function of FP may just be the unambiguous marking of narrow focus. This would be the difference between the preposed and the nonpreposed version in (31) and (32): in (32) the focus is unambiguously marked and, therefore, the OP is also unambiguously marked as given and salient. In (31), however, nothing is unambiguously marked. Notice that in a pair such as (33)–(34), the difference is minimal, since in both versions “she was” is clearly salient and the focus is clearly restricted also in both versions to “Korean.” (33)
I think she was Japanese. No, KOREAN she was
(34)
I think she was Japanese. No, she was KOREAN
Even if we consider the function of FP to be the marking of unambiguous narrow focus, we can still view the two features that Ward has noted are present in instances of FP as felicity conditions. For instance, consider the FPs in (28)-(29). Let us say that we are going to utter “BROCCOLI he wants now” when we want to restrict focus unambiguously to the noun. Why would we want to make it clear that it is broccoli that he wants and not something else? This is probably going to happen in a situation in which the fact that he wants this thing or the other is salient or known (Ward’s saliency of the OP “he wants”), and, therefore, several things that he may have wanted are also salient (Ward’s relationship between the preposed element and the set evoked in the discourse). So, the features noted by Ward could be viewed as discourse prerequisites for the unambiguous marking of narrow focus to be felicitous. From this perspective, FP can be viewed as a Focus-Background structure, where the focus is unambiguously marked, it excludes the verb, and it is placed in a preverbal position, different from the topic position. 6.3.1 FP and Focused Subjects We have seen that sentence-final focus can project. According to Selkirk (1984 and 1995), focus on an internal argument licenses focus
184 SYNTAX-INFORMATION STRUCTURE INTERFACE
on the head and focus on the head licenses focus on its phrase. For instance, the verb “cooked” in (35) does not have to be accented even when it is part of the focus: the focus on its internal argument licenses focus on its head. (35)
What did Mary do?
She [cooked LASAGNA]
However, as she notices, focus on a subject cannot project. In the sentence in (36) the verb cannot be understood as part of the focus, and, therefore, it cannot be an appropriate answer to a question such as “What happened with the lasagna?” (36)
MARY cooked the lasagna/it
In this case, if the verb is part of the focus, it needs to be accented. In this respect, subjects behave similarly to our focused preposed elements. Their focus is necessarily narrow, it cannot project to any other part of the sentence, just like our FPs. Both can be seen as instances of unambiguous Focus-background structures or what Lambrecht (1994) calls Argument-Focus, as opposed to cases such as (35) which have Predicate-Focus. As mentioned in chapter 5, subjects of unaccusative verbs seem to behave differently. Consider (37): (37)
My CAR broke down
(Lambrecht 1994)
In this case, the focus on the subject seems to be able to reach the verb. One could argue that focusing of an unaccusative subject is in fact focusing of a direct object and that, therefore, (37) is similar to (35) rather than (36). From this perspective, we would have to say either that focus is assigned prior to NP movement or alternatively that F-marking of a constituent licenses the F-marking of its trace, as in Selkirk (1995). However, this would predict a similar behavior of passive subjects and this prediction seems to be wrong. Rochemont points out that focus cannot project from a passive subject to the verb.3 If so, something else is going on with unaccusative subjects. Although I cannot examine this problem in depth here, I think that part of what is going on may be the fact that the verbs referred to as unaccusatives are verbs that take focused, nontopical subjects. Notice that as opposed to passive subjects, unaccusatives subjects appear postverbally in languages that allow for postverbal subjects. From this perspective, an alternative analysis,
FOCUS PREPOSING 185
which Selkirk also mentions although she does not adopt it, seems more appropriate: we can account for (37), if these focused subjects reconstruct in English, assuming that the appropriate level is LF. I will leave this open for further research. 6.3.2. FP and Language-Particular Features If we look at instances of FP in English and FP in languages like Spanish, we notice several important differences. First, FP in English does not trigger subject-verb inversion, as shown by (38)–(42) from Ward (1988): (38) (39)
SIX DOLLARS it costs I made a lot of sweetbreads. A COUPLE of POUNDS I think I made for her.
(40)
Are there black kids in that school now? Not many. I had two really good friends. DAMON and JIMMY their names were
(41)
The Contras worked with a new strategy REVOLUTIONARY they called it…
(42) I promised my father—on CHRISTMAS EVE it was—to kill… FP in Spanish, on the other hand, behaves similarly to wh-movement and triggers obligatory inversion of the subject Thus, while (43) with the order verb-subject is fine, (44) is ungrammatical. (43)
EN PRIMAVERA visitó Juan Leningrado in spring visited-3sg John Leningrad “IN SPRING John visited Leningrad” (Hernanz and Brucart 1987)
(44)
*EN PRIMAVERA Juan visitó Leningrado
As we mentioned in chapter 3, it is possible for a topical element to precede the focus-moved phrase, as in (45), but no phrase can intervene between the focal phrase and the verb.
186 SYNTAX-INFORMATION STRUCTURE INTERFACE
(45)
A sus padres MENTIRAS les cuenta siempre to his parents lies cl tell-3sg always “He always tells his parents lies”
As (46) shows, this is another similarity between FP and wh-movement in Spanish: (46)
El regalo ¿quién lo tiene? the present who cl has “The present, who has it?”
Another difference between FP in Spanish and other Romance languages, and FP in English is that FP in English tends to prepose nonreferential phrases. None of the preposed elements in (38)–(42) are referential. Thus, (47) would not be considered possible, except as an instance of YM: (47)
?the KNIFE we put in the drawer
However, notice that this becomes totally acceptable just by adding “even” to the preposed focused element, as in (48), that is, by making it clear that the preposed NP is the semantic focus: (48)
Even the KNIFE we put in the drawer
Another way of making (47) acceptable in English is by making the focused element be preceded by “it is,” making it into a Cleft: (49)
It is the KNIFE (that) we put in the drawer
The cleft in (49) also marks unambiguous narrow focus. This construction is much more productive in English than FP, while in Romance languages the opposite seems to be true. One might say that the preference in English for clefts over FPs may be related to English reluctance to have displaced elements. However, note that other mechanisms that displace elements such as Topicalization are not so restricted and are much more productive. It may be that the existence of Topicalization in English has something to do with the restrictions that this language has on FP. Notice that both constructions involve the preposing of an element that takes an accent in both cases, and lacks a
FOCUS PREPOSING 187
resumptive pronoun. In Romance languages like Spanish, however, there is no similar mechanism to FP which also involves an accent and no pronoun: there is no Topicalization. The equivalent of Topicalization in these languages is a Clitic-Left-dislocation and this construction has a resumptive pronoun. In addition to the presence of other preposing mechanisms in English such as Topicalization, the restrictions of FP in English might also be related to other phonological and syntactic features of this language. In English, a direct object can be unambiguously focused in situ as in (50): (50)
We put the KNIFE in the drawer
If the main function of FP is to mark unambiguous narrow focus, there may not be much need for FP in English. The unmoved version in (50) can be argued to be more economical. We would expect FP to be very productive in a language in which the option in (50) is not available. Vallduví (1990) points out that this is the case in Catalan. According to him, Catalan cannot shift accent to the left. That is, main accent has to be sentence-final and, therefore, the counterpart of (50) is ungrammatical in Catalan: (51)
*Fiquem el GANIVET al calaix
Furthermore, Catalan has restrictions concerning the order of complements and it cannot place the object in final position, where it would be interpreted as focal: (52)
*Fiquem al calaix el GANIVET
Therefore, the only way of expressing (50) in Catalan is through a Right-dislocation as in (53) or a FP construction, as in (54): (53) (54)
Hi fiquem el GANIVET t, al calaix El GANIVET, fiquem t al calaix (Vallduví 1990)
This raises a question: what happens if we want to focus “al calaix” as in (55)? (55)
Fiquem el ganivet al CALAIX
188 SYNTAX-INFORMATION STRUCTURE INTERFACE
This complies with all the rules: the focus is sentence-final and the order is fine. If there is any relationship, between the possibilities available and the productivity of FPs, we predict the preposing of PPs similar to “al calaix” in (56) to be less productive considering that now we have the nonmoved option in (55). (56)
Al CALAIX, fiquem el ganivet
However, this does not seem to be the case. That is, there is no sense in which (56) could be said to be less productive or more marked than (54). However, it should be taken into account that in Catalan (55) and (56) differ as to the possibility of having narrow focus on the PP. Catalan seems to have a further restriction having to do with the projection of focus. While the English counterpart of (55) could answer a question such as “Where did you put the knife?” In Catalan, (55) is not felt to be a felicitous answer to that question, but only to a question such as “What did you do?” That is, there is a clear preference for sentences such as (55) to be interpreted with VP focus rather than PP focus. According to Vallduví, nonfocused elements tend to be either left- or right-dislocated. Therefore, if the verb and the DO are not dislocated, as in (55), they are understood as part of the focus. This might explain the absence of the narrow focus interpretation in (54), the nonpreposed version, and also the productivity of (56). Needless to say, in order to reach any valid conclusions on the nature and function of FP crosslinguistically, a more comprehensive study is needed. 6.3.3 FP and Wh-Movement We have seen some language-particular differences concerning the nature and productivity of FP. I would like to add some tentative comments about the relation between FP and wh-movement, which may suggest that the correlation between a marked option and narrow focus is not a particular feature of FP. There seems to be a parallelism between wh-movement and FP in the following sense. In both English and Catalan and many other languages, wh phrases move to sentence-initial position in the unmarked case, as in (57): (57)
What did he say?
FOCUS PREPOSING 189
Under special contextual circumstances, the wh phrase can be left in situ, as in so-called echo-questions: (58)
He said WHAT?
This marked option seems to be marking narrow focus on the wh phrase, and, therefore, the rest of the utterance as the OP, as background information. The moved, unmarked version, on the other hand, can be ambiguous as to the extent of the focus and the background. The sentence in (57) could be uttered after “He said many interesting things,” where “he said” would be then part of the background, but this question could also be uttered after “He came late to the meeting” or “He called me yesterday,” etc., where the fact that he said something does not have to be salient or pragmatically presupposed. That is, while in the unmarked wh-moved question in (57) is not clear what is background and what is not, the echo-question in (58) “He said what” is only felicitous, if the fact that he said something is background information and probably also salient in the discourse. This would mean that in a language where there is obligatory wh-movement to sentence-initial position, there is the option of leaving the wh in situ to disambiguate and restrict the focus. If this is right, I think it parallels the difference between focus in situ and FP, taking into account that the unmarked case is the opposite. That is, while sentence-initial position is the unmarked position for wh phrases, focal elements are sentence-final in the unmarked case in these languages. As we have already seen, focus in situ is ambiguous. In its unmarked final position, as in (59), it can comprise just the NP, the VP, or the whole sentence. (59)
[He [said [many THINGS]]]
That is, it could be uttered after “What did he say?,” “What did he do?” or “What happened next?,” as in the case of (57). However, if we choose the opposite of its unmarked position, that is, if we place it in sentence-initial position, we make it unambiguously narrow: (60)
Many THINGS he said
From this perspective, both FP and wh-in situ would be mechanisms which disambiguate their unmarked counterparts.4 In both cases the opposite of the unmarked position is chosen: in situ for wh phrases and movement for focus. In addition, it may not be a coincidence that the
190 SYNTAX-INFORMATION STRUCTURE INTERFACE
unmarked option for focus is the opposite of the unmarked option for wh phrases in English and Romance languages. These are just tentative comments which will have to be tested against more data from other languages. It would be necessary to check first if wh-movement and sentence-final focus go together crosslinguistically, and second, if wh-in situ and FP are used to mark the disambiguated narrow option in the languages that do have those two features. In this chapter I have explored different analyses of Focus Preposing. First, I examined different possibilities to account for this construction taking the preposed element to be both focal and topical. These included a Reconstruction analysis, a two-dichotomy approach and the existence of a metalinguistic focus. Second, I suggested that this element does not have to be considered topical, and I proposed that Focus Preposing can be viewed as a Focus-Background structure which marks unambiguous narrow focus on the preposed element. From this perspective, I looked at some similarities between FP and focused subjects and I considered some language-particular features of FP. Finally, I tentatively suggested that there may be a general correlation between narrowness and marked constructions which is shown by both FP and wh in situ, as opposed to their unmarked counterparts: focus in situ and wh-movement. NOTES 1. The material in this chapter was presented at the Workshop on Focus held at the University of Massachusetts in 1995 and a slightly different version appeared in the proceedings of that workshop (Casielles 1998). 2. Ward also offers these examples to show that when there is no salient relevant scale, the FP is infelicitous.
(i)
(ii)
I was at this restaurant and # COFFEE I ordered (Cf. I was at this restaurant and I ordered coffee) I went grocery shopping. # CROWDED it was (Cf. It was crowded) (Ward 1988, examples (209) and (214))
3. He gives the following example:
(i)
My FRIEND was nominated
4. One could think that these mechanisms are precisely chosen to disambiguate their ambiguous counterparts. However, as Barbara Partee
FOCUS PREPOSING 191
(p.c.) points out, this idea would pose some questions. First, we would have to think about when one needs to disambiguate, if at all, in general. Second, in the particular case of FP, such as those in (2)-(7), it looks like in the contexts in which this construction is felicitous, there doesn’t seem to be any need for disambiguation, since the same context that makes this construction felicitous already disambiguates the potentially ambiguous focus.
192 SYNTAX-INFORMATION STRUCTURE INTERFACE
CHAPTER 7 Information Structure and Syntactic Structure
7.1 Introduction In previous chapters I have considered separately the notions of topic and focus and I have suggested keeping a unique notion of focus and distinguishing between different types of topical elements, especially between Sentence Topic and Background. I also examined different topicalizing and focusing mechanisms such as Clitic Left-dislocation, Left-dislocation, Topicalization and Focus Preposing. However, I have not said anything about the combination of these three primitives in an articulation. As mentioned in chapter 2, several different dichotomies (Theme-Rheme, Topic-Comment, Topic-Focus and Focus-Background/ Presupposition/Open proposition) have been suggested in the literature.1 None of these dichotomies, however, distinguishes between two different topical elements. The only articulation that, to my knowledge, distinguishes between two different topical elements is Vallduví’s hierarchical trichotomy. Vallduví points out that none of the dichotomies mentioned above can account for all the informational splits of sentences and suggests that what is needed is an articulation which maintains the focus-background distinction and at the same time separates a sentence-initial topic-like expression, My proposal regarding a Topic-Focus articulation is going to be based on Vallduví’s theory and will keep his idea of distinguishing between a sentence-initial topic and the Background. However, it will differ from Vallduví’s in important ways. I will suggest that although we should distinguish between these two topical elements, they do not form part of the same dichotomy. I will argue that sentences can have either a Topic-Focus structure or a Focus-Background structure
194 SYNTAX-INFORMATION STRUCTURE INTERFACE
and that no trichotomy is needed. Furthermore, I will show that in Spanish these two articulations have their own syntax. This chapter is organized as follows. In 7.2, I concentrate on the TFA. I offer an overview of previous proposals involving one or two dichotomies, I consider Vallduví’s theory and I argue for having two different dichotomies. In 7.3, I point out the way these two articulations are syntactically expressed in Spanish, as opposed to English, and I relate this to the analysis of BNs offered in chapter 4. Finally, in section 7.4 I offer some conclusions. 7.2 The Topic-Focus Articulation As the existence of so many different dichotomies proves, it has not been easy to decide how to divide a sentence into informational units. Some dichotomies emphasize the split between a sentence-initial topic and the rest of the sentence while others emphasize the split between the focus and the rest. In addition, as we will see, the use of a dichotomy to express the informational articulation of sentences has not been perceived as a totally satisfactory type of articulation, and other types have been proposed. Let us overview the main proposals with some examples. 7.2.1 Previous Proposals Two well-known dichotomies separate the topic from the rest of the sentence: Theme-Rheme (Firbas 1964, Halliday 1967, Contreras 1976) and Topic-Comment (Gundel 1988, Reinhart 1981). Although there are differences regarding each author’s definition of theme or topic, they all share the fact that the split is made between this topic element and the rest of the sentence. For instance, in the provided context, “John” would be considered the theme or topic and “teaches Spanish” the rheme or comment. (1)
What does John do? John teaches Spanish
However, other also well-known dichotomies separate the focus from the rest of the sentence, which is referred to as the Topic, Presupposition, Open proposition or Background: Topic-Focus
INFORMATION STRUCTURE AND SYNTACTIC STRUCTURE 195
(Sgall et al. 1986, Erteschik-Shir 1997) and Focus-Presupposition/Open proposition/Background (Chomsky 1971, Jackendoff 1972, Rochemont 1986, Prince 1981b, 1986, Selkirk 1984, Ward 1988). In this case, the split is made between the focus and the rest of the sentence. The relevance of this division can be seen in a example such as (2) where, in the context provided, “beer” is the focus and the rest (“John drinks”) is the Background or nonfocal elements. (2)
John drinks beer What does John drink?
Further, it has been suggested that it is not clear that there is only one split in the informational articulation of sentences. Sentences seem to have other things that are neither topics nor foci.2 As already mentioned in chapter 2, Dahl (1974a) proposes that an example such as that in (2) can be analyzed with two superimposed articulations. That is, he suggests using two distinct articulations—Topic-Comment and Focus-Background- which allows us to separate both the topic and the focus from the rest of the sentence, and have, therefore, two different splits. From this perspective, the example in (2) would be analyzed as in (3), where, on the one hand, “John drink” is the Background and “beer” the focus, and, on the other hand, “John” is the topic and “drinks beer” the comment.3 (3)
What does John drink?
background|focus John drinks beer topic|comment
In addition, it has been suggested that there are cases in which a topic, or part of a topical element is focused. As mentioned in chapter 6, this has also been accounted for by two distinct articulations. In this case, instead of having two superimposed articulations, as in (3) above, we can allow the Focus-Background articulation to occur both inside the Topic and inside the Comment, as Krifka (1991) suggests, following Jacobs (1984), offering the following example: (4)
What did Bill’s sisters do? [Bill’s [youngest]F sister]T [kissed John]C
Krifka (1991)
However, as we have suggested in previous chapters, the fact that some topics get an accent in English is not sufficient evidence that they are
196 SYNTAX-INFORMATION STRUCTURE INTERFACE
focused, since, as we have seen, accent and focus do not always coincide in this language. As Lambrecht (1994) has pointed out, accent in English can mark a topic relation or a focus relation. In addition, if we considered cases of accented topics to be cases of focused topics, this would be an English-particular notion of focus, which would not be valid for other languages where new topics are not marked with an accent. If, on the other hand, we restrict the notion of focus, as I have suggested in chapter 5, to Gundel’s Semantic Focus, in both English and Spanish, the focus in (4) would be “kissed John” and “Bill’s youngest sister” would be considered the topic. In English, since it is slightly different from the discourse-old “Bill’s sisters,” it gets phonologically marked as a new topic through a B accent. In Spanish, since there are implicit topics which can be used when the topic is discourse-old, a discourse-new topic does not need any special marking. An explicit topic in the preverbal subject position contrasts with an implicit discourse-old topic, and, constitutes, therefore, enough marking of its discourse-new status. No accent is necessary. Thus, putting aside this last possibility of allowing focused topics, which we have rejected, we still have several proposals about how topic and focus go together. On the one hand, it looks like we cannot do without separating a topic from the rest of the sentence, and, on the other hand, focus also needs to be separated. From this perspective, Dahl’s idea of doing both looks appealing. However, there is a certain redundancy in the fact that in (3), the verb is both Comment and Background and the DO is both focus and Comment. Maybe we do not need so many primitives. Vallduví notes this and suggests that we can separate both the topic and the focus, and avoid the redundancies in (3) by using a hierarchical trichotomy. That is, first we separate the focus from the rest and then, inside the unfocused part, we separate the topic (link, in his theory) from the rest. The complement of the focus, he calls ground, and the complement of the topic (inside the ground) he calls tail: (5)
S={Focus, Ground} Ground={Link, Tail}
Thus he has, on the one hand, the Focus-Ground separation, and inside the Ground, he can separate the sentence-initial topic (link) from other nontopic nonfocused material such as the verb (the tail). The following example shows the three primitives:
INFORMATION STRUCTURE AND SYNTACTIC STRUCTURE 197
(6)
ground focus L’amo l’ODIA, el link
ground bròquil tail
“The boss HATES broccoli”
Thus, Vallduví’s trichotomy distinguishes between topic or link and Background or ground. Since this is what we argued was needed in order to account for the distribution of BNs in Spanish and the differences between preverbal subjects and dislocated elements, let us see if his proposal can account for the Spanish data. We have already introduced his system in 5.3.4, where we considered in particular the notion of focus. Here I will concentrate on the other primitives: ground, link and tail. 7.2.2 Vallduví’s Link-Tail Division One of the goals of Vallduví’s trichotomy is to achieve two splits in the information articulation of sentences, to account for the fact that sentences seem to contain things that are neither topics nor foci. As he points out, he wants to keep both the Focus-Background split and the Topic-rest of the sentence split. In order to do this, and given the correlation between topic and the sentence-initial position, which we considered in chapter 2, and can be seen, for instance, in the topical nature of preverbal subjects and left-dislocated phrases, he restricts topics to the sentence-initial position and considers any other type of unfocused material to be a tail.4 I think Vallduví’s distinction between a sentence-initial topic (his link) and nonfocal material (his tail) is correct. In fact, I have already adopted it in chapter 4, where I argued to distinguish between Topic and Background. However, these notions are not equivalent to Vallduví’s link and tail. In 4.3, I suggested restricting the Sentence topic (STopic) to the preverbal subject position and Background to dislocated phrases. In Vallduví’s system both preverbal subjects (7) and left-dislocated phrases (8) are links, and right-dislocated phrases (9) are tails.5
(7)
L’amo ha TRUCAT “The boss CALLED”
link (preverbal subject)+focus
198 SYNTAX-INFORMATION STRUCTURE INTERFACE
(8) L’amo no l’EMPRENYARIA “The boss BOTHER”
I
link (left-dislocated phrase) +focus
wouldn’t
(9) L’amo I’ ODIA, el bròquil
link+focus+tail dislocated phrase)
(right-
“The boss HATES broccoli” While links can be identified by their sentence-initial position, this is not the case for tails. Vallduví points out that while both link and focus have some universal structural characteristics (sentence-initialness and intonational prominence respectively), there is no such correlate in the case of the tail, although it is true that it is never marked with prosodic prominence and structural properties of each individual language (basic word order, basic intonation contour, verb-secondness, etc.) may determine the position in which tails end up surfacing in the sentence. The tail, therefore, must be negatively identified as the nonfocal nonlink part of the sentence, although within each language particular informational/structural correlates should be found. (Vallduví 1990:61) The Catalan sentence in (9) and the English equivalent Vallduví gives shows that while tails appear right-dislocated in Catalan, de-accenting can mark tails in English. This possibility is not available in Catalan. As we saw in chapter 6, unfocused material cannot occur to the right of focused material unless it is syntactically dislocated. In addition, in Catalan we can have multiple links (10) and multiple tails (11): (10)
El bròquil l’amo l’ ODIA “The boss HATES broccoli”
link+link+focus
(11)
l’ODIA, l’amo el bròquil “The boss HATES broccoli”
focus+tail+tail
Thus, in this system, left-dislocated phrases pattern with preverbal subjects (they are both links) while right-dislocated phrases are instances of a different primitive (they are tails). I do not think this division can account for the Spanish data. We have seen that in Spanish left-dislocated phrases pattern with right-dislocated phrases rather than with preverbal subjects. Let us review the differences between preverbal subject and dislocated phrases.
INFORMATION STRUCTURE AND SYNTACTIC STRUCTURE 199
First, while we can have only one preverbal subject, we can have several dislocated phrases, as in (12): (12)
El niño a mí tonto no me lo parece The child to me stupid doesn’t seem
Second, as the contrast between (13) and (14) shows, preverbal subjects are restricted to DPs. (13)
(14)
Los niños jugaban en el parque “The children were playing in the park” *Niños jugaban en el parque “Children were playing in the park”
However, we can have any type of dislocated phrase, as in (15)-(18): (15)
Tonto no lo parece Stupid he doesn’t seem
(16)
Estudiando nunca está Studying he never is
(17)
Que fumas lo sabemos todos That you smoke we all know
(18)
Niños no creo que vengan Children I don’t think will come
Third, topical subjects occupy an specific preverbal position. As the contrast between (19) and (20) show, postverbal subjects cannot be topics. (19)
¿Dónde está Juan? “Where is John?”
Juan se ha ido “John has left”
#Se ha ido Juan has left John
Dislocated phrases, on the other hand, can appear either to the left or to the right: (20)
Las fresas las comí yo the strawberries cl ate I
Left-dislocation
200 SYNTAX-INFORMATION STRUCTURE INTERFACE
(21)
Las comí yo, las fresas cl ate I, the strawberries
Right-dislocation
Therefore, even though both preverbal subjects and dislocated phrases are interpreted as topical in Spanish, they differ as to the number, type and position of the topical element. We could summarize the differences between these two elements as follows: position only in a preverbal spec position Disl. Phrase: any number any phrase in right-or left-adjoined positions
(22) Prev. Subj.:
number only 1
type only a DP
From this point of view, I would like to suggest that in the division between two topical phrases, left-dislocated phrases pattern with right-dislocated phrases rather than with preverbal subjects. Therefore, on the one hand, we have the preverbal subject and on the other hand, left and right dislocations. Thus, instead of Vallduví’s division in (23), I propose the one in (24): (23)
(24)
Unfocused elements Link Preverbal subject Left-dislocated phrases Unfocused Elements Sentence Topic Preverbal subject
Tail Right-dislocated phrases (Other)
Background Left-& right-dislocated phrases
Thus, I have considered Vallduví’s primitives, link and tail, and I have argued that a different division of topical elements, that between STopic and Background is more appropriate, if we take into account the differences between preverbal subjects and dislocated phrases in Spanish. Now, how do STopic, Background and Focus form an articulation? I am going to argue that they do not.
INFORMATION STRUCTURE AND SYNTACTIC STRUCTURE 201
7.2.3 Two Dichotomies In Vallduví’s system, the two topical elements, link and tail, form a bigger part, the ground. That is, the unfocused part is divided into two elements, which can co-occur in a sentence. However, we never encounter the order expected by the structure in (5): Focus, link and tail, and, in fact, only syntactically and/or phonologically marked sentences have a tail. The three primitives (link-focus-tail) show up in Catalan right-dislocations and English nonrightmost, narrow focus, as in (25), while two primitives (focus-tail) form Focus movement structures in both languages, asin (26). (25)
L’amo l’ODIA, el bròquil “The boss HATES broccoli”
(26) No m’ho hagués cregut mai! L’amo està ben boig! BROQUIL, vol ara “I can’t believe this. The boss is going crazy! BROCCOLI, he wants now” We would like to know why it is that the trichotomy correlates with these syntactically and/or phonologically marked constructions and what the possible orders between them are. That is, do we ever get link-tail-focus, for instance? I think that we may be able to simplify the articulation in (5), give a unified analysis of right-, left-dislocations and Focus Preposing and get a less mysterious correlation between the information articulation and the syntax, if we just distinguish between two different articulations. As mentioned in chapter 5, tailful structures always involve cases of narrow focus, either syntactically marked as in the case of the Catalan right-dislocation in (25) or phonologically marked (through nonrightmost focus) as in the English translation of (25) or both, as in the Focus preposing construction in both languages (26). Syntactically and phonologically unmarked cases such as (7) above do not involve tails. I am going to suggest that, from an information articulation point of view, there are two types of sentences: STopic-Focus and Background-Focus. That is, there are three primitives, but they do not form a trichotomy. Every sentence has a focus and the complement of this focus (if explicit at all) is either a STopic or a Background.
202 SYNTAX-INFORMATION STRUCTURE INTERFACE
The former involve what Lambrecht refers to as Topic-comment sentences and the latter involve his Identificational sentences.6 From this point of view, and given our definition f STopic and Background, all instances of dislocated structures involve BackgroundFocus structures. This includes left-and right-dislocations, left- and right-multiple dislocations and focus preposing structures.7 Therefore, all the following constructions, which in Vallduví’s theory involve different primitives will be given a unified structure: Focus-Background Structures (27) L’amo no l’EMPRENYARIA link (left-dislocated phrase) “The boss I wouldn’t +focus BOTHER” (28) L’amo l’ ODIA, el bròquil ‘The boss HATES broccoli”
link+focus+tail dislocated phrase)
(29) El bròquil l’amo l’ ODIA “The boss HATES broccoli”
link+link+focus
(30) l’ODIA, l’amo el bròquil “The boss HATES broccoli”
focus+tail+tail
(31) BROQUIL, vol ara “BROCCOLI he wants now”
focus+tail
(right-
On the other hand, sentences involving a preverbal subject, will be instances of the STopic-Focus structure STopic-Focus structure (32)
L’amo ha TRUCAT “The boss CALLED”
link (preverbal subject)+focus
If this is right, the apparently similar sentences in (27) and (32), both link-focus structures in Vallduví’s system have a different information structure. The former is a Background-Focus and the latter is a STopic-Focus structure. The difference shows if we substitute a BN: (33)
Tomates no pondría en la ensalada “Tomatoes I wouldn’t put in the salad”
INFORMATION STRUCTURE AND SYNTACTIC STRUCTURE 203
(34)
*Tomates son caros “Tomatoes are expensive”
Interestingly, the analysis of all dislocated structures as Background-Focus structures would mean that in a sentence such as (28) the subject is also dislocated. That is, since this sentence has a right-dislocated element, it is an instance of a Background-Focus structure, which cannot have an STopic. Recall that in this theory STopic and Background do not co-occur, since it would not make sense to mark a particular element as the topic when the whole sentence excluding the focal element is being marked as topical. Thus, we would have to assume that the subject is also dislocated in (28). If this is right, the apparently similar subjects in (28) and (32) are different. I think this might be right. Although the contrast here may be more subtle, I think there is still a difference in grammaticality between (35) and (36): (35) (¿Cuánto uso tiene este ordenador entre estudiantes y profesores?) “How much use does this computer have among students and professors?” Estudiantes no lo usan mucho, este ordenador students don’t use it much, this computer (36) (¿Qué es lo que no usan *Estudiantes no usan este estudiantes?) ordenador “What is it that students don’t “Students don’t use this use?” computer” Furthermore, the English equivalent of structures such as (28) and (32) also shows the difference between the two structures. Thus, while (37) can be an answer to a question such as “What did the boss do?,” the sentence in (38) cannot. (37)
The boss CALLED
(38)
The boss CALLED his wife
That is, sentences such as (38) are only appropriate in a context in which everything except the focus is Background. In this case, the
204 SYNTAX-INFORMATION STRUCTURE INTERFACE
preverbal subject is not a STopic, neither in English nor in Catalan or Spanish. The Background-Focus structure of these sentences is marked phonologically in English (through nonrightmost focus, which automatically marks the rest of the sentence, both the subject and the object as Background) and syntactically in Catalan and Spanish. This can be done by a right-dislocation as in (39), but also, by left-dislocating both Background elements or right-dislocating both, as in (40) and (41) respectively. (39) El jefe la LLAMÓ, a su mujer (40) El jefe a su mujer la LLAMÓ/A su mujer el jefe la LLAMÓ (41) La LLAMÓ, el jefe a su mujer/La LLAMÓ, a su mujer el jefe That is, from this point of view, all these structures are instances of a Background-Focus articulation, where the focus is narrow and everything else is Background. This includes the preverbal subject in (39). Thus, sentences can have either an STopic-Focus structure or a Background-Focus structure. In English, this division coincides with a rightmost unmarked focus and a nonrightmost phonologically marked focus respectively. In Spanish and Catalan, which maintain rightmost unmarked focus, the difference is syntactically marked. But before I consider in more detail the correlation between the two information structures and the syntax, let us consider verbs. One of the reasons behind Dahl’s proposal of having two superimposed articulations considered above, was to account for discourse-old verbs such as the one in (42): (42)
What does John drink? John drinks beer
This is a potential problem, Since we cannot have both STopic and Background, we would have to say either that “John drinks” is Background or that “drinks beer” is Focus. Now, we suggested that Background-Focus structures are marked either phonologically (by nonrightmost focus) or syntactically (by a dislocation). So, from this point of view, the sentence in (42) cannot be a Background-Focus. Syntactically and phonologically, it looks like an STopic-Focus structure.
INFORMATION STRUCTURE AND SYNTACTIC STRUCTURE 205
We have two options: we can either say that “drink beer” is the focus or that the STopic-Focus structure does not have to classify ALL elements in a sentence. If we take the first option and suggest that “drink beer” is the focus we have to explain why “drink” is part of the focus, considering that it is not informative. Partly, it may be due to the fact that it is a conjugated verb. That is, maybe conjugated verbs can be part of the focus even if uninformative. As we are going to see below, verbs, and more specifically, conjugated verbs differ from other phrases in many ways. However, this cannot be the whole story, at least for English. As (43) shows, phrases other than verbs, can stay in the focus domain even if they are not informative: (43)
To whom did Mark give the books? Mark gave the books to the CHILDREN
That is, even though unmarked right-most focus can project up to the whole VP (the sentence in (43) is also possible as an answer to “What did Mark do?”), in English, it does not have to project. Therefore, in that context, the focus is reduced to “to the children.” Both the verb “gave” and the DO “the books” are uninformative but allowed to stay in the focus domain. Maybe we could say that this is an STopic-Focus structure where “Mark” is the STopic and “to the children” the only focused phrase in the focus domain, and we do not need to say anything else about “give the books.” In any case, verbs are still special, particularly in languages like Catalan and Spanish where unmarked right-most focus obligatorily projects and uninformative phrases such as the DO in (43) appear in a dislocated position to escape the focus feature. In these languages, conjugated verbs do not have to escape. That is, the counterpart of (42) is allowed: (44)
¿Qué bebe Juan? Juan bebe cerveza
An obvious difference between conjugated verbs and DOs is that verbs do not have a pronominal form. That is, we can replace “the children” with a pronoun in both English and Spanish, but we cannot replace the verb:
206 SYNTAX-INFORMATION STRUCTURE INTERFACE
(45) (46)
Mark gave them to the children What does John drink? *John does beer
Sometimes a verb can be marked as uninformative in Spanish, by repeating the verb in an infinitival form in a dislocated position. However, the conjugated form stays in the focus domain: (47)
Estudiar, no estudia nada to-study s/he doesn’t study much
Thus, conjugated verbs do have different characteristics, which may have a bearing on their being fine as uninformative elements inside the focus domain. As the projection of focus shows (Selkirk 1984), conjugated verbs do not need an accent even if they are part of the focus. That is, the verb in (48) does not need an accent to be part of the focus: (48)
What did John do? John cooked LASAGNA
However, this is also the case for the DO in (49): (49)
What did Mark do? Mark gave the books to the CHILDREN
That is, the whole VP is part of the focus in this context and neither the verb nor the DO need an accent. In any case, I think it is very plausible that the informative articulation treats conjugated verbs differently from other phrases. Lambrecht (1994:264) suggests that the discourse status of predicating expressions has a fundamentally different effect on focus prosody than that of referential expressions. In addition to the fact pointed out above that informative verb forms do not need an accent, he also points out that predicates, unlike nominal arguments, are not necessarily unaccented on second mention. He offers the following examples: (50)
a. He promised to go SHOPPING but he forgot to GO b. #He promised to go SHOPPING but he FORGOT to go
INFORMATION STRUCTURE AND SYNTACTIC STRUCTURE 207
In (50)a, the verb “to go” receives the focus accent even though it is anaphorically related to the VP “go shopping.” It would not be appropriate to accent “forgot” and deaccent “go” even though the former is more informative. Lambrecht suggests that “go” in (50) a receives the accent due to its final position. He points out that this behavior contrasts with that of nominal expressions and offers the examples in (51): (51) a. He promised to buy FOOD but he forgot to GET the stuff b. #He promised to buy FOOD but he forgot to get the STUFF In this case, the noun “stuff” which is anaphorically linked to “food” is necessarily unaccented. If accented, as in (51)b it would refer to something other than “food.” Lambrecht suggests that the status of verbs, from the point of view of the expression of information-structure distinctions, is in some sense less important than that of nouns. While lack of prosodic prominence on a referential expression is necessarily an indication of activeness or at least high accessibility of the referent, lack of prominence on a predicating expression does not necessarily have a similar implication. In addition, an informative verb must receive an accent only if it is located in a syntactic position which attracts an accent for independent reasons. Thus, he concludes that prosodic prominence on predicating expressions is generally assigned by default rather than by iconic motivation. I leave this open for further research.8 Thus, in this section I have suggested distinguishing between two different articulations: an STopic-Focus articulation and a Background-Focus articulation. The former is the unmarked one, syntactically, phonologically and pragmatically in both languages, Phonologically, it involves Cinque’s (1993) unmarked rightmost focus. Syntactically, it involves a preverbal subject and a predicate, and pragmatically it is felicitous as an out-of-the-blue utterance, both in English (52) and Spanish (53): (52)
What happened? What did John do? Who did John call?
John called his wife
(53)
¿Qué pasó? ¿Qué hizo Juan? ¿A quién llamó Juan?
Juan llamó a su mujer
208 SYNTAX-INFORMATION STRUCTURE INTERFACE
That is, the STopic-Focus articulation does not have any discourse requirements. It can start a discourse, and it is expressed by the same type of syntactic and phonological structures in both languages: a preverbal subject-predicate structure with sentence-final focus-related accent. Background-Focus structures, on the other hand, are pragmatically, phonologically and/or syntactically marked. Pragmatically, these structures are marked since they are only appropriate in a very restricted type of discourse, where everything except the focus, which is necessarily narrow, has to be discourse-old. Therefore, Background-Focus structures cannot start a discourse: they are not appropriate answers to a “What happened?” question: (54)
What happened? #John CALLED his wife
In English, these structures are phonologically marked with nonrightmost accent: (55)
Did John e-mail his wife? John CALLED his wife
In Spanish, they would be syntactically marked by the right-or left-dislocation of the Background elements: (56) a. A su mujer Juan la LLAMÓ/Juan a su mujer la LLAMÓ b. Juan la LLAMÓ, a su mujer c. La LLAMÓ, Juan , a su mujer/La LLAMÓ, a su mujer, Juan I would like to end this section with a brief comment about cases where no obvious Topic appears. Consider (57)–(58): (57)
What happened? JOHN called
(58)
¿Qué pasó? Llamó JUAN
It looks like in this case the event is not viewed as involving a topic and this is marked by accenting the preverbal subject, as in English, or having a postverbal subject, as in Spanish. That is, since an unaccented preverbal subject is interpreted as the STopic, if the speaker does not view the event as involving a topic, the subject needs some marking
INFORMATION STRUCTURE AND SYNTACTIC STRUCTURE 209
(either syntactic or phonological). Now, what is the information structure of these sentences. There are several possibilities that I would like to point out. One would be to have a third type of information structure, as Lambrecht does. As we saw in chapter 5 (section 5.3.5), in addition to Topic-Comment and Identificational sentences (our STopic-Focus and Background-Focus respectively), he also has Event-reporting sentences where the focus comprises the whole sentence. However, we may not need a third type of structure. Another possibility would be to consider these cases STopic-Focus structures, where the Topic is not overt. In fact, it is not clear that these structures involve no topic at all. Erteschik-Shir (1997), for instance, argues for the existence of what she calls a stage-topic, which refers to the here and now, to the Time/Place at which the event takes place (which she says corresponds to Kratzer’s spatio-temporal argument). This stage-topic can be implicit, as in (59).9 (59)
A man arrived
So, it may be that the structures in (57)–(58) do involve a topic of this kind and can be considered a special case of STopic-Comment structures.10 7.3 The TFA and Syntactic Structure In the previous section, I have proposed that there are two different information structures: an STopic-Focus and a Background-Focus structure. In addition, I have suggested that the two topical elements do not co-occur. That is, a sentence can have an STopic or a Background (if anything), but not both. Now, I would like to relate this to the connections suggested in chapter 4 and this chapter between these two elements and their syntactic positions. In chapter 4, I suggested that the two information primitives, STopic and Background, correlate in Spanish with two syntactic positions: a preverbal specifier subject position and left or right-adjoined positions respectively. Regarding the question of why these elements would be placed in those positions, I suggested that in the case of the STopic, we can assume that DPs check a topic feature in the preverbal specifier position. In the case of the Background, I argued that their position is not so much related with feature-checking but with avoiding a feature
210 SYNTAX-INFORMATION STRUCTURE INTERFACE
clash. From this perspective, uninformative elements are positioned outside the focus domain, the domain up to where the focus feature can project. If this is right, if STopics are restricted to the preverbal subject position and Background elements to adjoined positions, and if as suggested above, STopics and Background do not co-occur, it means that syntactically we can either have a preverbal specifier subject position or adjoined positions, but not both. I think this might be right. When we have dislocated elements, we do not seem to have a special subject position. The subject can appear intermingled in whatever order with the dislocated elements. (60)
A él el coche su madre no se lo deja to him the car his mother does not lend His mother doesn’t LEND him her car
(61)
Su madre a 61 el coche… El coche su madre a él… etc.
That is, it seems that if we start the sentence with an adjoined position, then the subject, if it does not stay inside the VP, is also in an adjoined position, and, therefore, can be a BN. The fact that subjects can appear dislocated intermingled with other dislocated phrases has been taken as evidence that in Spanish preverbal subjects do not occupy a specifier position, but an adjoined position (Contreras 1991). However, this does not take into account the contrast we have seen between grammatical dislocated BNs and ungrammatical (nondislocated) preverbal BN subjects. I think that what those examples show is that the subject CAN be occupying an adjoined position, not that it always does. More specifically, the subject has to occupy an adjoined position in cases like (60), because there are dislocated phrases. That is, it is a Background-Focus structure, and there is no specifier available. However this does not mean that subjects are always dislocated. In STopic-Focus structures, they are not. If this is right, the left periphery of the clause in Spanish can have two different incompatible structures:
INFORMATION STRUCTURE AND SYNTACTIC STRUCTURE 211
(62)
The structure in (62) is the syntactic structure of a Background-Focus information structure in Spanish, and the structure in (63) is the syntactic structure of an STopic-Focus information structure. (63)
That is, all the adjoined phrases in (62) are phrases that are considered uninformative, while the specifier position in (63) is occupied by a DP with a topic feature.11 Finally, I would like to mention two recent analyses, which also deal with the relation between syntactic structure and information structure: Rizzi (1997) and Zubizarreta (1998). Both works try to account for the correlation between certain syntactic positions and certain topical/focal interpretations. However, none of these works distinguish between topical vs. nonfocal or between nondislocated preverbal subjects and dislocated phrases. Rizzi (1997) notes that multiple left-dislocations are possible in Italian and proposes a variety of Topic phrases where topical elements move to satisfy his Topic-Criterion. I think this proposal faces some problems. First, it does not account for the differences we have seen here between preverbal subjects and dislocated phrases. Second, these cases of multiple left-dislocations do not behave like moved elements. These dislocated elements are not sensitive to island constraints and we
212 SYNTAX-INFORMATION STRUCTURE INTERFACE
can dislocate as many as we want and in any other.12 This is not how moved elements behave. As we know, movement processes have very specific requirements as to the ordering and number of moved elements. Thus, I think that these dislocated elements are better analyzed as base-generated adjunctions.13 Zubizarreta (1998) also considers the relation between syntactic and information structures in Spanish and argues that movement outside of VP may be due to phono-syntactic reasons. She also adopts Cinque’s theory and relates the position of dislocated phrases to the projection of focus and with the preference in Romance to maintain focus-related sentence-final accent. Thus, she suggests that defocalized constituents undergo a prosodically motivated movement (p-movement). As I mentioned above, I think there is evidence that shows that these are not moved elements. In addition, Zubizarreta (1998) also seems to equate topical with nonfocal. That is, although she relates the position of dislocated phrases to the fact that this is a position outside the domain of focus projection, she considers these phrases to be topics, and similarly to Rizzi (1997) argues for the existence of topic features that get checked in T(ense)P. She proposes that in Spanish TP is a syncretic category where features such as “topic,” “focus,” or “emphasis” can combine with the feature Tense. To account for what she considers to be multiple topics, she says: “To the extent that there may be more than one topic per sentence, it is reasonable to assume that there may be more than one ‘topic’ feature that participates in the feature checking algorithm. Thus, besides the ‘topic’ feature on T, there may be a ‘topic’ feature on a functional category above TP.” (Zubizarreta 1998:102) As we have seen above, dislocated elements behave more like Background elements rather than sentence topics. This means that if we want to relate the position of these elements with a feature, a [-focus] feature seems more appropriate than multiple topic features. This, in turn would explain the difference between nondislocated preverbal subjects and dislocated elements that we have seen here, which neither Zubizarreta (1998) nor Rizzi (1997) consider. While the position of dislocated elements is related with the theory of focus projection and the fact that these are nonfocal (or Background) elements, the position of preverbal subjects can be related with a Topic-Criterion a la Rizzi (1997), since in Spanish only topical subjects raise to the preverbal subject position. From this perspective, if IP-external elements get interpreted as nonfocal, we do not need multiple topic features or multiple topic phrases. In Spanish, an element in the preverbal subject
INFORMATION STRUCTURE AND SYNTACTIC STRUCTURE 213
position is interpreted as the sentence topic while IP-external elements are interpreted as Background elements. 7.4 Overall Conclusions The fact that some languages use different syntactic structures to express a different packaging of information has long been known and numerous studies have been devoted to investigating the appropriate articulation of this packaging, the informational primitives involved, and the syntactic correlations of different information structures in different languages. The notions of topic and focus (which are related to the classical Aristotelian division between subject and predicate) are not hard to grasp intuitively and they have been quite useful in pragmatic studies. However, it has proved very hard to identify these elements in a particular language, which is probably one of the reasons many syntactic works have denied a place for these notions in the grammar and have ignored the topical nature of dislocated phrases and preverbal subjects or the focal nature of postverbal subjects in languages like Spanish. From this point of view, different information structures have been often considered mere stylistic variations, outside the domain of syntactic analysis. On the other hand, many studies on the Topic-Focus articulation are based on particular conceptions of topic and focus, which are assumed to be valid crosslinguistically, and which do not take into account the syntactic characteristics of the language involved. Although all languages can be expected to express informational differences, we do not have to expect a universal notion of topic. Languages will probably vary in the way they express information structure and the role that these concepts play in their grammar will depend in part on their syntactic and phonological features. That is, the syntactic and/or phonological mechanisms available in a language will have a bearing on how information structure is reflected in that language. For instance, the correlation between topics and preverbal subjects is going to be different in a language such as English where all subjects have to be explicit and preverbal, regardless of their information status, than this correlation in a language such as Spanish, where subjects can be explicit or implicit and preverbal or postverbal, depending on their information status. This study has been an effort to bring together insights from both the literature on the TFA and current syntactic analysis in generative
214 SYNTAX-INFORMATION STRUCTURE INTERFACE
grammar, with a twofold purpose. One of the goals has been to contribute to a better understanding of information notions such as topic and focus, and their combination to form particular articulations. Another goal has been to account for the correlations that can be seen in Spanish between certain syntactic positions and certain informational interpretations (such as the contrast between preverbal vs. postverbal subjects, the analysis of dislocated phrases and focus-preposed elements, and the distribution of Bare Nominals). Thus, I have investigated the syntactic mechanisms available in Spanish to express informational differences. This investigation has shown that Spanish syntax differentiates between two types of “topical” elements. I have suggested that these elements correspond to two different primitives which have been independently postulated in the literature on information structure: Sentence Topic (STopic) and Background. In Spanish, while the sentence topic occupies a particular syntactic position (the preverbal subject position), Background elements appear either right-or left-dislocated. This has important consequences for the TFA, in general, and for the syntactic analysis of Spanish, in particular. Regarding the TFA in general, I have proposed that these two elements do not co-occur, and that, therefore, there are at least two different information structures: STopic-Focus and Background-Focus. In the STopic-Focus articulation (which is equivalent to the familiar Topic-Comment structure) a division is made between a unique discourse referent (chosen as the point of departure for the rest of the sentence) and the rest (which predicates some irrecoverable information about this element). This articulation can start a discourse. It does not have any discourse prerequisites and it is considered here unmarked. In the Background-Focus structure, the division is between a unique focus (the only irrecoverable and informative part of the sentence) and an open proposition (which is necessarily discourse-old or accommodated as such). This articulation has very specific discourse requirements. It is only felicitous in a very restricted context and it is considered marked here. From this perspective, three different primitives are distinguished: two topical elements (STopic and Background) and a Focus. However, this is different from other proposals which also distinguish these three primitives such as Vallduví’s hierarchical trichotomy in that focus cannot combine with both topical elements at the same time. Now, languages may vary in the way they express these two articulations. I have suggested that the pragmatically unmarked
INFORMATION STRUCTURE AND SYNTACTIC STRUCTURE 215
Topic-Focus articulation is also syntactically and phonologically unmarked both in English and Spanish. Syntactically, it involves a preverbal subject in a specifier position and phonologically sentence-final focus-related unmarked accent. The pragmatically marked Background-Focus articulation, on the other hand, is expressed differently in these two languages. In Spanish, it is syntactically marked: Background elements are syntactically detached through left-or right-dislocations. However, in English it is phonologically marked through non sentence-final, marked accent. It has been suggested that this different marking has to do with the way these two languages block focus projection, so that the focus is restricted to a particular element as the Background-Focus dichotomy requires. There are two ways of preventing focus from projecting. First, since only sentence-final focus can project, focus projection can be blocked by placing the focus-related accent in a non sentence-final position, as in English. Second, sentencefinal focus-related unmarked accent can be maintained by removing everything from the domain of focus projection, and placing it outside this domain in dislocated positions, as in Spanish. Furthermore, these insights from the TFA have helped us explain certain syntactic features of Spanish and recalcitrant data such as the distribution of BNs. Regarding the contrast between ungrammatical preverbal BN subjects and grammatical postverbal BN subjects, I have suggested that BNs in Spanish cannot leave the VP (which explains the fact that they only get the existential interpretation) due to the fact that NP-movement in this language is a topic-driven movement. Since the (optional) topic feature is a D feature NP-movement is restricted to DPs in this language. Second, the contrast between grammatical dislocated BNs and ungrammatical preverbal subjects has been accounted for as follows. As opposed to preverbal subjects, dislocated phrases are not topics but Background elements, elements that are placed outside the domain of focus projection (the domain which the focus feature can reach). Therefore, the contrast preverbal subject vs. dislocated phrase is, from a TFA point of view, a Topic/Background contrast. From a syntactic point of view, I have shown that we can have any type and number of dislocated phrases (in any order) and offered a base-generation analysis of dislocated structures. From this point of view, it is a specifier/adjoined position contrast, Finally, it has been proposed that the case of grammatical preverbal focused BN subjects is related to cases of Focus Preposing, which can be considered to be instances of an operator movement, similar to wh-movement.
216 SYNTAX-INFORMATION STRUCTURE INTERFACE
Although there remain many unanswered and even unraised questions, this study has attempted to contribute to the investigation of the TFA and the TFA-syntax connections with special reference to Spanish and English. I have suggested that some of the differences between English and Spanish seem to lie in the fact that the topic and focus features are syntactically active in Spanish but only phonologically active in English. Obviously, the hypotheses presented here will have to be tested against further syntactic, phonological and informational data from other languages. As I have pointed out in 3.4, I believe that STopic and Background are just two clear instances of different types of topical elements. It is expected that a comprehensive crosslinguistic analysis of the characteristics of nonfocal elements will help us discover the relevant features that will allow us to establish a feature-based typology of topics. NOTES 1. A very detailed discussion of the differences and similarities among these dichotomies, and a review of their insights and shortcomings is available in Vallduví (1990:35–56). He also considers an approach which does not consist of a bipartite structure. It is the notion of Dominance (Erteschik-Shir 1973, 1979, 1986 and ErteschikShir and Lappin 1979, 1983). 2. It should be noticed that some dichotomies which make use of the concept of Communicative Dynamism and treat the topic as a continuum rather than a discrete category, such as Firbas’s, can probably account better for the existence of nontopic, nonfocus elements than dichotomies which have a rigid topic-focus division. 3. What is interesting about this approach is that it can express the fact that sometimes the verb belongs to the background part without having to say that it is a sentence topic. 4. Vallduví points out that although links are restricted to the sentence-initial position, not all sentence-initial elements are links. This way we do not run into the problems we saw in chapter 2 that Halliday’s definition of theme had. 5. As we saw in chapter 5, since both links and tails are optional, in addition to the link-focus and link-focus-tail in (7)–(9), we can also have all-focus and focus-tail structures, as in the examples (73) and (75) in section 5.3.4. 6. Lambrecht (1994) distinguishes a third type, the event-reporting or presentational sentence type, which are analyzed as all-focus structures. This type includes sentences such as (i) which show a postverbal subject
INFORMATION STRUCTURE AND SYNTACTIC STRUCTURE 217
in the languages that allow for them (ii) and existential sentences such as (iii):
(i)
What happened?
(ii)
¿Qué pasó?
(iii)
There was accident
My CAR broke down Se me estropeó el coche
an
7. In Casielles (2000a) this is argued specifically for the analysis of multiple left-dislocations. 8. Although we have assumed that uninformative verbs such as those in (42)–(44) are in the focus domain, this may not be the case if we take into account that in the syntax, the verb is assumed to raise out of the VP to a higher position at some level both in English and Spanish. However, it all depends on the position it reaches and on the focus domain. Although we have been assuming a VP-domain for focus, it may be bigger; up to ASpP, according to Zubizarreta (1998). 9. She points out that only stage-level predicates can have stage topics. 10. Another possibility would be to say that the sentences in (57)–(58), which do not seem to involve an STopic, are instances of the other structure, the Background-Focus structure. However, I do not think that is the case. As opposed to Background-Focus structures, these topicless structures are fine discourse-initially: they can answer a “What happened?” question. In addition, they are not syntactically marked as Background-Focus structures; they do not involve dislocations in Spanish. Further, for certain predicates (most unaccusatives), the structure in (57)–(58) is the unmarked one. That is, these verbs have either an accented or postverbal subject by default. 11. Reinhart (1995) has argued that scrambling in Dutch is a discourse-motivated movement, too. If my analysis of dislocated phrases as Background elements is on the right track, it would mean that CLLDs in Spanish (and other Romance languages) have a lot in common with scrambled phrases in Dutch, if, as proposed by Reinhart (1995), the motivation for scrambling in this language is to allow scrambled phrases to escape the focus domain. 12. See Casielles (2001) for details. 13 On the other hand, I think that something like Rizzi’s (1997) Topic-Criterion seems very adequate to account for subject raising in Spanish, since only DPs with a Topic feature raise to the preverbal subject position, the position reserved for sentence topics.
218 SYNTAX-INFORMATION STRUCTURE INTERFACE
References
Abney, S. (1987). The English Noun Phrase in its Sentential Aspect. Ph.D. Dissertation, MIT. Abraham, W. and S.de Meij (eds.) (1986). Topic, Focus, and Configurationality. Amsterdam/Philadelphia: John Benjamins. Adjémian, C. (1978). “Theme, Rheme and Word order. From Weil to Presentday Theories.” Historiographia Linguistica 5.3, 253–273. Allerton, D.J. (1978). “The Notion of ‘Givenness’ and its Relations to Presupposition and to Theme.” Lingua 44, 133–168 . Beneš, E. (1959). “The Beginning of the German Sentence from the Point of View of Functional Sentence Perspective.” Casopis pro moderni filologii 41, 205–217. ——(1968). “On Two Aspects of Functional Sentence Perspective.” Travaux Linguistiques de Prague 3, 267–274. Bolinger, D. (1954). “English Prosodic Stress and Spanish Sentence Order.” Hispania 37, 152–6 . ——(1965). Forms of English: Accent, Morpheme, Order. Cambridge: Harvard University Press. Bonet, E. (1990). “Subjects in Catalan.” MITWPL 13, 1–26. ——(1991). Morphology after Syntax: Pronominal Clitics in Romance. Ph.D. Dissertation, MIT. Borer, H. (1984). Parametric Syntax. Dordrecht: Foris. Bosque, I. (ed.) (1996). El sustantivo sin determinación. La ausencia de determinante en la lengua española. Madrid: Visor. Campos, H. (1986). “Indefinite Object Drop.” Linguistic Inquiry 17, 354–359. Carlson, G. (1977). References to kinds in English. Ph.D. Dissertation, University of Massachusetts, Amherst. Carlson, G. and F.Pelletier (1995). The Generic Book, Chicago: The University of Chicago Press. Casielles, E. (1996). “On the Misbehavior of Bare Nouns in Spanish.” In C. Parodi et al. (eds.) Aspects of Romance Linguistics: Selected Papers from the XXIV Linguistic Symposium on Romance Languages, Washington D.C.: Georgetown University Press, 135–148. ——(1998). “FOCUS PREPOSING (it is called).” University of Massachusetts Occasional Papers 21, 51–64. ——(1999). “Notes on the Topic-Focus Articulation.” In J.Gutiérrez-Rexach and F.Martínez Gil (eds.) Advances in Hispanic Linguistics. Somerville, Mass.: Cascadilla Press, 346–363.
220 REFERENCES
——(2000a). “On the Topicalizing Nature of Multiple Left-dislocations.” Berkeley Linguistic Society 26, 53–64. ——(2000b). “El tratamiento del orden de palabras en algunas gramàticas de los siglos XIX y XX.” Historiographia Linguistica 27.2/3, 415–436 . ——(2001). “The Syntax and Semantics of Preverbal Topical Phrases in Spanish.” In J.Gutiérrez-Rexach and L.Silva-Villar (eds.) Current Issues in Spanish Syntax and Semantics. Berlin/New York: Mouton DeGruyter, 65–82. ——(2003a). “On the Interaction between Syntactic and Information Structures.” Bulletin of Hispanic Studies 80, 1–20 . ——(2003b). “Left-dislocated Structures in Spanish.” Hispania 86.2, 326– 338. Chafe, W.L. (1974). “Language in Consciousness.” Language 50.1, 111–133. ——(1976). “Givenness, Contrastiveness, Definiteness, Subjects, Topics, and Point of View ”In C. Li (ed.), 25–55 . ——(1987). “Cognitive Constraints on Information low.” In R.Tomlin (ed.) Coherence and Grounding in Discourse. Amsterdam: John Benjamins, 21– 52. Choi, H-W. (1999). Optimazing Structure in Context: Scrambling and Information Structure. Stanford, Calif.: CSLI Publications. ——(2001). “Binding and Discourse Prominence: Reconstruction in ‘Focus’ Scrambling.” In G.Legendre, J.Grimshaw and S.Vikner (eds.) Optimality Theoretic Syntax. Cambridge, Mass.: MIT Press, 143–169. Chomsky, N. (1965). Aspects of the Theory of Syntax. Cambridge, Mass.: MIT Press. ——(1971). “Deep Structure, Surface Structure, and Semantic Interpretation.” In D.Steinberg and L.Jacobovits (eds.) Semantics. Cambridge: CUP, 183–216. ——(1977). “On Wh-movement” In P.Culicover, T.Wasow, and A. Akmajian (eds.) Formal Syntax, New York: Academic Press, 71–132. ——(1980). Rules and Representations. New York: Columbia University Press. ——(1982). Some Concepts and Consequences of the Theory of Government and Binding. Cambridge: MIT Press. ——(1995). The Minimalist Program. Cambridge: MIT Press. ——(2001). “Beyond Explanatory Adequacy.” MIT Occasional Papers in Linguistics 20, 1–28. Chomsky, N. and M.Halle, (1968). The Sound Pattern of English. New York: Harper and Row. Cinque, G. (1977). “The Movement Nature of Left-dislocation.” Linguistic Inquiry 8, 397–412. ——(1990). Types of A’dependencies. Cambridge: MIT Press. ——(1993). “A Null Theory of Phrase and Compound Stress.” Linguistic Inquiry 24.2, 239–297.
REFERENCES 221
Clark, H. (1977). “Bridging.” In P.N.Johnson-Laird and P.C.Wason (eds.) Thinking: Readings in Cognitive Science. Cambridge: Cambridge University Press, 411–420. Clements, J.C. (1994). “Notes on Topicalization and Object Drop in Spanish.” In M.L.Mazzola (ed.) Issues and Theory in Romance Linguistics. Selected Papers from the Linguistic Symposium on Romance Languages XXIII. Washington D.C.: Georgetown University Press, 219–237. Contreras, H. (1976). A Theory of Word Order with Special Reference to Spanish, Amsterdam: North-Holland. ——(1986). “Spanish Bare NPs and the ECP.” In I.Bordelois, H.Contreras and K.Zagona (eds.) Generative studies in Spanish syntax. Dordrecht: Foris, 25–49. ——(1991). “On the Position of Subjects.” Syntax and Semantics 25, 63–79. Costa, J. (2001). “The Emergence of Unmarked Word Order.” In G.Legendre, J.Grimshaw and S.Vikner (eds.) Optimality-Theoretic Syntax . Cambridge, Mass.: MIT Press, 171–203. Crabb, D.M. (1955). A Comparative Study of Word Order in Old Spanish and Old French Prose Works. Washington D.C.: The Catholic University of America Press. Culicover, P. and L.McNally (eds.) (1998). “The Limits of Syntax.” Syntax and Semantics 29. New York: Academic Press Culicover, P. and M.Rochemont (1983). “Stress and Focus in English.” Language 59, 123–165. Dahl, O. (1974a). “Topic-Comment Structure Revisited.” In O.Dahl (ed.), 1–24. ——(1974b) (ed.) Topic and Comment, Contextual Boundness and Focus. Hamburg: Helmut Buske. Daneš, F. (1964). “A Three Level Approach to Syntax.” Travaux Linguistiques de Prague 1, 225–240. ——(1974a). “Functional Sentence Perspective and the Organization of the Text.” In F.Daneš (ed.), 106–28. ——(ed.) (1974b). Papers on Functional Sentence Perspective. Prague: Academia. Daneš et al. (1974). “Zur Terminologie der FSP.” In F.Daneš (ed.), 217–222. Diesing, M. (1992). Indefinites . Cambridge: MIT Press. Dwivedi, V. (1994). Syntactic Dependencies and Relative Clauses in Hindi. Ph.D. Dissertation, University of Massachusetts, Amherst. Erteschik-Shir, N. (1973). On the Nature of Island Constraints. Ph.D. Dissertation, MIT. ——(1979). “Discourse Constraints on Dative Movements.” In T.Givón (ed.) Discourse and syntax. Syntax and Semantics 12. New York: Academic Press. ——(1986). “Wh questions and Focus.” Linguistics and Philosophy 9, 117– 49. ——(1997). The Dynamics of Focus Structure. Cambridge: CUP.
222 REFERENCES
Erteschik-Shir, N. and S.Lappin (1979). “Dominance and the Functional Explanation of Island Phenomena.” Theoretical Linguistics 6, 41–85. ——(1983). “Under Stress: A Functional Explanation of English Sentence Stress.” Journal of Linguistics 19, 419–53. Fernández-Soriano, O. (1993) (ed.). Los Pronombres Átonos. Madrid: Taurus. Fintel, K.von (1994). Restrictions on Quantifier Domains. Ph.D. Dissertation, University of Massachusetts, Amherst. Firbas, J. (1964). “On Defining the Theme in Functional Sentence Perspective.” Travaux linguistiques de Prague 1, 267–280. ——(1966). “Non-thematic Subjects in Contemporary English.” Travaux linguistiques de Prague 2, 239–256. ——(1971). “On the Concept of Communicative Dynamism in the Theory of Functional Sentence Perspective.” Brno Studies in English 7, 12–47. Fontana, J. (1993). Phrase Structure and the Syntax of Clitics in the History of Spanish. Ph.D. Dissertation, University of Pennsylvania, Philadelphia. Gabelentz, G.von der (1891). Die Sprachwissenschaft. Leipzig: Weigel. ivón, T. (1983). “Topic Continuity in Discourse; an Introduction.” In T.Givón (ed.) Topic Continuity in Discourse: a Quantitative Cross-language Study. Amsterdam/Philadelphia: John Benjamins. Gómez-González, M.A. (2001). The Theme-Topic Interface: Evidence from English. Amsterdam/Philadelphia: John Benjamins. Goodall, G. (1992). “On the Status of Spec of IP.” WCCFL 10, 175–182 . Grimshaw, J. and V.Samek-Lodovici (1998). “Optimal Subjects and Subject Universals.” In P.Barbosa, D.Fox, P.Hagstrom, M.McGinnis and D. Pesetsky (eds.) Is the Best Good Enough? Optimality and Competition in Syntax. Cambridge, Mass.: MIT Press, 193–219. Gundel, J. (1988) [1974]. The Role of Topic and Comment in Linguistic Theory. New York: Garland. ——(1994).“On Different Kinds of Focus.” In P.Bosch and R.Van der Sandt, (eds.) Focus and Natural Language Processing. Heidelberg: Institute for Logic and Linguistics, 457–466. Hajičová, E. (1987). “Focusing -A Meeting Point of Linguistics and Artificial Intelligence.” In P.Jorrand and V.Sgurev (eds.), 311–321. Hajičová, E., B.H.Partee and P.Sgall (1998). Topic-Focus Articulation, Tripartite Structures, and Semantic Content. Dordrecht: Kluwer. Halliday, M.A.K. (1967) “Notes on Transitivity and Theme in English.” Journal of Linguistics 3, 37–81 and 199–244. Hatcher, A.G. (1956). “Theme and Underlying Question: Two Studies in Spanish Word Order.” Word 12 (Monograph 3), 234–250. Heim, I. (1982). The Semantics of Definite and Indefinite Noun Phrases, Ph.D. Dissertation, University of Massachusetts, Amherst. Hernanz, M.L. and J.M.Brucart (1987). La Sintaxis. Barcelona: Crítica. Herring, S. (1990). “Information Structure as a Consequence of Word Order Type.” Berkeley Linguistic Society 16, 163–174.
REFERENCES 223
Hockett, C.F. (1958). A Course in Modern Linguistics. New York: Macmillan. Horn, L.R. (1989). A Natural History of Negation. Chicago: University of Chicago Press. Horvath, J. (1986). FOCUS in the Theory of Grammar and the Syntax of Hungarian. Dordrecht: Foris. Jacobs, J. (1984). Funktionale Satzperspektive und Illokutionssemantik. Linguistische Berichte 91, 25–8. ——(1999). “The Dimensions of Topic-Comment.” Ms. Universität Wuppertal. Jakendoff, R. (1972). Semantic Interpretation in Generative Grammar. Cambridge, Mass.: MIT Press. Jorrand, P. and V.Sgurev (eds.) (1987). Artificial Intelligence II: Methodology, Systems, Applications, Amsterdam: Elsevier. Kamp, H. (1981). “A Theory of Truth and Discourse Representation.” In J. Groenendijk, T.Janssen, and M.Stokhof, (eds.) Formal Methods in the Study of Language. Amsterdam: Mathematical Centre Tracts 135, 277–322. King, T. (1995). Configuring Topic and Focus in Russian. Stanford: CSLI Publications. Kiss, K.E. (1981). “Structural Relations in Hungarian, a ‘Free’ Word Order Language.” Linguistic Inquiry 12, 185–213. ——(1998). “Identificational Focus versus Information Focus.” Language 74.2, 245–273. Koopman, H. and D.Sportiche (1991). “The Position of Subjects.” Lingua 85, 211–258. Kratzer, A. (1989). “Stage-Level and Individual-Level Predicates.” In E.Bach, A.Kratzer and B.Partee (eds.) Papers on Quantification. Umass, Amherst. Krifka, M. (1991). “A Compositional Semantics for Multiple Focus Constructions.” SALT 1, 127–158. ——(1992) “Focus, Quantification and Dynamic Interpretation.” Ms. University of Texas at Austin. Kroch, A. (1989). “Reflexes of Grammar in Patterns of Language Change.” Language Variation and Change 1, 199–244. Kuno, S. (1972). “Functional Sentence Perspective: A Case Study from Japanese and English.” Linguistic Inquiry 3.3, 269–320. ——(1976). “Subject, Theme and the Speaker’s Empathy A Reexamination of Relativization Phenomena.” In Ch. Li (ed.), 417–444 . Kuroda, S-Y. (1965) Generative Grammatical Studies in the Japanese Language. Ph.D. Dissertation, MIT. Laca, B. (1990). “Generic Objects: Some More Pieces of the Puzzle.” Lingua 81,, 25–46. Laka, M.I. (1990). Negation in Syntax: On the Nature of Functional Categories and Projections. Ph.D. Dissertation, MIT. Lambrecht, K. (1994). Information Structure and Sentence Form. Cambridge: CUP
224 REFERENCES
Lee, H. (2000). “The Emergence of the Unmarked Order in Hindi.” NELS 30, 469–483. Levin, B. and M.Rappaport (1995). Unaccusativity at the Syntax-Lexical Semantics Interface. Cambridge: MIT Press Lewis, D. (1979). “Scorekeeping in a Language Game.” In R.Bäuerle, U.Egli, and A.von Stechow (eds.) Semantics from Different Points of View . Berlin: Springer Verlag. Li, C.N. (ed.) (1975). Word Order and Word Order Change. Austin: University of Texas Press. ——(ed.) (1976). Subject and Topic. New York: Academic Press. Longobardi, G. (1994). “Reference and Proper Names: A Theory of NMovement in Syntax and Logical Form.” Linguistic Inquiry 25.4, 609–665. Mathesius, V. (1915). “O passivu v moderní angličtinĕ. Sbornik filologický 5, 198–220. ——(1939). “On the So-called Functional Sentence Perspective.” Slovo a Slovesnost 5, 171–174. ——(1942). “Language and Style.” Ctení o Jazyce a Poezii, 13–102. ——(1975). A Functional Analysis of Present Day English on a General Linguistic Basis. Transl. by L.Dušková, ed. by J.Vachek. Prague: Academia. (originally published in Czech under the title Obsahovy rozbor soucasné anglictiny na zaklade obecne lingvistickem, Praha, 1961). Nash, L. (1995). Argument Scope and Case Marking in SOV and in Ergative Languages: the Case of Georgian. Ph.D. Dissertation, Univ. Paris. Olarrea, A. (1996). Pre and Postverbal Subject Positions in Spanish: A Minimalist Account. Ph.D. Dissertation, University of Washington. Ordoñez, F. (2000). The Clausal Structure of Spanish. A Comparative Study. New York: Garland. Partee, B.H. (1989). “Speaker’s Reply”In S. Allen (ed.) Possible Worlds in Humanities, Arts and Sciences: Proceedings of Nobel Symposium 65. Berlin/New York: Walter de Gruyter, 152–161 . ——(1991). ‘Topic, Focus and Quantification.” SALT 1, 159–187. Paul, H. (1886). Prinzipien der Sprachgeschichte. Halle: Niemeyer. Peregrin, J. (1996). “Topic and Focus in a Formal Framework.” In B.Partee and P.Sgall (eds.) Discourse and Meaning: Papers in Honor of Eva Hajičová Amsterdam: John Benjamins. Polinsky, M. (1999). Review of Lambrecht: Information Structure and Sentence Form. Language 75.3,, 567–582. Portner, P. and Y.Katsuhiko (1998). “The Semantics and Pragmatics of Topic Phrases.” Linguistics and Philosophy 21.2, 117–57. Postal, P. (1971). Crossover Phenomena. New York: Holt, Rinehart and Winston.
REFERENCES 225
Prince, E.F. (1981a). “Topicalization, Focus-Movement and Yiddish-Movement: a Pragmatic Differentiation.” Berkeley Linguistics Society 7, 249–264. ——(1981b). “Toward a Taxonomy of Given/New Information.” In P.Cole, (ed.) Radical Pragmatics . New York: Academic Press, 223–255. ——(1984). “Topicalization and Left-dislocation; a Functional Analysis.” In S.J.White and V.Teller (eds.) Discourse in Reading and Linguistics Annals of the New York Academy of Sciences 433, 213–25 . ——(1986). “On the Syntactic Marking of Presupposed Open Propositions.” CLS 22, 208–22 . ——(1998). “On the Limits of Syntax, with Reference to Left-Dislocation and Topicalization.” In P.Culicover and L.MacNally (eds.), 281–302. Reinhart, T. (1981). “Pragmatics and Linguistics: an Analysis of Sentence Topics.” Philosophica 27, 53–94 . ——(1995). “Interface Strategies.” OTS Working Papers. Utrecht University. Rivero, M.L. (1980). “On Left-dislocation and Topicalization in Spanish.” Linguistic Inquiry 11, 363–93. Rizzi, L. (1982). Issues in Italian Syntax. Dordrecht: Foris. ——(1997). “The Fine Structure of the Left Periphery.” In L.Haegeman (ed.) Elements of Grammar. Dordrecht: Kluwer, 281–337 . Roca, F. (1992). “Object Clitics in Spanish and Catalan.” Catalan Working Papers in Linguistics 1992, 245–280. Universitat Autònoma de Barcelona. Rochemont, M. (1986). Focus in Generative Grammar. Amsterdam; John Benjamins. Rochemont, M. and P.Culicover (1990). English Focus Constructions and the Theory of Grammar, Cambridge: CUP. Rodman, R. (1974). “On Left-dislocation.” Papers in Linguistics 7, 437–66. Rooth, M. (1985). Association with Focus, Ph.D. Dissertation, University of Massachusetts, Amherst. ——(1992). “A Theory of Focus Interpretation.” Natural Language Semantics 1, 75–116. Ross, J. (1967). Constraints on Variables in Syntax. Ph.D. Dissertation. MIT. Runner, J. (1993). “Quantificational Objects and Agr-o.” MIT Working Papers in Linguistics 21. Cambridge: MIT ——(1994). “A Specific Role for Agr.” UMOP 17, 153–177. Selkirk, E.O. (1984). Phonology and Syntax. The Relation between Sound and Structure. Cambridge: MIT Press. ——(1995). “Sentence Prosody: Intonation, Stress, and Phrasing.” In J. Goldsmith (ed.) The Handbook of Phonological Theory. Oxford: Basil Blackwell. Sgall, P., E.Hajičová and J.Panenová. (1986). The Meaning of the Sentence in its Semantic and Pragmatic Aspects. Dordrecht: Reidel. Soh, H.L. (1998). Object Scrambling in Chinese. Ph. D. Dissertation, MIT.
226 REFERENCES
Sportiche, D. (1996). “Clitic Constructions.” In J.Rooryck and L.Zaring (eds.) Phrase Structure and the Lexicon. Dordrecht; Kluwer, 213–76 . Stalnaker, R. (1978). “Assertion.” In P.Cole (ed.) Pragmatics. Syntax and Semantics 9. New York: Academic Press, 315–32. Stechow, A.von and S.Uhmann (1986). “Some Remarks on Focus Projection.” In W.Abraham and S.de Meij (eds.), 291–320. Strawson, P.F. (1971). “Identifying Reference and Truth Values.” In D. Steinberg and L.Jakobovitz (eds.) Semantics. Cambridge: CUP, 86–99. Suñer, M. (1982). Syntax and Semantics of Spanish Presentational Sentence Types. Washington: Georgetown University Press. ——(1988). “The Role of Agreement in Clitic-doubled Constructions.” Natural Language and Linguistic Theory 6, 391–434. Tateishi, K. (1991). The Syntax of Subjects. Ph.D. Dissertation, University of Massachusetts, Amherst. Trávníček, F. (1962). “On So-called Functional Sentence Perspective.” Slovo a slovesnost 22, 163–71. Uriagereka, J. (1991). “Determiner Clitic Placement.” In R.Freidin (ed.) Principles and Parameters in Comparative Grammar. Cambridge, Mass.: MIT Press. ——(1992). “A Focus Position in Western Romance.” In K.Kiss (ed.) Discourse Configurational Languages. Oxford: Oxford University Press. ——(1995). “Aspects of the Syntax of Clitic Placement in Western Romance.” Linguistic Inquiry 26.1, 79–123. Vallduví, E. (1988). “Functional Load, Prosody and Syntax: Left-detachment in Catalan and Spanish.” CLS 24, 391–404. ——(1990). The Informational Component. Ph.D. Dissertation, University of Pennsylvania, Philadelphia. ——(1992). “A Preverbal Landing Site for Quantificational Operators. Catalan Working Papers in Linguistics. 1992, 319–344. Universitat Autònoma de Barcelona. Ward, G. (1988). The Semantics and Pragmatics of Preposing. New York: Garland. Ward, G. and E.F.Prince (1991). “On the Topicalization of Indefinite NPs.” Journal of Pragmatics 15.8, 167–178. Wegener, Ph. (1885). Untersuchungen über die Grundfragen des Sprachlebens. Halle: Niemeyer. Weil, H. (1879) [1844]. De I’ordre des mots dans les langues anciennes comparéees aux langues modernes. Paris: Joubert. Translated as The Order of Words in the Ancient Languages compared with that of the Modern Languages. Boston: Ginn & Co. 1887 by Charles W.Super. Reedited: Amsterdam/Philadelphia: John Benjamins 1978.
REFERENCES 227
Wilkinson, K. (1991). Studies in the Semantics of Generic Noun Phrases. Ph.D. Dissertation, University of Massachusetts, Amherst. Zubizarreta, L. (1998). Word Order, Prosody and Focus. Cambridge: MIT Press
228 REFERENCES
Index
A accent, 144, 145, 146, 163, 164 aboutness, 9, 14, 18, 21, 22, 25, 27, 42, 45, 87 accent, 12, 28, 41, 46, 49, 50, 88, 125–33, 139, 142–50, 162–64, 170, 176, 178, 186, 187, 194, 196, 206, 207, 208, 211, 214 accessible referent, 29, 46 active referent, 46, 50 Adjémian, 44, 96 adjunctions, 6, 84, 98, 211
Chomsky, 13, 21, 23, 56, 75, 96, 121, 121, 136, 162, 164, 168, 194 Cinque, 4, 9, 45, 72, 74, 75, 76, 78, 79, 85, 87, 96, 97, 98, 111, 114, 117, 156, 162, 207, 211 cleft, 186 Clements, 97 clitic doubling, 59, 60 CLLD (Clitic Left-dislocation), 4, 75, 78, 81, 82, 83, 84, 97, 98, 111, 114, 119, 123, 217 contextual boundness, 30 Contreras, 6, 8, 18, 20, 43, 80, 96, 97, 98, 101, 109, 111, 115, 123, 162, 193, 210 Czech, 17, 43, 89, 94, 162
B accent, 144, 145, 146, 148, 163, 164, 178, 196 Background, 10, 12, 21, 93 Background-Focus, 12, 200, 202, 203, 204, 207, 208, 209, 210, 213, 214, 217 base-generated, 4, 12, 84, 98, 103, 114, 118, 121, 211 BLC, 24, 45, 170, 171, 173, 173, 174, 181 BN (Bare Nominals), 5, 6, 7, 10, 81, 101–23 BNMC (Bare Noun Movement Constraint), 107, 114, 115, 118 Bolinger, 44, 96
Dahl, 11, 21, 32, 44, 194, 196, 204 Daneš, 17, 43 Diesing, 7, 13, 105, 107, 116, 117, 168 discourse referent, 14, 22, 25, 32, 40, 42, 49, 158, 213 discourse-oldness, 9, 22, 28, 87 DP (Determiner Phrases), 8, 10, 70, 107, 112, 118, 121, 123, 199, 209, 214, 217 DP movement, 10, 117, 121, 122
Catalan, 80, 85, 96, 129, 131, 152, 161, 181, 182, 187, 188, 198, 200, 203, 204, 205 Chafe, 18, 20, 28, 29, 44, 46, 96, 153
echo-question, 189 Emphatic Stress, 144, 145, 146, 148
229
230 INDEX
Erteschik-Shir, 1, 11, 19, 21, 27, 40, 125, 148, 150, 159, 168, 171, 179, 180, 194, 208, 216 existential interpretation, 7, 70, 104, 105, 106, 107, 109, 118, 214 Firbas, 17, 20, 43, 47, 193 Focus, 125–64 AI (artificial intelligence), 141 argument-, 125, 157, 158, 159, 184 contrastive, 125, 128, 132–43, 142, 150, 165, 166, 168, 178, 179, 180 identificational, 160, 161 in situ, 171, 189 information, 125, 160 metalinguistic, 148–53, 179–80 narrow, 92, 140, 151, 153, 157, 158, 159, 160, 161, 165, 180, 182, 183, 186, 187, 188, 190, 200 predicate, 158, 184 presentational, 133, 134, 135, 138, 140, 165, 166 psychological, 141, 142, 146 relation, 28, 127, 157, 163, 169, 178, 196 retrieve-add, 11, 155, 156, 159 retrieve-substitute, 11, 155, 156, 157, 159 semantic, 141–50, 150, 163, 178, 179, 196 sentence, 158 sentence-final, 65, 66, 119, 121, 130, 183, 189, 207, 214 VP, 65, 188 wide, 140, 158, 161 Focus Preposing, 5, 8, 170–90 Focus-Background, 11, 12, 14, 32, 47, 66, 140, 152, 157, 171, 176, 180, 183, 190, 191, 193, 194, 196, 202 Focus-Presupposition, 13, 14
generic interpretation, 104, 105, 106, 107, 108, 109, 118, 123 German, 26, 86, 92, 105, 106, 107, 109 given information. See old information Givón, 44 ground, 13, 16, 20, 21, 32, 152, 153, 154, 155, 157, 173, 196, 196, 200 Gundel, 11, 19, 20, 23, 33, 34, 44, 47, 84, 89, 96, 125, 141, 150, 161, 163, 178, 179, 180, 193, 196 Hajičová, 141, 144 Halliday, 18, 20, 33, 34, 44, 47, 50, 96, 141, 142, 193, 216 Hatcher, 44, 96, 97 hierarchical trichotomy, 32, 173, 191, 196 Hockett, 18 Italian, 45, 72, 74, 75, 76, 77, 78, 79, 80, 87, 97, 123, 161, 211 Jackendoff, 21, 28, 88, 127, 144, 145, 168, 178, 194 Jacobs, 11, 92, 194 Japanese, 44, 89, 94, 98, 128, 173, 174, 183 Kiss, 11, 94, 125, 160, 161, 163, 169 Krifka, 11, 13, 96, 108, 146, 176, 194 Lambrecht, 21, 25, 30, 40, 157 59, 176, 184 Latin, 16, 89 Left-dislocation, 4, 35, 70, 72 LF (Logical Form), 1, 7, 8, 103, 106, 107, 117, 118, 184 mapping hypothesis, 7, 105, 107 Mathesius, 17, 20, 22, 27, 43, 50
INDEX 231
new information, 29, 125, 125, 128, 129, 131, 132, 133, 134, 135, 136, 142, 143, 157, 163, 164 NP-movement, 10, 107, 214 null subjects, 51, 55, 55, 58 old information, 19, 27, 128, 129, 133 Open Proposition, 171, 175, 182 Partee, 13, 43, 45, 95, 96, 117, 168, 178, 179, 190 partitive clitic, 80 Pierrehumbert, 178 postverbal subjects, 3, 6, 9, 12, 13, 36, 37, 38, 49, 68, 94, 101, 103, 104, 118, 199, 212, 213 Prince, 1, 23, 28, 44, 46, 79, 91, 96, 100, 116, 123, 153, 181, 194 Reinhart, 1, 10, 19, 20, 23, 26, 27, 44, 45, 162, 193, 217 Right-dislocation, 9, 53, 72, 84, 85, 187, 199 Rizzi, 1, 56, 84, 98, 211, 217 Rochemont, 11, 21, 36, 125, 132 Rodman, 78, 96 Russian, 44, 89 Selkirk, 125, 136, 137, 162, 166, 169, 183, 184, 194, 206 Sgall, 19, 20, 30, 31, 43, 47, 168, 194 Spec, IP, 6, 7, 10, 36, 105, 106, 113, 116, 118, 121, 151 specifier, 6, 12, 93, 98, 112, 113, 114, 160, 209, 215 SPLD (Strong Pronoun Leftdislocation), 77, 78, 79, 91, 98 STopic (Sentence Topic), 200 STopic-Focus, 11, 12, 200, 202, 204, 205, 207, 208, 209, 210, 213 Structure dislocated, 72, 81, 87, 200, 203, 215
information. See TFA syntactic, 209 subject-verb inversion, 5, 185 TFA (Topic-Focus Articulation), 118, 193, 209 Theme-Rheme, 1, 14, 191, 193 Topic, 16 accented, 127, 146, 147, 163, 170, 196 acceptability scale, 30, 6, 109, 110, 111, 112, 114 definition(s), 20 discourse-new, 196 dislocated, 6, 101, 103, 110, 111 features, 88 internal/external, 67 link, 196 null, 56, 58 relation, 25, 28, 157, 169, 170, 178, 196 sentence-initial, 3, 31, 191, 193, 196, 196 STopic (Sentence Topic), 10, 12, 53, 93, 98, 118, 121, 196, 209, 210, 213, 214, 215, 217 subjects, 3, 38, 55, 59, 69, 121, 122, 147, 184, 199, 212 tail, 196 tests, 23 theme, 1, 13, 14, 17, 18, 23, 33, 34, 43, 44, 47, 49, 92, 123, 193, 216 Topicalization, 4, 28, 70, 78, 171 topicalizing devices, 4, 9, 49, 53, 87, 70, 88, 131 Topic-Comment, 13, 14, 32, 39, 92, 127, 158, 159, 176, 191, 193, 194, 208, 213 Topic-Criterion, 211, 212, 217 Topic-Focus, 1, 4, 5, 11, 14, 130, 131, 152, 171, 173, 179, 191, 193, 194, 212, 214 truth conditions, 178
232 INDEX
unaccusative, 50, 137, 166, 184 Vallduví, 1, 11, 21, 24, 32, 152–58, 173, 196–200 verb-subject order, 74, 76, 174 Ward, 5, 21, 23, 24, 28, 45, 46, 78, 88, 90, 123, 128, 170, 171, 173, 173, 175, 181, 182, 183, 185, 190, 194 wh-in situ, 189 wh-movement, 9, 11, 82, 117, 128, 171, 174, 175, 185, 186, 188, 189, 190, 215 wh-phrase, 34, 51, 83, 174 Yiddish Movement, 181 Zubizarreta, 1, 10, 13, 98, 124, 162, 211, 212, 216