SYNTAX AND SEMANTICS VOLUME 33
EDITORIAL BOARD
Series Editors BRIAN D. JOSEPH AND CARL POLLARD Department of Linguis...
114 downloads
1117 Views
8MB 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
SYNTAX AND SEMANTICS VOLUME 33
EDITORIAL BOARD
Series Editors BRIAN D. JOSEPH AND CARL POLLARD Department of Linguistics The Ohio State University Columbus, Ohio
Editorial Advisory Board JUDITH AISSEN University of California, Santa Cruz
PAULINE JACOBSON Brown University
PETER CULICOVER
MANFRED KRIFKA University of Texas
The Ohio State University
ELISABET ENGDAHL University of Gothenburg
WILLIAM A. LADUSAW University of California, Santa Cruz
JANET FODOR City University of New York
BARBARA H. PARTEE University of Massachusetts
ERHARD HINRICHS University of Tubingen
PAUL M. POSTAL Scarsdale, New York
A list of titles in this series appears at the end of this book.
SYNTAX and SEMANTICS VOLUME 33 Long-Distance Reflexives Edited by Peter Cole Department of Linguistics University of Delaware Newark, Delaware Gabriella Hermon Department of Linguistics University of Delaware Newark, Delaware C.-T. James Huang Department of Linguistics University of California Irvine, California
ACADEMIC PRESS San Diego London Boston New York Sydney Tokyo Toronto
This book is printed on acid-free paper. Copyright © 2001 by Academic Press All Rights Reserved. No parts of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the Publisher. The appearance of the code at the bottom of the first page of a chapter in this book indicates the Publisher's consent that copies of the chapter may be made for personal or internal use of specific clients. This consent is given on the condition, however, that the copier pay the stated per copy fee through the Copyright Clearance Center, Inc. (222 Rosewood Drive, Danvers, Massachusetts 01923), for copying beyond that permitted by Sections 107 or 108 of the U.S. Copyright Law. This consent does not extend to other kinds of copying, such as copying for general distribution, for advertising or promotional purposes, for creating new collective works, or for resale. Copy fees for pre-2000 chapters are as shown on the title pages. If no fee code appears on the title page, the copy fee is the same as for current chapters. 0092-4563/00 $35.00 Explicit permission from Academic Press is not required to reproduce a maximum of two figures or tables from an Academic Press chapter in another scientific or research publication provided that the material has not been credited to another source and that full credit to the Academic Press chapter is given.
Academic Press A Harcourt Science and Technology Company 525 B Street, Suite 1900, San Diego, CA 92101-4495 http://www.academicpress.com
Academic Press Harcourt Place, 32 Jamestown Road, London NWI 7BY, UK http://www.academicpress.com Library of Congress Catalog Card Number: 0092-4563 International Standard Book Number: 0-12-613533-9 PRINTED IN THE UNITED STATES OF AMERICA 00 01 02 03 04 05 SB 9 8 7 6 5 4 3 2 1
CONTENTS
Contributors Introduction
xi xiii
Grammatical and Discourse Conditions on Long Distance Reflexives in Two Chinese Dialects
1
PETER COLE, GABRIELLA HERMON, AND CHER LENG LEE 1. Introduction: Approaches to Long Distance Reflexives 2. Logophoric Conditions on LD Reflexives in Mandarin and Teochew 3. Subcommanding Antecedents of LD Reflexives and SOURCE/SELF Requirements on Antecedents 4. Modifying Phrases as Tests for SOURCE/SELF 5. Additional Evidence on the Role of SELF 6. Pivot Restrictions 7. Logophoricity and Attitudes De Se 8. The Syntactic and Semantic Analysis of Attitudes De Se 9. The Binding Theoretical Status of LD Reflexives 10. The C-Command Requirement 11. VP Ellipsis 12. The Blocking Effect Revisited 13. Counter Examples to the Grammatical Account of the Blocking Effect 14. Conclusion Acknowledgments Notes References
v
1 4 6 8 10 11 16 19 23 25 26 28 35 36 38 39 44
Vi
Contents
Long-Distance Anaphors in Hindi/Urdu: Syntactic and Semantic Issues ALICE DAVISON 1. 2. 3. 4. 5. 6. 7. 8.
Introduction: Nonlocally Bound Anaphoric Expressions Lexical Forms of Anaphors and Inflectional Morphology Overview of Binding Relations A Head-Raising Account for Long-Distance The Functional Projection Host for Reflexives Syntactic/Semantic Factors Summary and Conclusion Appendix: Object Agreement and Subject Orientation Acknowledgments Notes References
Reflexive Anaphor or Conjunctive Operator: Riau Indonesian Sendiri DAVID GIL 1. 2. 3. 4. 5.
Why Do Only Some Languages Have Long-Distance Reflexives? Riau Indonesian Sendiri as a Reflexive Anaphor Sendiri as a Conjunctive Operator Generalizing to Other Forms Acknowledgments Notes References
Syntactic vs. Logophoric Binding: Evidence from Norwegian Child Language ARILD HESTVIK AND WILLIAM PHILIP 1. 2. 3. 4. 5.
Introduction Local vs. Nonlocal (Logophoric) Binding Using Child Language as Evidence: Avrutin and Cunningham (1997) Experimental Evidence from Norwegian Child Language Conclusion References
47
47 48 52 56 60 68 75 77 78 78 81
83
83 85 88 98 106 111 112 114
119
119 120 126 128 138 139
Contents
Logophoricity, Attitudes, and Ziji at the Interface C.-T. JAMES HUANG AND C.-S. LUTHER LIU 1. 2. 3. 4. 5. 6.
vii
141
Introduction Previous Analyses Ziji as a Logophor Ziji as a (Syntactic) Anaphor The Long Distance Ziji: Syntax and Semantics Summary Acknowledgments Notes References
141 143 154 166 174 184 185 185 192
Local and Long-Distance Reflexives in Turkish JAKLIN KORNFILT
197
1. 2. 3. 4. 5. 6. 7.
Introduction: Types of Morphologically Free Reflexives Differences Between the Bare and the Inflected Reflexive The Nature of the Inflected Reflexive: A LD-Reflexive or a Pronoun? Elements of a Third (or Fourth?) Kind The Bare Reflexive: Anaphor or Emphatic Pronoun? Complex Reflexives Conclusion Acknowledgments Notes References
Anti-Antilocality JEFFREY LIDZ 1. 2. 3. 4. 5. 6.
Introduction Anaphora in Kannada Why Apparent Antilocality is not Syntactic Predictions of Condition R Alternative Analyses Conclusion Acknowledgments Notes References
197 200 201 205 212 217 219 220 220 224
227
227 229 232 237 242 250 251 251 253
viii
Contents
Long-Distance Reflexivization in Chechen and Ingush
255
JOHANNA NICHOLS 1. 2. 3. 4. 5.
Introduction Local Reflexivization Logophoric Reflexivization Ordinary Long-Distance Reflexivization Conclusion Acknowledgments Notes References
Why the Blocking Effect?
255 257 259 261 274 275 276 277 279
HAIHUA PAN 1. 2. 3. 4. 5. 6. 7.
Introduction The Blocking Effect The Blocking Effect is not Symmetrical Previous Analyses and their Limitations Self-Ascription and Long-Distance Bound Ziji Discussion Conclusion Acknowledgments Notes References
Syntactic and Nonsyntactic Constraints on Long-Distance Reflexives
279 280 282 286 293 305 308 309 309 314 317
CARL POLLARD AND PING XUE 1. 2. 3. 4.
Introduction Syntactic and Nonsyntactic Reflexives Nonsyntactic Conditions on Ziji Conclusion Acknowledgments Notes References
317 320 328 335 338 338 340
Contents
Anaphors, Logophors, and Binding
ix
343
ERIC REULAND 1. 2. 3. 4. 5. 6.
Introduction Interpreting Sig Non-C-Commanding Antecedents The Interpretive Process What Makes a Pronominal Interpretation of Sig Logophoric? Summary and Conclusion Acknowledgments Notes References
Index
343 346 348 350 357 363 364 364 368 371
This page intentionally left blank
CONTRIBUTORS Numbers in parantheses indicate the pages on which author's contributions begin.
Peter Cole (xiii, 1), Department of Linguistics, University of Delaware, Newark, Delaware 19716 and Department of Linguistics, Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany Alice Davison (47), Department of Linguistics, University of Iowa, Iowa City, Iowa 52242 David Gil (83), Department of Linguistics, Max Planck Institute for Evolutionary Anthropology, Leipzig 04103, Germany Gabriella Hermon (xiii, 1), Department of Linguistics and School of Education, University of Delaware, Newark, Delaware 19716 and Department of Linguistics, Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany Arild Hestvik (119), Department of Linguistics and Comparative Literature, University of Bergen, N-5007 Bergen, Norway C.-T. James Huang (xiii, 141), Department of Linguistics, University of California, Irvine, California 92697 Jaklin Kornfilt (197), Department of Languages, Literatures, and Linguistics, Syracuse University, Syracuse, New York 13244 Cher Leng Lee (1), Department of Chinese Studies, National University of Singapore, Kent Ridge, Singapore 119260 Jeffrey Lidz (227), Institute for Research in Cognitive Science, University of Pennsylvania, Philadelphia, Pennsylvania 19104 C.-S. Luther Liu (141), Department of Foreign Languages and Literatures, National Chi-Nan University, Puli, Nantou, Taiwan Johanna Nichols (255), Department of Slavic Languages, University of California at Berkeley, Berkeley, California 94720 Haihua Pan (279), Department of Chinese, Translation, and Linguistics, City University of Hong Kong, Kowloon, Hong Kong
xi
xii
Contributors
William Philip (119), Utrecht Institute of Linguistics OTS, Utrecht University, The Netherlands Carl Pollard (317), Department of Linguistics, The Ohio State University, Columbus, Ohio 43210 Eric Reuland (343), Utrecht Institute of Linguistics OTS, Utrecht University, The Netherlands Ping Xue (317), Applied Research and Technology, The Boeing Company, Seattle, Washington 98124
INTRODUCTION LONG-DISTANCE REFLEXIVES: THE STATE OF THE ART
1. LONG-DISTANCE REFLEXIVES AND UNIVERSAL GRAMMAR The papers in this volume examine the locality conditions on reflexives across languages. In languages around the world the distribution of anaphoric elements appears roughly like that found in English: certain forms occur only with antecedents that c-command them and occur in the same local domain (roughly, clause) as the anaphoric elements, while other forms cannot have c-commanding antecedents within that same domain. The occurrence of many genetically unrelated and typologically dissimilar languages with anaphoric elements that conform to the same distributional restrictions constituted powerful early support for universalist, generative approaches to the syntax of what has come to be known as Binding. While a universalist approach to Binding as instantiated in the Binding Theory has provided strong support for Universal Grammar, it has long been known that in many languages some forms that can be used as local reflexives can also take antecedents outside their local domain (hereafter, long-distance reflexives). For instance in Chinese, the reflexive ziji in (1) can be interpreted as referring to the matrix subject, the intermediate subject or the lowest subject: xiii
xiv
Long-distance Reflexives: The State of the Art
(1) Zhangsani renwei Lisij zhidao Wangwuk xihuan ziji i/j/k Zhangsan think Lisi know Wangwu like self 'Zhangsan thinks Lisi knows Wangwu likes self.' The question naturally arises whether these "exceptions" to Binding Theory are arbitrary (thereby seriously undercutting the universality of Binding Theory and leaving unexplained why so many anaphoric forms in unrelated languages show the same distribution), or whether violations of Binding Theory occur in a delimited domain, calling for the refinement and revision of the Binding Theory rather than its total abandonment.
2. TYPOLOGICAL PROPERTIES OF LONG-DISTANCE REFLEXIVES Approaching the problem from the perspective of syntax, it was soon noticed that long-distance reflexives appear to share a number of characteristics across languages (Pica, 1987):1 (2) Apparent Typological Characteristics of Long-Distance Reflexives: a. Long-distance reflexives are monomorphemic. b. They take subject antecedents. c. Their occurrence can, in many languages, be restricted to environments in which the antecedent and reflexive are found in specific domains (i.e., specific types of IPs such as infinitival or subjunctive). Furthermore, it was noted in the Chinese literature (Y.-H. Huang, 1984; Tang, 1985,1989) that long-distance reflexives are subject to a "Blocking Effect": When the subjects of the matrix and the subordinate clauses share features for person, either NP is a possible antecedent for ziji. When, however, the subjects of these clauses differ in person, only the subject of the subordinate clause is a possible antecedent for ziji: The subordinate subject blocks the matrix subject (and any higher subject regardless of person) from anteceding ziji: (3) Zhangsani renwei woj zhidao Wangwuk xihuan ziji*i/*j/k. Zhangsan think I know Wangwu like self 'Zhangsan thinks I know Wangwu likes self.' In (3) the presence of third-person Wangwu as subject of the lowest clause blocks ziji from referring to either the intermediate subject wo or the main clause subject Zhangsan.2 No similar restriction was noted in the literature on European languages with long-distance reflexives. This raised the question of how the cross-linguistic variation with respect to the existence of the Blocking Effect should be explained and of whether Blocking correlates
Introduction
XV
with any property of the languages in which it occurs or of the forms that are subject to Blocking, a question to which we shall return below. In addition to the restrictions just mentioned, which were widely interpreted as syntactic in nature, many authors noted that long-distance reflexives are frequently limited to taking antecedents that are "prominent" in the discourse. The discourse requirements appeared to vary from language to language, to at least some extent, but they could, in most cases, be reduced to what came to be referred to as "logophoric" requirements, on analogy with the system of logophoric pronouns occurring in some African languages. This raised the question of how logophoricity should be related to long-distance reflexives. Are long-distance reflexives simply logophoric pronouns masquerading as reflexives, or is the relationship between logophoricity and long-distance reflexives more complex than simple equivalence?
3. THE GRONINGEN AND CORNELL WORKSHOPS In order to assess the current state of knowledge and provide an impetus for the further study, a workshop on long-distance anaphors was held by the Department of Linguistics of Groningen University in June of 1987. This workshop resulted in a volume that appeared in 1991 (Koster and Reuland). While there was considerable diversity in the interests of the contributors to that volume, two themes preoccupied many of the authors: 1) What is the nature of the domain restriction on long-distance reflexives? 2) What sorts of logophoric or other discourse conditions are necessary for long-distance anaphora? While there was some brief discussion of Asian languages in several of the chapters in the volume, with the exception of the extended discussion of Mandarin in Huang and Tang (1991), nearly all of the detailed analysis was of European languages. This, of course, reflected the fact that in the 1980s most of the research on long-distance anaphora had been on European languages (especially Italian and Scandinavian languages). The present volume grows out of the explosion of interest in long-distance anaphora that followed the publication of Koster and Reuland. A workshop similar to the Groningen workshop was organized at the Cornell LSA Linguistic Institute in July of 1997 to commemorate the tenth anniversary of the Groningen workshop and to assess progress over the last ten years. The contributions to the Cornell workshop provided the basis for most of the chapters in this volume. How do the results of the Cornell workshop differ from those ten years earlier? The most obvious difference between the 1991 volume and this volume is that the focus in the present volume is on the anaphoric systems of languages that had received little or no attention in the literature ten years earlier. In addition to this volume's four chapters containing extensive discussions of Mandarin (Cole, Hermon, and Lee; Huang and
xvi
Long-distance Reflexives: The State of the Art
Liu; Pan; and Pollard and Xue),3 there is a chapter comparing a variety of Mandarin with a southern Chinese dialect (Teochew) and (to a lesser extent) with Malay (Cole, Hermon and Lee), and chapters on Hindi-Urdu (Davison), Riau Indonesian (Gil), Turkish (Kornfilt), Kannada (Lidz), Chechen and Ingush (Nichols). In fact, the only languages discussed in this volume that figured prominently in Koster and Reuland are Norwegian (Hestvik and Philip) and Icelandic (Reuland). Not surprisingly, the long-distance reflexives found in the languages described here are considerably more diverse than those that were the focus of research of Koster and Reuland; but, despite the diversity, the major themes of the earlier volume are prominent in this volume as well.
4. PRONOUN OR BOUND ANAPHOR A fundamental and recurring topic in many of the chapters in this volume is whether the "long-distance reflexives" in the language under study have the distribution of pronouns or of bound anaphors. While in a number of studies it is taken as a given that long-distance reflexives are anaphors (e.g. Lidz), the question is in fact a difficult one. The study of the typological characteristics of long-distance reflexives has been frustrating because every proposed characterization of the construction appears to have many counterexamples. If, however, "long-distance reflexives" are not a single type of morpho-syntactic entity, it would not be expected that they would have the same properties or the same analysis across languages. In fact, a comparison of the papers in this volume suggests that long-distance reflexives are, in fact, several different sorts of entities, which differ with regard to whether they have the distribution of bound anaphors or of pronouns.4 If this is correct, it is not surprising that they differ in both syntactic and discourse properties: Bound anaphors are generally assumed to require (or strongly favor) a binding relationship with their antecedents, thereby explaining why they require c-commanding antecedents, do not allow extra sentential antecedents, and require (or strongly favor) sloppy readings in VP ellipsis. In contrast, pronouns are assumed to enter into both binding and coreference relations. Thus, c-command is not required, extra sentential antecedents are possible, and both strict and sloppy readings are easily available under VP ellipsis. It may be useful to illustrate the range of variation found in the languages discussed in this volume. Inflected reflexives in Turkish (Kornfilt this volume and 1997) provide an example of a form that can be used as a reflexive within the local domain, but which is a pronominal when used beyond that domain. Another such example is Malay (Cole and Hermon, 1998a, 1998b; and, to a lesser extent, Cole, Hermon, and Lee, this volume). In these lan-
Introduction
xvii
guages apparent long-distance reflexives show none of the characteristics of bound anaphors: outside of the local domain the antecedent does not need to c-command the "reflexive," extra sentential antecedents are possible, and strict readings are easy to come by under VP ellipsis. This is illustrated from Malay: (4) Ahmadi tahu Salmahj akan membeli baju untuk diri-nyai/j/k.5 Ahmad know Salmah will buy clothes for self-3SG 'Ahmad knows Salmah will buy clothes for him/herself.' (5) a. [NP Bapa Sitij]i tidak suka dirinyai/j/k. father Siti not like self.3SG 'Siti's father does not like her/himself/him.' b. [NP Ibu Alijli telah mengenalkan dirinyai/j/k. mother AH already introduce self.3SG kepada kawan-kawan saya. to friends my 'Ali's mother has introduced him/herself/her to my friends.' In (4) dirinya can be interpreted either as a local reflexive, as referring to the matrix subject, or as referring to some third individual in the discourse. The examples of (5) show that a c-commanding or non-c-commanding antecedent, or an extra sentential, discourse antecedent are all possible. Turning to VP ellipsis, we see that dirinya allows both strict and sloppy readings, even in local contexts, as is expected with a pronoun: (6) John nampak dirinya di dalam cermin; Frank pun. John see self-3psg at inside mirror Frank also 'John saw himself/him in the mirror and Frank did too .' (='Frank saw Frank in the mirror' or 'Frank saw John in the mirror') (7) AH cukur dirinya di dalam bilik air; Bill pun. Ali shave self-3psg at inside bathroom Bill also 'Ali shaved himself/him in the bathroom. Bill did so too.' (='Bill shaved Bill' or 'Bill shaved Ali') To conclude, the relationship between the seeming long-distance reflexive and its antecedent in Malay would appear to be either coreference or binding, as is the case for personal pronouns.6 In contrast, in Kannada long-distance reflexives require c-commanding antecedents, cannot take extra sentential antecedents, and require sloppy readings under VP ellipsis (Lidz, this volume; and Amritavalli, 1999). Similar facts are found in Hindi-Urdu (Davison, this volume, and 1999). Example (8) illustrates the VP ellipsis facts in Hindi-Urdu (Davison, 1999): (8) Guatami [apnee (aapj-koo caalaak] samajhtaa hai, aur Guatam self's self-dat smart consider-Impf is and
xviii
Long-distance Reflexives: The State of the Art
vikramj bhii f. Vikram also 'Guatam considers himself smart, and so does Vikram' (=Vikram considers Vikram smart; *Vikram considers Guatam smart) In these languages long-distance reflexives have the properties associated with bound anaphors, and, like other bound anaphors, appear to require a binding relationship with antecedents. The situation is more complex in Icelandic (Reuland, this volume). In Icelandic the reflexive form sig can occur with long-distance antecedents in both infinitival and subjunctive clauses. However, sig has different properties in each environment: When sig and its antecedent are found within an infinitival clause, sig behaves like a bound anaphor, but when the domain is a subjunctive clause, it behaves like a pronominal. Thus, in some environments sig must be related to its antecedent by binding, while in other environments a relationship of coreference is possible. Finally, in Chinese (both Mandarin and Teochew) the simplex reflexive (ziji in Mandarin/kaki in Teochew) normally shows the distribution of a bound anaphor: The antecedent must c-command (or subcommand) the reflexive, and VP ellipsis yields only (or predominantly) sloppy readings, as in Hindi-Urdu (Cole, Hermon, and Lee, this volume; and Huang and Liu, this volume; as well as earlier works by many authors). However, in limited discourse contexts ziji can take extra sentential discourse antecedents, as is illustrated in (9), taken from Pollard and Xue (this volume):7 (9) Zhangsarii zhidao neijian shi yihou hen qifen; Lisii shuo Zhangsan know that-CL thing after very angry Lisi say neixie hua mingming shi zai he ziji i/j zuodui. those word obviously is being with self against 'Zhangsani was very angry when he learned that. By saying those words Lisij was obviously acting against himselfjj/himi.' The Icelandic and Mandarin facts suggest that the same reflexive form can be a bound anaphor in one grammatical and discourse context and a "free anaphor" with pronominal properties in another. To summarize, there appear to be at least the following types of "longdistance reflexives": 1) long-distance bound anaphors, which show the distribution of bound variables (illustrated by Chechen/Ingush, Chinese, Hindi-Urdu, Kannada), 2) forms which are used as reflexives locally and as pronominals non-locally (illustrated by Turkish and Malay) and 3) forms that are "primarily" bound anaphor reflexives, but which can be used nonlocally in specific syntactic and discourse contexts (Chinese "free anaphors," Icelandic subjunctives and long-distance uses of English reflexives
Introduction
xix
(Zribi-Hertz). If "long-distance reflexives" do in fact display the diversity just claimed, several conclusions follow: 1) The distribution and properties of each type of "long-distance reflexive" (anaphor, pronominal, mixed) need to be investigated separately. 2) The factual contradictions both in and across languages may be resolved by distinguishing properly the various types of "long-distance reflexives." 3) A major problem in understanding the phenomenon of long-distance anaphora is the determination of the grammatical and discourse conditions that license each type of longdistance anaphor. The successful solution to this problem may provide the basis for the development of a theory of long-distance anaphora with sufficient power to predict the properties of a reflexive from its grammatical and discourse context. We shall return in Section 10 to a consideration of what might constitute the basis for such a theory.
5. THE ROLE OF DISCOURSE 5.1. Variation among Languages and Dialects with Bound Anaphor long-distance Reflexives A second theme that recurs in many of the chapters is an attempt to characterize the role of discourse/pragmatics versus syntax in the licensing of Long-distance reflexives. While many aspects of the problem remain unclear, many new facts about the relationship of discourse factors and long-distance reflexives emerge in the contributions to the volume. First of all, the role of discourse varies considerably from language to language and from form to form within some languages. Nichols shows that Chechen and Ingush exhibit bound anaphor-type long-distance reflexives. In these languages, however, there is no requirement that the antecedent satisfy logophoric or other discourse conditions. In many other languages discourse adds restrictions over and above those of the grammar. For instance, Amritavalli (1999) notes that in Kannada (which also exhibits bound anaphor long-distance reflexives) the antecedent for the reflexive tannu must be aware of the state or event described, leading to a preference for the pronoun avannu over the long-distance reflexive tannannu in sentences like (10) (Amritavalli's 119): (10) Sittei j ?tannannuii*j*k/avanannui kaaNalu bandaaga Sita self-ace he-acc to see came then raajui sattu hoogidda Raju was dead 'Rajuj was dead when Sitaj came to see ?selfij*j*k / hinij.'
XX
Long-distance Reflexives: The State of the Art
Even within the same language, different dialects and varieties may vary with regard to the discourse properties of long-distance reflexives. For example, Cole, Hermon, and Lee (this volume) show that there is considerable variation in the discourse conditions on long-distance reflexives in the Chinese dialects of Singapore. In these dialects (in most contexts) the reflexive has the distribution of a bound anaphor, but the logophoric conditions on Singapore Teochew are much stricter than those on Singapore Mandarin: In Teochew the antecedent of a long-distance reflexive must be aware that he, himself, is carrying out the action described (a de se requirement), while in Singapore Mandarin self-awareness on the part of the antecedent is not required. However, weaker "PIVOT" conditions (Sells, 1987) hold in both Singapore Mandarin and Singapore Teochew. Turning to other chapters that deal with Mandarin in the volume, Huang and Liu claim that a de se requirement constrains the relationship between matrix subject antecedents and object ziji in complement clauses. In contrast, Pollard and Xue claim that such examples (illustrated by (11)) are not subject to logophoric requirements:8 (11) Zhangsarii zhidao \Lisi-j renwei [Wangwuk zui xihuan ziji i/j/k ]]. Zhangsan know Lisi think Wangwu most like self 'Zhangsanj knows that Lisij thinks that Wangwu likes himself/himi/j most.' Pollard and Xue explicitly claim with respect to (11) that "the key fact about such cases is that any commanding subject qualifies as the antecedent for ziji on the strength of its syntactic prominence alone [emphasis ours]." We take such apparent factual contradictions among authors writing about Chinese as indicative of the fact that the discourse conditions vary in subtle ways from dialect to dialect, and from speaker to speaker within a single dialect. It would appear that the extent of the variation has not been recognized adequately in the literature. 5.2. Absence of Logophoric Conditions on Pronominal Long-distance Reflexives A second fact about discourse that emerges from the chapters in this volume is that, while logophoric conditions may (but need not) restrict bound anaphor-type long-distance reflexives, they do not affect pronominal longdistance reflexives. Thus, in Malay "long-distance" dirinya does not require any special perspective or self-awareness (Cole and Hermon, 1998a, 1998b; and Cole, Hermon and Lee, this volume). Rather, it has the same discourse properties as a personal pronoun. We believe that a similar lack of logophoric restrictions occurs for long-distance inflected reflexives in Turkish
Introduction
xxi
(as shown by Kornfilt, this volume, and 1997, as well as for Middle English and Old English pronouns (Keenan, 2000 and 1976), which, as discussed in footnote 6, can also be used for local coreference). Thus, forms that are indeterminate between pronouns and reflexives do not seem to be subject to logophoric conditions. 5.3. Logophoric "Conversion" of Reflexives to Pronominals In contrast to the situation in Malay and Turkish, as well as to that in Chechen/Ingush, Hindi-Urdu, Kannada etc., in languages like English, and in Icelandic subjunctives, a form that in most environments is exclusively a bound anaphor is given special license by a combination of syntax and discourse to function as a pronominal. As was shown by Reinhart and Reuland (1993 and earlier works) and by Zribi-Hertz (1989), this occurs mostly when a reflexive occurs in a non-argument position (though the syntactic conditions vary greatly from language to language) and when the antecedent satisfies logophoric conditions:9 (12) a. Philipj was supposed to be fooling (...), because Desiree (...) had undoubtedly explained to them the precise nature of her relationship with himself. (Zribi-Hertz 1989: (43b)) b. But Rupert; was not unduly worried about Peter's opinion of himselfi. (Zribi-Hertz 1989: (46b)) In the examples of (12) the reflexive occurs in a position in which it is not an argument of the verb. From a discourse perspective, the sentences present the situation as pictured in the mind of the antecedent for the reflexive: the use of the reflexive indicates to the reader that the scenes are being viewed from Philip and Rupert's perspective. Note that in the case of sentences like (12b) VP ellipsis allows either a strict or a sloppy interpretation:10 (13) Rupertj was not unduly worried about Peter's opinion of himself; nor was Fredj. The elliptical clause of (13) can be understood to mean either (14) or (15): (14) Nor was Fred unduly worried about Peter's opinion of Rupert (strict reading). (15) Nor was Fred unduly worried about Peter's opinion of Fred (sloppy reading). The fact that both strict and sloppy interpretations are easily available shows that in English, as Icelandic, the reflexive form has taken on the referential properties of a pronoun.
xxii
Long-distance Reflexives: The State of the Art
Although the reflexives show the distribution of a pronoun in both Malay and English, the use of reflexives differs in an important way. In English and in Icelandic subjunctives, as discussed by Reuland (this volume), a reflexive can only be used as a pronominal in the specific licensing environments just described, while in Malay, Turkish, etc. no special licensing conditions are necessary. This suggests that the Malay/Turkish and the English/Icelandic cases should not be conflated, and that the English/Icelandic type of long-distance reflexive might usefully be viewed as a reflexive that is "converted" to a pronominal under special syntactic and discourse circumstances,11 while the Malay/Turkish cases are inherently pronominal (or are indeterminate between an anaphor and a pronominal). (We will discuss and extend Reuland's analysis of the Icelandic facts further in Section 10.)
6. LOGOPHORIC PRONOUNS AND LOGOPHORIC RESTRICTIONS ON LONG-DISTANCE REFLEXIVES We have seen that the various types of long-distance reflexives (bound anaphor, pronominal, and "de-anaphoric" pronominal reflexives) differ with regard to whether special discourse conditions apply to their use: bound anaphors may be subject to "logophoric" discourse requirements but need not be, pronominal (or indeterminate) forms are not subject to these requirements, and "converted" reflexives are (we hypothesize) always subject to such requirements. We would like to turn to an examination of whether the logophoric requirements on long-distance reflexives are the same as those on logophoric pronouns. The term "logophoric pronoun" was coined by Hagege (1974) to describe a special subgroup of personal pronouns used in reported speech in some African languages to refer to the individual whose speech is reported or to the addressee of that speech. For example, in Mupun, as described by Frajzyngier (1993), there are three sets of personal pronouns as well as a set of reflexive anaphors:12 (16) Personal Pronouns in Mupun
Set A 3MSG 3FSG 3PL
Subj wu wa mo
Set B Obj wur war
Subj. cfi de du
Set C
Obj. din de dun
Subj. gwar paa nuwa
Introduction
xxiii
(17) Reflexives in Mupun Singular Plural Ip sen sun 2M sak suk 2F sik 3M sin sut 3F set In complements of the verb sat, pronouns of Set A indicate disjoint reference with the subject of the matrix clause: (18) wu/wa/mo sat n wu/wa/mo ta dee n-jos. he/she/they say COMP he/she/they stop stay PREP-Jos 'Hel/shel/theyl said that he2/she2/they2stopped over in Jos.' In contrast, the Set B pronouns are used in indirect speech to indicate coreference with the matrix subject:
(19) wu sat n di n nas an 3M say COMP 3M FUT beat 1SG 'Hel said he1 will beat me.' Set C is used to indicate that the pronoun refers to the addressee of the matrix clause. If the reference is to other persons, the pronoun must be drawn from Set A: (20) a. n- sat n-wur na wur ji ISG-say PREP-3SG COMP 3SG come 'I told himl that he2 should come.' b. n-sat n-wur ns gwarji ISG-say PREP-3SG COMP 3SG come 'I told him! that hej should come.' The logophoric pronouns are used to indicate coreferentiality or disjoint reference with the subject or addressee of superordinate verbs of speaking. Whenever coreference is within the same clause, a reflexive pronoun is employed: (21) n-seet nfaar n-sen ISG-buy shirt PREP-REFL:1SG 'I bought a shirt for myself.' (22) mu-sdrep nhsr-mo n-sun IPL-buy-PL shirt-PL PREP-REFL:1PL 'We bought shirts for ourselves.'
xxiv
Long-distance Reflexives: The State of the Art
(23) wu cet mbise n-sin 3M cook food PREP-REFL:3M 'He cooked for himself.' To summarize, in a "classic" logophoric system like that described by Frajzyngier for Mupun, logophoricity is a property of pronouns rather than of bound anaphors, and is licensed by whether the matrix verb is a verb of saying: In Mupun verbs of thinking like pan 'think, remember' do not permit Set B and C (logophoric) pronouns, but rather require pronouns from Set A (non-logophoric): (24) wu ben «9 wu pan an 3M think COMP 3M remember 1SG 'He thinks he remembers me.'
According to Frajzyngier (1993, chapter 20), Set A pronouns are required in (24), and can be used to indicate either coreference or disjoint reference. This is unlike the case when the matrix verb is a verb of saying. In that case, the use of a Set A pronoun indicates disjoint reference:
(25) wu sat ns wu nas an he said COMP 3M beat 1SG 'Hel said that he2 beat me' According to Frajzyngier, the same pattern obtains with Set C pronouns, but examples showing the distinction between verbs of saying and verbs of mental activity like 'think' are not provided in the grammar.13 Logophoric conditions on long-distance reflexives differ in a number of ways from classic logophoric pronominal systems like that illustrated by Mupun. First, as was shown above, the elements in question belong to different grammatical classes. Pronominal long-distance reflexives (like those in Malay and Turkish) are not subject to logophoric requirements. Thus, it cannot be true that when reflexive forms have "pronominal" usage, they become logophoric pronouns. The logophoric requirements appear to be optionally present for bound anaphor long-distance reflexives and obligatorily present for reflexives that have undergone "conversion" to pronominals, as in the case of Icelandic subjunctives and English long-distance reflexives. Second, the discourse restrictions on reflexives are different from those on true logophoric pronouns. The choice of logophoric pronouns is determined by whether the higher verb is a verb of saying and whether the antecedent is the subject of that verb or the addressee. In contrast, long-distance reflexives are well formed when the higher predicate is one of saying or of thinking. While a special set of pronouns (Set C) exists in Mupun to indicate coreference with the addressee, no such possibility exists in the case of long-distance reflexives: long-distance reflexives never indicate coreference
Introduction
XXV
with the addressee. These facts suggest that the analogy between long-distance reflexives and logophoric pronouns collapses when the two are compared in a fine grained fashion, and the origin of the discourse conditions on long-distance reflexives is distinct from that of logophoric pronouns. Furthermore, there will be separate sources for the "logophoricity" of bound variable and "conversion" long-distance reflexives. These questions are explored in Section 10. To conclude this section, while the term "logophoricity" appears to be too well established to banish it from discussions of long-distance reflexives, it is important to recognize that there is strong evidence against the hypothesis that long-distance reflexives are covert logophoric pronouns. Furthermore, the system of logophoricity found with "classic" logophoric pronouns is quite different from that found with longdistance reflexives of various types.14
7. LOGOPHORICITY AND ATTITUDES DE SE We would like to turn now to the question of whether there appears to be any unified use of the concept of logophoricity with respect to bound variable long-distance reflexives. Sells (1987) claims expressly that the term has been used for three different though related concepts. Sell's SOURCE is similar to the use of "logophor" in the literature on logophoric pronouns. The SOURCE is the source of speech, e.g., the subject of a higher verb of saying. Sell's second type of logophoricity, SELF, is the individual whose mental state the sentence describes (subject of verbs of thinking), and PIVOT is the center of deixis or perspective for the sentence (the reference point for indexicals). What these notions share is that they refer to the individual whose point of view or general state of consciousness is expressed by the sentence. According to Sells, languages differ with regard to which type of logophoricity licenses long-distance reflexives, though there are claimed to be implicational relations among different types of logophoricity. In the spirit of Sells (1987), Huang and Liu attempt a unified treatment of these types of logophoricity in the domain of long-distance reflexives by proposing that the core meaning of logophoricity is SOURCE. This can be extended in some languages to also include SELF and in other languages to PIVOT as well: (26) SOURCE C SELF C PIVOT Thus, in some languages only SOURCE will license the use of a longdistance reflexive (e.g., with verbs of saying); in others both SOURCE and SELF will license long-distance reflexives (with verbs of saying or verbs of thinking); in yet others PIVOT, in addition to SOURCE and SELF, may
xxvi
Long-distance Reflexives: The State of the Art
also license long-distance reflexives (e.g., with the aid of deictic expressions like come, go, here, there, etc.). According to Huang and Liu, the logophoricity condition (being a SOURCE, SELF, or PIVOT as required in a given language) is a necessary but not a sufficient condition for the antecedent of a long-distance reflexive. Thus, as noted by Pollard and Xue (this volume), there are instances of SOURCE that do not license long-distance ziji in Chinese: (27) Zhangsarii cong Lisi} chu tingshuo Wangwuk bu xihuan zijii/*j/k. Zhangsan from Lisi place hear Wangwu not like self 'Zhangsanj heard from Lisij that Wangwuk does not like himi/*j/himselfk.' In (27) Lisi is the SOURCE of communication, but this is not sufficient to permit the use of long-distance ziji. According to Huang and Liu, the relation of a long-distance reflexive to its antecedent is mediated through a structure of predication that normally requires the reflexive to be c-commanded by the antecedent. In (27) the SOURCE Lisi does not license the long-distance ziji because it does not stand in a predication relation with a suitable predicate that includes the reflexive. The range of cross-linguistic variations allowed for under (26) is illustrated by the two Chinese dialects presented in Cole, Hermon, and Lee. Cole, Hermon, and Lee show that, whereas in the Teochew dialect spoken in Singapore, the long-distance ziji requires its antecedent to be SOURCE or SELF, in the Mandarin spoken in the same city PIVOT alone may license long-distance reflexives. Huang and Liu try to unify the various types of logophoricity by taking them to represent a progressive liberation of the notion of 'core logophoricity'—SELF being an extended (or "virtual") SOURCE, and PIVOT being an extended (or "virtual") SELF. This view they share, in part, with Cole, Hermon, and Lee, and with Pan. All these authors explore the possibility that SOURCE and SELF (and for Huang and Liu, also PIVOT) can be reduced to the notion of de se. Developing an idea suggested by Chierchia (1989) for Italian, Huang and Liu note that the restriction on ziji in the variety of Mandarin they report on, is not merely that the antecedent for ziji must be the SOURCE of communication or the individual whose mind the sentence describes (SELF), but, rather, there is a stronger requirement: The individual in question must be aware that the sentence is a description of an event in which he himself is a protagonist (a de se restriction), or more precisely, that the individual actually ascribes, or is disposed to ascribe, to himself/herself the property denoted by the predicate containing the reflexive. Following Chierchia, the de se restriction is taken to be applicable to long-distance reflexives generally. For instance, Chierchia exam-
Introduction
XXV11
ines the distinction between long-distance reflexives and pronouns in Italian, and notes that sentence (28) is a contradiction, while (29) is not: (28) Pavarotti crede che i propri pantaloni siano in fiamme. Pavarotti believes that the self pants are in flame Ma non si e' accorto che i pantaloni sono i propri. but not realize that the pants are the own 'Pavarotti believes that self's pants are on fire, but he hasn't realized that the pants are his own.' (29) Pavarotti crede che i suoi pantaloni siano in fiamme. Ma Pavarotti believes that the his pants are in flame but non si e' accorto che i pantaloni sono i propri. not realize that the pants are the own 'Pavarotti believes that his pants are on fire, but he hasn't realized that the pants are his own.' The contradictoriness of (28) disappears when the personal pronoun suoi is substituted for proprio. That is, the use of the long-distance reflexive proprio requires that Pavarotti realize that it is his own pants that are on fire. No such requirement holds for the personal pronoun suoi: In (29) Pavarotti must realize of a certain individual that the individual's pants are on fire. It is not necessary, however, that Pavarotti identify that individual with himself. A de se restriction similar to (28) applies in Mandarin, as discussed by Huang and Liu and by Pan, as well as in Teochew discussed by Cole, Hermon, and Lee. The existence of a de se restriction in Italian and in certain Chinese dialects suggests that neither SOURCE nor SELF adequately characterizes the discourse requirement in these languages. Rather, an apparent SOURCE/SELF discourse restriction may in fact be an artifact of a stronger de se requirement. While the chapters in this volume show that languages differ with respect to whether bound anaphor long-distance reflexives are subject to a de se restriction, the presence of such a restriction may explain instances of apparent SOURCE/SELF restrictions in languages in which they occur. We shall return to the origin of de se below.
8. THE DIVIDING LINE BETWEEN LOCAL AND LONG-DISTANCE REFLEXIVES We will turn now to a new topic, one which figures in several of the chapters. It is taken as given in discussions of long-distance reflexives that there is a well defined notion of what constitutes a local reflexive and what anaphoric
xxviii
Long-distance Reflexives: The State of the Art
relations are non-local. There are in fact two widely held notions of locality in the generative literature on reflexives. The traditional Binding Theory (Chomsky, 1981, and subsequent works) treats the local domain ("governing category") as, roughly, consisting of the clause plus the subject of a non-finite complement clause. In a radical departure from the Chomskyan approach, Reinhart and Reuland (1993 and earlier work) have argued that the core use of reflexives is to mark coargumenthood, not intraclausal coreference.15 Thus, the expected division would be between anaphors used to mark coargumenthood and other anaphors, rather than between clause internal (or governing category internal) and clause external (or governing category external) anaphora. Reinhart and Reuland's proposal is provided immediate support by the fact that many Germanic languages distinguish between different anaphoric forms along the lines predicted by Reinhart and Reuland's theory. For example, in Dutch zichzelf 'himself' is restricted to coargument positions:16 (30) Max haat zichzelf. Max hates selfself 'Max hates himself.' (31) *Max haat zich. Max hates self 'Max hates himself.' (32) Max hoorde mij over zich praten. Max heard me about self talk 'MAX heard me. talk about him '
(33) *Max hoorde mij over zichzelf praten. Max heard me about selfself talk 'Max heard me talk about him.' (34) Jan{ zag Pietj de spullen naast zich{ / *zichzelf i neerleggen. Jan saw Piet the gear next to self selfself put 'Jan saw Piet put the gear next to him.' Examples (30)-(34) show that the "true reflexive" zichzelf 'himself is restricted to coarguments while the "pronominal reflexive" zich is permitted for both intraclausal non-coargument reflexives and for long-distance reflexives. Similar facts occur outside Germanic as well. See Lidz (this volume and 1996). While the pattern just illustrated appears to provide prima facie support for Reinhart and Reuland's division of anaphora into coargument and non-coargument anaphora, the authors of several chapters provide arguments against the definition of locality in terms of co-argumenthood. First, Lidz argues from Kannada examples much like (30)-(34) that the distinc-
Introduction
xxix
tion between zich and zichzelf-type anaphors is not a distinction between coargument and non-coargument anaphora. Rather, the two types of anaphors differ in meaning. The apparent anti-locality of the zich-type anaphor is due to a distinction between "pure reflexivity" (zich-type) and "near reflexivity" (zichzelf-type). Lidz argues that pure reflexives are compatible only with inherently reflexive predicates or with predicates that have been marked as reflexive by a verbal reflexive marker (as in Kannada). Thus, zich-type reflexives are predicted to occur only with reflexive predicates like 'shave' or in cases of non-coarguments, for which the distinction between pure reflexivity and near reflexivity is not relevant. Lidz argues that this semantic distinction removes an apparently strong empirical support for the division between coargument and non-coargument anaphora. In evaluating the force of Lidz's arguments, it is not clear that the facts he adduces are not tangential to the issue of the nature of locality. In any case, Lidz provides additional evidence that the distinction between zich-type and zichzelf-type anaphors is not devoid of semantic content. The question of whether the dividing line between local and non-local should be the governing category or the coargument/non-coargument distinction is considered in two additional chapters. Hestvik and Philip present evidence from Norwegian child language that binding of non-coarguments within the simple clause (locative PPs and possessive reflexives) constitutes an instance of "core binding" while binding into picture NPs is an instance of "logophoric" binding. Thus, on the assumption that "core binding" is equivalent to the local domain, from the perspective of child language the local domain appears to include non-coarguments within the simple clause. A similar point is made by Huang and Liu, who show that "local" reflexives in Mandarin are not subject to any logophoric restrictions (neither de se requirements nor PIVOT-based restrictions like the Blocking Effect). What counts as "local" for this purpose includes not only non-coarguments like locatives and benefactives, but also reflexives with subcommanding rather than c-commanding antecedents in the same clause: In both Mandarin and other Chinese dialects, when the head of a noun phrase is inanimate (and, hence, in Chinese, not a potential antecedent for a reflexive), the specifier of that noun phrase can serve as antecedent for the reflexive: (35) Zhangsani de jiaoao ziji-le ziji^. Zhangsan's DE arrogance hurt-ASP self 'Zhangsanj's arrogance harmed himi;.' In (35) the non-coargument Zhangsan is the antecedent for ziji. There is, however, persuasive evidence that ziji does not need to meet any logophoric restrictions: In (35) the antecedent is neither SOURCE nor SELF. Furthermore, the Blocking Effect is not exhibited:
XXX
Long-distance Reflexives: The State of the Art
(36) Zhangsani de biaoqing gaosu wo-j [ziji i/*j shi wugude]. Zhangsan DE expression tell me self is innocent 'Zhangsan's expression tells me that he is innocent.' Thus, the absence of logophoric effects provides strong reason to believe that the relationship between ziji and its antecedent is local—despite the fact that Zhangsan is not an argument of the same predicate as ziji. We conclude that the contributions to this volume provide arguments that there is a division between local and long-distance reflexives along the lines of the locality domain as defined in Chomsky (1981) and related work, the traditional notion of "governing category." We would emphasize, however, that evidence for the need for the notion governing category does not in itself provide evidence against the relevance of co-argumenthood for explaining the distribution of zich and zichzelf-type reflexives, as proposed by Reinhart and Reuland (1993). Thus, the evidence discussed in this section does not constitute a general argument against the use of coargumenthood to define one type of locality. It does, however, appear to show that coargumenthood does not constitute a replacement for the notion of the governing category.
9. TYPOLOGICAL CHARACTERISTICS OF LONG-DISTANCE REFLEXIVES We would like to turn now to a review of the status of the typological properties traditionally associated with long-distance reflexives: monomorphemicity, c-command/subject orientation and the Blocking Effect. 9.1. Monomorphemicity It was noted early in the literature that long-distance reflexives are typically monomorphemic while local reflexives tend to consist of more than one morpheme (Pica, 1987): (37) Credo [che Marioi sostenga [che tu abbia parlato di I-believe that Mario claims that you have spoken of set e delta sua famiglia in TV]]. self and of-the his family on TV 'I believe that Mario claims that you spoke about him and his family on TV (Giorgi 1984) (38) *Giannii pensava [che quella casa appartenesse Gianni thought that that house belonged
Introduction ancora a se
XXXI stessoi].
still to self self 'Gianni thought that that house still belonged to him.' (Giorgi 1984) In (37) from Italian the monomorphemic reflexive se takes a long-distance antecedent while in (38) the bimorphemic se stesso is ill formed with a long-distance antecedent. Similar facts continue to be discovered as new languages are examined. For instance, Davison (1999) summarizes the facts of Hindi-Urdu as follows: "If the simple reflexive [apnee] is contained within a nonfinite embedded clause, it may be coindexed with subject antecedents in higher clauses or with the local subject. ... In a nonfinite clause, the complex reflexive apnee aap and the reciprocal have only a local c-commanding subject antecedent. ..." There are, however, a variety of counterexamples to the claim that only simplex reflexives can have long-distance interpretations. For instance, the Malay reflexive dirinya consists of two morphemes, diri- and a third-person marker -nya. Yet, as was seen above, long-distance interpretations are possible. Similarly, the inflected reflexives kendis Ikendileri in Turkish (Kornfilt, this volume, and 1997) are clearly bimorphemic, consisting of kendi 'self plus -si 'third-person singular" or -led 'third-person plural'. Furthermore, long-distance uses of English reflexives are clearly bimorphemic. We do not, however, interpret such examples as counterexamples to Pica's generalization regarding the monomorphemicity of long-distance reflexives. Rather, we take examples of this type to show that the generalization that long-distance reflexives are monomorphemic applies only to those "long-distance reflexives" that are bound anaphors, and not to those that are pronominals. To the best of our knowledge, all bound anaphor longdistance reflexives are monomorphemic.17 It does not, however, appear to be true that all monomorphemic reflexives are long-distance. For example, the German reflexive sich consists of a single morpheme like the Icelandic reflexive sig or the Italian reflexive se. While these forms are all restricted to third-person antecedents, there does not appear to be a division into two morphemes, one meaning 'self and the other 'him' as is the case for English himself or Malay dirinya. However, German sich is local while Icelandic sig and Italian se take long-distance antecedents. Thus, monomorphemicity appears to be a necessary but not a sufficient condition for bound anaphor long-distance reflexives. 9.2. C-command and Subject Orientation We saw earlier that c-command and subject orientation are restricted to bound anaphor long-distance reflexives. It has been noted in the literature (e.g., Xu, 1994 inter alia and Yu, 1996 inter alia) that there are a variety of
XXxii
Long-distance Reflexives: The State of the Art
apparent counterexamples to subject orientation. Such counterexamples might be taken to indicate that subject orientation has a functional rather than a grammatical explanation. In this section we would like examine four questions related to this general issue: (1) Is subject orientation derivable from logophoric discourse conditions on long-distance reflexives, or does it appear to be a grammatical requirement? (2) Can subject orientation be reduced to a requirement that (bound anaphor) reflexives refer to the noun phrase with the highest rank on the following thematic hierarchy? (39) Thematic Hierarchy Agent < Experiencer < Theme < Other (3) In the event that the answer to (2) is negative, can subject orientation be reduced to the requirement that (bound anaphor) long-distance reflexives refer to the noun phrase with the highest grammatical function on a hierarchy of grammatical functions? (40) Hierarchy of Grammatical Functions Subject < Direct Object < Indirect Object < Oblique (4) In the event that these functional explanations fail, can subject orientation can be reduced to c-command? We shall argue that there is evidence against (1) and (2) and (3), but that (4) is consistent with the available evidence, at least if c-command is understood as c-command/subcommand. The main evidence that subject orientation is not reducible to a discourse requirement is typological. We saw earlier that languages with bound anaphor long-distance reflexives differ regarding whether and which logophoric requirements are respected. Thus, in the languages discussed in this volume, Chechen and Ingush (Nichols) do not manifest any logophoric requirements. In contrast, in Teochew (Cole, Hermon, and Lee, this volume) there are strong logophoric restrictions on the relationship between the reflexive and its antecedent. Despite this striking difference between Chechen/Ingush and Teochew, both manifest subject orientation. A similar point can be made on the basis of the comparison of Singapore Teochew and Mandarin in Cole, Hermon, and Lee (this volume and 2000). Although Singapore Mandarin and Teochew differ greatly in the nature of the logophoric requirements found, long-distance reflexives are subject oriented in both cases. Thus, there is no correlation between the existence of subject orientation and logophoric requirements. We turn next to the question of whether subject orientation could be reduced to the requirement that bound anaphor long-distance reflexives must refer to the noun phrase that is highest on the thematic hierarchy in the domain chosen. (We assume that in a multiclausal structure the choice of the
Introduction
XXX111
clause in which the antecedent occurs is independent of the choice of which noun phrase in that clause serves as antecedent.) This hypothesis is given some support from the fact that non-nominative experiencers can serve as antecedents for bound anaphor long-distance reflexives. There is, however, clear evidence that this hypothesis cannot be correct, at least for Mandarin: Cole and Wang (1996) show that while objects in postverbal position cannot serve as antecedents for long-distance ziji, preverbal objects are possible antecedents. (41) Wangwui shuo Zhangsan j zengsong gei Lisik yipian guanyu Wangwu say Zhangsan give to Lisi one about zy/i/j/*k de wenzhang. self DE article 'Wangwu says that Zhangsan gave an article about him/himself to Lisi.' (42) Zhangsan^ yiwei Lisi^ hui ba Xiao Mingk dai hui Zhangsan thought Lisi will BA Xiao Ming take back zijimde jia. self's DE home 'Zhangsan thought Lisi would take Xiao Ming back to his home.' Example (41) shows that ziji cannot refer to the addressee, the postverbal object of zengsong gei 'give', but rather only to the matrix subject, Wangwu, or the complement subject, Zhangsan. In contrast, in (42), in which the theme occurs in preverbal position, ziji can refer to the matrix subject, Zhangsan, the complement subject, Lisi or the theme, Xiao Ming. The fact that within the domain of the complement clause either the agent or the theme can be the antecedent for ziji shows that "subject orientation" cannot be reduced to a requirement that the antecedent must be the highest noun phrase on the thematic hierarchy. The same examples provide evidence against the proposal that subject orientation can be reduced to the requirement that the reflexive must refer the highest noun phrase on a hierarchy of grammatical functions (at least for Mandarin). In (42) the complement subject, Lisi, outranks the complement object, Xiao Ming, on a hierarchy of grammatical functions. Despite that fact, the object is a possible antecedent for ziji. Turning now to the hypothesis that subject orientation can be reduced to c-command, there appears to be evidence that this is the case. Under current analyses of the ba 'preverbal object' construction, ba is a functional head and the "object" of ba is the specifier of the maximal projection which is the complement of ba, as shown in the tree (adapted from Cole and Wang, 1996) below:
XXxiv
Long-distance Reflexives: The State of the Art
(We show the complement of ba as AgrP for the sake of concreteness, but the choice of projections is not relevant to our point here. Another possibility is PREDP.) We shall assume for the moment 1) that TP is the local domain in which Binding is determined, and 2) that some version of the movement theory of long-distance reflexives applies, under which the reflexive moves to a position in which its antecedent is within the local domain of the reflexive. We shall also, for the moment, assume 3) that ziji adjoins to the head of AgrP (as in Cole and Wang, 1996), though this assumption will be dispensed with below:
Given the resulting tree (roughly (44)), c-command alone would predict that the possible antecedents for ziji are the subject (the specifier of Tense/ AspectP) or the specifier of AgrP (the noun phrase immediately following ba). Thus, both subject orientation and the failure of subject orientation in
Introduction
XXXV
examples like (42) can be shown to be predicted from c-command alone if a movement analysis for ziji is assumed. Furthermore, these predictions are not dependent on the assumption that ziji adjoins to the head of AgrP (the head movement analysis). If, instead, it is assumed that ziji adjoins to VP, the same predictions are made:
Thus, we conclude that under either version of the movement analysis of bound anaphor long-distance reflexives, the reduction of subject orientation to c-command makes more accurate predictions about the distribution of ziji than does an analysis in which subject orientation is a primitive (but see Pollard and Xue, this volume, for an analysis in which subject orientation is argued to be a primitive). We shall discuss the status of the movement analysis in Section 10 below, but clearly the facts just reviewed constitute an argument in favor of movement for Chinese. It remains to be seen, however, whether the argument can be generalized to bound anaphor long-distance reflexives in other languages. 9.3. The Blocking Effect The existence of a Blocking Effect based on person has been discussed primarily in the literature on long-distance reflexives in Chinese. Although the Blocking Effect was analyzed as syntactic in nature in most early studies of Chinese reflexives (e.g., Battistella, 1989; Cole, Hermon, and Sung, 1990; Huang and Tang, 1991; Sung, 1990). It was originally noted by Y.-H. Huang (1984), who suggested that Blocking has a functional origin. The controversy regarding Blocking continues in this volume. While there is general agreement that the Blocking Effect is due, at least in part, to discourse/semantic/pragmatic factors, it remains controversial whether there is a grammatical component to Blocking.
xxxvi
Long-distance Reflexives: The State of the Art
In particular, one question that has been left open by the functional/ discourse account is why the Blocking Effect is observed only in some languages (e.g., Mandarin and Teochew) but not in others (e.g., Chechen/ Ingush, Kannada, Hindi-Urdu, Icelandic or Norwegian). Is it a typological difference in the discourse structure among languages that accounts for the cross-linguistic variations, or is it a typological difference in syntax? Proponents of the functional account have not touched upon this issue. There is some prima facie strong evidence that syntax plays a significant role in accounting for the cross-linguistic distribution of the Blocking Effect, as shown by Cole, Hermon, and Lee (this volume), who relate Blocking in Chinese to the fact that Chinese lacks verb agreement. We will not examine the Chinese internal evidence for and against the agreement theory of Blocking here. This topic is debated in detail in several papers in the volume (Cole, Hermon, and Lee; Huang and Liu; Pollard and Xue; and Pan). We also will not review technical details here. There is, however, a typological prediction made by the agreement theory of Blocking that we would like to examine on the basis of the known cross-linguistic data: The agreement theory of Blocking predicts that syntactic Blocking will be found only in languages with bound anaphor long-distance reflexives that lack verb agreement. This typological prediction is not, to the best of our knowledge, made by any other theory of Blocking. The typological prediction is corroborated by Chinese languages, which lack agreement and manifest Blocking. It is also corroborated by HindiUrdu, Icelandic, Italian, and Kannada, which have verb agreement and, as predicted, lack Blocking. Furthermore, strong support is provided by a comparison of Kannada (Amritavalli, 1999; and Lidz, this volume, and 1996) and Malayalam (Jayaseelan, 1999). The facts regarding long-distance reflexives in these two languages are very similar except that Malayalam has lost verb agreement. As predicted by the agreement theory of Blocking, Malayalam manifests a Blocking Effect like that seen in Chinese, while none is found in Kannada or any of the other Dravidian languages, none of which, other than Malayalam, has lost verb agreement. Thus, the Dravidian facts suggest that the absence of agreement and presence of Blocking in Chinese are not coincidental. While the Dravidian facts provide corroboration for the agreement theory of Blocking, other languages present problems for the theory. First, the Mainland Scandinavian languages have lost verb agreement, but long-distance reflexives are not subject to Blocking. This would appear to be a strong counterexample to the agreement theory. This evidence, however, is not as persuasive as might appear. As is pointed out in Sung (1990), the Mainland Scandinavian languages exhibit agreement in predicate adjective constructions:18
Introduction
XXXV11
(46) Predicate Adjective Agreement a. Gutten er stor. the boy is big 'The boy is big.' b. Huset er stort. the house is big 'The house is big.' c. Husene/Guttene er store. 'The houses/boys are big.' We assume that agreement in predicate adjective constructions is indicative of the presence of agreement features on AGR, which are then transmitted to the predicate adjective. Thus, the Mainland Scandinavian languages do not appear to constitute a genuine counterexample to the agreement theory of Blocking. Finally, Korean may constitute an additional language in which verbs do not agree with their subjects, but for which Blocking is absent.19 However, Korean exhibits honorific agreement: (47) atu-nim-i ttokttokha-si-ta. son-NOM smart-HON-Decl 'The son is smart.' Sentence (47) illustrates obligatory honorific agreement in Korean. As in the case of Mainland Scandinavian predicate adjective agreement, honorific agreement may indicate that abstract agreement features are present on AGR. The presence of such features would be expected to result in the absence of syntactic Blocking in the language. While the apparent counterexamples to the agreement theory of Blocking are inconclusive, it must be recognized that the typological argument in favor of the agreement theory is not entirely convincing. So far the only languages that illustrate the correlation between the absence of verb agreement and the Blocking Effect are Chinese and Malayalam. The empirical evaluation of the hypothesis, therefore, must await the examination of additional languages that lack verb agreement and that exhibit bound anaphor long-distance reflexives. We conclude that Blocking clearly has a discourse component, but the typological facts are not yet available that would determine whether Blocking has a syntactic component as well.
10. THEORIES OF LONG-DISTANCE REFLEXIVES We would like to end this chapter by examining the ways in which the facts described in previous sections might contribute to our understanding
XXXviii
Long-distance Reflexives: The State of the Art
of how long-distance reflexives fit into the anaphoric possibilities for natural language. We have seen that there appear to exist three kinds of entities that might be considered "long-distance reflexives" in a pre-theoretical sense: long-distance bound anaphors, forms that are indeterminate between local reflexives and pronouns, and reflexives that undergo "conversion" from bound anaphor to pronominal in specific syntactic and pragmatic contexts. Reuland (this volume) discusses two types of reflexives (local reflexives and those that have undergone "logophoric conversion" to pronominals), and notes that they are in complementary distribution: local reflexives undergo "conversion" to pronominals when 1) anaphoric binding is blocked in the syntax and 2) they satisfy certain logophoric conditions. This distribution suggests that local reflexives are given pronominal interpretation only if, given the specific syntactic properties of the language in question, they lack an eligible local antecedent as bound anaphors (e.g., per Reuland, binding is blocked by subjunctive mood in Icelandic). That is, the complementarity derives from a principle of Avoid Pronoun, which requires that a form inherently capable of interpretation as either a pronominal (i.e., by coreference) or as an anaphor (by binding) must receive interpretation as a bound anaphor in contexts where an anaphoric interpretation is possible. Why might such a preference for anaphoric interpretation exist? Reuland argues that the complementarity can be explained if we assume that local reflexives constitute a chain, a single syntactic object, perhaps as a result of movement of the reflexive to an inflectional projection. Reuland proposes that interpretation in which syntax and semantics work in tandem (i.e., by chain formation, in which a binding relationship is forced by the existence of the chain) is inherently "cheaper" than interpretation by coreference (in which real world information plays a role in the interpretive process).20 Thus, whenever a single form can either form a chain (and, hence, receive a bound anaphor interpretation) or be interpreted pronominally, the possibility of the more economical chain interpretation should eliminate the possibility of pronominal interpretation. Pronominal interpretation should occur only when, for some reason, chain formation is blocked.21 We would like to extend this proposal from local reflexives to bound anaphor long-distance reflexives. Let us assume that this type of long-distance reflexive is derived by movement. (We abstract away from the issue of whether the movement is head movement or phrasal movement, whether it is adjunction to IP or to AGR and what drives the movement. These questions are dealt with in some detail in Cole, Hermon, and Lee and Huang, and Liu and in the references cited in those chapters.)22 If bound anaphor long-distance reflexives are derived by movement, chain formation must also occur in the case of bound anaphor long-distance reflexives. This predicts that in languages in which long-distance chain formation is possible, a
Introduction
XXXIX
c-command relationship would hold between the reflexive and its antecedent and the reflexive would receive a sloppy rather than a strict interpretation under ellipsis. This extension of Reuland's analysis would predict that only when chain formation is somehow blocked a pronominal interpretation would become possible. This prediction appears to be correct, at least with respect to Chinese. As was pointed out by Yu (1992,1996) and others, when ziji is unbound syntactically it receives a pronominal interpretation that is subject to logophoric requirements:23 (48) Bu qingchu ziji shenme shihou neng qu Meiguo nian shu; not clear self when can go U.S. read book Xiao Li ye zheme juede. little Li also thus feel 'It is not clear when I can go to the U.S. to study. Little Li feels the same way.' In (48) there is no syntactic binder for ziji. As predicted, the absence of a syntactic binder allows ziji to receive a pronominal interpretation, as is shown by the fact that the elliptical clause can receive either a strict or a sloppy interpretation ('Xiao Li says it is unclear when I can go' or 'Xiao Li says it is unclear when he can go').24 Thus, Reuland's theory of economy together with the movement theory of bound anaphor long-distance reflexives predicts that both local and bound anaphor long-distance reflexives will be able to receive a pronominal interpretation only when there is no syntactic binder available. We will next examine the predictions of Reuland's analysis with regard to long-distance reflexives like those in Malay and Turkish forms that are interpreted as pronominals when they are long-distance. Are such forms counterexamples to Reuland's theory of economy? In the context of Reuland's theory, the fact that such forms have the properties of pronominals rather than bound anaphors tells us that, unlike bound anaphor longdistance reflexives, chain formation must be impossible. There is, in fact, evidence with respect to local uses of Malay dirinya that chain formation has failed to apply. Example (49) shows that even in a local context c-command is not required for dirinya: (49) [NP Bapa Siti^ tidak suka dirinyai/j//k. father Siti not like self.3SG 'Siti's father does not like her/himself/him.' Examples with dirinya are to be contrasted with the complex reflexive dirinya sendiri, which requires a local antecedent and which is ill-formed when the potential antecedent does not stand in a c-command relations to the reflexive:
xl
Long-distance Reflexives: The State of the Art
(50) [Ibu Alijli telah menyiramair di dirinya sendirii/*j. mother Ali already splash water on self.3SG alone 'All's mother already splashed water onto herself.' We, therefore, conclude that dirinya, unlike dirinia sendiri, cannot participate in chain formation.25 Since chain formation is blocked with this form, there will not be competition between pronominal and anaphoric dirinya. Thus, it is expected that apparent long-distance reflexives employing forms like dirinya will have the distribution of a pronominal rather than that of a bound anaphor. We have shown that an extension of Reuland's analysis successfully predicts the syntactic properties of each type of apparent long-distance reflexive. We will turn now to the discourse properties. We shall start with bound anaphor long-distance reflexives. These have been argued to consist of chains between the surface site of the anaphor and a position in which a c-commanding antecedent is located within the local domain (roughly, the governing category) of the reflexive. Such a chain may be long-distance (interclausal) when otherwise permitted by the structure of the language and by principles of universal grammar (such as principles of chain formation).26 The problem raised by discourse for this analysis is why logophoric conditions might apply to the relationship between the head of the chain and the antecedent, with variation from language to language. We would like to suggest that the answer may be provided by viewing the chain as uninterpretable unless it is related to an antecedent by a predication relation along the lines suggested in Huang and Liu's contribution to this volume. While a variety of technical problems remain unsolved with respect to this proposal, we propose that discourse principles, which may vary from language to language, determine whether the predication relationship would be subject to logophoric requirements. Thus, in the case of bound anaphor long-distance reflexives, logophoricity may restrict the relationship between the chain and the ultimate antecedent. However, the relationship between the head and tail of the chain would be a syntactic relation that would not be subject to logophoric requirements. In Huang and Liu's terms, predication of a chain structure on the ultimate antecedent establishes a relation at the interface level between syntax and discouse. It is then natural that such a mediation process is subject to both discourse and syntactic restrictions. But the relation between elements internal to a chain structure would be strictly syntactic, not subject to extra-sentential considerations. We turn next to those reflexives that can be used long-distance only when they lack a c-commanding antecedent in their local domain (e.g., English and Icelandic subjunctives). Why are such reflexives obligatorily subject to logophoric requirements when logophoric requirements are optional in the case of movement chains (predication, according to Huang and Liu) and ab-
Introduction
xli
sent entirely in the case of the Malay/Turkish type of reflexive, in which chain formation is blocked? Reuland proposes that the answer can be found by examining the fundamental semantic content of pronouns. The semantic contribution of pronouns is to express orientation vis-a-vis the speech act (speaker, perceiver). This orientation is the lexical content of personal pronouns like / and you. In the case of anaphors that undergo "conversion" to pronominals, orientation is not lexically specified since these forms are radically deficient in lexical content. While "classical" personal pronouns express only speech act orientation, "de-anaphoric" pronouns lack even that lexical content. In order to be well formed, they must take their orientation from the context. That is, they must satisfy conditions of logophoricity. Turning, finally, to "pronominal" long-distance reflexives like those found in Malay and Turkish, Reuland's theory would predict that, unlike sig, they must include a specification of orientation in their lexical content, as must third-person personal pronouns. It is not immediately apparent how the specification of orientation could be demonstrated without circularity, so we will leave this question open for future discussion. To conclude, we have argued that the properties of "long-distance reflexives" derive from a variety of sources, and that economy considerations are, as Reuland (this volume) suggests, responsible for the partial complementary distribution of the forms. We have also argued that "logophoricity" does not have a unified origin. First, the logophoricity found with longdistance reflexives is entirely separate from that found in classical logophoric pronoun systems (Frajzyngier, 1993). Second, bound anaphor long-distance reflexives manifest logophoricity due to de se restrictions, which may themselves be reducible to discourse constraints on predication. Third, the logophoric restrictions on "conversion" long-distance reflexives may be due to the need for pronominals to include a specification of orientation along the lines discussed by Reuland. If this is correct, we would expect that a fine-grained comparison of the conditions on bound anaphor and conversion long-distance reflexives would show that they do not have identical logophoric conditions. Such a study could be conducted fruitfully in Chinese dialects that have both "free anaphor" ziji and de se conditions on the use of ziji as a bound anaphor long-distance reflexive.
ACKNOWLEDGMENTS We would like to thank Martin Everaert, Zygmunt Frajzyngier, David Gil, Eric Reuland, and Li-May Sung for their very helpful comments and suggestions regarding various parts of this chapter. This work was supported by the National Science Foundation (grants SBR-9729519, INT-9423291 and BCS-9121167), by the
xlii
Long-distance Reflexives: The State of the Art
Max Planck Institute for Evolutionary Anthropology, Leipzig) and by a grant from the Netherlands Organization for Scientific Research (NWO) to the University of Utrecht.
NOTES 1
See Cole and Hermon (1998a) for a summary of the typological characteristics of long-distance reflexives. 2 It is a matter of controversy in the Chinese literature whether a third-person subject blocks reference to a first- or second-person subject, or whether blocking occurs only when the intervening subject is first- or second-person. (It is uncontroversial that the Blocking Effect is stronger when the blocker is first- or second-person.) For differing positions on the facts, see Cole, Hermon, and Lee (this volume) on the one hand, and Huang and Liu (this volume) and Pan (this volume) on the other. 3 These chapters display the diversity of views on both the facts and the analysis of Mandarin. 4 But note that this position is denied by Pollard and Xue, who argue against a division between anaphors and pronouns, at least with respect to Chinese ziji. 5 The index k on dirinya indicates that dirinya can refer to someone in the discourse other than Ahmad or Salmah. 6 It is argued in Cole and Hermon (1998b) that dirinya is not lexically specified as either a pronoun or an anaphor. Rather, dirinya lacks specification for the feature [aanaphor]. The existence of forms that are not specified as either pronominal or anaphoric is not unique to Malay. Keenan (1976 and 2000) argues that in both Old English and Middle English there is only a single form that is used for local and interclausal anaphora: (i)
Beowulf line 1473 c.750 sy Pan he hine to gu ege gyred hcefde once hei(nom) hinii (ace) for battle girded had 'Once he had girded himself for battle.'
(ii)
Chaucer, "The Knight's Tale", verse 1384 At Thebes, in his contree, as I sayde, Upon a nyght in sleep as hej hynij leyde
Similarly, in Javanese, while there exist forms that are exclusively used as local reflexives, the personal pronouns can be employed either for clause-internal anaphora (including coarguments) or for interclausal anaphoric reference. 7 This is example (6) in Pollard and Xue. 8 Example (1) in Pollard and Xue (this volume). 9 The following examples are from Zribi-Hertz (as cited by Pollard and Xue (this volume)). The nature and origin of the logophoric conditions are discussed in greater detail in Section 10.
Introduction
xliii
We would like to emphasize that the syntactic environments making possible a free interpretation of anaphoric forms may be quite different cross-linguistically. As was pointed out to us by Eric Reuland, whereas in English the relevant condition is whether the anaphor is a syntactic argument of a predicate, in Icelandic the crucial factor is whether sig is in the domain of a subjunctive verb. As discussed in Reuland's contribution to this volume, the crucial factor is whether something prevents the anaphor from being hooked up to its antecedent by a syntactic process. In Icelandic this is the blocking of chain linking by the subjunctive. In English, this could be, according to Reinhart and Reuland (1993), the impossibility for SELF to covertly move to the predicate from a non-argument position. (Of course, to obtain well-formedness, appropriate discourse requirements must be met as well.) 10 There seems not to be any way to construct an analogous test sentence in the case of (lla). n lt is not important for us whether or not the pronominal reflexive is literally "converted" from an anaphoric reflexive by a process analogous to derivational morphology, or whether this is simply a marked use of the reflexive form. We shall use the term "converted" simply to mean that the pronominal use of the reflexive is somehow regarded as "marked" or "secondary." 12 These tables are adapted from Frajzyngier (1993). All Mupun examples are taken from Frajzyngier as well. Some of the discussion of the examples is based on a presentation made by Frajzyngier at the Max Planck Institute for Evolutionary Anthropology in Leipzig and on personal communications. 13 Examples are, however, provided for verbs of perception: (i)
n-naa wur wur ps ddm n-kaano ISO-see 3sg 3sg PrEP go PREP-K. 'I saw him going to Kano.'
Note that a Series A pronoun wur is used despite the fact that the two instances of wur are coreferential. 14 A very similar point is made by Culy (1997), who argues that true logophoric pronouns are primarily devices for indirect discourse and that the representation of point of view is not a central use of these forms. 15 Our presentation is a considerable simplification of Reinhart and Reuland's position. See Reinhart and Reuland (1993) for a full exposition of their proposal. 16 Examples (29)-(32) are from Lidz this volume. Example (33) is from Everaert (1991). 17 In some cases it is not clear if a form is a bound anaphor reflexive or a pronominal. The string ta ziji in Mandarin appears to be ambiguous between a local bound anaphor and a pronoun ta modified by an emphatic reflexive ziji. Thus, Tang (1985, 1989), Sung (1990) and others have argued that all instances of apparent longdistance ta ziji are in fact instances of the^pronoun ta modified by an emphatic reflexive rather than a complex anaphor. Ittnust be recognized, however, that there are no completely convincing syntactic tests that demonstrate that all cases of apparent long-distance ta ziji are really the pronoun ta modified by ziji. One argument in favor of the proposal that long-distance ta ziji involves pronominal ta, is found in Cole, Hermon, and Lee (2000), published in Chinese, who show that long-distance ta ziji (and its Teochew counterpart i kaki) manifest the
xliv
Long-distance Reflexives: The State of the Art
discourse conditions of the personal pronouns ta and i, and not those of the longdistance reflexives ziji and kaki: While ziji and kaki are subject to various logophoric conditions (see Cole, Hermon, and Lee, this volume), no such conditions apply to long-distance ta ziji and i kaki. This state of affairs would be expected if ta ziji and i kaki are really the third-person pronoun modified by an intensive reflexive, but not if they are bound anaphor long-distance reflexives. (Similar considerations apparently hold for Cantonese as well.) 18 Example (77), from Norwegian, was provided to Li-May Sung by Arild Hestvik (personal communication). 19 The facts about Blocking in Korean are less clear than those regarding Blocking in Chinese. While some speakers have told us that Blocking occurs with the reflexive casin, others have disagreed. Contrary to earlier work on this topic by Cole, Hermon, and Sung (1990) and subsequent papers, we assume here that Blocking does not hold in Korean. 20 Reuland phrases his account in terms of cross-modular operations. A syntactic chain requires only two cross-modular operations while coreference requires four. See Reuland (this volume) for the details of how the evaluation is conducted. 21 There is a point of similarity here in Pollard and Xue (this volume), who propose that reflexives are interpreted on discourse/pragmatic considerations once they are not "bound" syntactically. 22 There are a number of technical problems with the movement analysis of long-distance reflexives, the most troubling of which is the fact that the distribution of reflexives, at least in Chinese, does not respect islands. For possible solutions to this problem for various versions of the movement analysis, see Huang and Liu (this volume), Cole and Sung (1997), and Sung (1990). 23 The presence of logophoric requirements does not distinguish "conversion" long-distance reflexives from bound anaphor long-distance reflexives in those varieties of Chinese in which roughly the same logophoric requirements apply to bound variable ziji as to the use of ziji as a free anaphor as in (79). However, it is notable that even in Singapore Mandarin, the use of ziji as a free anaphor is subject to logophoric interpretation. 24 The presence of the two readings is especially clear if the following dialog is imagined: Speaker: Bu qingchuziji shenme-shihou neng qu Meiguo nian shu. not clear self when can go U.S. read book Xiao Li: Wo ye yi-yang. I also the same Sentence (79) reports this dialog. Reuland speculates that number plays a critical role in determining whether a form is able to undergo chain formation, and, hence, show the distribution of a bound anaphor. Dirinya appears to be a counterexample to the proposal that number is the only feature that could block chain formation: While the reflexive root din- is inflected for person, inflection for number is not possible. It should be noted that number appears to distinguish diri- from personal pro25
Introduction
xlv
nouns. In the pronominal system it is possible to distinguish dia '3P pronoun' (singular or plural) from mereka or dia orang '3PPL pronoun' (obligatorily plural). No such distinction exists for diri-: The only possible form is diri-nya '3P', and number cannot be indicated. We have seen, however, that dirinya does not permit chain formation. Thus, it cannot be the presence or absence of number that is critical for distinguishing between forms that allow chain formation and those that do not. 26 That is, long-distance chains would be permitted when the reflexive is monomorphemic and when no domain restrictions block the chain. For instance, in Icelandic a finite clause boundary would block chain formation.
Peter Cole Gabriella Hermon C.-T. James Huang
xlvi
Long-distance Reflexives: The State of the Art
REFERENCES Amritavalli, R. (1999). Lexical anaphora in Kannada. In B. Lust, K. Wali, J. Gair, and K.V. Subbarao (eds.), Lexical Pronouns and Anaphors in Some South Asian Languages: A Principled Typology. Mouton de Gruyter, Berlin. Battistella, E. (1989). Chinese reflexivization. Linguistics 27,987-1012 Chierchia, G. (1989). Anaphora and attitudes de se. In R. Bartsch, J. van Benthem, and P. van Emde Boas (eds.), Semantics and Contextual Expression, 1-31. Dordrecht: Foris. Chomsky, N. (1981). Lectures on Government and Binding. Dordrecht: Foris. Cole, P., and G. Hermon (1998a). Long-distance reflexives in Singapore Malay: An apparent typological anomaly. Linguistic Typology 2(1) (Spring 1998). Cole, P., and G. Hermon (1998b). VP ellipsis and Malay reflexives. In IATL 5: Adam Zachary Wyner (ed.), Proceedings of the Thirteenth Annual Conference of the Israeli Association for Theoretical Linguistics, 39-54. Cole, P., and C. Wang (1996). Antecedents and blockers of Long-distance reflexives. Linguistic Inquiry 27,357-390. Cole, P., G. Hermon and L-M. Sung (1990). Principles and parameters of long-distance reflexives. Linguistic Inquiry 21,1-22. Cole, P., G. Hermon and C-L. Lee (2000). Logophoric conditions on long-distance reflexives and pronouns in Singapore Mandarin and Singapore Teochew. To appear in Studies in Chinese Linguistics, Volume 14. Culy, C. (1997). Logophoric pronouns and point of view. Linguistics 35,845-859. Davison, A. (1999). Lexical anaphora in Hindi-Urdu. In B. Lust, K. Wali, J. Gair, and K.V. Subbarao (eds.), Lexical Pronouns and Anaphors in some South Asian Languages: A Principled Typology. Berlin: Mouton de Gruyter. Everaert, M. (1991). Contextual determination of anaphor/pronominal distinction. In Jan Koster and Eric Reuland (eds.), Long-Distance Anaphora. New York: Cambridge University Press. Frajzyngier, Z. (1993). A Grammar of Mupun. Berlin: Reimer. Giorgi, A. (1984). Toward a theory of long-distance anaphora: A GB approach. Linguistic Review 3,307-359. Hagege, C. (1974). Les pronoun logophorique. Bulletin de la Societe de Linguistique de Paris 69,287-310. Huang, C.-T. J.; and C.-C. J. Tang (1991). The local nature of the long-distance reflexive in Chinese. In Jan Koster, and Eric Reuland, (eds.), Long-distance Anaphora. Cambridge: Cambridge University Press. Huang, Y.-H. (1984). Reflexives in Chinese. Studies in Literature and Linguistics 10. Jayaseelan, K.A. (1999). Lexical anaphora in Malayalam. In B. Lust, K. Wali, J. Gair, and K.V. Subbarao (eds.), Lexical Pronouns and Anaphors in Some South Asian Languages: A Principled Typology. Berlin: Mouton de Gruyter. Keenan, E.L. (1976) "Logical expressive power and syntactic variation in natural language," in E.L. Keenan (ed.), Formal Semantics for Natural Languages, Cambridge: Cambridge University Press. Keenan, E.L. (2000). An historical explanation of English Binding Theory. Presented at SHEL-1: Studies in the History of the English Language 1, UCLA, 2000.
Introduction
xlvii
Kornfilt, J. (1997). Turkish. London: Routledge. Koster, J., and E. Reuland, (eds.) (1991). Long-Distance Anaphora. Cambridge: Cambridge University Press. Lidz, J. (1996). Dimensions of reflexivity, Ph.D. dissertation, University of Delaware. Pica, P. (1987). On the nature of the reflexivization cycle. NELS 17(2), 483-99. Reinhart, T. and E. Reuland (1993). Reflexivity. Linguistic Inquiry 24,657-720. Sells, P. (1987). Aspects of logophoricity. Linguistic Inquiry 18,445-479. Sung, L-M. (1990). Universals of reflexives. Doctoral dissertation, University of Illinois at Urbana-Champaign. *„ \ Tang, C.-C. J. (1985). A study of reflexives in Chinese. M.A. thesis, National Taiwan Normal University. Tang, C.-C. J. (1989). Chinese reflexives. Natural Language and Linguistic Theory 7,93-121. Xu, L. (1994). The antecedent of ziji. Journal of Chinese Linguistics 22,115-137. Yu, X.-F. William (1992). Challenging Chinese reflexive data. Linguistic Review 9, 285-294. Yu, X.-F. W. (1996). A study of Chinese reflexives. Unpublished doctoral dissertation. University of London, School of Oriental and African Studies. Zribi-Hertz, A. (1989). Anaphor binding and narrative point of view: English reflexive pronouns in sentence and discourse. Language 65,695-727.
This page intentionally left blank
SYNTAX AND SEMANTICS VOLUME 33
This page intentionally left blank
GRAMMATICAL AND DISCOURSE CONDITIONS ON LONG DISTANCE REFLEXIVES IN TWO CHINESE DIALECTS PETER COLE* GABRIELLA HERMON+ CHER LENG LEE *Department of Linguistics University of Delaware Newark, Delaware +
Department of Linguistics and School of Education University of Delaware Newark, Delaware Department of Chinese Studies National University of Singapore Kent Ridge, Singapore
1. INTRODUCTION: APPROACHES TO LONG DISTANCE REFLEXIVES There has been wide discussion of the fact that in some languages reflexives must occur in a "local" relation to their antecedents while in others locality requirements are much more relaxed. For instance, in English, himself is usually restricted to, roughly, the same clause as its antecedent, (1) Johni thinks [Tomj knows [Billk likes himself*i/*j/k]]. Syntax and Semantics, Volume 33 Long-Distance Reflexives
1
Copyright © 2001 by Academic Press All rights of reproduction in any form reserved. 0092-4563/00 $35.00
2
Peter Cole et al.
while in Mandarin ziji 'self can occur indefinitely far from its antecedent: (2) Zhangsani renwei [Lisij zhidao [Wangwuk xihuan ziji i/j/k ]]. Zhangsan think Lisi know Wangwu like self. 'Zhangsan thinks that Lisi knows that Wangwu likes himself.' We shall refer to reflexives which take "non-local" antecedents as "long-distance (LD) reflexives". Three sorts of accounts for the differences in locality requirements among languages are found in the linguistic literature: grammatical accounts, discourse accounts, and "mixed" accounts.1 According to the grammatical accounts, LD reflexives occur when certain specific grammatical conditions obtain. Thus, according to one current grammatical treatment, the head movement analysis (Pica, 1987; Battistella, 1989; Cole, et al., 1990; Hestvik, 1992 inter alia; Cole and Sung, 1994; Cole and Wang, 1996), LD reflexives involve LF movement of the reflexive to an INFL (or AGR) position in which the antecedent both c-commands the reflexive and is within the governing category of the reflexive. Under this analysis, a reflexive will be LD when it is an X° rather than an XP form. The antecedent must c-command the reflexive, and, therefore, in most cases, must be in subject position.2 Somewhat different grammatical analyses have been presented by Tang (1989 inter alia), Huang and Tang (1991 inter alia), Progovac (1992 inter alia) and a variety of other authors. What these analyses have in common is that they attempt to specify the necessary and sufficient conditions for LD reflexives in purely syntactic terms, without reference to discourse.3 In contrast, a variety of authors have attempted to characterize the environments in which LD reflexives occur in terms of discourse and/or pragmatics alone. For instance, Chen (1992) argues that the distribution of Mandarin ziji can be explained in terms of two discourse properties, [+pivot] and [+high topicality]. Similarly, Yan Huang (1997) argues that the properties of LD reflexives can be explained fully in terms of neo-Gricean pragmatic principles like those in Levinson (1987,1995 inter alia). Unlike both the purely grammatical and the purely discourse approaches, we shall argue that neither grammatical nor discourse conditions alone can predict when LD reflexives will be possible. Rather, both grammatical and discourse conditions must be satisfied. Further, we will argue that while the grammatical conditions on LD reflexives are universal and are specified by Universal Grammar (UG), the ways in which discourse affects the distribution of LD reflexives vary from dialect to dialect and from language to language. Our discussion will be based primarily on a comparison of the distribution of LD reflexives in two Chinese dialects, the Mandarin spoken in Singa-
Grammatical and Discourse Conditions
3
pore (hereafter, Mandarin) and the dialect of Teochew spoken in Singapore (hereafter, Teochew).4 At first glance, Mandarin and Teochew appear to exhibit very different distributional restrictions on the use of reflexives, restrictions that might appear to be grammatical in nature. We shall show, however, that while both Mandarin and Teochew LD reflexives are subject to a combination of grammatical and discourse requirements (e.g., those associated with the notion of "logophoricity"), it is the discourse requirements rather than the grammatical requirements that differ from one dialect to the other. Contrary to appearances, the grammar of the dialects with respect to Binding seems to be identical. Our analysis of discourse conditions on LD reflexives is based primarily on the taxonomy of logophoricity of Sells (1987), who, like many authors, has argued that the notion of logophoricity plays a crucial role in understanding the felicity conditions on the use of LD reflexives. Unlike most researchers on this topic, Sells argues that logophoricity is not a unified notion, and that logophoric requirements are of three types, those relating to SOURCE (the source of communication), SELF (the one whose mental state the sentence describes) and PIVOT (the center of deixis or the perspective of the sentence).5 In a given sentence, the SOURCE, SELF, or PIVOT can be external to the sentence (i.e., the speaker) or internal (one of the protagonists of the sentence). Logophoric requirements are requirements that the antecedent for an anaphoric form (e.g., an LD reflexive) must be the SOURCE, SELF, or PIVOT of the sentence. Sells takes a position different from that of other writers in that he claims that languages differ with regard to the type of logophoricity required of the antecedent of an LD reflexive. For instance, he argues that in Japanese antecedents for LD reflexives must be internal PIVOTS while in Icelandic they must represent the individual whose mental state is described (i.e., internal SELF).6 Sells's proposal that logophoricity is not a unified notion provides the basis for a taxonomy of differing types of logophoric conditions that may be relevant for determining whether a noun phrase is a possible antecedent for a long distance reflexive. We shall show that Sells's taxonomy provides a useful tool for explaining the differences between the conditions on LD reflexives in Mandarin and Teochew. Using Sells's diagnostics for PIVOT, SOURCE, and SELF, we shall demonstrate that Mandarin and Teochew differ with regard to which logophoric requirements apply in the dialect: In both Mandarin and Teochew, PIVOT (perspective) requirements apply to the choice of antecedent for a LD reflexive.7 In Teochew, however, but not in Mandarin, SOURCE and/or SELF requirements apply as well. We argue that SOURCE/SELF may turn out to reflect a single, unified condition, but that PIVOT is distinct from SOURCE/SELF Following Chierchia (1989), we shall contend that Sells's taxonomy can, in part be reduced to a self-as-
4
Peter Cole et al.
cription (de se) requirement on long distance reflexives. While Sells's taxonomy provides a useful means of subdividing various types of logophoricity, we follow Chierchia, who suggested that Sells's SELF condition may derive from the requirement that LD reflexives must be associated with a de se interpretation.8 While both SOURCE and SELF may be derivable from a de se requirement, we shall argue that the PIVOT restriction cannot be reduced to de se? While the dialects differ with regard to the discourse requirements on LD reflexives, they share the same grammatical requirements. In both dialects LD reflexives show the grammatical characteristics of bound anaphors rather than those of pronouns. Thus, not only should LD reflexives in Chinese not be conflated with logophoric pronouns in African languages, for example, which have been shown to be pronouns, but Chinese dialects differ from each other much less than initial appearance would suggest with regard to the grammar of LD binding.
2. LOGOPHORIC CONDITIONS ON LD REFLEXIVES IN MANDARIN AND TEOCHEW We shall now turn to a comparison of the discourse conditions on LD reflexives in Mandarin and Teochew. In Mandarin, LD reflexives are possible even when the antecedent for the reflexive is neither described as the SOURCE of saying nor as the individual whose state of consciousness is described (SELF). (3) Antecedent neither SELF nor SOURCE Zhangsani wangji le Lisij hen taoyan ziji i/j de gege. Zhangsan forget perf Lisi very hate self 's brother 'Zhangsani forgot that Lisij hates
brother.'
(4) Zhangsani bu xiao de Lisij hen taoyan ziji i/j Zhangsan not aware Lisi very hate self 'Zhangsan was not aware that Lisi hates himi/j.' We show in (3) that the antecedent need not be either SOURCE or SELF. In (3) the possible antecedents for ziji are the local subject, Lisi, and the matrix subject, Zhangsan. But the matrix subject is neither the SOURCE of saying nor does the sentence report on the state of the world as pictured in the mind of the matrix subject (SELF): Note that the matrix verb, wangji
Grammatical and Discourse Conditions
5
'forget', indicates that the matrix subject is unaware of the information conveyed by the subordinate clause.10 A further example is given in (4). Zhangsan is a possible antecedent for ziji despite the fact that Zhangsan is neither the SOURCE of the statement that Lisi hates him, nor is this fact represented in his consciousness (SELF). These examples show that antecedents for LD reflexives in Mandarin do not need to be either SOURCE or SELF.11 Turning to Teochew, however, LD reflexives in analogous sentences are anomalous: (5) a. Ah Mengi m kitek Ah Lengj toryam kaki??i/j gai ah hia. Ah Meng not remember Ah Leng hate self 's brother 'Ah Meng forgot that Ah Leng hates her brother.' b. Ah Mengi m zai da Ah Lengj zeng toryam kaki ??i/j Ah Meng not know that Ah Leng really hate self 'Ah Meng does not know that Ah Leng hates herself.' The ungrammaticality of (5a)-(5b) is explained, however, if, in contrast to Mandarin, when the antecedent for LD ziji is neither SOURCE nor SELF, speakers find it difficult or impossible to interpret the reflexives nonlocally. This is despite the fact that the pragmatics of the situation described in (5a) and (5b) makes a LD interpretation much more natural than a local interpretation (speakers find it peculiar for someone to hate herself rather than someone else). The infelicity of the LD interpretation of (5a) and (5b) might seem to suggest that LD reflexives are not possible in Teochew, but this is not the case. Speakers find them felicitous when the LD antecedent fills the role of internal SOURCE or SELF: (6) Antecedent as Internal SOURCE a. Ah Lengi ga Ah Mai da Ah Mengk m zai kakii/*j/k Ah Leng with grandmother say Ah Meng not know self hiao zo. knows how do 'Ah Leng told her grandmother that Ah Meng didn't know she/he can do it.' b. Ah Lengi da Ah Mengi toi dio kakii/j. Ah Leng say Ah Meng see asp self 'Ah Leng says Ah Meng saw
6
Peter Cole et al.
LD reflexives in (6a) and (6b) are wellformed because the antecedent, Ah Leng is the internal SOURCE.12 The presence of an internal SELF also licenses the use of kaki with a nonlocal antecedent: (7) Antecedent as Internal SELF a. Ah Lengi siangseng Ah Meng j do kia lai toi dio kakii/j. Ah Leng believe Ah Meng at mirror in see asp self 'Ah Leng believes Ah Meng saw
in the mirror.'
b. Ah Leng jengwui Ah Meng bo suka kakii/j. Ah Leng feel Ah Meng not like self 'Ah Leng feels that Ah Meng dislikes In (7) Ah Leng is the individual whose state of mind the sentence describes. Thus, Ah Leng is the internal SELF. Just as in sentences in which the potential LD antecedent is the SOURCE, when the antecedent fulfills the role of SELF, that noun phrase can be an LD antecedent for kaki.13
3. SUBCOMMANDING ANTECEDENTS OF LD REFLEXIVES AND SOURCE/SELF REQUIREMENTS ON ANTECEDENTS We shall next show that Mandarin and Teochew differ with regard to whether "subcommanding" antecedents are possible for LD reflexives. These differences corroborate our claim that Mandarin and Teochew differ with respect to the discourse conditions on antecedents for LD reflexives. The phenomenon of "subcommand" was first investigated by Tang (1989 inter alia), who noted that antecedents for reflexives in Mandarin appear to be possible in environments in which subcommand rather than c-command holds of the relationship between the antecedent and the reflexive. Subcommand relaxes the requirements of c-command by allowing the specifier of a c-commanding noun phrase to count as a possible antecedent for a reflexive when the c-commanding nominal is inanimate,14 and, hence, is not itself a possible antecedent. The subcommand relation is intended to account for the grammaticality of examples like (8) [Zhangsani dechezi]jhaile ziji i/*j Zhangsan 's car harmed self 'Zhangsan's car harmed
Grammatical and Discourse Conditions
7
in which Zhangsan can be the antecedent for ziji despite the fact that Zhangsan does not c-command ziji. Tang notes that examples like (8) are grammatical just when (1) the phrase containing Zhangsan is inanimate, and, therefore, is not a potential antecedent for ziji, and (2) Zhangsan is the specifier rather than the complement of the head noun, chezi.15 Example (9) illustrates the ungrammaticality of sentences like (8) when the c-commanding noun phrase is animate, and, therefore, is a potential antecedent for ziji.16 (9) informed Subcommand with Animate Head [Zhangsani de taitai]j haile ziji*\iy Zhangsan 's wife harmed self 'Zhangsan's wife harmed
.'
Example (10) shows that the subcommanding antecedent must be the specifier, and not the complement, of a phrase c-commanding the reflexive: (10) Zhangsani get Lm-} dequangao bangzhu le tai/jl zijii/*j. Zhangsan give Lisi 's advice help perf him self 'That Zhangsan gave Lisi advice helped him.' In (10) the subject of the sentential subject, Zhangsan, is a possible antecedent for ziji, but the object of the sentential subject is not.17 Not only are subcommanding nominals possible antecedents for local reflexives in Mandarin, as was shown in (8)-(10), but they are possible antecedents for LD reflexives as well.18 (11) a. Zhangsani de xin shuo Mali} renwei ziji i/j shi wugude. Zhangsan 's letter say Mary think self is innocent 'Zhangsan's letter says that Mary thinks
is innocent.'
b. Zhangsani xie xin de shi zhengming le Lisij renwei Zhangsan write letter's matter prove perf Lisi think ziji i/j shi wugude. self is innocent "The fact that Zhangsan wrote the letter proves that Lisi thinks that he is innocent.' What is of interest here is that the subcommanding antecedents are typically neither SOURCE nor SELF Although in (lla) Zhangsan might be viewed as an indirect SOURCE (something said by Zhangsan's letter was ultimately said by Zhangsan), in a mystery novel sentence like (lib), Zhangsan's letter might be on a topic entirely unrelated to his innocence. It is also not the case that Zhangsan is indicated as SELE Thus, sentences like
8
Peter Cole et al.
those of (11) show that SOURCE and SELF are not essential for antecedents of ziji in Mandarin. Turning now to Teochew, unlike Mandarin, sentences like (12) are informed: (12) a. Ah Mengi gai chiaj hai liao kaki*i/*j. Ah Meng 's car harm perf self 'Ah Meng's car harmec b. Ah Mengi sia seng gai si zengmeng liao Ah Leng-} Ah Meng write letter 's matter prove perf Ah Leng jengwui kaki*i/j bor salah. thinks self innocent 'The fact that Ah Meng wrote the letter proves that Ah Leng t h i n k s i s innocent.' There is a clear contrast between sentences like (9) and (11) in Mandarin, in which it is felicitous for ziji to refer to an LD antecedent, and those of (12) in Teochew, in which reference to a LD antecedent is infelicitous. However, the infelicity of (12) is to be expected if the antecedents of LD reflexives in Teochew are subject to SELF/SOURCE requirements which do not apply in Mandarin. Thus, the contrast between (9)/(ll) and (12) constitutes additional evidence that Mandarin and Teochew differ in terms of whether there is a SOURCE/SELF condition on LD reflexives.
4. MODIFYING PHRASES AS TESTS FOR SOURCE/SELF Sells (1987) suggests some additional tests for determining whether a nominal is an internal SOURCE or SELF. He notes that the use of "speaker-evaluative" modifying phrases like that fool can assist in determining whether the SOURCE is external (the speaker) or an internal protagonist of the discourse. For example, in (13) John told Fred that that fool Alan was harming him. the phrase that fool can be understood as either the speaker's description of Alan or that of John. In the former case, despite the fact that the matrix verb is a verb of saying, the sentence is taken to have an external SOURCE, while in the latter case, it is understood to have an internal SOURCE.
Grammatical and Discourse Conditions
9
Returning to Mandarin, just as in English, in (14) Zhangsani gaosu Lisij net .ge shagua Wangwuk shanghai zijiy i/*j/k. Zhangsan tell Lisi that CL fool Wangwu harm self 'Zhangsan told Lisi that that fool Wangwu h a r m e d . ' the phrase nei ge shagua 'that fool' can be understood as either the speaker's description or as Zhangsan's description. However, even when neige shagua 'that fool' is attributed to the speaker (i.e., to an external SOURCE), it is possible for ziji to refer to Zhangsan. This constitutes additional evidence that it is not necessary in Mandarin for the antecedent of ziji to be an internal SOURCE. Turning now to Teochew, when a speaker-evaluative phrase is attributed to an external SOURCE, the reflexive kaki cannot be interpreted as referring to an LD antecedent: (15) Ah Mengi ga Ah Lengj da hi gai sagua Ah Soik hai kaki*i/*j/k. Ah Meng with Ah Leng say that CL fool Ah Soi harm self 'Ah Meng told Ah Leng that that fool Ah Soi harmed
.'
However, the judgments are altered when the speaker-evaluative phrase is attributed to the matrix subject: when higai sagua 'that fool' is attributed to Ah Meng, kaki can be understood as referring to Ah Meng. Thus, the interaction of speaker-evaluative phrases like higai sagua 'that fool' and LD reflexives constitutes further evidence that in Teochew antecedents of LD reflexives must be instances of an internal SOURCE (or, as shown further following, SELF), a requirement that does not hold in Mandarin.19 Turning now to SELF, the adverbs mysteriously and inexplicably are a diagnostic for SELF. In (16) John thinks that Mary
hates him.
mysteriously/inexplicably can be understood as mysterious or inexplicable to John, making John the internal SELF, or as mysterious or inexplicable to the speaker, making the speaker the external SELF. Turning now to Mandarin, in (17) Zhangsai renwei [Wangwuj momingqimiao de xihuan ziji^. Zhangsan think Wangwu inexplicably DE like self 'Zhangsan thinks that Wangwu inexplicably likes
10
Peter Cole et al
momingqimiao 'inexplicably' can be taken to refer to the speaker's view of the world, making the speaker the external SELF. Despite this, however, ziji can refer to Zhangsan. But in (18), the Teochew equivalent to (17), (18) Ah Mengi jengwui Ah Lengj mok meng ki miao gai suka kakii/j . Ah Meng feel Ah Leng inexplicably DE like self 'Ah Meng thinks that Ah Leng inexplicably likes kaki can only refer to Ah Meng when mok meng ki miao 'inexplicably' expresses Ah Meng's perspective, i.e., when the situation is inexplicable to Ah Meng rather than to the speaker. When the mok meng ki miao 'inexplicably' is understood as expressing the speaker's perspective, kaki must be interpreted locally, that is, as referring to Ah Leng. Thus, the interaction of the interpretation of mok meng ki miao 'inexplicably' and the possible reference for kaki constitutes additional evidence that the antecedent for kaki must be SELF (or, as shown previously, SOURCE), a requirement holding for Teochew kaki but not for Mandarin ziji.
5. ADDITIONAL EVIDENCE ON THE ROLE OF SELF In this section we shall provide additional evidence that SELF plays a central role in the wellformedness of LD reflexives in Singapore Teochew, but not in the Mandarin of Singapore. Consider the following situation: Zhangsan's father has been asked by the owner of a firm to evaluate the performance of the employees of the firm on the basis of the firm's records of performance and to recommend a reward for exceptional employees. The employees are identified only by number. Zhangsan is unaware that his own child is one of the employees that he has identified as deserving an award. In Mandarin the following sentence is wellformed: (19) Zhangsani qing laoban jiangshang ziji i de haizi. Zhangsan ask owner reward self 's child 'Zhangsani asked that the owner reward hisi own child.' Sentence (19) is wellformed despite the fact that Zhangsan is unaware of the fact that the person rewarded is his own child. Thus, in (19), consciousness on the part of the antecedent is not necessary in order for ziji to be employed.
Grammatical and Discourse Conditions
11
A similar example is given in (20): (20) Zhangsani chengsan houlai sha le ziji i de ren. Zhangsan praise later kill perf self 's man 'Zhangsan praised the man who later killed him.' In (20) the antecedent for ziji, Zhangsan, is not the internal SELF because Zhangsan was unaware that the man in question would later kill him. Thus, in Mandarin internal SELF is not a necessary condition for the use of LD ziji. The situation in Teochew, however, is quite different. In contrast with Mandarin, the Teochew equivalent of (19) is informed (in the context described): (21) *Ah Mengi kio towkang sio kakii gai kia. Ah Meng ask owner reward self 's child 'Ah Meng asked the owner to reward his own son.' This sentence is wellformed, however, in a context in which Ah Meng knows that the person being rewarded is his child. Similarly, the Teochew equivalent of (20) is also informed: (22) *Ah Mengi orlor kawao tai liao kakii gai nang. Ah Meng praise later kill perf self 's man Ah Meng praised the man who later killed him.' The contrast between (19) and (20) and (21) and (22) shows that, in Mandarin, a long distance reflexive is wellformed even when the antecedent is not conscious of the event described employing the reflexive, but in Teochew, consciousness (i.e., internal SELF) is a necessary condition on the use of the long distance reflexive.
6. PIVOT RESTRICTIONS We have shown that Mandarin and Teochew LD reflexives differ in the discourse conditions which must hold of their antecedents. While in Teochew the antecedent of an LD reflexive must be an internal SOURCE or SELF,20 this logophoric requirement on antecedents does not hold in Mandarin. We would like to turn now to the PIVOT condition. We shall show that in both Mandarin and Teochew, in order for the antecedent of an LD reflexive to be fully felicitous, it must be the PIVOT, the center of deixis or role perspective for the sentence. As was first noted by Y.H. Huang (1984), LD reflexives in Mandarin are subject to a Blocking Effect, a fact that has recently been claimed to be due
12
Peter Cole et al
to logophoric restrictions on antecedents of LD reflexives (e.g., Huang and Liu 1997 and this volume). While sentences like (22) (which repeats (2)) are fully grammatical, (22) Zhangsani renwei [Lisij zhidao [Wangwuk xihuan zijii/j/k]]. Zhangsan think Lisi know Wangwu like self 'Zhangsan thinks that Lisi knows that Wangwu likes himself.' sentences like (23) are informed: (23) Zhangsani renwei [Wangwuj zhidao [wok xihuan ziji *i/*j/k ]]. Zhangsan think Wangwu know I like self. 'Zhangsan thinks that Wangwu knows that I like myself.' In such works as Tang (1989), Huang and Tang (1991), Battistella (1989), and Cole et al, (1990), the illformedness of examples like (23) was analyzed as due to a grammatical requirement that all subjects between ziji and its antecedents must agree in person features. It has been argued, however, in such works as Huang and Liu (1997 and this volume), that the Blocking Effect is not due to grammatical requirements, but rather to a discourse restriction that the antecedent for ziji satisfy a logophoric requirement, which we take to be the requirement that the antecedent constitute an internal PIVOT. That is, the antecedent for ziji must be the center of deixis or role perspective for the sentence. The occurrence of a first or second person pronoun is taken to indicate that the speaker, rather than some internal protagonist of the sentence, is the center of deixis. The grammatical view of the Blocking Effect was first called into question by such writers as Xu (1992,1993) and Yu (1989,1991), who noted that the grammatical analysis of Blocking would predict that only subjects that intervene between ziji and its antecedent should induce the Blocking Effect. The logophoric analysis, however, would predict that any external PIVOT in the sentence would induce Blocking since the occurrence of an external PIVOT would mean that the antecedent for ziji is not the internal PIVOT. Since a first or second person pronoun anywhere in the sentence would constitute an external PIVOT, according to the PIVOT analysis, Blocking could be induced by the presence of a first or second person pronoun anywhere in the sentence, and not just in a subject position between the antecedent and the reflexive. We shall argue that in fact both the grammatical and the logophoric analyses are correct (though incomplete): nonsubjects can induce Blocking.21 However, there exists evidence that Blocking by subjects is in fact a process distinct from Blocking by nonsubjects, and that both logophoric and grammatical Blocking exist. Turning now to the facts in Mandarin, as mentioned previously, Xu and
Grammatical and Discourse Conditions
13
Yu each provide evidence that first and second person nonsubjects can induce a Blocking Effect: (24) a. Zhangsani shuo [[ni j zuo de chunshi] haile ziji*i/j] Zhangsan say you do silly deeds harmed self 'Zhangsan says that the silly things you have done have harmed you.' b. Zhangsani shuo [[Lisij zuode chunshi] haile zijiyj] Zhangsan say Lisi do silly deeds harmed self 'Zhansan says that the silly things Lisi has done have harmed him/you.' In (24a) the subcommanding specifier of the subordinate subject (ni 'you') differs in person features from the matrix subject, but the subordinate subject itself (ni zuo de chunshi 'silly deeds you have done') does not: Both are third person. However, the subcommander prevents the matrix subject from anteceding ziji. In (24b), however, the subcommander and the matrix subject are both third person, and the matrix subject is a possible antecedent for ziji. Furthermore, going beyond Xu and Yu's examples, at least for many speakers, the Blocking Effect can be induced by a non-c-commanding nominal that is not a possible antecedent for ziji: (25) a. Zhangsani Zhangsan 'Zhangsan b. Zhangsani Zhangsan 'Zhangsan
cong nij nar ting shuo Malik hen taoyan ziji?i/*j/k. from you there hear say Mary very hate self heard from you that Mary hates herself.' cong Lisij nar ting shuo Mal\ hen taoyan ziji i/*j/k from Lisi there hear say Mary very hate self heard from Lisi that Mary hates herself.'
While neither ni 'you' in (25a) nor Lisi in (25b) is a possible antecedent for ziji (since they fail to c-command/sub-command ziji),22 the fact that ni is second person prevents the matrix subject from anteceding ziji (or, for some speakers, makes it more difficult for the matrix subject to antecede ziji). In contrast, when ni is replaced by a third person nominal (Lisi), it is not problematic for the matrix subject to be the antecedent for ziji. Reduced felicity is also found when a first or second person pronoun occurs in a position in the sentence which the grammatical analysis would predict to be irrelevant to Blocking: (26) a. Zhangsan{ zhidao Mali} gen ziji??i/??j shuoguo ni xiang qu Zhangsan know Mary with self said you want go Taiwan. Taiwan
14
Peter Cole et al.
'Zhangsan knows Mary told him/herself that you want to go to Taiwan.' b. Zhangsani zhidao Mali-j gen ziji-i/j j shuoguo Xiao Ming xiang qu Zhangsan know Mary with self said Xiao Ming want go Taiwan. Taiwan 'Zhangsan knows Mary told him/herself that Xiao Ming wants to go to Taiwan.' The second person nominal in (26a) does not intervene between ziji and its antecedent. Despite this fact, (26a) is less felicitous than (26b), in which ni 'you' is replaced by a third person nominal, Xiao Ming. While explanations for (24) are possible within the framework of such grammatical analyses as head movement,23 examples like those of (25) and (26) cannot be explained by any of the current grammatical accounts. These facts, however, are predicted to occur on the hypothesis that (1) the antecedent for a LD reflexive in Mandarin must be the internal PIVOT and (2) the presence of a first or second person pronoun anywhere in the sentence constitutes an external PIVOT, which blocks the possibility of an internal PIVOT. Turning to Teochew, we see that the same PIVOT requirement holds in that dialect as well. Just as in Mandarin (23), the Blocking Effect is found: (27) Ah Mengi jengwui Ah Lengj zai wak suka kaki* i/j/k Ah Meng think Ah Leng know I like self 'Ah Meng thinks Ah Leng knows that I like
(28) Ah Mengf jengwui Ah Soij zai Ah Lengk suka kakiirj/k. Ah Meng think Ah Soi know Ah Leng like self 'Ah Meng thinks Ah Soi knows that Ah Leng like;
As is seen in (27) and (28), the presence of a first person subject between the local subject and the matrix subject blocks reference to both the matrix subject and the intervening first person subject. In addition, Blocking occurs in Teochew in sentences analogous to Mandarin (24) and (25):
Grammatical and Discourse Conditions
15
(29) a. Ah Mengi da [le-} zor gai sa sih]k hai liao fc0/cz??i/j7*k.24 Ah Meng say you do 's silly deed harm perf self 'Ah Meng says that the silly thing you have done harmed b. Ah Mengi da [Ah Lengj zor gai sa si]k hai liao kakii/j*/kk. Ah Meng say Ah Leng do 's silly deed harm perf self 'Ah Meng says that the silly thing Ah Leng has done harmed
'
(30) a. Ah Meng^ dor lej hio kor tia da Ah Soik toryam A:fl/c/??i/*j/k.25 Ah Meng from you there hear that Ah Soi hate self 'Ah Meng heard from you that Ah Soi hates
.'
b. Ah Meng{ dor Ah Lengj hio kor tia da Ah Soik toryam Ah Meng from Ah Leng there hear that Ah Soi hate kakiytjfc. self 'Ah Meng heard from you that Ah Soi hates
.'
Thus, we conclude that the antecedents for LD reflexives in both Mandarin and Teochew are subject to a logophoric restriction that the felicity of an antecedent is reduced if that nominal is not the internal PIVOT. To summarize, both Mandarin and Teochew LD reflexives are subject to logophoric conditions on the antecedent of the reflexive. These conditions, however, differ in the two dialects. In Teochew the antecedent must be SOURCE or SELF and PIVOT. That is, the antecedent must be the source of communication or the individual in whose mind the described event is pictured. Furthermore, felicity is reduced if there is a center of deixis/role perspective anywhere in the sentence other than the antecedent of the LD reflexive. In contrast, in Mandarin, at least in the Mandarin of Singapore, SOURCE and SELF are not critical. The PIVOT requirement, however, holds in Mandarin as well. If a nominal other than the antecedent for an LD reflexive is the center of deixis, the felicity of the non-PIVOT nominal as an antecedent is re-
16
Peter Cole et al
duced. First and second person pronouns are inherently relational vis-a-vis the speaker. (/ is the speaker; you is the person the speaker is addressing.) Thus, when a first or second person pronoun occurs anywhere in the sentence, the speaker is the external PIVOT. Thus, a Blocking Effect is found whenever the sentence contains non-third person pronouns. This occurs in both Teochew and Mandarin. This is a strong argument that the Blocking Effect is, at least in part, due to logophoric restrictions on discourse rather than to grammar. We shall return later to the question of whether all instances of the Blocking Effect are due entirely to principles of discourse or whether there is a class of cases for which a grammatical explanation is necessary.
7. LOGOPHORICITY AND ATTITUDES DE SE In the discussion so far, we have employed Sells's taxonomy of logophoricity as a tool in differentiating the discourse properties associated with Mandarin and Teochew. We saw that both Mandarin and Teochew respect a requirement that the antecedent for an LD reflexive must be the internal PIVOT. However, Teochew differs from Mandarin in requiring that the antecedent be the internal SOURCE or SELF. The fact that it is either SOURCE or SELF that must be satisified (and not, for instance, both simultaneously) suggests that SOURCE and SELF might be derived from a single more abstract condition on the antecedent. We shall argue in this section that Chierchia's de se restriction is a promising candidate for such a condition. Following Lewis (1979), Chierchia (1989) proposes that sentences involving self-ascription differ semantically from those that do not. Thus, a sentence like (31) Pavarotti believes that his pants are on fire. has a de se and a de re interpretation.26 Under the de se interpretation, Pavarotti might notice that his pants are on fire and perhaps say (32) My pants are on fire. Under this interpretation, Pavarotti ascribes to himself the property of having burning pants. The second interpretation is the de re interpretation. Under this interpretation, (31) could be uttered truthfully when, for instance, during a Three Tenors concert, Pavarotti is looking at a television monitor of the ongoing concert as seen from behind. He notices in the monitor that the pants of one of the singers are on fire. He does not realize, however, that it is he himself
Grammatical and Discourse Conditions
17
who has burning pants. Under these circumstances, it could be truthfully said of Pavarotti that he believes that a certain individual, who, in fact, is Pavarotti, has his pants on fire. That is, (31) is true. Thus, although on the de re interpretation, Pavarotti believes that his pants are on fire, he does not ascribe having pants on fire to himself. Chierchia examines whether sentences like (31) are genuinely ambiguous between two interpretations rather than vague with respect to whether self-ascription takes place. He argues that if, in natural language, there exist grammatical constructions which are limited to the de se interpretation, this constitutes evidence that de se interpretations are, in fact, distinct from de re interpretations. Chierchia notes that, in Italian, LD reflexives are possible only if the antecedent ascribes the proposition which is the object of belief, thought, etc. to himself. Thus, (33) is a contradiction: (33) Pavarotti crede che i propri pantaloni siano in fiamme. Pavarotti believes that the self pants are in flame Ma non si e' accorto che i pantaloni sono i propri. but not realize that the pants are the own 'Pavarotti believes that self s pants are on fire, but he hasn't realized that the pants are his own.' Note that the contradictoriness of (33) is due to the use of the LD reflexive propri. When the personal pronoun suoi is substituted for propri, the contradiction disappears: (34) Pavarotti crede che i suoi pantaloni siano in fiamme. Ma non Pavarotti believes that the his pants are in flame but not si e' accorto che i pantaloni sono i propri. realize that the pants are the own 'Pavarotti believes that his pants are on fire, but he hasn't realized that the pants are his own.' Returning to Teochew, let us consider the possibility that the apparent SOURCE/SELF requirement on antecedents for kaki is, in fact, a de se requirement. Let us assume that for (8b) to be felicitous (repeated as (35)). (35) Ah Lengj'engwui Ah Meng} bo suka kaki^. Ah Leng feel Ah Meng not like self 'Ah Leng feels that Ah Meng dislikes
.'
the antecedent, Ah Leng, must ascribe to herself the property of Ah Meng not liking her. This state of affairs would entail that Ah Leng pictures in her mind that Ah Meng does not like her. Thus, de se entails SELE
18
Peter Cole et al
Similarly, assuming a de se restriction on the antecedent in (35), if Ah Leng were to speak truthfully on the subject, she would have to be disposed to state (36): (36) Ah Meng bo suka wa. Ah Meng not like me 'Ah Meng does not like me.' Thus, (35) entails (37) and implicates (38): (37) Ah Meng would say that Ah Meng does not like her. (38) Ah Meng says that Ah Meng does not like her. Therefore, de se implicates SOURCE and entails something very close to SOURCE, namely, the disposition for the antecedent to be the SOURCE if the antecedent speaks truthfully. While de se entails SELF and strongly implicates SOURCE, neither SOURCE nor SELF entails de se. As we saw, the fact that Pavarotti believes that a certain individual has burning pants does not mean that he believes that he is that individual. But having this belief means that Pavarotti is the internal SELF. Similarly, the fact that Pavarotti says that a certain individual's pants are on fire does not constitute his ascribing that property to himself. It does, however, mean that Pavarotti is the SOURCE of communication. Although neither SELF nor SOURCE entails de se, both implicate de se. Thus, if, as in (35), Ah Leng feels that Ah Meng does not like her, it would be natural to assume, in the absence of evidence to the contrary, that Ah Leng is aware that the individual whom Ah Meng does not like is she herself. Similarly, if, as in (7b) (repeated as (39)) (39) Ah Leng{ da Ah Meng j .toi dio kakii/j]. Ah Leng say Ah Meng see asp self 'Ah Leng says Ah Meng saw Ah Leng says that Ah Meng saw her, it is natural to assume that Ah Leng is aware that it is she herself that Ah Meng saw. Thus, while self-ascription is not entailed by SELF or SOURCE, self-ascription is, in fact, implicated by either SELF or SOURCE. These facts suggest that SOURCE and SELF are not the primitives employed by Teochew in determining the well-formedness of LD antecedents for kaki, but rather that Sells's taxonomy of logophoricity is derivable in part from self-ascription (as Chierchia claimed). While there are a number of problems to be overcome by this analysis, many of these are discussed
Grammatical and Discourse Conditions
19
by Huang and Liu (this volume), and we shall assume that satisfactory solutions are provided. Thus, we shall henceforth take it as demonstrated that in Teochew the SELF/SOURCE restriction can be successfully reduced to a de se requirement on antecedents of LD reflexives. (The de se restriction would not, however, apply to Mandarin or, at least, not to the dialect of Mandarin under discussion in this paper.) In the next section, we shall consider some syntactic and semantic implications of the analysis of de se.
8. THE SYNTACTIC AND SEMANTIC ANALYSIS OF ATTITUDES DESE In this section we would like to examine some syntactic and semantic consequences of the analysis of attitudes de se. We shall argue, pace Chierchia, that the self-ascription requirement in Teochew should be analyzed as a felicity condition on the use of the reflexive kaki rather than as a semantic condition on verbs of propositional attitude, as was suggested by Chierchia. It may be useful to summarize Chierchia's analysis of de se. Chierchia proposes that, on the de se reading, a sentence like (31) would have a logical form like (40): (31) Pavarotti believes that his pants are on fire. (40) believe (P,l x [x's pants are on fire]) According to (40), on the de se interpretation "believe" is taken to be a self-ascriptive relation between an individual (Pavarotti) and a property. Note that in order to insure that the relation between Pavarotti and the property in question is self-ascriptive, it is necessary for Chierchia to stipulate something along the lines of (41):27 (41) a. x stands in the belief relation with property Q (i.e., x self-ascribes Q, in Lewis's terms) if x believes that x has Q and furthermore K(x,x), where K is the cognitive access that we have to ourselves, b. K(x,x) = df x is disposed to describe the relevant belief by referring to x by means of the first person pronoun. On the basis of (41), Pavarotti has the necessary awareness that the property specified in (40) is one that he himself possesses, rather than a property of some individual that merely turns out to be him.
20
Peter Cole et al
Let us now apply Chierchia's analysis to a Teochew sentence like (35): (35) Ah Lengi jengwui Ah Meng j bo suka kakii/j. Ah Leng feel Ah Meng not like self 'Ah Leng feels that Ah Meng dislike: It will be remembered that a self-ascriptive relation is required between Ah Leng and kaki for the sentence to be well-formed in Teochew. Following Chierchia, the semantic representation for (35) would be (42): (42) jengwui (AL, Xx [AM bo suka x\) feel AL AM not like According to Chierchia's proposal, an appropriate syntactic structure corresponding to (42) in the Principles and Parameters tradition would be along the lines of (43): (43) Ah Leng jengwui [CP [IP ziji-OPi{ [ipAh Meng suka ?;]]] Ah Leng feel self-OP Ah Meng like In (43) ziji is adjoined to IP, and is treated as the natural language instantiation of the lambda operator (hence, ziji-OP). 28 Structure (43) is essentially that proposed by Huang and Tang (1991) for LD reflexives. In Huang and Tang's analysis, the reflexive ziji is treated as an XP operator that adjoins to IP. While Chierchia's proposal shows clear similarities to that of Huang and Tang, it is also compatible with the head movement analysis, first proposed for Chinese by Battistella (1989) and later adopted by Sung (1990), Cole et al. (1990), and other authors. The head movement analog of (43) is (44): (44) Ah Leng jengwui [CP ziji i -C [ipAh Meng t{-I [VP suka t i ]]] Ah Leng feel self-C Ah Meng like In (44), as in (43), ziji is analyzed as the instantiation of the lambda operator which converts the subordinate CP from a proposition to a property. Thus, both XP adjunction and head movement are compatible with Chierchia's semantics for sentences like (42). Abstracting away from the differences between the two proposals, we shall take (43) and (44) to be the application of Chierchia's analysis of de se to LD reflexives in Teochew and other Chinese languages. To what extent are the structures in (43) and (44) compatible with the syntax of LD reflexives in Chinese languages? Structure (44) departs from the usual head movement analyses of LD reflexives in that the LF position
Grammatical and Discourse Conditions
21
of ziji is the C of the complement clause rather than the I or AGR of the matrix clause, as in (45): (45)
[AGRPAh Len [ A G R ' - z i j i I I jengwui [CP t'i-C [lPAh Meng trl Ah Leng AGR-self feel Ah Meng [vpsukatfi]]]] like
In structure (45) we follow the phrase structure proposed in Cole and Wang (1996).29 In (45), unlike (44), ziji moves to a pre-VP position. Thus, according to (45), the complements of the matrix verb do not c-command (or subcommand, etc.) ziji at LF. An XP-movement analog of (45) is (46), in which ziji is adjoined to VP. (46)
Ah Leng [VP ziji-OP, [VPjengwui [CP [IP t'{ [IP Ah Meng suka ti]]]] Ah Leng self-OP feel Ah Meng like
As in (45), ziji asymmetrically c-commands the complement of the verb (in this case, CP). Structures (45)-(46) make different predictions from (43)-(44) with regard to which noun phrases can be the antecedent for ziji. According to (43)-(44), the antecedents for ziji should include both matrix subjects and matrix objects, but according to (45)-(46) postverbal objects would not be potential antecedents for ziji. Sentences like those of (47) provide a test for these differing predictions: (47) a. Sengsehj ga hakseng} [kakii/*j zornimue siosim]. teacher teach student self how careful 'The teacher teaches students how he should be careful.' b. Ah Mengi gaodai Ah Leng-} kaki^ oi kih. Ah Meng inform Ah Leng self will go 'Ah Meng informed Ah Leng that he/*she would go.' As can be seen from the sentences of (47), postverbal objects like hakseng and Ah Leng cannot be the antecedent for kaki. This is so despite the fact that hakseng can be viewed as the individual in whose mind kaki zornimue siosim is pictured (i.e., hakseng can be viewed as the internal SELF), and hakseng can be viewed as attributing kaki zornimue siosim to himself. Similarly, Ah Leng can be viewed as the internal SELF in (47b), and Ah Leng is most naturally understood as standing in a self-ascriptive relationship to kaki. The illformedness of the sentences of (47) on the reading in which kaki is understood to be hakseng or Ah Leng is explicable if these sentences have a structure like (45)-(46), but not if they have a structure along the lines proposed by Chierchia, structures (43)-(44): According to (45)-(46), hakseng
22
Peter Cole et al
and Ah Leng fail to c-command kaki at LF, and, therefore are not potential antecedents for kaki. In contrast, in (42)-(43), hakseng and Ah Leng c-command kaki. Thus, under (42)-(43), there is no explanation for why hakseng and Ah Leng fail to be possible antecedents for kaki. As a result, if it is assumed that Chierchia's analysis of the semantics of de se is correct, and that de se interpretation is due to lambda abstraction of the embedded clause (and not, crucially, of the matrix VP), there appears to be a contradiction between the structures required by the syntax and the semantics for LD reflexives in Teochew. The semantics would appear to require (43)-(44) while the syntax appears to require (45)-(46). The problem goes beyond Chinese and would hold of all languages in which LD reflexives are both "subject-oriented" and are subject to a de se condition on the choice of the antecedent. It is, of course, desirable that the structures posited for the syntax be the same structures needed for semantic interpretation, so we will explore two possible approaches to resolve this contradiction. The first option would be to assume that Chierchia's proposed semantic structures are, in fact, correct, but that the syntactic derivation includes an additional movement which is "invisible" to the semantics. Thus, the input to semantic interpretation would include only the movement indicated by the boldface, solid arrows in (48), and not the final movement indicated by a dashed arrow:
According to (48), the final movement from C to AGR is invisible to semantics, but visible to LF syntax. Thus, (48) would predict both subject orientation and the occurrence of the de se interpretation. While (48) does predict correctly that subject orientation and de se interpretation can co-occur, the prediction is made at the cost of a disassociation of the structures employed for syntax and semantics. According to (48), "reconstruction" applies to the output of the LF syntax, restoring the structure immediately prior to the last stage of movement. This reconstruction is completely ad hoc and should be rejected if a less objectionable solution is available. A second option would be to assume that the input to semantic interpretation is (45) or (46), and, therefore, that Chierchia's semantics for de se must be modified. Clearly, lambda abstraction over the matrix VP would not yield the necessary self-ascriptive relation between an individual entity and a property, so an alternative approach to de se is necessary.
Grammatical and Discourse Conditions
23
A possibility that springs to mind is that self-ascription is not due to the semantics of the matrix verb, but, rather, to the semantics of the reflexive itself. According to such an alternative account, the reflexive kaki would carry with it the presupposition that its antecedent ascribes to himself the proposition of which kaki is an argument. Thus, in (35) Ah Leng would ascribe Ah Meng's dislike to herself and would be disposed to say, "Ah Meng dislikes me." The attribution of the de se interpretation to the reflexive rather than to the matrix predicate has a number of advantages. First, and most important, the contradiction between syntax and semantics is resolved (in favor of (45)-(46)). Second, there exist a variety of cases in which long-distance reflexives show de se effects but which do not involve a verb of prepositional attitude like "believe". For example, in (49) (49) Ah Leng bor suka hi gai toi dio kaki gai nan. Ah Leng not like this clas see asp self rel person Ah Leng does not like the person who saw her.' (Kaki can refer to Ah Leng.) In order for (49) to be wellformed, it is necessary for Ah Leng to be aware that the person whom she does not like looked at her. It cannot be the case that the man in question was looking at her without her being aware of the fact. Thus, (49) is subject to a de se requirement. There is, however, no way to interpret (49) as a relation between Ah Leng and a property, along the lines of (42). Thus, (49) seems prima facie unsuited to a Chierchia-style analysis of de se. In order to extend Chierchia's approach to cases like these, it is necessary to introduce considerable complications.30 We shall, therefore, assume pace Chierchia, that de se interpretation is a property of the anaphor rather than one of the predicate.
9. THE BINDING THEORETICAL STATUS OF LD REFLEXIVES We showed in Section 7 that the logophoric conditions on Teochew (SOURCE/SELF) are reducible in part to a de se condition on LD reflexives in that dialect, but that the logophoric conditions on Mandarin (PIVOT) are not. We would now like to turn to the question of what the existence of logophoric conditions tells us with regard to the binding theoretic status of LD ziji and kaki. Are these forms, when used with LD antecedents, bound anaphors and hence, presumably, subject to some version of Condition A of the Binding Theory, or are they pronouns and hence, subject to some version of Condition B? The fact that LD reflexives are subject to
24
Peter Cole et al
logophoric conditions might suggest that forms like kaki and ziji are ambiguous between local reflexives and logophoric pronouns. (Logophoric pronouns were first reported for certain African languages, e.g., Ewe (Clements, 1975), Mundang, and Tuburi (Hagege, 1974).) This, in fact, is the position taken by Cole and Hermon (1998) for Malay and by Reuland (this volume) for Icelandic LD reflexives. Reuland argues that in certain environments (e.g., subjunctives), reflexive forms are used as logophoric pronouns (as shown by the fact that reflexives in subjunctive clauses can take non-c-commanding antecedents, etc.). A variant on the proposal that LD reflexives are reducible to logophoric pronouns would be to claim that while reflexives are, in and of themselves, bound anaphors, they can occur in two different structures. In one structure, the reflexive is bound by a phonologically null pronoun, and, hence, takes as its antecedent the antecedent of that pronoun, while in the other structure, the reflexive is directly bound by a local antecedent. This tack is taken, for instance, by Yu (1996), who proposes that ziji can occur in two different structures: (50) Local Ziji
(51) Long Distance Ziji
In Yu's analysis, ziji is a predicate (meaning, roughly, "same as") that takes two arguments. Since ziji is a bound anaphor, its arguments must c-command ziji and be within its Governing Category. In (50) only one argument (y) is bound by a c-commanding antecedent within the Governing Category for ziji (i.e., internal to the NP). The second argument of ziji (x) requires a local binder, but none is present inside NP. A local binder is provided by adjoining ziji to the verb:
Grammatical and Discourse Conditions
25
In (52) the specifier of IP c-commands ziji and is within its Governing Category. Thus, ziji is bound by the specifier of IP (the local subject). In contrast to (50), in the LD structure (51), both arguments of ziji are bound within the DP. Since the binder for ziji is an instance of pro, the possible reference of ziji is predicted to be whatever pro can refer to. Thus, the possible reference for ziji in (51) is that of a pronoun (modula any differences between overt pronouns and pro). Thus, approaches like that of Yu (1996) predict that LD ziji (and, presumably, its Teochew counterpart, kaki) will have the referential possibilities of a pronoun. In this section, we examine whether the predictions made by this class of analyses are correct for Mandarin and Teochew. We shall argue that the referential properties of ziji and kaki are not those of pronouns, but rather those of bound anaphors. (A similar position is argued for by Pollard and Xue (this volume)). Thus, we shall conclude that ziji and kaki are not logophoric pronouns (or anaphors bound by logophoric null pronouns), but rather bound anaphors which are subject to logophoric conditions. This contrasts with the situation found in Icelandic, in which there is reason to think that at least some "long-distance reflexives" are really logophoric pronouns. It also contrasts with the facts of Turkish as described by Kornfilt (this volume), who provides evidence that a structure roughly like that proposed by Yu for Chinese is, in fact, correct for Turkish. Furthermore, the Turkish facts are parallel to those reported earlier by Cole and Hermon (1998) for Malay. In that language, there is also evidence that seeming "LD reflexives" are, in fact, pronouns. We, therefore, shall argue that languages differ with respect to whether reflexive forms that take long-distance antecedents are pronouns or bound anaphors. While apparent LD reflexives in some languages are pronouns, those in other languages (e.g., ziji and kaki in Mandarin and Teochew) are true LD bound anaphors.
10. THE C-COMMAND REQUIREMENT There are a number of differences between pronouns and bound anaphors that can be used in determining whether a given form is a pronoun or an ana-
26
Peter Cole et al.
phor. The first is c-command (and/or subcommand, etc.). Across languages, reflexives require a c-commanding antecedent, while pronouns do not: (53) [John's; brother]j looked at him^ in the mirror. (54) [John'si brother]j looked at himself*i/j in the mirror. Examples (53)-(54) show that in English a reflexive must be c-commanded by its antecedent but a pronoun need not be. The situation is somewhat more complicated in Chinese since, as was discussed in Section 3, the antecedent for a reflexive may, under appropriate circumstances, subcommand, rather than c-command, the reflexive. However, as was shown by (10), (repeated), (10) Zhangsani gei Lisij de quangao bangzhu le tai/j /ziji i*j Zhangan give Lisi 's advice help perf him self 'That Zhangsan gave Lisi advice helped him.' a nominal which neither c-commands nor subcommands ziji (e.g., Lisi) cannot be the antecedent for ziji: Although Lisi cannot antecede ziji, this nominal can antecede the pronoun ta.31 Thus, sentences like (10) make it is possible to test whether c-command/subcommand is required for the antecedent of LD ziji. It is not entirely clear whether (10) should be viewed as local or LD. There is no such uncertainty with respect to (55), which is clearly LD: (55) Zhangsani gei Lisij de xin shuo Malik renwei ziji i/*j/k /tai/j/k shi Zhangsan to Lisi 's letter say Mary think self he is wugude. innocent r^ 'Zhangsan's letter to Lisi says that Mary thinks \ \ is innocent.' [she) In (55), the specifier of gei, Zhangsan, can serve as the LD antecedent of ziji. However, the non-c-commanding/non-subcommanding complement of gei, cannot. Turning to ta, the complement of gei is wellformed as the antecedent. Thus, we conclude that LD ziji requires c-command/subcommand, and, hence, has the referential possibilities of an anaphor rather than that of a pronoun.
11. VP ELLIPSIS A second test which distinguishes between pronouns and bound anaphors is VP ellipsis. Under VP ellipsis a pronoun can be shown to be ambiguous between a constant and a bound variable:
Grammatical and Discourse Conditions
27
(56) [slMaryi; believes that John saw he^ at the movies] and [S2Susan does too]. Oversimplifying somewhat, in order for ellipsis to take place, the covert VP of S2 must be identical with the overt VP of SI. S2 of (56) can mean either that Susan believes John saw Mary at the movies (the strict reading) or that Susan believes John saw Susan at the movies (the sloppy reading). Under the strict reading, the pronoun of SI has been interpreted as a constant referring to Mary. Thus, under that reading, the covert VP of S2 is [yp saw M], where M is a constant referring to Mary. Under the sloppy reading, however, the pronoun of SI is taken to be a variable (x), picking out whatever individual is the subject of the verb phrase "believes that John saw x"32 Since the subject of the verb phrase is Susan (and not Mary), the VP is interpreted ultimately as "believes that John saw Susan." While the pronoun her can be interpreted as either a constant or as a bound variable, this is not the case for the reflexive herself. (57) [S1 Mary saw herself in the mirror] and [S2 Susan did too]. At least for many speakers, (57) is restricted to the sloppy reading of S2. This would follow if reflexives, unlike pronouns, are only interpretable as bound variables and not as constants. Thus, in English, VP ellipsis provides a way to distinguish reflexives and pronouns. While pronouns can have either a strict or a sloppy interpretation under ellipsis, reflexives have only a sloppy interpretation.33 Returning to Chinese, it is somewhat difficult to find a construction with properties like those of English VP ellipsis. The use of ye yiyang 'also the same' seems to provide a similar test: (58) [S1 Zhangsan kanjian ziji]', [s2 Lisi ye yiyang]. Zhangsan see self Lisi also the same 'Zhangsan saw himself and so did Lisi.' In (58), S2 can only mean that Lisi saw himself and not that he saw Zhangsan. Thus, local reflexives in Mandarin, like those in English, are interpreted as bound variables rather than constants. The same holds true in Teochew, the judgments for which match those in Mandarin: (59) Ah Meng toidio kaki; Ah Leng a si be yd. Ah Meng see self Ah Leng the same 'Ah Meng saw himself in the mirror and so did Ah Leng.' Turning now to LD reflexives, sentences with ziji receive only a sloppy interpretation, indicating that ziji is a bound variable rather than a constant:
28
Peter Cole et al.
(60) Zhangsan^ shuo Lisi changchang kuidai ziji-i, Wangwu ye Zhangsan say Lisi always mistreat self Wangwu also yiyang. the same 'Zhangsan; says that Lisi always mistreats himi; so does Wangwu [say Lisi mistreats Wangwu].' On the LD interpretation of (60), and given the indexing found in the antecedent clause, ye yiyang is interpreted to mean that Wangwu says Lisi mistreats Wangwu, not that he mistreats Zhangsan. That is, ziji is treated as a bound variable rather than as a constant. In contrast to (60), when the pronoun ta is substituted for ziji, Wangwu can be interpreted to say that Lisi mistreats Zhangsan, the strict reading: (61) Zhangsani shuo Lisi changchang kuidai tai; Wangwu ye yiyang. Zhangsan say Lisi always mistreat him Wangwu also the same 'Zhangsanj says that Lisi always mistreats him;; so does Wangwu.' The same facts are found in Teochew: (62) Ah Meng i da Ah Leng diam diam pel kakii//i; Ah Soi a si be yd. Ah Meng say Ah Leng always pressure self/him Ah Soi the same 'Ah Meng says that Ah Leng always pressures himself/him. So does Ah Soi.' On the interpretation in which kakili refers to Ah Meng, when kaki is used a sipe yd means that Ah Leng pressures Ah Soi (the sloppy or bound variable interpretation). But when the pronoun / is used, Ah Leng may be pressuring either Ah Soi (the sloppy interpretation) or Ah Meng (the strict or constant interpretation). To summarize, the LD reflexives zijilkaki patterns with the local reflexive and is treated as a bound variable while the pronoun tali is ambiguous between a variable and a constant. This constitutes an additional reason to believe that long-distance reflexives in Chinese are anaphors rather than pronouns.
12. THE BLOCKING EFFECT REVISITED We showed in Section 6 that the Blocking Effect is due, at least in part, to the discourse requirement that the antecedent must be the PIVOT, the center of deixis in the sentence. We shall show here that in addition to the
Grammatical and Discourse Conditions
29
Blocking due to discourse restrictions, there is also grammatical Blocking that does not appear to derive from logophoric conditions on antecedents. Grammatical Blocking follows naturally from current grammatical analyses of LD reflexives (e.g., Battistella, 1989; Cole, et al, 1993; Cole and Wang, 1996; Huang and Tang, 1991; Tang, 1989). Thus, the existence of grammatical Blocking provides evidence that LD reflexives in Chinese are derived along the lines suggested in these works (e.g., by LF movement). We will present three arguments in favor of the existence of a grammatical Blocking Effect. The first argument is crosslinguistic. As was discussed in detail in Sung (1990); Cole et al. (1993); and Cole and Sung (1994) inter alia, the head movement theory of LD reflexives predicts that Blocking will occur only in languages in which overt verb agreement is not found. (In addition to such well-known cases as Chinese versus Italian, Blocking is found in Malayalam (no verb agreement), but not in Kannada (agreement present).34 This is because Blocking is interpreted as due to an agreement conflict. In a sentence like (63) (63) Zhangsani renwei [woj zhidao [Wangwuk xihuan zij i / * j / * k ]]. Zhangsan think I know Wangwu like self 'Zhangsan thinks that I know that Wangwu likes himself.' ziji adjoins successive cyclically to each AGR until it adjoins to the AGR of the clause of which its antecedent is the subject. When ziji adjoins to the AGR of the lowest clause, it must agree in person features with the specifier of AGR, the subject of that clause, in this case, Wangwu. Thus, in order to avoid an agreement conflict in the lowest clause, ziji must be third person:35
Note that in (64) the person features of the adjoined reflexive ziji percolate to the AGR node created by adjunction. This is because AGR in Chinese lacks inherent person features. If AGR had inherent person features, the features of AGR itself (the head) would percolate to the adjoined AGR node. This is illustrated in detail below. Ziji subsequently adjoins to the AGR of the next clause up.36
30
Peter Cole et al
As in (64), AGR itself lacks person inherent features. Thus, the person features of ziji are percolated to the adjoined AGR node. In the case of (65), however, the subject of the clause is wo T. Thus, a person conflict exists between the specifier of AGR and AGR. This agreement conflict, according to the account in Cole et al. (1993) and Cole and Wang (1996), explains why ziji in (63) can agree with the local subject, Wangwu, but not with the intermediate subject, wo, or the matrix subject, Zhangsan. What is significant for our argument is that this analysis predicts that the person conflict seen in (65) will not occur in languages in which AGR has inherent person features (e.g., Italian, Icelandic, Kannada). This is because, according to the feature percolation principles discussed in Cole et al. (1993), a reflexive adjoined to AGR will percolate its person features to AGR only if AGR itself does not have inherent person features. This follows from the principle that when there is a feature conflict between the head of a projection and some other daughter of that projection, the features of the head will percolate to the mother node. It is only when there is no conflict with the features of the head that those of a nonhead daughter will percolate to the mother node. Thus, in a language with verb agreement (i.e., with inherent person features on AGR), structure (66) will occur.
In (66), AGR has inherent person features (first person). As a result, the person features of the reflexive are not percolated to the adjoined AGR projection and no agreement violation occurs. Thus, the head movement analysis predicts that the Blocking Effect will be found only in languages like Chinese and Malayalam in which verb agreement does not occur. It will not occur in languages like Italian, Icelandic, and Kannada which do mani-
Grammatical and Discourse Conditions
31
fest verb agreement. As was shown in the works cited at the beginning of this section, this appears to be a typologically correct prediction.37 The hypothesis that the Blocking Effect is entirely logophoric provides no explanation for these typological facts. While languages might differ with respect to whether PIVOT is significant in that language, it would not be expected that the presence or absence of Blocking by subjects would correlate with the presence or absence of verb agreement in the language. Thus, the typological facts provide reason to believe that Blocking by subjects has a grammatical explanation and is not due simply to logophoric conditions on antecedents of LD reflexives. Our second argument in favor of a grammatical component for Blocking by subjects is based, in part, on facts discussed in Cole and Wang (1996). Cole and Wang note that while both subjects and the nominals following ba/bei are possible antecedents for LD reflexives, subjects create strong Blocking Effects while the nominals following ba/bei do not: (67) a. Zhangsani Zhangsan 'Zhangsan b. Zhangsan^ Zhangsan 'Zhangsan home.'
yiwei woj hui ba nik dai hui ziji i/j/k de jia. thought I will BA you take back self 's home thought I would take you back to *his/my/your home.' yiwei wo-} hui bei nik dai hui ziji*-^ de jia. thought I will by you take back self 's home thought I would be taken by you back to *his/my/your
(68) a. Zhangsan^ yiwei Lisi hui ba nik dai hui ziji i/j/k de jia. Zhangsan thought Lisi will BA you take back self 's home 'Zhansan thought Lisi would take you back to his/my/your home.' b. Zhangsani yiwei Lisi hui bei nik dai hui ziji[/j/k de jia. Zhangsan thought Lisi will by you take back self 's home 'Zhansan thought Lisi would be taken by you back to his/my/your home.' The sentences of (67) show that a second person nominal following ba/bei is a possible antecedent for ziji. The nominal following ba/bei, however, does not block reference to the subject of that clause. A first person subordinate subject, however, blocks reference by ziji to the matrix subject. In contrast, in (68), in which the subordinate subject is third person, reference to the matrix subject is possible. This pattern is explained if only subjects are relevant for Blocking (since only subjects occur as the specifier of AGR), but any c-commanding nominal (including the nominals following ba/bei) is a potential antecedent for ziji.38 The facts of (67) and (68) appear to contradict the claim made in Section 5 that the occurrence of a first or second person pronoun anywhere in the
32
Peter Cole et al.
sentence will cause a Blocking Effect. In fact, however, a comparison of (68) with (69) shows that a mild Blocking Effect has occurred in (68): (69) a. Zhangsan^ yiwei List hui ba Xiao Mingk dai hui ziji i/j/k de Zhangsan thought Lisi will BA Xiao Ming take back self 's jia. home 'Zhangsan thought Lisi would take Xiao Ming back to his home.' b. Zhangsani yiwei Lisi hui bei Xiao Mingk dai hui ZJ/fyj/k de Zhangsan thought Lisi will by Xiao Ming take back self 's jia. home 'Zhangsan thought Lisi would be taken by Xiao Ming back to his home.' While in (68) it is possible to understand ziji as referring to Zhangsan, this reading is much more salient in (69). That is, a mild Blocking Effect occurs in (68) due to the occurrence of a second person pronoun following balbei. The strength of the Blocking, however, is much less than in (67), in which a first person pronoun occurs in subject position. This is precisely the pattern that would be expected if grammatical Blocking is independent of logophoric Blocking. This pattern would have no explanation if all Blocking were due to logophoric constraints on discourse. Interestingly, while the same pattern is found in Teochew, the effect of logophoric blocking is stronger. In (70), a first or second person passive agent reduces the possibility of reference to the matrix subject by kaki more than in (68b), the equivalent Mandarin sentence: (70) Ah Meng i sio da Ah Lengj kol Ah Soik kiao tng kakii/j/k gai Ah Meng think that Ah Leng give Ah Soi bring back self 's lai. home 'Ah Meng thinks that Ah Leng was brought back to his/her home by Ah Soi.' In (70), which contains only third person nominals, kaki can refer to Ah Meng, Ah Leng, or Ah Soi. In (71), however, in which the passive agent is a second person pronoun, (71) Ah Mengi s^° da Ah Lengj ko lek kiao tng kaki/???i/j/k gai Ah Meng think that Ah Leng give you bring back self 's lai. home
Grammatical and Discourse Conditions
33
'Ah Meng thinks that Ah Leng was brought back to her home by you.' the possibility of reference to Ah Meng is markedly reduced, much more than is the case in the equivalent Mandarin examples (68b). The possibility of reference to the subject of the subordinate clause is not reduced, which suggests that this relationship must be viewed as local and, hence, not subject to logophoric conditions. While Blocking by the passive agent is stronger in Teochew than in Mandarin, Blocking by an intermediate subject is stronger still: (72) Ah Mengi sio da le ko! Ah Soik kiao tng kaki*^ gai Ah Meng think that you give Ah Soi bring back self 's lai. home 'Ah Meng thinks that you were brought back to her home by Ah Soi.' The pattern of Blocking in Teochew passives can be explained if (1) grammatical Blocking in Teochew is determined by the same principles as in Mandarin and (2) the logophoric effects in Teochew are stronger than they are in Mandarin. Such a picture would be unsurprising if, in general, the grammar of Teochew reflexives is the same as that of Mandarin, but the logophoric conditions in Teochew are stronger. We saw previously that an additional logophoric condition (the de se requirement) holds in Teochew that does not hold in Mandarin. The facts surrounding ba and passive sentences suggest that the logophoric condition shared by Teochew and Mandarin (PIVOT) is stronger in Teochew than in Mandarin. It would be of interest to know whether the somewhat "bleached" picture of Mandarin which emerges from this analysis is peculiar to Singapore, where Mandarin was introduced artificially a relatively short time ago, or whether it is true as well in the Mandarin heartland, northern China. As Xu Liejiong pointed out to us, there has been considerable variation in the description of the facts in Mandarin as reported by different authors, suggesting that not all Mandarin speakers share the same discourse/semantic requirement for LD ziji. The existence of this variation is to be expected if subdialects and idiolects differ with respect to the strength of various discourse principles, which interact with invariant principles of grammar. Indeed, while our attention in this chapter has been focused on differences among Chinese dialects with regard to one class of discourse principles, those concerned with the role of logophoricity, principles of logophoricity may well be only one class of discourse principles that affects the distribution of LD reflexives. Extending our perspective beyond Chinese dialects,
34
Peter Cole et al
Jayaseelan (1998), in a study on the Blocking Effect in Malayalam, shows that a variety of discourse factors can affect Blocking, among them distance from the antecedent and salience. A detailed investigation of crossdialectal and crosslinguistic difference in discourse conditions on LD reflexives is of pressing importance for future progress. Our third and final argument that the Blocking Effect has a grammatical as well as a discourse basis is provided by a close examination of sentences in which a third person subject blocks a first or second person antecedent, as in (73). (73) Zhangsani renwei [wj zhidao [Wangwuk xihuan ziji*i/'/k]]. Zhangsan think I know Wangwu like self 'Zhangsan thinks that I know that Wangwu likes himself.' Let us consider (73) from the perspective of discourse. The presence of wo T in (73) is predicted to force the speaker to be the external pivot in (73), thereby preventing Zhangsan from being the internal pivot. Since, according to the theory of discourse employed in earlier sections of this paper, only the PIVOT can be the antecedent for LD ziji, it is predicted that the possible antecedents for ziji in (73) are the local subject, Wangwu (since local subjects are not affected by logophoric conditions, or the intermediate subject, wo (since wo is the PIVOT), but not the matrix subject, Zhangsan (which is neither local nor the PIVOT). That is, first and second person are predicted to block third person (since first and second person induce an external PIVOT), but third person is predicted not to block first and second person.39 It is true that blocking of first person by third person is somewhat weaker than blocking of third person by first person (as in (23), repeated): (23) Zhangsani renwei[WangwUj zhidao [wok xihuan ziji*i/j/k]], Zhangsan think Wangwu know I like self 'Zhangsan thinks that Wangwu know that I like myself.' However, the hypothesis that blocking is entirely due to discourse predicts that reference by ziji to the intermediate subject in (73) would be equally well-formed as reference to the intermediate subject in (2) (repeated): (2) Zhangsani renwei [Lisij zhidao [Wangwukxihuan ziji i/j/k ]]. Zhangsan think Lisi knows Wangwu like self 'Zhangsan thinks that Lisi knows that Wangwu likes himself.' This, however, is not the case; reference to the intermediate subject is worst in (23), bad in (73), and good in (2). These gradations of grammaticality are not predicted by either the grammatical or the discourse hypotheses alone. However, if both discourse and grammar play a role in the Blocking Effect, the gradations of wellformedness
Grammatical and Discourse Conditions
35
are predicted correctly: Sentence (2) is best because it does not violate either grammatical or discourse restrictions. Sentence (73) is somewhat informed because it violates grammatical but not discourse conditions. Finally, (23) is worst because it is informed in terms of both grammar and discourse. To conclude this section, Blocking appears to be the result of the interaction of a grammatical Blocking Effect, which is due to specifier head agreement conflict in languages that lack inherent agreement features, and logophoric Blocking due to PIVOT requirements. The apparent confusion with respect to speaker judgments on Blocking follows from the nonunitary nature of the causes of Blocking.
13. COUNTER EXAMPLES TO THE GRAMMATICAL ACCOUNT OF THE BLOCKING EFFECT We argued in the previous section that in both Mandarin and Teochew two types of Blocking occur, grammatical Blocking and logophoric Blocking. The latter is responsible for those cases of Blocking that are caused by a non-third person nominal which occurs in nonsubject position. When the blocker is in an intervening subject position, both grammatical and logophoric blocking occur, resulting in a more pronounced Blocking Effect. These claims appear to be undermined by a class of cases which we would like to discuss now, those in which our knowledge of the world demands a reflexive/antecedent relationship which is not licit according to the grammatical theory of Blocking. In such well-known sentences as those of (74), a local antecedent for the reflexive is ruled out by our knowledge of the world (Yu, 1991): (74) a. !Zongtoni qing woj zuo zai ziji i/*j de shenbian. president ask me sit at self 's side 'The President asked me to sit at his/*my side.' b. IZhangsariipa woj chaoguo ziji i/*j. Zhangsan fear I surpass self 'Zhangsan feared I would surpass him/*myself.' In the sentences of (74), the matrix subject is clearly more felicitous as an antecedent than is the complement subject. This appears to be due to our understanding of the real-world context in which the discourse takes place: The president would not ask me to sit beside myself, an impossible action. Therefore, the sentence is interpreted as an invitation to sit next to the president. Similarly, Zhangsan would have no reason to fear my surpassing myself. Thus, the natural interpretation would be for him to fear my surpassing
36
Peter Cole et al.
him. As a result, the sentences of (74) have been taken to demonstrate that our knowledge of the world can override the Blocking Effect. This is not expected if the Blocking Effect is grammatical. Thus, sentences like those of (74) have been interpreted as constituting a counterexample to the grammatical theory of Blocking. While this argument appears impressive, upon reflection, it is not entirely satisfying. First, while the matrix subject is clearly preferred as antecedent for ziji over the subordinate subject, these sentences are not perfect even on the preferred reading. Second, the sentences improve considerably when ta ziji or ta is substituted for ziji: (75) a. Zongtorii qing woj zuo zaita zijii/*j deshenbian. president ask me sit at he self 's side 'The President asked me to sit at his/*my side.' b. Zhangsan^ pa woj chaoguo ta zijii/*j. Zhangsan fear I surpass he self 'Zhangsan feared I would surpass him/*myself.' (76) a. Zongton^ qing wo-} zuo zai ta^ de shenbian. president ask me sit at his side 'The President asked me to sit at his/*my side.' b. Zhangsan^ pa woj chaoguo tai/*j. Zhangsan fear I surpass him 'Zhangsan feared I would surpass him/*myself.' On the assumption that "LD ta ziji" is really the pronoun ta, which, like any noun phrase, can be modified by ziji used as an intensifier, examples like those of (75) and (76) differ from (74) in that in (75) and (76) pronouns rather than LD reflexives are used for interclausal coreference.40 The marginality of (74) is predicted if the Blocking Effect is due to both grammar and discourse. While the real world context might favor the LD use of ziji in (74), the fact that (74) violates the grammatical constraints on Blocking would lead to reduced grammaticality. Thus, both the reduced grammaticality of (74) and the contrast in grammaticality between (74), on the one hand, and (75) and (76) on the other, are predicted if there exists a grammatical Blocking Effect in addition to a discourse-based Blocking Effect.
14. CONCLUSION We have argued that the distribution of the LD reflexive ziji in Mandarin and kaki in Teochew is determined by two sets of principles, discourse and
Grammatical and Discourse Conditions
37
grammar. Where these two Chinese dialects appear to differ is in terms of the discourse conditions on the reflexive. The grammatical conditions, however, appear to be the same in the two dialects. We have shown that both Mandarin and Teochew LD reflexives are subject to logophoric conditions on their antecedents. In Teochew, the antecedent must be PIVOT and also SELF or SOURCE. The SELF/SOURCE requirement, we have argued, derives from a de se restriction on the antecedent of the reflexive: The antecedent must ascribe to himself the proposition which is the object of belief, thought, etc. In Mandarin, in contrast, the antecedent need only be PIVOT. There is no de se requirement, at least in the variety of Mandarin under discussion in this chapter. While Mandarin and Teochew differ in terms of the discourse requirements on LD reflexives, in both Mandarin and Teochew, LD zijilkaki obeys the grammatical conditions on bound anaphors rather than those on pronouns. Thus, while subject to logophoric conditions, zijilkaki do not constitute covert instances of logophoric pronouns like those found in some African languages, as do the seeming LD reflexives in Turkish (Kornfilt, this volume), Icelandic subjunctives (Reuland, this volume), and Malay (Cole and Hermon, 1998). Rather, the LD reflexives in Mandarin and Teochew are a type of bound anaphor which is subject to discourse/semantic conditions specific to LD reflexives. The peculiarities of the distribution of LD reflexives in Mandarin and Teochew are of special interest because the differences in distribution between the two dialects appear, on initial examination, to suggest that there are major differences in the grammar of Binding between the two dialects, differences that might seriously undercut universalist approaches to language (e.g., Chomsky, 1995), at least as applied to Binding. On closer examination, however, the ways in which the dialects differ are seen not to be due to grammar (in the narrow and technical sense in which the term "grammar" is restricted to sentence grammar). Rather, the distributional differences are due to different discourse properties associated with specific lexical items (zijilkaki). This result is one which is unexpected when the data are examined in the absence of a strong theory of grammar, but it is what is predicted by universalist approaches to grammar: What all languages share is a universal grammatical core. Languages differ outside that core, e.g., in the "grammar of discourse" and in the semantic properties associated with lexical items. This may appear surprising because it is often tacitly assumed that the discourse properties of language are the reflection of general cognitive principles of how information is structured in language. If discourse is the direct reflection of general cognitive principles, then it should be universal. Thus, languages should not differ in terms of whether a certain discourse principle applies in that language.
38
Peter Cole et al.
While the differences between Mandarin and Teochew seem surprising from this perspective, our results may appear less surprising when they are viewed from a different angle. The differences between the two dialects can be reduced to differences in the connotations of specific lexical items (zijilkaki), whether the item in question carries with it a presupposition that the antecedent ascribes the object of belief, etc. to himself. Differences in the connotations of lexical items are frequently found in closely related dialects/languages. Thus, it should not be surprising that such differences should occur in Chinese dialects. While these lexical differences play a significant role in the distribution of the lexical items in question, the existence of these differences does not entail that languages differ radically in the more fundamental ways in which information is structured in discourse. It is interesting to note that the distinction between the universal syntactic core and the language specific discourse conditions has been the focus of investigation in recent studies of pronouns and reflexives in child language. It has been widely argued in the acquisition literature that children have more difficulties with pronouns, which are referentially dependent, than with anaphors (which are syntactically bound; see, for example, Avrutin and Wexler, 1992). More recently, it has also been argued that children exhibit difficulties with logophoric reflexives in languages like English, Dutch, and Norwegian, since their interpretation also requires extrasyntactic, discourse-based knowledge (Avrutin and Cunningham, 1997; Coopmans and Avrutin, 1999; and Hestvik and Philip, this volume). In other words, while young children exhibit near-perfect knowledge of bound pronouns and local reflexives (which are subject only to grammatical Binding conditions), they have difficulties with the interpretation of lexical items (like logophors) which require access to the discourse level of representation. The acquisition data then crucially relies on the distinction between a universal syntactic core and a nonuniversal set of discourse conditions.43
ACKNOWLEDGMENTS This chapter is one of a series of works reporting on research on reflexives and other topics in the grammar of the dialect of Teochew Chinese spoken in Singapore. The authors are listed alphabetically and participated equally in all aspects of the preparation of the chapter. The research on which the chapter is based was supported in part by funding from the National Science Foundation (grants SBR-9121167, SBR-9729519 and INT-9423291), the Max Planck Institute for Evolutionary Anthropology (Leipzig) and by a grant from the Nederlandse Organisatie voor Wetenschappelijk Onderzoek (the Netherlands Organization for Scientific Re-
Grammatical and Discourse Conditions
39
search) to the Utrecht Institute of Linguistics. We would like to express our thanks for their generous support. We received helpful comments from many colleagues while working on this paper. We would especially like to thank David Gil, Jim Huang, Xu Liejiong, Eric Reuland and Ken Safir, whose suggestions, counter proposals etc. had a profound influence on our thinking on this topic.
NOTES 1
As will be seen in later sections of this paper, it is not certain whether "discourse" properties of LD reflexives are to be analyzed as discourse or semantic properties. 2 We oversimplify here. See the discussion of c-command, subcommand, and o-command below. For discussion of an important class of examples in which the head movement theory predicts correctly that nonsubject antecedents will be possible, see Cole and Wang (1996), an analysis of cases in which the objects of ba "preverbal affected object" and bei "passive" can antecede ziji. 3 In reality, all of the previous authors were aware that there exist instances of LD reflexives that require characterization in terms of discourse conditions. However, these authors did not incorporate discourse conditions into their analyses. Thus, we will treat these analyses as purely grammatical. 4 Teochew (Chaozhou Hua) is a dialect of South Min. The variety of Teochew spoken in Singapore and Malaysia has undergone considerable lexical influence from Malay. We have not examined in any detail the extent to which Singapore Mandarin and Singapore Teochew differ from the Mandarin and Teochew spoken in other areas with respect to reflexives. We know of no evidence, however, to believe that the conditions on the use of reflexives in Singapore Mandarin and Teochew differ greatly from those on the Mandarin and Teochew spoken elsewhere. Thus, we shall refer to the Singapore varieties simply as Mandarin and Teochew. We shall return to the question of dialectal and idiolectal variation below. 5 We understand PIVOT in a less concrete way than did Sells as the role perspective taken by the sentence, that of the speaker or that of one of the protagonists of the sentence. Sells restricts PIVOT to the physical deixis. 6 Sells proposes implicational universals of logophoricity, e.g., internal SELF —» internal PIVOT. We do not adopt these implicational universals. In fact, the data of Singapore Mandarin provide counterexamples to Sells's implicational hierarchies. 7 Pollard and Xue (this volume) point out that there exist examples in Mandarin which suggest that if the notion of PIVOT is defined in a very physical sense as deixis (which is how Sells defined it), Mandarin is not subject to a PIVOT requirement: (i) Zhangsani hen gaoxing yinwei Lisi yao lai kan zij\ Zhangsan very happy because Lisi want come see self 'Zhansanj is very happy because Lisi will come to see himi.'
40
Peter Cole et al.
(ii) Zhangsani hen gaoxing yinwei List yao qu kan ziji-r Zhangsan very happy because Lisi want go see self 'Zhansan; is very happy because Lisi will go to see him.' In (ii) the deixis of qu can be from the perspective of Lisi, not Zhangsan. It is still possible, however, to understand Zhangsan as the antecedent of ziji. While ziji can be understood as coreferential with Zhangsan in (ii), we find (ii) less felicitous than (i). This suggests that when the antecedent for ziji is not PIVOT the wellformedness is reduced, but this is not sufficient to make the sentence completely informed. Thus, deixis seems to influence the perspective of the sentence, but not to determine it entirely. 8 In Chierchia's analysis, this requirement is semantic rather than pragmatic or discourse-based. Thus, if Chierchia's treatment of de se is correct, the use of the term discourse for SELF/SOURCE is incorrect, and the distinction would need to be captured in the semantics. There are, however, syntactic problems with Chierchia's analysis, so we shall take no firm position on whether de se is due to semantics or to properties of the discourse. 9 The proposal that attitudes de se differs semantically from attitudes de re is due to Lewis (1979). Subsequent to Chierchia, this proposal has been applied to long-distance reflexives by Pan (1997 and this volume) and Huang and Liu (1997 and this volume). 10 Note that according to Sells, SELF is defined as the one whose mind is being reported on rather than the one in whose mind the event is pictured. Our departure from Sells is motivated by the contrast between Mandarin and Teochew which will be discussed below. In Section 7, we shall argue that SELF is, in fact, derivative of de se, so the departure from Sells's definition is not harmful to our ultimate account of the facts. 11 Our analysis of these examples departs from that of Sells, who treats examples like those of (4) as constituting instances of internal SELF. For example, for Sells, a sentence of the form i.
John was not aware that Mary liked SELF,
could be viewed as having the logical form of ii.
->(John was aware that Mary liked SELF)
In (ii), "Mary liked SELF" is described as pictured in John's mind and the proposition as a whole is negated. That is, negation is external in (ii). Thus, the occurrence of matrix negation does not rule out the possibility that the matrix subject is SOURCE or SELF. We shall show shortly that there is empirical reason to believe that in Chinese sentences of the form of (4), the matrix subject is neither SOURCE nor SELF. As noted previously, the notions of SOURCE and SELF will later be argued to be derivative of de se. 12 AH Ma is ruled out as a potential antecedent because it does not c-command kaki. There are, however, a number of environments in which c-command as such does not hold, e.g., see the discussion of subcommand below. Some apparent viola-
Grammatical and Discourse Conditions
41
tions of c-command may involve differences between S-structure and LF. See, for instance, Sung's (1990) discussion of psych-predicates. Pollard and Xue (this volume), Pollard and Sag (1992), and Pollard and Xue (1998) provide an alternative approach to syntactic prominence, in which c-command is replaced by o-command (based on the Obliqueness Hierarchy of Pollard and Sag, 1992). 13 The examples in this section have been explained on the assumption that in sentences like those of (5), negation prevents the matrix subject from constituting an instance of internal SELF. This state of affairs is inexplicable if the logical structure of (5b) is (i)
->(Ah Meng know that Ah Leng hates SELF)
In (i), Ah Meng is a possible SELF because negation is external. If, however, the only logical form possible for (5b) is (ii)
Ah Meng -i(know that Ah Leng that hates SELF)
in which negation is internal to the VP, Ah Meng would not constitute an internal SELF. Why might external negation be excluded in Teochew and in Chinese languages generally? It is well known that, in Chinese, logical operators are interpreted in situ. For instance, as was shown in Huang (1982), in English, sentences like (iii) I didn't see many students, can be understood either as (iv) -i (many (x) (I saw x) or as
(v)
many (x) -i (I saw x)
According to (iv), the number of students that I saw was not many (perhaps 5). According to (v), however, the number of students that I saw may have been many or few so long as the number of students that I didn't see was many. Interpretation (iv) corresponds to the order of logical operator (-1, many) in (iii), but the order of logical operators in interpretation (v) is the opposite of the order of logical operators in (iii). Both interpretations are possible. Turning to Chinese, in (vi) Wo meiyou kanjian hen duo xuesheng. I not see very many student T did not see many students.' the only possible interpretation is (vii) -i (many (x) (I saw x) in which the order of logical elements in LF is isomorphic to the order in the sentence. What can be concluded from these examples is that, in Chinese, an operator appears in the same position in surface structure as in logical form.
42
Peter Cole et al
We shall now return to why an external negation interpretation might not be possible for sentences like those of (5). We see that in Chinese negation occurs in surface structure in an immediately pre-VP position. On the assumption that negation occurs in the same position in LF as in surface structure in Teochew as well as Mandarin, it would be expected that negation in Chinese is VP negation (i.e., of type <<,e,t,>,<e,t») and not sentential negation (type ). In the absence of external negation, the logical form of (5b) is (ii) (repeated), (ii)
Ah Meng -i(know that Ah Leng that hates SELF)
thereby predicting that in Chinese Ah Meng cannot be the internal SELF in (6b). 14 Unlike English, in Mandarin and other Chinese dialects, the antecedent for a reflexive must be animate. 15 Subcommand is a recursive relation so that a specifier of a specifier, etc. may constitute a wellformed antecedent for a reflexive. Tang defines the subcommand requirement as follows: (i) (ii)
(iii) (iv)
The antecedent of a reflexive must be animate. a reflexive can be bound by 6 iff a. 6 is coindexed with a, and b. 6 subcommands a, and c. 6 is not contained in a potential binder of a. A potential binder for a is any NP which satisfies all conditions of being a binder of a except that it is not yet coindexed with a. 6 subcommands a iff a. 6 c-commands a, or b. 6 is an NP contained in an NP that c-commands a or that subcommands a, and any argument containing B is in subject position.
For an alternative analysis of subcommand, see Sung (1990) and Cole, et al. (1993), where this phenomenon is treated in terms of feature percolation rather than subcommand. While Sung (1990) and Cole, et al. (1993) present arguments that the principles of feature percolation independently needed for other phenomena also account for the core cases that motivate subcommand, it should be noted that under certain definitions of c-command like that of Kayne (1993), subcommand would constitute an instance of c-command. Thus, we leave open whether a treatment in terms of feature percolation or subcommand is to be preferred. 16 For an alternative view, see Pollard and Sag (1992). 17 In contrast to ziji, the pronoun ta can refer to either the specifier or the complement of the sentential subject. 18 Huang and Liu (1997 and this volume) take as a fact that subcommanding antecedents are not possible for LD reflexives. However, our informants found them grammatical but slightly less felicitous than (10). There would appear to be considerable variation in how acceptable such sentences are to Mandarin speakers. This variation does not exist with regard to local reflexives. It may be the case that the variation is due to differences in discourse requirements on antecedents of LD reflexives for
Grammatical and Discourse Conditions
43
speakers of different dialects of Mandarin. What appears to be universal for speakers of Mandarin is that these discourse requirements hold of LD reflexives and not of their local counterparts. Thus, regardless of the absolute grammaticality of sentences like (11), they constitute confirmation of Huang and Liu's claim that LD reflexives are subject to logophoricity requirements while local reflexives are not. 19 It can be objected that the fact that the speaker is understood as an external SOURCE does not prevent Ah Meng from being internal SOURCE as well (Ah Meng is the source of speech.) A solution suggested by one reader is that the requirement should be rephrased as requiring that an LD reflexive in Teochew must have a SOURCE/SELF antecedent in the minimal domain of the perspective role. For reasons that we shall discuss in Section 7, we shall take a somewhat different approach, and shall argue that the SOURCE/SELF requirements discussed in this section can be reduced to a self-ascription requirement. 20 But see Section 7, where this formulation of the restriction is revised in terms of a de se requirement on LD reflexives in Teochew. 21 For additional arguments that nonsubjects can induce blocking, see Pollard and Xue (1998). 22 But see Sung (1990) for a discussion of well-known cases in which c-command does not hold (e.g., psych-predicates). 23 See Cole, et al. (1993) for such an analysis. 24 Reference to index k is impossible both for pragmatic reasons and because le zor gai sa si "the silly thing you have done" is inanimate. It will be remembered that reflexives in all Chinese dialects are restricted to animate antecedents. 25 Le "you" is not a possible antecedent for kaki because it fails to c-command kaki. 26 An interpretation de se is ipso facto an interpretation de re, but an interpretation de re is not necessarily an interpretation de se. 27 Chierchia considers various alternatives to (k). The differences among these alternatives would not affect the discussion which follows. 28 Chierchia uses the symbol O rather than ziji-OP. 29 A variety of details which are not relevant here are left out for the sake of simplicity. 30 Huang and Liu (this volume) develop an extension of Chierchia's treatment of de se in which the use of a LD reflexive is licensed if the sentence in question entails a self-ascriptive relation between an individual and a property. For (49) to be true, it must also be true that Ah Leng believes that she has the property that the individual in question was looking at her. 31 There are a number of apparent counterexamples in the literature (e.g., Yu, 1996) to the claim that LD reflexives must be c-commanded/subcommanded by their antecedents. These include experiencer nominals and the objects of ba and bei, all of which are wellformed antecedents for ziji, but which, under some analyses, fail to c-command (or appear to fail to c-command) LD ziji. It has, however, been argued elsewhere (e.g., Cole et al., 1990; Sung, 1990) that experiencers are, in fact, LF subjects. There is also an extensive literature supporting the claim that ba and bei are not prepositions but rather functional predicates. The apparent objects of ba and bei are, in fact, the specifiers of the projection which is the complement of ba or bei. As specifiers, the seeming objects of ba and bei are claimed
44
Peter Cole et al
to stand in a c-command relation to the VP (pace Yu inter alia). See Cole and Wang (1996), Sybesma (1992), and Zou (1993). Another approach to these problems is that of Pollard and Xue (this volume and earlier works), in which c-command is replaced by o(bliqueness)-command. While the replacement of c-command by o-command would complicate this argument, the argument could be reconstructed as one that bound anaphora require o-command while pronouns do not. Thus, a variant on this argument is possible employing the rather different principles employed by Pollard and Xue. 32 That is, the VP is interpreted as x. x believes that John saw x. 33 This is an oversimplification of the facts. A number of authors (e.g., Safir (1992) and Hestvik (1995)) have argued that strict readings are possible for reflexives under special circumstances. The argument below does not require that reflexives cannot have strict readings but only that reflexives and pronouns be distinguishable by the ease with which strict versus sloppy readings are available. 34 See Jayaseelan (1998). 35 We assume that ziji is generated freely in any person and that derivations with clashes in person features are ruled out by feature checking. 36 Presumably, the movement path of ziji obeys the Head Movement Constraint and ziji also adjoins to all heads between its original position and its ultimate position. These intermediate steps are not relevant to our discussion so we will ignore them. 37 Korean appears to present a problem for this analysis. Korean seems not to exhibit the Blocking Effect, but it also appears not to manifest verb agreement for person. However, Korean does have honorific agreement. Thus, the absence of person agreement may not constitute a counterexample. 38 It will be remembered that the nominals following balbei are analyzed as specifiers of their projections rather than as objects of prepositions. Thus, they c-command ziji. 39 This is similar to the predictions made earlier by Pan (1995,1997, and this volume), who bases his predictions on a variant of the de se hypothesis. 40 Evidence for this position is provided in Lee and Cole (to appear). 41 In their study of the acquisition of LD reflexives, Chien, et al. (1993) have found that Mandarin-speaking children (in Taiwan) have considerable difficulty with using the LD reflexive ziji. We would predict this to be due to the fact that children may have trouble figuring out the discourse conditions associated with the LD reflexive, since these conditions are not universal but language specific.
REFERENCES Avrutin, S., and Cunningham, J. (1997). Children and reflexivity. In Proceedings of the Annual Boston University Conference on Language Development 21,13-23. Avrutin, S., and Wexler, K. (1992). Development of Principle B in Russian: Coindexation at LF and coreference. Language Acquisition 2,259-306.
Grammatical and Discourse Conditions
45
Battistella, E. (1989). Chinese reflexivization: A movement to Infl approach. Linguistics 27,987-1012. Chen, P. (1992). The Reflexive Ziji in Chinese: Functional vs. formalist approaches. Research on Chinese in Hong Kong 1-36. Chien, Y.-C, Wexler, K. and Chang H.-W. (1993). Children's development of long-distance binding in Chinese. Journal of East Asian Linguistics 2,229-259. Cherchia, G. (1989). Anaphora and attitudes de se. In R. Bartsch, J. van Benthem and P. van Emde Boas (Eds.). Semantics and Contextual Expressions, 1-31. Dordrecht, Foris Chomsky, N. (1995). The Minimalist Program. Cambridge, MA. The MIT Press. Clements, G. N. (1975). The logophoric pronoun in Ewe: Its role in discourse. Journal of West African Languages 10,141-177. Cole, P. and Hermon, G. (1998). Long distance reflexives in Singapore Malay: An apparent typological anomaly. Linguistic Typology 2,221-258. Cole, P., Hermon, G. and Sung, L.-M. (1990). Principles and parameters of long-distance reflexives. Linguistic Inquiry 21,1-22. Cole, P., Hermon, G. & Sung, L.-M. (1993). Feature percolation. Journal of East Asian Linguistics 2,91-118. Cole, P., and Sung, L.-M. (1994). Head movement and long distance reflexives. Linguistic Inquiry 25,355-406. Cole, P., and Wang, C. (1996). Antecedents and blockers of long distance reflexives. Linguistic Inquiry 27,357-390. Coopmans, P., and Avrutin, S. (1999). A syntax-discourse perspective on the acquisition of reflexives in Dutch. Presented at GALA, Potsdam University, October 1999, Potsdam, Germany. Hagege, C. (1974). Les pronoun logophorique. Bulletin de la Societe de Linguistique de Paris 69,287-310. Hestvik, A. (1992). LF movement of pronouns and antisubject orientation. Linguistic Inquiry 23,557-594. Hestvik, A. (1995). Reflexives and ellipsis. Natural Language Semantics 3(2), 211-237. Hestvik, A., and Philip, W. (this volume). Syntactic vs. logophoric binding: Evidence from Norwegian child language Huang, C.T. James (1982) Logical relations in Chinese and the Theory of Grammar, MIT Doctoral dissertation; also published by Garland Publishing, New York (1998). Huang, Y.-H. (1984). Reflexives in Chinese. Studies in Literature and Linguistics 10. Huang, Y. (1997). Interpreting long-distance reflexives: A Neo-Gricean pragmatic approach. Presented at the Workshop on Long-Distance Reflexives, Cornell University. Huang, C.T. J., and Liu, C.-S. L. (1997). Logophoric versus anaphoric Ziji. Presented at the Workshop on Long-Distance Reflexives, Cornell University. Huang, C.T. James and C.-S. Luther Liu (this volume). Logophoricity, attitudes, and ziji at the interface. Huang, C.-T. I, and Tang, C.-CJ. (1991). The local nature of the long-distance reflexive in Chinese. In J. Koster and E. Reuland (Eds.), Long-distance anaphora. Cambridge: Cambridge Univ. Press.
46
Peter Cole et al.
Jayaseelan, K. A. (1998). Blocking effects and the syntax of Taan. Yearbook of South Asian Languages and Linguistics. Kayne, R. S. (1993). The Antisymmetry of Syntax. MIT Press, Cambridge. Lee, C. L. and Cole, P. Xinjiapo Huayu he Chaozhou Hua Chang Ju Li Fansheng Daici he Daici de "Luoge Zhaoying Tiaojian" (Logophoric conditions on long-distance reflexives and pronouns in Singapore Mandarin and Teochew). Studies in Chinese Linguistics 14. Levinson, S. C. (1987). Pragmatics and the grammar of anaphora: A partial pragmatic reduction of binding and control phenomena. Journal of Linguistics 23, 379-434. Levinson, S. C. (1995) Three levels of meaning. In F. R. Palmer (Ed.), Grammar and Meaning: Essays in Honor of Sir John Lyons, Cambridge Univ. Press, Cambridge, UK. Lewis, D. (1979). Attitudes de dicto and de se. The Philosophical Review 88,513-543. Pica, P. (1987). On the nature of the reflexivization cycle. In Proceedings ofNELS 17, Vol. 2, 483-499. Amherst: Graduate Linguistic Student Association, Department of Linguistics, University of Massachusetts. Pollard, CJ. and Sag, I.A. (1992). Anaphors in English and the scope of binding theory. Linguistic Inquiry 23,261-303. Pollard, CJ. and Xue, P. (1998). Chinese reflexive ziji: Syntactic reflexives versus nonsyntactic reflexives. Journal of East Asian Linguistics 7,287-318. Progovac (1992). Relativized SUBJECT: Long distance reflexives without movement. Linguistc Inquiry 23,671-680. Safir, K. (1992). Implied non-coreference and the pattern of anaphora. In Linguistics and Philosophy: An International Journal 15(1), 1-52. Sells, P. (1987). Aspects of logophoricity. Linguistic Inquiry 18,445^79. Sijbesma, R. (1992). Causatives and accomplishments: The case of Chinese Ba. Doctoral dissertation, University of London, UK. Sung, L.-M. (1990). Universals of reflexives. Doctoral dissertation, University of Illinois at Urbana-Champaign. Tang, C. T. J. (1989). Chinese reflexives. Natural Language and Linguistic Theory 7, 93-121. Xu, L. (1992). The long-distance binding of ziji. Journal of Chinese Linguistics 21,1, 123-142 Xu, L. (1993). The antecedent of Ziji. Journal of Chinese Linguistics 22,1. Yu, X.-F. W. (1989). Chinese reflexive verbs. Unpublished manuscript. Yu, X.-F. W. (1991). Logophoricity with the Chinese Reflexives. Presented at the Third North American Conference on Chinese Linguistics. Yu, X.-F. W. (1996). A study of Chinese reflexives. Unpublished doctoral dissertation. School of Oriental and African Studies, University of London, UK. Zou, K. (1993). The syntax of the Chinese Ba construction. Linguistics 32,715-736.
LONG-DISTANCE ANAPHORS IN HINDI/URDU: SYNTACTIC AND SEMANTIC ISSUES ALICE DAVISON Department of Linguistics University of Iowa Iowa City, Iowa
1. INTRODUCTION: NONLOCALLY BOUND ANAPHORIC EXPRESSIONS In Hindi/Urdu, reflexive pronouns may have both local antecedents within the minimal clausal domain and an antecedent in a higher clause: (1) a. siitaai -nee raamj-koo [PRO j apnee (aap)-koo deekh-nee]-kee liyee Sita-Erg Ram-Dat self's seelf -Dat see-inf-Gen-for majbuur kiyaa force do-Pf 'Sitaj forced Ranij [PROj to look at self: apnee^ I apnee aap*^.]' b. raami-nee [Syaam j -koo apniii/j I khud-kiiii/j I usi/j nindaa kar-tee Ram-Erg Shyam-Dat self's/ self-Gen criticism do-Impf hu-ee] sun-aa be-Pfhear- Pf-ms 'Ranij heard [Shyanij criticizing selfj/j /hinii/j].' The simplex reflexive in (la,b) is ambiguous for local and long-distance antecedents, with preference for the local reading. The complex reflexive apnee aap in (Ib) has only a local antecedent. Pronouns are locally subject-free (lb). Genitive anaphors (1b) are only monomorphemic and have Syntax and Semantics, Volume 33 Long-Distance Reflexives
47
Copyright © 2001 by Academic Press All rights of reproduction in any form reserved. 0092-4563/00 $35.00
48
Alice Davison
both local and long-distance antecedents. Genitive anaphors may be theta-marked arguments (Ib) or adjuncts. There has been extensive discussion of the option of a reflexive antecedent outside the local domain in a number of languages, originally proposed by Pica (1984), Battistella (1987), and Huang and Tang (1989) (see Cole, Hermon, and Sung (1990) and Cole and Sung (1994) for discussion). The extension of binding domain has been based in a number of proposals on the premise that the reflexive itself undergoes head-movement to an inflectional category. Successive cyclic applications of head-movement yields both the local and the long-distance readings. This chapter explores a version of the head-movement account of long-distance anaphor binding (Cole et al. (1990) and Cole and Sung (1994)) for Chinese zijl The central idea is that lexical long-distance reflexives like Chinese ziji have the same properties as the Romance se reflexives, cliticized in overt syntax to the TENSE-V complex (LeBeaux, 1983, Battistella, 1989). The se type of reflexive is a minimal projection X°, unlike the phrasal soi-meme "self-same", and it cliticizes to a head, INFL or the V-INFL complex formed by adjunction. In this article, I revise it somewhat to account for the local and long-distance binding found in Hindi-Urdu, an Indie language of north India and Pakistan. The head movement account makes the correct prediction for this language that X° reflexives will be subject-oriented, both locally and long-distance bound, and that XP anaphors will be only locally bound. I focus on three issues here: (1) the conditions on LF movement of a reflexive clitic to a functional head, yielding both local and long-distance binding, (2) the definitions of domains and their inflectional heads which allow head-adjunction, and (3) the semantic or morphological factors which motivate this movement. I conclude by noting some problems for the head adjunction analysis, as well as data which appears to rule out for this language the alternative idea of a relativized SUBJECT proposed by Progovac (1993).
2. LEXICAL FORMS OF ANAPHORS AND INFLECTIONAL MORPHOLOGY Questions of local vs long-distance binding in Hindi/Urdu focus crucially on the lexical forms of anaphoric expressions in this language and on the inflectional categories of tense, aspect, and agreement associated with particular clausal domains. I will begin with a brief survey of the anaphoric expressions in 2.1, noting what is lexically encoded within each category and whether it is morphologically complex or not. I then give a brief overview of the inflectional morphology in Hindi/Urdu in 2.2, as these are the functional
Long-distance Anaphors in Hindi/Urdu
49
categories which are potential reflexive hosts of reflexive clitics. Their presence is determined by selection by the matrix predicate. 2.1. Pronouns and Reflexives Pronouns and anaphors contrast lexically in the features which they express. Pronouns (la) express distinctions of person (I, II, III),number (singular/plural), and orientation to the speaker (proximal/distal). They do not express gender through lexical contrast (he/she), though gender agreement is expressed in verbal agreement. Pronouns also are distinguished for morphological case. There are lexically distinct forms for direct (0 or nominative) case and for oblique morphological case. Oblique case is governed by postpositions, which serve as Case markers (like the genitive -kaa) and also express other semantic relations, such as locative relations (-see 'from, with'). (2) a. Pronouns (lexically distinguished for number, person, and proximal/distal for 3p) Distal woo '3s-Nom 'that' Proximal yah '3s-Nom 'this.' aap '2p-Hon-Nom.' turn '2p-Fam-Nom maiN 'ls-Nom'
us-kaa 'that-Gen' us-koo/usee 'that-Dat' is-kaa 'this-Gen' is-koo/isee 'this-Dat' aap-kaa '2p-Gen' aap-koo '2p-Dat' tumhaaraa '2p-Gen' tum-koo/tumheeN '2p-Dat meeraa 'ls-Gen' mujh-koo/mujhee 'ls-Dat'
b. Reflexives (not distinguished for person, number, proximal/distal) Simplex: apnaa 'self-Gen khud-kaa 'self-Gen' swayam-kaa 'self-Gen' Complex: apnaa 'self-Gen'
apnee-koo 'self-Dat' khud-koo 'self-Dat' swayam-koo 'self-Dat' apnee-aap-koo 'self s-self-Dat'
c. Reciprocals eek duusree-kaa eek duusree-koo 'One-second, each other-Gen; Dat' aapas-meeN 'among each other mutually' d. Pronoun-anaphor: Genitive pronoun, all persons with apnaa 'self-Gen' e.g., us-kaa apnaa '3s-Dis-gen. self-Gen' Reflexives (Ib) have no nominative form (except as emphatics) and are not lexically distinguished for the categories which distinguish pronouns, except for morphological case. They have no distinctions for person, number, or gender. In principle, the reflexive forms have antecedents of any number, person, or gender, though often the reflexive is used with a discourse antecedent to refer to the speaker or hearer. In this chapter, most examples involve binding relations in which the antecedent is third person.
50
Alice Davison
There is a formal distinction in the class of anaphors which does have consequences for binding and interpretation. Some forms are clearly monomorphemic (3a), while others are phrasal (3b), including the reciprocal eek duusraa "each other ('s)." The distinction breaks down in the genitive form apnaa "self's," which fuses the stem and the genitive case marker. This form is the genitive of both the simplex reflexive apnee and the complex apnee aap.l (3) a. Monomorphemic: (Reflexives) apnee-Post, khud, swayam 'self, PRO, (Recip.) aapas b. Polymorphemic: (Reflexive) apnee-aap-Post., Reciprocal) eek duusree-Post. 2.2. Inflectional Categories Verbal tense and agreement inflection are summarized in (4): (4) Verb inflection: ASPECT: imperfective, perfective, progressive, other aspect compounds TENSE: (finite) past/future/contingent/present (non finite) infinitive AGR: (finite) Person, number (non-finite) gender, number, inflected case These inflectional suffixes occur on the rightmost verb in a clause, which may be the main verb (1) or an auxiliary compounded with the main verb. Tensed sentences, whether finite or nonfinite, may project ASPECT as well. ASPECT is required in present and past finite tensed sentences, but it is not required in the imperative, future, or contingent. Tense is not always overtly represented; I assume that, like AGR, it may be fused with ASPECT as a hybrid category (Giorgi and Pianesi, 1997). The sentence projections below have the following tense/aspect structure:
Both finite TENSE and perfective ASPECT are required to license ergative subject case (Davison, 1998); if subjects originate in VP, then the subject DP
Long-distance Anaphors in Hindi/Urdu
51
passes through the specifier positions of both TENSE and ASPECT to check the [D] feature of (both) functional projections. Either TENSE or ASPECT, if projected alone, serves as a binding domain with a specifier position for the antecedent. 2.3. Interaction of Reflexive Antecedents and Verbal Inflection in Hindi/Urdu The categories of tense, aspect, and agreement are illustrated in sentences (6)-(8) following. TENSE in these sentences is realized in the copula hai/haiN 'be-present-3s/pl' and the future inflection (shown below in (16) following). ASPECT is realized as the perfective suffix on the main verb in (7) and on both the main and embedded clause verbs in (8). (6) w00i bacceej-koo apnee-aapi/*j-see kaisee alag kar 3s-Erg child-Dat self's self-from how separate do sak-tii hai? be-able-Impf is 'How can shei separate/remove the childi from selfi/*j?' (7) bacceeinee [duusree bacceei -see] apneei/*j us*i/j/k-kee khilaunee child-ms-Erg second child-ms-from self's/ 3s-Gen-mpl toy-mpl chiin li-ee (haiN) snatch take-Pf-mpl are '[One childi] snatched/has snatched from [another childj] self'si/*j toys].' The agreement relations in these sentences are different. Hindi-Urdu has just one obligatory agreement relation per clause, which is controlled by a nominative argument (subject, direct object, nominal in N-V predicates). If more than one nominative argument is available, agreement is controlled only by the one which c-commands the others. The nominative subject controls agreement as well as the reflexive in (6). The singular ergative subject in (7) is the antecedent of the reflexive, but not the controller of agreement. Instead, the plural nominative object controls agreement.2 Unlike the simple clauses in (6)-(7), the sentence in (8) contains an embedded clause. (8) kusumi-nee [mohiniii-koo apneei/j kamree-meeN baiTh-ee Kusum-fs-Erg Mohini-fs-Dat self's room-in sit-Pf-ms hu-ee] deekh-aa (hai) be-Pf-ms see-Pf-ns be-pres 'Kusumi saw/has seen [Mohinij sitting/seated in self'si/j room.' (Montaut, 1990:202)
52
Alice Davison
The matrix verb deekh-naa 'see' selects an aspectual participle and a perfective participle form huaa of the copula hoo-naa 'be-inf.' The perfective participle forms of the embedded clause verb in (8) has number-gender agreement, but here it does not show agreement with the (feminine) subject moohinii. Because the embedded clause is an example of Exceptional Case Marking, the subject is marked with a dative position -koo. Since all postpositions in Hindi/Urdu block verb agreement, the verbal forms have default morphological agreement, masculine singular (oblique). The main clause verb also has the agreement default values, 3rd person masculine singular, because neither the matrix subject nor the embedded subject is nominative. Nevertheless, both the matrix and the embedded clause subjects are possible antecedents for the reflexive apnee. Domains which have nonfinite tense/aspect morphology selected by the matrix verb do allow local reflexive binding. The sentence in (9), in addition to (8), illustrates a property of subjects: they may be marked for case by a postposition (-nee 'ergative', -koo 'dative', but they are possible antecedents of reflexive, though without also controlling verb agreement morphology. (9) moohani-koo [apnaai/*j usi/j -kaa mitr
bahut acchaa] lag-taa
Mohan-Dat self-Gen 3s-Gen friend-Norn much good strike-Impf hai is 'Mohani likes self si/*j/ hisi/i friend very much.' (Yamabe, 1990:156) Agreement is instead determined by the nominative object NP (see Appendix for discussion of some implications of these data for other accounts of X° binding). In summary, reflexives are bound by nominative subjects (6), ergative subjects (7)-(8), and dative subjects (9). If the subject is not nominative, then it does not control verb agreement. Verbs have both tense and aspectual inflection, and show Person-Number agreement (finite tense) and Number-Gender agreement (nonfinite tense/aspect). Finite tense and perfective aspect have a special status in this language, as together they license ergative case marking on transitive subjects (7)-(8). 3. OVERVIEW OF BINDING RELATIONS Before going into greater technical detail, I begin with an overview of the generalizations about binding in Hindi-Urdu which a formal description must derive. All of them are derived by the analysis I will propose, a modification of the Cole/Sung LFraising of reflexives combined with specific assumptions about agreement, tense/aspect, and the semantic composition of anaphors.
Long-distance Anaphors in Hindi/Urdu
53
3.1. Conditions on Domains, Antecedents, and Form of Anaphors 3.1.1. LOCAL BINDING Monomorphemic and polymorphemic reflexives are syntactically interchangeable in the local (simple clause) domain. The XP reflexive is often preferred in the local domain, but the X° version is also possible.3 The difference is more a question of markedness, leading to contrasts such as (10a,b): (10) a. apnee-koo deekhoo! self-Dat see -Imperative-2s-Fam 'Look at yourself (you are a mess).' (Montaut, 1994) b. apnee-aap-koo deekhoo! self's- self-Dat see-Imperative-2s-Fam 'Look out for yourself (don't worry about others).' (Ibid) The contrastive meaning in (lOb) has syntactic consequences, which will be discussed (6.2). This contrast is not possible with the possessive apnaa alone, as the genitive form is the same for both local and long-distance reflexives. 3.1.2. LOCAL COMPLEMENTARITY Pronouns and reflexives are complementary in the local domain. (11) bacceeinee [duusree bacceej-see] apnaai/*j lus*i/j/k-kaa khilaunaa child-Erg second child-from self's/ 3s-Gen toy-ms chiin liyaa snatch take-Pf-Ms 'One childi snatched from another childj self si/*j toy.' In (11), apnaa 'self's refers to the subject; us-kaa '3sg-Gen' refers to nonsubjects. 3.1.3. SUBJECT ORIENTATION Both monomorphemic and polymorphemic reflexives are subject-oriented in all domains; pronouns are disjoint from the local subject. (12) wooi bacceej-koo apnee-aapi/*j -see kaisee alag 3s child- Dat self's self -from how separate kar sak-tii hai? do be-able-Impf-Fs is 'How can shei separate/remove the childj from selfi/*j?' The sentence in (12) shows that XP reflexives are locally bound and subject oriented. The subject orientation of the X° reflexive is shown in examples (13)-(14). Pronouns are not coindexed with the subject—see (Ib) and (7).
54
Alice Davison
3.1.4. NONFINITE BINDING DOMAINS Arguments (13) and adjuncts (14) in nonfinite clauses may have either local or long-distance antecedents, though often a local antecedent is preferred. Phrasal anaphors have only local antecedents (Gurtu, 1992). Semantic and contextual information may cause one subject antecedent to be preferred cover the other. The binding of PRO (control) is stable: the matrix indirect object is consistently the antecedent of PRO in (13) and (14). (13) maaNi-nee raamj-koo [PRO j apnee-aapi/*j-koo gumnaam patr mother-Erg Ram-Dat. self-(self)-Dat anonymous letter likh-nee] kee liyee manaa kiyaa write-Inf.- for forbid do-Pf 'Motheri forbade Ranij [PROj to write seli/j anonymous letters].' [Simplex] apnee-koo = aap-koo=Ram/*mother
Ram(preferred)/mother
[Complex]
apnee
(14) maaNi-nee baccooNJkoo [PRO J apnee-i/j kamree-meeN mother-Erg children-Dat self's room-in kitaabeeN paRh-nee diiN book-Nom-Fpl read- Inf give-Pf-Fpl 'Mother allowed the children [PRO to read books in her/their room]' Reflexive binding is subject oriented even when the verb agreement varies (15)-(16).
3.1.5. DIVERGENCE OF BINDING AND AGREEMENT Reflexive coindexing and agreement do not always refer to the same antecedent (14)-(15). (15) baccooNi nee apniii/j kitaab-eeN parh-tiN children-3mpl-Erg self's book-3fpl-Nom read-Pf-fpl 'The children read self'si/*j books.' (16)
bacceei apntii/*j kitaab-eeN parh-eeN-gee children-3mpl-Nom self's book-3fpl-Nom read-Fut-3mpl 'The childreni will read self'si/*j books.'
In many cases, agreement and reflexive binding have the same antecedent (16). Nominative arguments other than the subject may trigger agreement (14)-(15), when the subject is postpositionally marked (15) or has no phi features because the subject is PRO (14). In the same sentence (14) are two separate coindexing chains, shown schematically in (17):
Long-distance Anaphors in Hindi/Urdu
55
(17) a. reflexive binding (multiple subjects) maaNi-nee baccooNj-koo [PROj apneei/j-. V-Inf..].. V-Finite] mother-Erg children-Dat self's b. Verbal agreement: (object of embedded clause): maaNi-nee baccooNj-koo |PROj [apnee] kitaabeeN mother-fs-Erg children-mpl-Dat book-fpl-Nom V-Inf] diiN give-Pf-fpl Examples like these show that agreement, reflexive binding, and control cannot be instances of the same coindexing processes, but must be independent of one another. 3.2. The Blocking Effect and the Role of AGR In Chinese, neither agreement nor tense is overtly represented, nor are there overt case distinctions, such as nominative contrasting with other cases. The account of long-distance ziji in Chinese in Cole et al (1991) and Cole and Sung (1994) is based on covert movement of ziji to AGR, that is, on the functional head which checks phi-features. The AGR projection is marked for phi-features by Specifier-head agreement with the (subject) specifier of AGR. Head to head movement of AGR places ziji in higher AGR projections, where other antecedents are found. This analysis explains the Blocking Effect (Tang, 1989), which rules out possible syntactic antecedents on the basis of nonmatching phi features of person and number. An instance of ziji with multiple syntactic antecedents should have multiple readings, but it fails to have the full set of readings, just in case an intermediate antecedent differs in person features from the outermost antecedent. Only a local antecedent is possible. In Hindi/Urdu, there is no blocking effect caused by different PNG features of multiple antecedents. (18) [hareek baat-par ravij-kaa apneej/?i-koo dooS dee-naa] each thing-on Ravi-Gen self -Dat blame give-Inf maiNi pasand nahiiN kar-taa I liking not do-Impf 'Ii don't like [Ravij's blaming self?i/j for everything].' (cf. Subbarao, 1984,54) (19) prasaadfkoo [[[meeraa^ [[apnaa ? ? i / j wahaaNjaa-naa] ucit] Prasad -Dat my self's there go-Inf proper nahiiN samajh-naa] Thiik] nahiiN lag-aa] not consider-Inf right not strike-Pf
56
Alice Davison
'It did not appear right to Prasadj [that Ij did not consider proper [self s??i/j going there]].' (Subbarao, 1984,88) The Blocking Effect would predict that there should be a clear contrast between (18) and (19): (18) should allow both local and long-distance coindexing, while (19) should not because the local antecedent is first person. But this contrast is not found. Nonfinite inflection allows some optional morphological agreement, though not for person. So if binding involves nonovert person features, we might expect that first and second person phi-features would block nonlocal readings with abstract person agreement. Rather, another principle is seen, favoring local binding independently of the PNG phi features of the antecedents. All the conditions in 3.1.2 are shared by Hindi/Urdu and some other language which has long-distance reflexives. Chinese is similar to Hindi/Urdu in points 3.1.1, 3.1.2, and 3.1.3, but different in points 3.1.4, 3.1.5, and 3.2. Russian and Italian are like Hindu/Urdu in points 3.1.4,3.1.5, and 3.2.
4. A HEAD-RAISING ACCOUNT FOR LONG-DISTANCE In this section, I adapt the head-raising account for Chinese outlined and discussed in Cole and Sung (1994), on the analogy of the se type of reflexive in Romance languages. It is a minimal projection X°, unlike the phrasal soi-meme Lit. "self-same"), and it cliticizes to a head, INFL or the V-INFL complex formed by adjunction. An extension of this idea to languages without overt clitics makes the X° reflexive cliticize to INFL in Logical Form, schematically shown in (20). INFL is realized by TENSE, ASPECT, and AGR, about which I will say more in detail following (see the summary in (25)). The reflexive is identified by its antecedent, the subject of the clause, by Specifier-Head agreement with respect to properties of the Spec NP. In (20) a-c, INFL has V adjoined to it, and then REFL.
Long-distance Anaphors in Hindi/Urdu
57
INFL stands for the complex which consists of semantically interpretable TENSE and ASPECT, as well as the formal features of agreement. Inflection is shown in more detail in the next section. 4.1. LF Movement to Tense and Aspect The LF of sentence (13) has the derivation in (21).4
58
Alice Davison
Recall that this sentence contains an embedded infinitive clause, which is a binding domain. The finite matrix clause is also a binding domain, and the result is that the sentence as a whole is ambiguous for reflexive binding of a simplex reflexive apnee: both the embedded subject PRO and the matrix subject maaN 'mother' are possible antecedents. The complex reflexive apnee aap has only local binding in the embedded clause. The X° reflexive apnee in the embedded nonfinite clause is successively adjoined to the heads in its own clause (V, Tense2, C), then to matrix V and Tense1 It enters into a head-Specifier relation with the Specifier of TENSE, PROJ in the embedded clause and with maaN 'mother' in the matrix clause, giving the ambiguity of antecedent found in (7) for apnee-koo 'self-Dat'. The unambiguous reading for the phrasal reflexive apnee aap-koo 'self's self-Dat' allows only a local antecedent PROj. In the Principles and Parameters version of this analysis, the Head-Movement Constraint (cf. Baker, 1988) does not allow an XP category to adjoin to the X° category Tense2 (or other heads) and to undergo further adjunction to Tense1 This account derives some of the important properties of anaphors in this language, the subject orientation of both local and long-distance reflexives (22), and the difference between X° and XP anaphors; XP anaphors have only local antecedents (21). Surface word order does not disturb the binding relation. Like other objects, reflexive objects can be preposed (see Mahajan, 1990, Jones, 1993, and Kidwai, 1998). Reflexive binding remains local if the direct object is in an outer specifier position above and to the left of the subject.5 Not all semantic clausal domains allow local binding. Causative complements do not count as embedded domains (Baker, 1988). Though raam is a reflexive antecedent in (22a), it ceases to have subject status if embedded in a causative construction (22b). The X° reflexive apnee-koo 'self-dat' has only a matrix subject antecedent pulis 'police' in (22b): (22) a. raami-nee [apnee-^-.-koo/usee*i/j paisaa curaa-tee hu-ee] Ram-Erg self -Dat / 3s-Dat money steal-impf be-Pf deekh-aa see-Pf 'Rami saw [himselfi/^/him*i/j stealing the money]', (e.g., on a videotape) b. pulisi-nee raamj-koo [apneei/*j-koo paisaa curaa-tee hu-ee] police-Erg Ram-Dat self -Dat money steal-impf be-Pf dikh-aa-yaa see-Caus-Pf 'The policei showed Ranij [themselvesi/*himselfj stealing the money.]'
Long-distance Anaphors in Hindi/Urdu
59
Reciprocals, in contrast with reflexives, are typically phrasal; eek duusree 'one -second/other.' They do not have to have subject antecedents and are only locally bound: (23) [raarrii aur Syaamj _kaa Ram and Shyaam-Gen nindaa kar-naa] vijayk-koo criticism do-Inf Vijay-Dat 'It's all right with Vijayk [for otheri+j/*i+k/*j+k's clothes.].'
eek duusreei+j/*k-kee kapRooN-kii one-second -Gen clothes-Gen acchaa lag-taa good strike-Impf Rani and Shyanij to criticize each
The only antecedent which is possible for eek duusree 'one-other' in (23) is the plural antecedent within the minimal clause, made up of Ram and Shyam. This plural antecedent cannot include Vijay, the referent of the matrix clause dative subject, so that the excluded readings involve Ram and Vijay, Shyam and Vijay, or all three together. 4.2. Differences from the Cole/Sung Analysis I have made the LF representation in (21) somewhat different from the equivalent in the account of Chinese in Cole and Sung (1994) First, I propose that the host functional category to which the reflexive is cliticized is not AGR, but TENSE. Following, I will offer some arguments for why TENSE (and ASPECT) are the relevant functional heads in Hindi/Urdu, and in the appendix I give some reasons why AGR plays no essential role in reflexive binding in this language. Second, I come to a different conclusion about how cliticization takes place in the matrix clause. Cole and Sung point out that there are two options for moving the REFL cliticized to the embedded TENSE. The reflexive clitic might move independently from the (embedded) TENSE 1 to (matrix) TENSE 1, as in (24a), or does the whole TENSE2 complex move, resulting in (24b)?
Cole and Sung opt for (24a), independent movement of ziji, because Chinese makes no distinction in verbal inflection between finite and nonfinite tense. In Hindi/Urdu, finite tense inflection is lexically distinguished from various types of nonfinite inflection in clauses, and the distinction has both syntactic and semantic consequences. Long-distance reflexives in Hindi/
60
Alice Davison
Urdu are only possible out of embedded domains not marked for finite tense (5.2 following). The nature of Tense suggests that (24b) is correct for Hindi/Urdu. Finite tense is deictic (Partee, 1984), but nonfinite tense is not. Nonfinite, tense and aspect are not independently referential, but must be interpreted relative to the reference time specified in the highest TENSE projection (Borer, 1986; Enc, 1997); see also an account of tense and aspect in the context of the Miminalist Program in Giorgi and Pianesi (1997) If (24b) is the result of reflexive movement in Hindi/Urdu, the embedded tense and the reflexive are both referential expressions which must be identified in a local relation with an antecedent. The derivation in (21) includes (24b) as the result of clitic raising to the matrix CP.
5. THE FUNCTIONAL PROJECTION HOST FOR REFLEXIVES In this section, I argue that TENSE and ASPECT are essential to the interpretation of reflexives by serving as the host projection, but AGR is not, whether or not AGR is an independent projection.. The status of AGR as an inflectional head within the assumptions of the Minimalist Program has undergone considerable evolution (Chomsky, 1995). Whether AGR is a separate feature-bearing head or a set of [-Interpretable] case and agreement features on TENSE/ASPECT, it captures an association between nominative case and agreement of phi-features, person, number, and gender. This association is well supported empirically in a number of languages with overt agreement, including Italian (Giorgi and Pianesi, 1997, 54-55), Russian (see Appendix), and Hindi/Urdu (see preceding examples). 5.1. The Blocking Effect and the Role of AGR/Phi Features My claim that AGR does not play an essential role in reflexive binding is based on the different morpho-syntactic status of the inflectional categories in Chinese and Hindi/Urdu. The explanation proposed for Chinese by Cole and Sung captures the subject orientation of ziji, the possibility of multiple subjects by an operation of reflexive movement at LF to an inflectional head, whose specifier is a possible antecedent of the reflexive. This account can be transferred to Hindi/Urdu if some inflectional head with a specifier is available as host. The account is particularly plausible if this inflectional head enforces the Extended Projection Principle requirement for a sentence subject. AGR is a plausible choice of this inflectional head in Chinese, in which neither agreement nor tense is overtly repre-
Long-distance Anaphors in Hindi/Urdu
61
sented. There are also no overt case distinctions, nominative contrasting with other cases, which would be the result of overt checking of [-Interpretable] Case features. The covert movement of ziji to AGR checks covert phi-features., by Specifier-head agreement with the (subject) specifier of AGR. Head to head movement of AGR places ziji in higher AGR projections, where other antecedents are found. This analysis explains the Blocking Effect (Tang, 1989), which rules out possible syntactic antecedents on the basis of nonmatching phi features of person and number (see 3.6, preceding, and Jayaseelan, 1997, for discussion of the Blocking Effect in a language with tense, but not overt agreement). In the Cole/Sung analysis, syntactic binding is not a matter of person-number feature agreement, not strictly of referential indexing, which might be a post-LF matter of interpretation. To explain subject orientation in Hindi/Urdu, all that is needed is some functional projection whose specifier is plausibly a subject; in other words, some projection which has a [D] feature which enforces the requirement of the Extended Projection Principle and checks subject Case. AGR features do not check all subject cases, and in fact may be present when it expresses only the 3ms default features in the clausal types in which AGR features are obligatory. In the next section, I narrow down the possible choices of inflectional category which can be host for local and long-distance binding of subject-oriented reflexives.
5.2. Binding Domains and Functional Heads in Hindi/Urdu With a comparison of possible binding domains, it is possible, I believe, to show that TENSE/ASPECT must be present, even when AGR is not, for local binding to be possible in an embedded domain. If none of the above are possible, then it may be the case that an event head or PredP head is a host (Bowers, 1993). The functional categories summarized in 2.2 have many possibilities of combination. The host may be Finite TENSE, which has independent reference to times relative to the time of speech, must be combined with Person/Number agreement. Nonfinite tense expressed by -naa has nominal properties and can host oblique case or gender agreement. ASPECT suffixes (perfective and imperfective suffixes, progressive participle rahaa) are associated with adjectival agreement morphology: oblique case, number, and gender. The conjunctive participle -kar expresses perfective ASPECT without tense or agreement. It marks an event as completed with respect to the matrix reference time (Davison, 1981; see general discussion in Giorgi and Pianesi, 1997).6 The facts are summarized in (25), with reference to example sentences preceding and to follow:
62
Alice Davison
(25) Domains, functional categories and binding possibilities a. Finite tense: Finite Tense, Person-Number agreement, optional Aspect and Number-Gender Agreement Local binding only, no antecedents in matrix clause (26)-(27) b. Nonfinite tense -naa: infinitive, optional number/gender agreement and aspect. Local and long-distance antecedents, local antecedent preferred, (la), (13), (14), (18) c. Nonfinite Aspect: Perfective/imperfective participle, number/gender agreement Local and long-distance antecedents, local antecedent preferred. Perfective -yaa/imperfective -taa, number/gender AGR (Ib), (8), (22a,b), (28) Conjunctive participle -kar; no AGR (32) d. Small clause: With optional nonfinite copula participle: local and long-distance antecedent (9), (28) Without optional copula participle: stage level predicate (?) local and long-distance binding (35); individual level predicate: no local binding (29) e. Causative: -aa/-vaa Matrix subject only, (Mohanan, 1990; Jones, 1993) (22b) f. NP Picture nouns: local and long-distance antecedents (31) Event nominals: no local antecedent within NP. (30) The binding properties of these domains differ. Finite clauses, unlike nonfinite clauses of all types, allow only local binding. Finite clauses do not allow long-distance binding, even when a finite clause is a complement of a matrix verb. Only a pronoun can be coindexed across a finite clause boundary(26a),(27): (26) a. raadhaai (yah) pasand nahiiN kar-tii Radha this liking not do-Impf-f [ki usi/j-kaa/ apnaa*i/*j bhaaii aisee loogooN-see baat that 3s-Gen/ self s-Gen brother such people-with talk kar-ee] do-Conting-3s 'Radhai does not like (it) [that heri/j/self s*i*j brother should talk to such people.' b. raadhaai [apneei/*j bhaaii-kaa aisee loogooN-see baat kar-naa] Radha self's brother-Gen such people-with talk do-Inf
Long-distance Anaphors in Hindi/Urdu
63
pasand nahiiN kar-tii liking not do-Impf 'Radhai does not like [self'si/*j brother to talk to such people.' (Subbarao, 1984,53) (27) raami samajh-taa hai [ki useei/j /* apneei/j (aap)-koo [PROi/j Ram think-Impf is that 3s-Dat/ self's (self)-Dat caar bajee pahuNc-naa] hai four o'clock arrive-Inf is 'Rani thinks [that hei/j /*selfi/j has to arrive at 4 o'clock.'7 This restriction is true of all finite clauses, whether paratactically joined in discourse, complements of matrix verbs, relative clauses, or adverbial clauses. Exactly why this is the case is beyond the scope of this paper, but these facts suggest that embedded finite tense does not need to move to the matrix tense for interpretation, while nonfinite dependent tense does (see preceding discussion in 4.1 concerning head-movement (20a,b). It is possible that finite tense cannot move without violating a general condition (such as the CED) if finite clauses are syntactic adjuncts and also islands for wh-in situ. Nonfinite clauses are inflected for tense (-naa 'infinitive') or aspect (perfective/imperfective). Many of the examples given have shown that these domains are hosts for local binding, and allow long-distance binding (1), (8), (14), (18). These clauses have overt tense and aspect. Some verbs such as (NP-dat) lag-naa "to seem to NP, strike NP" select an embedded participial clause (28), which allows both local and matrix binding of the reflexive. Local binding only is preferred for the XP apnee aap. But it also selects a Small Clause, which allows no overt inflectional category. Compare (28) with (29) and (30), which differ in that local binding is not possible in (29b) and (30a), but is possible in (28) and (30b): (28) Nonfinite clause with aspect, and Agr: raami-koo [moohan-j apneei/j (aap)-see Sarmindaa (huaa)] Ram-Dat Mohan self's self-from ashamed (be-Pf) lag-taa strike-Impf 'Rami regards/sees [Mohanj as ashamed of himselfi/j]'; apnee aap=Mohan/??Ram, apnee=Ram/Mohan (29) Small clause with no tense or aspect: a. raami [moohanj-koo apnee aapi/*j-see Sarmindaa] samajh-taa Ram Mohan -Dat self's self -from ashamed consider-Impf hai is 'Rani regards/sees Mohan: as ashamed of himselfi/*j'.
64
Alice Davison
b. raami [moohanj-koo apnaai/*j sab-see baRaa dufman] Ram Mohan -Dat self's all-from big enemy maan-taa hai consider-Impf is 'Ram considers [Mohan his own worst enemy]'. apnaa= *Mohan/Ram us-kaa = Ram/*Mohan (Kachru and Bhatia, 1975) Other verbs like samajh "think, consider" maan "consider, regard" select small clause complements with neither Agr nor Tense/Aspect (29a,b). As Kachru and Bhatia (1975) noted, local binding within this small clause is not possible. Note that the genitive form apnaa in (29b) is the same for both X° and XP reflexives. But the nongenitive forms in (29a) show the local/nonlocal distinction very clearly.8 Local binding is not possible in (29a) within the small clause, though the XP reflexive is locally bound within the whole sentence (compare this sentence with (28), which has optionally expressed aspect huaa "be-pf" and Agr.9 There are other domains which lack local binding. On the assumption that causative predicates are bi-clausal (Baker, 1988), we might expect that local binding within the causative complement would be possible, between the reflexive and the cause. But only the causative agent is a possible antecedent (Mohanan, 1990, Jones 1994) in the standard judgements. The embedded causative complement lacks tense and aspect (and presumably AGR) (cf. Baker, 1988). NPs lack tense and perhaps aspect, and certainly have no internal AGR relation to a subject. Some NPs, such as dhookaa 'deception' are excluded as binding domains on their own: (30) Noun Phrase: (No local binding) a. *[raami -kaa apnee-koo/kee liyee dhookaa] kaanuun-kee Ram-Gen self -Dat/for sake deception law -Gen xilaaf nahiiN hai against not is 'Ram's deception of himself is not against the law.' b. [raami-kaa khudi/*j-koo dhookaa dee-naa] kaanuun-kee Ram -Gen self -Dat deception give-Inf law -Gen xilaaf nahiiN hai against not is 'It's not against the law [for Rami to deceive himself].' Binding is possible only when the N dhookaa 'deception' combines with a transitive light verb like dee 'give' or kar 'do' (30b) (though an "arbitrary" index is possible if there is a reflexive object of dhookaa}. An explanation is that the NP lacks a functional head as a host for the reflexive, which can be
Long-distance Anaphors in Hindi/Urdu
65
found only in the noun-verb combination. This is not quite the right explanation, as some "picture" types of noun do allow local binding: (31) [joon i kii apneei-baaree-meeN raay] meerii-koo Thiik nahiiN John-Gen self's about opinion Mary-Dat good not lag-ii strike-Pf 'Mary didn't like [Johni's opinion about himselfi'] (Gurtu, 1992) Picture nouns notoriously have a more complex internal structure than the ordinary common noun, and are more like event and result nominalizations, which may project an event variable. 5.3. Distinguishing Tense/Aspect from AGR as the Host Head in a Local Binding Domain We have already seen that some domains which lack overt inflectional properties do not form local binding domains (such as causative complements) or are quite variable (such as small clauses and NPs). I argue that there is local binding even when AGR is absent, and that when it is present, it is associated with nominative cause. The reflexive antecedent often does not have nominative case and does not control agreement. Agreement is often determined by an object or predicate N. If it were the reflexive clitic host, then its agreement and referential properties would often be at odds with the actual antecedent of the reflexive. The first argument concerns the conjunctive participle, marked with an uninflected suffix -kar 'perfective' (25e). The bare verb stem is combined with a suffix meaning "having V-ed," a form of perfective aspect. Agreement is impossible in this constituent: agreement morphology is completely absent, and the local PRO subject has no phi features (14/34). The conjunctive participle serves as a local binding domain, as we see from the well-formedness of the XP reflexive: (32) [PRO i/*j pahlee apnee (aap)i/*j- -koo deekh-kar] tab e-i mujheej first self's (self) -Dat see-Prt then I-Dat _ik_aa doo education give-Imper-2s-fam 'PROi/*j having first looked at yourself, then ei give mej advice.' Both forms of the reflexive are possible inside the conjunctive participle clause. The complex form apnee aap, which has only local binding, is bound by the local subject PRO. The simplex form apnee can be locally or long-distance bound by PRO or the matrix subject (the null subject of the imperative verb doo 'give-imperative-2ps-familiar'). This argument depends on the
66
Alice Davison
assumption that XP reflexives, like simplex reflexives, undergo local cliticization to an inflectional head. I justify this assumption in section 6.2, with the goal of explaining the strict subject orientation of both types of reflexive, and a semantic/pragmatic contrast between them. The second argument is based on the inconsistency of agreement and reflexive binding when ASPECT, TENSE, and AGR are all present. (33) maaNi-nee raamj-koo [PRO j apnee-aap*i/j-koo gumnaam patr mother-Erg Ram-Dat self-(self)-Dat anonymous letter likh-nee]-kee liyee manaa ki-yaa write-Inf-Gen sake forbid do-Pf 'Motheri forbade Rami [PROj to write selfi/j anonymous letters]' [Simplex] apnee-koo = Ram (preferred)/mother [Complex] apnee aap-koo=Ram/*mother The embedded clause is the object of a postposition in (33), so that no agreement is possible on the -naa TENSE suffix. Agreement is possible in sentence (34) below; it is triggered by the embedded clause object. (34) maaN i -nee baccooNj-koo [PROj apneei/j kamree-meeN kitaabeeN mother-Erg children-Dat self's room-in book-fpl. paRh-nee] diiN read-Inf give-Pf-fpl 'Motherj allowed the childrenj [PROj to read books in self Si/j (her/their) room]' The matrix subject maaN 'mother' is not available for agreement because it is marked with a postposition. The embedded clause subject PRO is not marked with a postposition, but it is not nominative either; it does not control agreement. In this sentence, the embedded clause object controls agreement, but the reflexive has two antecedents, neither of which is related to agreement. We see in (35) that embedded nominative subjects correct do control agreement: (35) raamj-koo [siitaa-j apneei/j (aap*i/j)-see naaraaz] Ram-3m-Dat Sita-Nom-3f self's -self -from angry lag-tii hai seem-Impf-f be-Pres-3s 'Sitaj seems to Rami [angry with selfi/j]' (H. Khanna) The simplex reflexive in (35) is ambiguous; but the (feminine) agreement in lag-tii is with only one subject siitaa. Here, the matrix subject fails to control agreement because it is marked with a postposition. The third argument is based on a reindexing process common to both Chinese (Tang, 1989) and Hindi/Urdu. A reflexive may have an antecedent
Long-distance Anaphors in Hindi/Urdu
67
which does not strictly c-command it, but is a specifier of a subject NP ("sub-command"). (36) [raami-kee hRday]j-meeN apnee(aap) i/*k -see ghRNaa hai Ram-gen heart -in self's self -from hatred is 'In Ram'Sj heartj is hatred of himselfi/*j; Ram hates himself in his heart.' (37) [PRO, yah baat sun-kar] [siitaai-kaa dil]j khus this matter hear-CPrt Sita-fs-gen heart-ms-NOM happy hoo ga-yaa/* ga-ii be go-pf-ms/go-pf-fs '[PRO; having heard this] [Sita'siheart ]j because happy; Sita became happy.' The reindexing of siitaai -kaa dilj with the index of siitaai does not affect the agreement of the verb. If the relevant indices consisted of PNG phi features, which are checked by AGR, then reindexing the masculine noun dil 'heart' with the features of siitaa, a feminine name, then the nominative siitaa-kaa dil 'Sita's heart' should have the feminine feature of siitaa instead of the masculine feature of the NP head dil. But this is clearly not the case. In Chinese, the reindexing could well be by AGR features, since there is no overt agreement. 5.4. Summary The preceding discussion of the binding domains in (25) has made the following points: (a) Clauses with finite tense prohibit long-distance binding. A property of Tense, not Agr, distinguishes possible long-distance binding domains. (b) Agreement is indexed independently of reflexives in local and long-distance domains (8)-(9). (c) The domains which lack local coindexing of reflexives are small clauses, causative complements, and NPs, whose primary feature is that they lack Tense/Aspect and other functional projections. The absence of Agr is a consequence of the lack of Tense/Aspect. Some local binding may be possible if there is an event argument (or PredPhrase (Bowers, 1993) but, in that case, there is no justification for Agr being present. (d) Finally, there is one clausal domain which lacks Agr but has an aspectual marker on the verb, the conjunctive participle -kar 'perfective.' Local binding is possible for both simplex and XP reflexives
68
Alice Davison
The differences of Chinese and Hindi/Urdu morphosyntax seem to entail a difference in the choice of the functional heads with may host reflexive clitics. If there is such parametric variation, it is not arbitrary, but determined by what overt morphology there is in a language and whether the functional heads are "scattered" or fused. For example, Giorgi and Pianesi (1997) argue that Italian projects AGR, but may omit TENSE (in the present). If so, then reflexives are bound by the Specifier of AGR rather than TENSE. Within the choices of host found in a given language, reflexive hosts must be the projections which require referential specifiers, to serve as antecedents. If so, then negation is not a possible reflexive host, but TENSE and other heads are, as they have a [D] feature for the Extended Projection Principle requirement, guaranteeing a subject. More speculatively, the event variable may be a host in some minimal domains, as it expresses a relation between a subject and a property. In the preceding section, I have used a version of the Cole-Hermon-Sung account of simplex reflexives as head movement, to account for the properties of apnee and apnee aap in Hindi/Urdu. The head-raising account derives the subject condition on reflexive binding, because the NP specifier of TENSE or ASPECT to which the reflexive is adjoined is also the clause subject. It also derives the multiple antecedents of the simplex reflexive, which combines with nonfinite TENSE. To get an interpretation, nonfinite TENSE raises to the matrix TENSE.10 Long-distance binding is possible only across nonfinite clause boundaries, as in Russian (cf. Progovac, 1993). This version differs from the account of Chinese in another respect. Simplex reflexives move covertly to adjoin to TENSE or ASPECT, rather than AGR. The possibilities of local binding for both simplex and phrasal reflexives depend on the availability of the right kind of host functional projection. Local binding is not possible in causative complements and event/result NP. It is not generally possible in small clauses without functional projections, as opposed to participles, which have ASPECT and AGR. Local binding is available, however, in -kar conjunctive participles which lack AGR, but do have ASPECT. AGR in Hindi/Urdu plays no direct role in reflexive binding, and there is no Blocking Effect which would result from a mismatch of phi-features between AGR and the potential antecedents, as in Chinese (Cole and Sung, 1994).
6. SYNTACTIC/SEMANTIC FACTORS There remain some properties of Hindi/Urdu reflexives which the head-raising analysis does not fully account for. These include the animacy condition on both kinds of antecedents, the subject orientation of local XP
Long-distance Anaphors in Hindi/Urdu
69
reflexives (and the lack of it on syntactically bound reciprocals), and the motivation for successive adjunctions to functional heads. 6.1. The "Animacy" Condition One very general property of X° reflexives in Hindi/Urdu, as well as Chinese, is that they typically require an animate antecedent (though this condition has exceptions in Hindi/Urdu which will be discussed). It would not be surprising that the long-distance antecedent must be animate, since the matrix verb tends to be a verb of perception/prepositional attitude, like deekh'see' or a verb like mazbuur kar- 'force' which involves a person intentionally bringing about some state of affairs. It is not a surprise that the subject of such verbs would be animate. But a local antecedent must be animate also. A representative sentence which fails this requirement is (38): (38) *kaar i nee apnee aapi/j-koo diiwaar-par tooR diyaa car-Erg self's self -Dat wall -on smash give-Pf 'The cari smashed itselfi/j against the wall.' (also: bicycle, scooter, most inanimate objects) The kind of inanimate reflexives which are commonplace in English and the impersonal reflexive in Romance languages are apparently not possible: (39) a. The plane smashed itself into the ground b. At this point the Vermilion River divides itself into two branches. (40) Le restaurant se trouve au coin de la rue. The restaurant self finds at-the corner of the street. 'The restaurant is/ is located on the corner.' Other means exist to express the changes of state for inanimate themes, like the intransitive verbs derived from nouns in (41).11 The animacy condition is met, however, by as insignificant an animate being as a bacterium (42) : (41) kaari diwaar-par zoor-see laR/bhiR ga-ii car wall-with force-with clash/collide go-Pf 'The car hit the wall with force.' (42) jiivaaNui-nee apnee aapi/*j- koo doo hissee-meeN baaNT diyaa bacteria-Erg self's self -Dat two parts-in divide give-Pf 'The bacteriumi divided itselfi/*j in two parts.' (43) aadhii raat-koo kampyuTar i apnee appi/*j-koo naST kar half night-Dat computer-Nom self's self -Dat destroyed do dee-gaa give-Fut-3 'The computei will destroy itselfi/*j at midnight.'
70
Alice Davison
The condition is also met by subject referents which are not, strictly speaking, alive, but which like a computer (43), have some sort of autonomy of action because of their complex internal structure. It is unclear whether the animacy condition also holds for reciprocals (Gurtu, 1992): (44) unhooNi -nee eek duusreei/*j-par banduuk-eeN-j taan-iiN 3pl -Erg one second-on gun-Fpl aim-Pf-Fpl 'Theyj aimed the gunsj at each otheri/*j.' (Gurtu, 1992:5) Some speakers find that eek duusree refers either to the guns themselves or those (animate) agencies which aimed them. If so, then this is another way in which reciprocals are different from reflexives. Spatial relations in general may be exempt from the condition. There is a large of sentences like (34) which are grammatical with an inanimate antecedent for the reflexive apnee, perhaps also for the reciprocal eek duusree (46). See also Mahajan, (1990).12 (45) kitaab i /(??) patthari apniii/*j jagah-see gir gaii/gayaa book-F /stone-M self's place-from fall go-Pf-F/M 'The booki/stonei fell from itsi/*j place.' (46) yee imaarteeNi+j eek duusreei+j-kee saamnee khaRii haiN these buildings-Fpl-Nom one another-Gen facing standing are 'These buildingsi+j face each otheri+j.' The condition on animacy is a clearly a semantic one. If it were a purely syntactic condition, then sentences like (45)-(46) should not be possible, nor (47), compared with (38): (47) kaar i apnee app i koo kaisee tooR sak-taa hai? car self's self -Dat how smash can-Impf is 'How can the cari destroy itselfi (it's not possible)?' Peter Hook (personal communication) suggests that it is a selectional condition on the external argument, or subject, of agentive predicates. Pronouns are disjoint from subjects and do not have this requirement. The animacy requirement appears to be one of the factors in the semantic identification of reflexives at LF, influenced by the properties of subject referents as well as the lexical properties of the verbs. 6.2. XP Reflexives In this section, I consider the fact that both X° and XP are subject oriented, but XP reflexives are only locally bound. By contrast, the reciprocal
Long-distance Anaphors in Hindi/Urdu
71
is locally (syntactically) bound but not subject oriented. I suggest that both X° and XP reflexives undergo covert movement to TENSE/ASPECT, but reciprocals do not. (I note, however, that reciprocals are anaphors, in that they require syntactic antecedents.)
6.2.1. SUBJECT ORIENTATION OF XP REFLEXIVES The head movement account correctly predicts that the simplex reflexive apnee will be subject-oriented in both local and long-distance domains, whether arguments or adjuncts. XP locally bound reflexives are also subject-oriented, though it is harder to find the right examples to demonstrate this property definitively. We need a predicate with three or more arguments, so that we can ask if the two nonsubject arguments can be coindexed. Further, the arguments must refer to animate beings, as stated previously. Sentences with morphological causative verbs meet the criterion of the number of arguments and in some cases, but as we have seen in (22b), reflexives necessarily refer to the causer, the syntactic subject in causative sentences. So the verb in the test case must be another type, such as one formed from the light verb kar-naa 'do' and a transitive adjective alag 'separate (from)': (48) wooi baccee-j-koo apnee-aapi/*j-see kaisee alag kar 3s child-Dat self s self-from how separate do saktii hai? be-able-Impf is 'How can shei separate/remove the childj from selfi/*j?' This sentence shows that the only antecedent for the XP reflexive is the syntactic subject; the other possible coindexing, though somewhat odd, is clear, yet not possible. The subject condition is a syntactic condition on reflexive binding even for XP reflexives in Hindi/Urdu (unlike in English, for example).
6.2.2. THE HEAD MOVEMENT CONSTRAINT ON LF MOVEMENT The older form of the INFL-clitic analysis assumes that the Head Movement Constraint applies to both overt and LF movement. The HMC prevents XP reflexives from undergoing head-raising to the higher clause. But if we assume that covert movement involves only formal features, the phrasal/head distinction is impossible. So there is nothing which prevents the apnee aap combination from adjoining covertly to TENSE, either the phrase in toto or just the subconstituent apnee. The problem then is to explain why further movement is not possible.
72
Alice Davison
To propose a tentative answer to this question, let us consider another property of XP reflexives. There is a semantic/pragmatic contrast between simplex and XP reflexives (10/49): (49) a. (tum) apnee-koo deekhoo! 2s-Fam self-Dat see-Imperative-2s-Fam 'Look at yourself (you are a mess).' b. (tum) apnee-aap-koo-deekhoo! 2s-Fam self self-Dat see-Imperative-2s-Fam 'Look out for (only) yourself (don't worry about others).' (Montaut, 1994:96) The XP reflexive in (49b) has a contrastive-focus sense which is not found in the pragmatically neutral simplex reflexive. There are several simplex reflexives which are neutral for Focus: apnee, khud, swayam, but only one XP reflexive apnee aap (the possessive pronoun-reflexive is also emphatic). The source of the Focus feature may be derived compositionally from the constituent aap [+ Focus] which combines with the neutral reflexive apnee.13 I propose that the focus property is what limits binding to the embedded local domain in sententially complex sentences, following a suggestion of Speas (1991). Speas resolves some puzzles of movement and pronoun binding by introducing a A' focus chain which blocks reconstruction. Here, I propose a focus chain which blocks covert head movement to the matrix clause. The reflexive XP cliticizes to the local TENSE. Then the [+Focus] feature is checked by movement to a FOCUS head above TP. Further movement is impossible, either from Focus or from the embedded TENSE.
In this representation, the reflexive apnee aap[+FOC] 'self's self is copied and adjoined to TENSE. Then the focused constituent aapf+FOCUS] moves to the Specifier of the FOCUS projection (cf. Rizzi, 1997) to check the focus feature. The reflexive adjoined to TENSE could, in principle, undergo further head-raising to a matrix tense, in a sentence like (13/21). But if apnee aap 'self's self undergoes movement over the FOCUS projection to
Long-distance Anaphors in Hindi/Urdu
73
the matrix TENSE, the Focus projection intervenes between the embedded Tense Phrase and a matrix projection of Tense. Its Head is closer to the attractor or landing site than the matrix TENSE, an instance of a Relativized Minimality violation in the terms of Rizzi (1990) or Strict Cyclicity (Collins, 1997,81ff). The strict cycle governs two instances of covert movement: adjunction of the reflexive apnee aap[+FOCUS] to embedded TENSE and adjunction of the focused element to FOCUS to check the [+Interpretable] FOCUS feature. What should be prohibited is a return to TENSE to copy the original reflexive clitic and adjoin it to matrix TENSE, to get a long-distance reflexive reading. Though covert movement is necessarily anti-cyclic because it follows all instances of overt movement (Collins, 1997,88-89), we could impose a Strict Cycle condition on all covert movement, ruling out the unwanted further movement of the XP reflexive because focus movement has intervened. Interestingly, the XP pronoun-reflexive combination us-kaa apnaa '3s-selfs' is emphatic with some contrastive focus, but does not have a strictly local or subject-oriented interpretation. It seems to be a genuine compound, one which has the binding properties of pronouns and reflexives which are consistent with one another. It has no discourse antecedent, for example, because 3p pronouns allow 3p discourse antecedents, while the reflexive allows a l/2p discourse antecedent. It is not necessarily subject oriented. So I propose that it has a + FOCUS feature, but does not undergo adjunction to TENSE. Reciprocals eek duusree (lit.) 'one-second/other' also are not required to be subject oriented and are not contrastive. So I will assume that the reciprocal is a referentially dependent expression which is not covertly moved to TENSE, nor is it + FOCUS. It is, however, required to have a syntactic antecedent. The component duusree alone means 'the other, the second one', interpreted with a discourse antecedent. The indefinite eek 'one' would seem to rule out the link to discourse, requiring referential identification within the local clause. (For discussion of the meaning inherent in various kinds of anaphoric expressions, see Safir, 1996). 6.3. Reflexive Movement Some other technical questions arise, when the original Principle and Parameters analysis is stated in more recent theoretical terms. For example, must the path of reflexive movement strictly observe the Shortest Move condition and go head to head, regardless of category? The only relevant moves are to functional projections TENSE/ASPECT, because only these moves result in the interpretation of the reflexive. We might regard the other links in the movement path as harmlessly irrelevant, as they do not af-
74
Alice Davison
feet interpretation, or else have some relativized Shortest Move condition which constrains movement to the nearest head position which could affect interpretation. I will not resolve this question here. Another technical question was discussed previously. As the head/phrase distinction disappears at LF, the Head Movement Constraint or some sort of Chain Uniformity condition cannot be what blocks further movement of XP reflexives. Instead, I propose that XP reflexives have a more complex internal structure, including a + FOCUS feature, which induces additional movement. If this movement takes place, then some version of the Strict Cycle prevents successive cyclic movement of the reflexive. The last technical question concerns the motivation for movement of reflexives. From the point of view of the theory laid out in general terms in Chomsky (1995), both overt and covert movement have to be motivated for licensing of purely formal features like Case or interpretable features like interrogative meaning or Focus. In the terms of the somewhat earlier assumptions about movement made by Cole et al, 1990, and Cole and Sung, 1994, as in Chomsky (1986), no motivation for movement was required, but if a reflexive did not cliticize to a head and enter into a head-specifier relation, then it would receive no index and would not be interpretable at LF. The account of reflexive movement for indexing in remains problematic in later theoretical terms (Chomsky, 1995). Reflexive movement does not fall within any of the categories of movement to check features. Clearly, movement is covert in Hindi/Urdu as well as in Chinese, and the result involves some Interpretable property relevant for interpretation, rather than some morphological feature. In fact, movement may be predictably covert because it affects only interpretation, and not surface form, following a proposal in Kidwai (1998). A referentially dependent expression like a reflexive, which requires a syntactic antecedent under structurally based condition (local coindexing, subject orientation) would violate Full Interpretation at the LF interface if it were not identified in some way with another referential expression. By this reasoning, reflexives move to satisfy a particular requirement of reflexives, such as "Binding Requirement," not to check a feature of the host head. Tense/Aspect in Hindi/Urdu have the EPP [D] category feature which attracts a nominal phrase by overt movement to the Specifier position. But these functional heads do not appear to require a reflexive clitic; rather, the reverse: a reflexive requires a functional category head as its host for referential identification, more in line with the GREED principle, which requires upward movement to satisfy internal feature checking requirements. Rather than reinstate GREED in the face of arguments against it (Chomsky, 1995, Chapter 4.10; Collins, 1997), we should explore the ramifications of the EPP and the internal semantic properties of reflexives.
Long-distance Anaphors in Hindi/Urdu
75
Referential identification of reflexives in Hindi/Urdu appears from the evidence reviewed previously (5.3) to be based on referential indices rather than on phi-features. The Chinese evidence from the Blocking Effect supports the claim that referential identification is syntactically encoded as phi-feature agreement, which then forms the basis for introducing the appropriate referential indices when the syntactic object is interpreted within a discourse context. Hindi/Urdu has no Blocking Effect, and agreement relations are governed by different constraints than is reflexive binding. Further, Hindi/Urdu appears to make less use of D(iscourse)-linking (Pesetsky, 1987) than does Chinese, suggesting that the LF representation encodes more syntactic/semantic information than does Chinese.14 The Cole/Sung explanation could be recast as a requirement that ziji must check some abstract, arbitrarily assigned phi features. For the kind of referential indexing found in Hindi/Urdu, we might expand the EPP constraint: the [D] feature of Tense/Aspect might have additional consequences. After attracting a nominal category to its Specifier, which adds a referential index to the highest Tense/Aspect head which is projected in a syntactic structure, this head covertly attracts all referentially dependent expressions which require a subject antecedent, however that property might be expressed. The reflexives in Hindi/Urdu differ in etymology, as noted in endnotes 2. The common semantic content involves a "self" concept, normally required to be animate or capable of autonomous acts. A related word aapaa means "self-control," and is used to refer to humans able to control their own acts (not small children, for example). The referential dependent property of at least some of these words, therefore, may stem from the reference to an animate autonomous being, explaining the central importance of a referential index rather than agreement features, in the syntactic object at LF. All this discussion remains speculative, however.
7. SUMMARY AND CONCLUSION In this chapter, I have discussed how reflexives and reciprocals in Hindi/Urdu would be accounted for using a somewhat altered version of the head movement account proposed for Chinese. I first note the properties of X° and XP anaphors in Hindi/Urdu, pointing out the cluster of similarities between the two languages with respect to anaphors: subject orientation, long-distance binding of X° but not XP anaphors, requirement of an animate antecedent in most cases. There are some important differences. Hindi/Urdu allows long-distance binding only if the embedded clause
76
Alice Davison
is nonfinite, suggesting that dependent Tense/Aspect also moves along with the reflexive, contrary to what is concluded by Cole and Sung (1994) about Chinese. Further, reflexive binding and overt morphological agreement in Hindi/Urdu are independent, though overlapping in some sentences. Because the movement of dependent Tense is required in any case, and because Hindi/Urdu has no Blocking effect explained in terms of Agr and phi-features, I propose that the host functional projection for the moved reflexive is Tense or Aspect, not Agr, in this language. Local binding would not be possible if there were no host in a particular domain. We see that local binding of both types of anaphor is possible within the conjunctive participle, which lacks Agr, but does have Aspect. Local binding is absent for the most part in domains lacking Tense/Aspect, as well as Agr, such as small clauses, NP, and causative complements. The NPs and Small Clauses which do allow local binding may project an event variable, which is a default host. In the normal case, the reflexive must adjoin to Tense, the enforcer of the Extended Projection Principle, which insures that a subject is in its Specifier or subject position. Reflexive movement is covert movement of an entire, semantically interpretable form in order to be identified referentially by a subject, which must be animate with exceptions already noted. The older version of the analysis says nothing about motivation for movement, and why it must be covert in some languages and not others. In more recent theory, movement is typically motivated by feature checking, though it is hard to see how that kind of explanation can apply to reflexives. I propose speculatively that movement of reflexives is an extension of the Extended Projection principle, and its effect is to make possible Full Interpretation. Covert movement allows only movement of features, which in the case of reflexives must be lexical or semantic, not formal. No distinction is possible between X° and XP constituents which move covertly. The local subject binding of XP reflexives is something of a problem—if it or a part of it can be head-adjoined to local Tense to derive the subject requirement, how is further movement blocked? I propose, again speculatively, that the pragmatic difference of the two forms indicates a different structure. The XP reflexive contains a +FOCUS component, which is checked by a FOCUS head above Tense, after adjunction of the reflexive to Tense. This covert movement blocks subsequent movement of the reflexive as this movement would violate the Strict Cycle condition. A still unresolved technical question is the kind of chain formed by adjunction. If there is a series of short moves to phrasal heads, then most of the moves are unmotivated, though perhaps harmless; if the moves do not result in movement to Tense, then there is no interpretation. The condition on shortest move might be relativized: movement must be to the
Long-distance Anaphors in Hindi/Urdu
77
nearest functional head which is referentially active, at which identification is possible.
8. APPENDIX: OBJECT AGREEMENT AND SUBJECT ORIENTATION An alternative account is proposed in Progovac (1993) for long-distance reflexives in languages as various as Chinese, Korean, Icelandic, and Russian. No LF movement is assumed. Instead, the classic GB binding theory is restated using relativized SUBJECTS: XP antecedents are allowed (only) for XP reflexives, and a X° antecedent, Agr, only for X° reflexives. XP reflexives always have a local antecedent, which may or may not be subject-oriented. The simplex reflexives have subject-oriented long-distance antecedents, and these properties follow from the nature of the Agr antecedent, under the assumptions that (a) Agr is coindexed only with syntactic subjects and (b) clauses without overt tense contain Agr which is dependent on a matrix Agr. In Russian, nonfinite clauses lack overt Agr, and are dependent on matrix Agr in finite clauses (cf. Borer, 1980). Finite tense/Agr is never dependent in Russian (though it may be dependent in Chinese. Hence, long-distance readings in Russian do not cross finite clause boundaries, though they may do so in Chinese. This analysis does not transfer to Hindi/Urdu, because Agr is controlled by objects as well as subjects (6)-(8), (13)-(16) preceding, and because Agr is overt in some nonfinite clauses: (51)
raarrii -koo [PRO i/*j apniiyi/*j saaikal calaa-nii] aa-tii hai Ram-3sm-Dat self's bicycle-3sf ride-Inf-f come-Impf-f 'Rami knows [(how) PROi/*j to ride self'si/*j bicycle.'
The reflexive apnii is indexed with the PRO null subject, which, in turn, is controlled by the matrix subject raam-koo 'Ram-Dat.' Yet the overt agreement on both the embedded infinitive and the matrix tensed verbal complex is controlled by the embedded clause object saaikal 'bicycle'; both postpositional NPs like raam-koo 'Ram-Dat' and PRO lack agreement features. The nominative object saaikal 'bicycle' has a full array of phi-features, which are expressed on the verbs. Nevertheless, the Borer/Progovac approach correctly represents nonfinite Tense as dependent on matrix Tense, so that PRO is controlled, in most cases, by a matrix subject. We might suggest that the EPP feature of Tense is what refers to subjects and serves as a link, independently of Agr, at
78
Alice Davison
least in Hindi/Urdu. But the relation of Agr to agreeing subjects is called into question in Russian by sentences like the following: (52) a. mnye nravitsya etat film I-ls-Dat. appeal-Pres-3s this film-Nom-3s. 'I like this film.' b. mnye nravyatsya eti filmi I-ls-Dat appeal-Pres-3pl this film-Nom-3pl. 'I like these films.' Here, the verb agreement is controlled by the nominative theme, not by the dative experiencer. A (subject-oriented) reflexive as well as a pronoun is possible in such a construction:15 (53)
reZisyorUi nravitsya yevoi/j/svoyi/j/*k director-3s-Dat. appeal-Pres-3s his /self's 'The director likes hisi/j /selfsi/j film.'
film film-Nom-3s
In the version with the reflexive, the agreement is controlled by a NP which is not also coindexed with the reflexive. The finite Agr should be the SUBJECT coindexed with the reflexive in Progovac's proposal. The independence of agreement and reflexive coindexing and control limits the applicability of the Borer-Progovac use of Agr to define or identify subjects and tense.
ACKNOWLEDGMENTS This paper is a revised and expanded version of a paper presented at the Workshop on Long-distance Anaphors, LSA Institute, Cornell University, July 5-6,1997.1 would like to thank the following: James Gair, Peter Cole, Barbara Lust, K. V. Subbarao, Kashi Wali: NSF conference Delhi U. 1996; Ajay Bharati and Rashmi Gupta, University of Iowa, for invaluable help as language consultants, and Olga Petrova and Anastasia Priyanikova for the Russian data. Part of the research for this paper was supported by a sabbatical leave from the University of Iowa and a fellowship from the American Institute of Indian Studies (1992). Thanks to Jan Parkin for her valuable help in editing.
NOTES 1
Forms ofapnaa 'self(s)' and eek duusraa 'one another' have adjectival inflection and combine like an N with post-positions. In addition to the most common reflexives apnee (aap), khud (of Perso-Arabic origin) and swayam (from Sanskrit) are also regularly used. The compound form apnee-aap-kaa 'self's-self-of is apparently pos-
Long-distance Anaphors in Hindi/Urdu
79
sible but not much used. The pronoun-anaphor (1c) has only a genitive form. The reflexive apnaa 'self's' is combined with the genitive form of any one of the pronouns, matching the antecedent in person and number. This form pronoun-GEN - apnaa_ has mixed properties (both local and long-distance binding); it requires a syntactic antecedent. See Davison (2000) for more discussion. 2 In the inflectional paradigm of masculine nouns, there is homophony which creates certain difficulties of exposition here: the nominative plural inflection -ee is identical in form to the singular oblique form. Only the former can control agreement on the verbal complex. 3 If so, then the Xo form apnee-koo 'self-dat' is not a syntactically bound pronoun, as opposed to a reflexively marked anaphor (cf. Reinhart and Reuland, 1993). 4 If AGR is projected as in Giorgi and Pianesi (1997), then the form of the tree would be (schematically):
I argue below that AGR is not a host for REFLEXIVE, and thus need not be projected. 5 As pointed out in Mahajan (1990), a scrambled direct object can be an antecedent for a possessive reflexive within the subject (i), though not an antecedent for a reflexive which is the subject (ii). (i)
moohani-koo [[apneei/*j/us-keei/j]baccooN-nee ghar-see nikaal diyaa Mohan-Dat self's/ 3s-Gen children-Erg house-from eject give-pf 'Hisi children threw Mohani out of the house.' (Mahajan, 1990,33)
(ii) *moohari -koo [[apnee i/*j -nee ghar-see nikaal diyaa Mohan-Dat self s/ 3s-Gen children-Erg house-from eject give-pf 'Mohani himselfi threw out of the house.' The sentence in (i) would be ungrammatical in the normal subject-object order, suggesting that the antecedent moohan-koo in (i) is in an outer specifier position or higher projection. If we assume that the reflexive is cliticized to the Tense project and that the antecedent for a reflexive is in the same projection, then the "scrambled" position must be an outer specifier position (Chomsky, 1995,4.10).
80
Alice Davison
In that case, the difference between (i) and (ii) reduces to a difference of c-command between the reflexive and the clitic. In (i), the clitic asymmetrically c-commands the surface site of the reflexive, because the reflexive is a proper subpart of the subject DP. In (ii), the subject phrase is identical to the reflexive DP. 6 This view of TENSE and ASPECT is very similar to the categories TENSE1 and TENSE2 in Giorgi and Pianesi (1997). Their TENSE2 corresponds to ASPECT in this chapter, and their AGR1 and AGR2 correspond to Person-Number agreement on finite tense and Number-Gender agreement on verbs with ASPECT inflection, respectively. 7 Gurtu notes that a similar sentence is grammatical with the simplex reflexive (1992:45), but my consultant finds both kinds of reflexive ungrammatical if bound across a finite clause boundary. It is possible that apnee-koo has an arb reading which is not available for apnee-aap-koo because of the contrastive property of the latter (cf. (49) following). 8 The inconsistency in small clauses may hinge on the nature of the predicate, whether it denotes a stable property of the subject (individual level) or is temporary (stage level). Examples like (29a,b) can be regarded as having individual level predicates. The absence of local binding in the Small Clause would be explained if individual level predicates project no event variable, but this suggestion remains speculative (see Raposo and Uriageraka, 1995, for critical discussion. 9 J. Gair suggests that the small clause "subject" in (23) may be not be a syntactic subject, in which case it would be ineligible as a reflexive antecedent. If so, then this NP has a different status from the SC subject in (22), perhaps because of the presence of huaa (be-pf) in (22) and its absence in (23). 10 If the matrix TENSE itself is nonfinite, then it too must raise to a matrix TENSE, and so on, until a finite TENSE is reached, with independent, deictic time reference. This characterization implies that long-distance syntactic reflexive binding is not possible across finite clause boundaries; neither apnee nor apnee aap in a ki finite clause is possible with its antecedent in the matrix, as shown previously in (20)-(21). Reflexive binding is possible across any nonfinite clause boundary; there are as many potential antecedents as there are c-commanding subjects. 11 These denominal verbs are, such as bhiR-naa 'come into conflict, collide,' N. bhiiR 'crowd,' and laR-naa 'fight, collide,' N. laRaaii 'fight, conflict.' There are also intransitive verbs related to transitives, such as Tuut-naa 'become broken,' related to tooR-naa 'break (something),' phaT-naa 'to come apart, explode' related to phaaR-naa 'split, tear,' etc. 12 Rajesh Bhatt suggest that the common factor for inanimate reflexive antecedents is a change of state or unaccusative verb. The question as a whole deserves further investigation. 13 The XP reflexive apnee aap can also be used as a emphatic marker only on subjects. (This observation is originally due to Y. Kachru, 1975.) (i)
baccaa apnee aap siiRiyaaN utar ga-yaa child-Norn self's self stairs-Nom descend go-Pf "The child got down the stairs (by) himself.'
Long-distance Anaphors in Hindi/Urdu
(ii)
81
bhaaii-nee baccee-koo apnee aap siiRiyaaN utaar di-yaa brother-Erg child-Dat self s self stairs descend-Caus give-Pf 'Brother took the child down the stairs (by) himself (refers only to brother).'
14
The arguments for this view are complex, but in summary, we note that the interpretation of wh- in situ within a clause is strictly constrained by morphosyntactic factors and is not affected by discourse information in the ways illustrated in Pesetsky, 1987. 15 Interestingly, Hindi also allows both pronouns and reflexives to be coindexed with dative experiencer subjects (Saxena, 1985).
REFERENCES Baker, M. (1988). Incorporation. Chicago: University of Chicago Press. Battistella, E. (1989). Chinese reflexivization. Linguistics 27:987-1012. Borer, H. (1986). I-Subjects. Linguistic Inquiry 17,375-416. Bowers, J. (1993). The syntax of predication. Linguistic Inquiry 24,591-656. Chomsky, N. (1986). Knowledge of Language. New York: Praeger. Chomsky, N. (1995). The Minimalist Program. Cambridge: M.I.T. Press. Cole, P., Hermon, G., and Sung, L.-M. (1990). Principles and parameters of long-distance reflexives. Linguistic Inquiry 21,1-22. Cole, P., and Sung, L.-M. (1994). Head movement and long-distance reflexives. Linguistic Inquiry 25,355-85. Collins, C. (1997). Local Economy. Cambridge: M.I.T. Press. Davison, A. (1981). Syntactic indeterminacy resolved: A mostly pragmatic account of the Hindi conjunctive participle. In P. Cole (ed.), Radical Pragmatics. New York: Academic Press. Davison, A. (1998). Licensing ergative subject case. Unpublished manuscript. Davison A. Lexical anaphora in Hindi-Urdu. In K. Wali, K.V. Subbarao, B. Lust, and J. Gair (eds.) (2000), Lexical Pronouns and Anaphors in Some South Asian Languages: A Principled Typology. Berlin: Mouton de Gruyter. Enc,, M. (1987). Anchoring conditions for Tense. Linguistic Inquiry 18,633-657. Giorgi, A., and Pianesi, F. (1997). Tense and Aspect: From Semantics to Morphosyntax. Oxford: Oxford University Press. Gurtu, M. (1992). Anaphoric Relations in Hindi and English. Delhi: Munshiram Manoharlal. Huang, C.-T. J., and Tang, C.-C. J., (1989). The local nature of long-distance reflexivization in Chinese. NELS 19,191-206. Jayaseelan, K. A. (1997). Blocking effects and the syntax of taan. Paper presented at the South Asian Language Analysis conference, Jawarharlal Nehru University, January 4-6,1997. Jones, D. (1993). The binding of Hindi anaphora: Reflexives vs reciprocals and the implications for A-binding in scrambling and subject orientation. Paper pre-
82
Alice Davison
sented at the XV South Asian Language Analysis conference, University of Iowa. Kachru, Y. and Bhatia, T. (1975). Evidence for global constraints: The case of reflexivization. Studies in the Linguistic Sciences 5.1,42-73. Kidwai, A. (1998). XP-adjunction in universal grammar: Scrambling and binding in Hindi/Urdu. Unpublished manuscript. Lebeaux, D. (1983). A distributional difference between reciprocals and reflexives. Linguistic Inquiry 14,723-730. Mahajan, A. (1990). The A/A-bar distinction and movement theory. Unpublished dissertation, M.I.T. Mohanan, T. (1990). Arguments in Hindi. Dissertation, Stanford University. Montaut, A. (1990). La reflexivisation en hindi/urdu: "Empathie" et syntaxe. LINX 22, Paris: Universite de Paris V-Nanterre. Montaut, A. (1994). Reflexivisation et focalisation en hindi/ourdu. Bulletin de la Societe Linguistique de Paris, 89 83-120. Partee, B. (1984). Nominal and temporal anaphora. Linguistic and Philosophy 7, 243-286. Pesetsky, D. (1987). Wh-in-situ: Movement and unselective binding. In E. Reuland and A. ter Meulen (eds.), The representation of (in)definiteness. Cambridge: M.I.T. Press. Pica, P. (1987). On the nature of the reflexivization cycle. NELS 17, vol. 2 483-499. Progovac, L. (1993). Long-distances reflexives: Movement to Infl vs relativization of SUBJECT. Linguistic Inquiry 24,755-772. Raposo, E., and Uriageraka, J. (1995). Two types of small clauses (towards a syntax of Theme/Rheme relations). In A. Cardinaletti and M. Guasti (eds.), Syntax and Semantics 28: Small clauses. New York: Academic Press. Reinhart, T., and Reuland, E. (1993). Reflexivity. Linguistic Inqiuiry 24,657-720. Rizzi, L. (1990). Relativized Minimality. Cambridge: M.I.T. Press. Rizzi, L. (1997). The fine structure of the left periphery. In L. Haegeman (ed.), Handbook of Generative Syntax. Dordrecht, Netherlands: Kluwer Academic Publisher. Safir, K. (1996). Semantics atoms of anaphora. NLLT 14,545- 589. Saxena, Anju (1985). Reflexivization in Hindi: A reconsideration. International Journal of Dravidian Linguistics 14,225-237. Speas, M. (1991). Generalized transformations and the D-structure position of adjuncts. In S. Rothstein (ed.), Syntax and semantics 25, Perspectives on phrase structure: Heads and licensing. New York: Academic Press, Subbarao, K.V. (1984). Complementation in Hindi syntax. Delhi: Academic Publications. Tang, C-C. J. (1989), Chinese reflexives. Natural Language and Linguistic Theory 7, 93-121. Yamabe, J. (1990). Dative constructions in Indie languages. Tokyo: University of Tokyo M.A. thesis.
REFLEXIVE ANAPHOR OR CONJUNCTIVE OPERATOR: RIAU INDONESIAN SENDIRI DAVID GIL Department of Linguistics Max Planck Institute for Evolutionary Anthropology Leipzig, Germany
1. WHY DO ONLY SOME LANGUAGES HAVE LONG-DISTANCE REFLEXIVES? Why is it the case that some languages, such as Hindu/Urdu (Davison, this volume), Chechen-Ingush (Nichols, this volume), and Mandarin Chinese (Huang and Liu, this volume), have long-distance reflexives, while other languages, such as English, Hebrew, and Riau Indonesian do not? In principle, the distribution of long-distance reflexives across languages may be an outcome of any or all of the following kinds of factors: (1) a. chance b. general properties of particular languages b. specific properties of particular forms As suggested in (la), whether or not a language has long-distance reflexives may be due to chance; or, in other words, it may be the consequence of so many interacting factors of different types, synchronic, and perhaps also diachronic, that no general systematic patterns can be discerned. Alternatively, as proposed in (lb), the occurrence of long-distance reflexives in different languages may be governed by other, more general grammatical properties of the respective languages. For example, cross-linguistic surveys may reveal implicational universals to the effect that if a language Syntax and Semantics, Volume 33 Long-Distance Reflexives
83
Copyright © 2001 by Academic Press All rights of reproduction in any form reserved. 0092-4563/00 $35.00
84
David Gil
has some property P, then it has long-distance reflexives, or alternatively, if a language has some other property Q, then it does not have long-distance reflexives. Although a logical possibility, such accounts have not yet, to the best of my knowledge, been proposed in the literature.1 What is clear, however, is that such explanations cannot be the whole story. This is because in some languages, for example Mandarin, local and long-distance reflexives are known to co-occur side by side. In such cases, no global property of the language can be invoked to account for the fact that one reflexive is local while the other is long-distance. The co-occurrence of local and long-distance reflexives within the same language suggests that in at least some cases, as put forward in (1c), the distribution of long-distance reflexives may be due to specific properties of the particular forms which constitute the long-distance reflexive constructions in different languages. Indeed, one such proposal has been widely discussed, namely, the implicational universal to the effect that, if a reflexive is long-distance, it is monomorphemic (equivalently, if a reflexive is polymorphemic, it is local)—see, for example, Pica (1984), Faltz (1985), Reinhart and Reuland (1991), Cole and Sung (1994), and Cole and Wang (1996). Accounts such as these accordingly recast the question posed at the beginning of this chapter. Rather than asking about languages, they ask about forms: Why is it that some reflexives are long-distance while others are local? This chapter suggests one possible mode of addressing the above question, accounting for the distribution of long-distance reflexives in terms of specific properties associated with particular forms, but with a twist. Specifically, it is argued that some reflexive forms are not long distance because—contrary to presupposition—they are not reflexives. Like other construction types, reflexives are defined in terms of a combination of semantic and formal properties. Consider the following: (2) a. John taught himself Icelandic b. John learned Icelandic alone Although both sentences describe roughly the same situation, only (2a) is a reflexive. Sentence (2b) lacks a characteristic feature of a reflexive construction, namely, an anaphor, that is to say, an expression whose interpretation is determined by another expression, called its antecedent. Together, (2a) and (2b) show that meaning alone does not suffice to determine whether or not a given construction may be characterized as reflexive. It follows that a translation of an English reflexive construction into some other language is not necessarily a reflexive construction in that language. In order to show that a certain construction in a particular language is reflexive, it is necessary to examine the syntactic and semantic properties of the construction within the grammatical system of the language in question.
Reflexive Anaphor or Conjunctive Operator
85
More generally, any cross-linguistic typology of reflexive constructions must, as a prerequisite, make sure that all of the constructions under consideration are indeed true reflexives within their respective languages, and not mere translational look-alikes. This chapter provides a case study of the issues involved in determining whether a given form in a certain language constitutes a bona fider reflexive construction. It is concerned with one language, Riau Indonesian, and with one particular form, sendiri. A brief introduction to Riau Indonesian is given in Section 2. Translations of English reflexive constructions into Riau Indonesian typically contain an occurrence of the form sendiri. However translations of long-distance reflexive constructions from other languages into Riau Indonesian typically do not contain the form sendiri. Thus, a superficial examination of sendiri suggests that it is a reflexive form, with local but not long-distance interpretations. However, a closer look at sendiri suggests that, contrary to first appearances, it is not a reflexive form at all. Three arguments to the effect that sendiri is not a reflexive anaphor are provided in Section 3. Rather, it is argued that sendiri is an instance of another, quite different construction type, namely, a conjunctive operator. Conjunctive operators, often referred to alternatively as focus operators, are items such as English also, too, only, just, and others. A characterization of the semantics of conjunctive operators is provided in Gil (1993; 1994a,b,c; 1995a,b). In this chapter, it is shown that this analysis can be extended straightforwardly to account also for the interpretation of sendiri. And in fact, the analysis of sendiri as a conjunctive operator provides a straightforward explanation for the unavailability of those interpretations that would—under an alternative, reflexive analysis—be characterized as long distance. The analysis of sendiri as a conjunctive operator is presented in Section 4. What this chapter argues, then, is that—in the case of Riau Indonesian sendiri, at least—an apparent reflexive anaphor lacks long-distance interpretations because, in actual fact, it is not a reflexive form at all, but, rather, a completely different kind of object, associated with a very different set of properties, specifically, those of a conjunctive operator. In Section 5, the possibility is raised that various other supposed reflexive constructions in other languages may also be amenable to such an analysis.
2. RIAU INDONESIAN When most people think of Malay /Indonesian, they generally have in mind one of the two "standard languages" of Malaysia or of Indone-
86
David Gil
sia—taught in elementary and high schools, and used in various formal or official situations, in politics, education, the media, and so forth. In particular, most of the Malay and Indonesian data cited in the linguistic literature is from one of the standard languages. However, Standard Malay and Indonesian are artefacts of conscious, politically motivated language engineering, rarefied registers which few people speak "properly" (whatever that means), and nobody acquires through the natural processes of first-language acquisition. As a result, much of the Malay/Indonesian data cited in the literature is probably more appropriately viewed as reflecting various imperfect approximations to the standard languages, produced by speakers of different regions and ethnicities. Thus, for example, the only previous study with which I am familiar of Malay/Indonesian reflexives, that of Cole and Hermon (1997,1998), describes a near-standard version of Malay used in Singapore. Alongside Standard Malay and Indonesian is a wide range of Malay/ Indonesian language varieties, acquired natively, as a first language, by a majority of the inhabitants of the Malay/Indonesian archipelago, and spoken colloquially in most everyday contexts, at home, at work, in the marketplace, indeed almost everywhere—see Gil and Tadmor (to appear) for detailed discussion. These myriad vernacular dialects or languages are of a low degree of mutual intelligibility: in their diversity they are perhaps comparable to the different varieties of Arabic or Slavonic. Sociolinguistically, and simplifying somewhat, one may roughly distinguish between two main types of Malay/Indonesian dialects: (a) "classic" dialects, spoken by ethnically homogeneous speech communities, typically small localized populations of rural Malays; as opposed to (b) "koine" dialects, associated with ethnically heterogeneous speech communities, often consisting of large rural or urban populations which are partly or entirely non-Malay. The variety described in this chapter, namely, Riau Indonesian, is a dialect of the latter koine type. Riau Indonesian is the dialect of Malay/Indonesian spoken in informal situations by the inhabitants of Riau province in east central Sumatra. Contrary to what is suggested in various linguistic atlases, such as Moseley and Asher (1994), the population of Riau is linguistically and ethnically heterogeneous. Although the indigenous population is Malay, a majority of the present-day inhabitants are migrants from other provinces, speaking a variety of languages—mostly Minangkabau, but also Batak, Bugis, and others. Riau Indonesian is acquired as a native language by most or all children growing up in Riau province, whatever their ethnicity; and it is the language most commonly used as a lingua franca for interethnic communication. In addition, like other colloquial varieties of Indonesian, it is gradually replacing other languages and dialects as a vehicle of intraethnic communication.
Reflexive Anaphor or Conjunctive Operator
87
To this point, the only descriptions of Riau Indonesian in the linguistic literature are those provided in Gil (1994d, 1999, to appear a,b). Riau Indonesian differs in many respects from Standard Malay/Indonesian. From a Eurocentric perspective, its most noteworthy feature is the pervasiveness of underspecification: the absence of obligatory overt grammatical expression for a wide variety of categories, including number, (in)definiteness, tense, aspect, thematic role, and ontological type. Consider, for example, a simple two-word expression such as the following: (3) Makan ayam eat chicken 'An association of eating and chicken' In the above sentence, ayam 'chicken' is unmarked for number and (in)definiteness, while makan 'eat' is unspecified for tense and aspect. In addition, the semantic relation between the thing word and the activity word is indeterminate: the chicken could bear any thematic role whatsoever, such as agent, patient, and so forth. Finally, the expression as a whole can be associated with any ontological category: it can denote an activity, for example, 'The chicken is eating'; a thing, for example, 'The chicken that is eating'; a time, for example, 'When the chicken is eating'; a place, for example, 'Where the chicken is eating,' and so forth. Sentence (3) thus has a single underspecified meaning, involving eating and chicken, associated with each other in an arbitrary manner. Only in actual discourse is the range of possible interpretations dramatically reduced, by the context in which the sentence is uttered. In Gil (1994d, 1999, to appear a,b) it is argued that such facts reflect a grammatical organization that does without many of the staple categories of syntactic theory. In particular, it is argued that in Riau Indonesian, there is no distinction among categories of noun, adjective, verb, and sentence, nor between lexical categories and their phrasal projections. Instead, almost all words belong to a single open syntactic category, corresponding roughly to the traditional notion of Sentence. (A theory of syntactic categories within which such claims can be made more rigorous is provided in Gil, 2000.) A corollary is that Riau Indonesian is also lacking in traditional grammatical relations, such as subject and direct object. Indeed, the reader may note that everything that needs to be said about Riau Indonesian in this chapter can be couched within a very limited technical vocabulary, without recourse to distinct syntactic categories or grammatical relations. A word on methodology. When working with Riau Indonesian, it is often difficult or impossible to elicit reliable judgments from native speakers. What happens all too often is that the moment speakers realize that they are being questioned in a "learned" context, they switch from whatever col-
88
David Gil
loquial variety they were just using into the standard language, or, rather, their sometimes imperfect variant thereof. And when speakers do provide judgments for ordinary or everyday language, they frequently make claims that are in gross conflict with their actual linguistic behavior, for example, by characterizing as ungrammatical forms or constructions that they use all the time. Faced with such obstacles, the study of Riau Indonesian reported on here makes use of an alternative method of data collection, based on the gathering of spontaneous speech specimens: actual utterances produced by native speakers in real-life situations, either jotted down right away into a notebook or recorded and subsequently transcribed. Most of the data presented is of such a character. In particular, many of the constructions containing sendiri were gathered by means of a particular ruse. On repeated occasions and in different places, I would seat myself in an outdoor coffee shop with an open laptop computer showing a drawing of a man holding a gun to his head, and I would pretend to be engrossed in working on the drawing. Often, a crowd of curious onlookers, children and adults, would gather around, and describe the picture to each other, frequently producing spontaneous translational equivalents of reflexive constructions.
3. SENDIRI AS A REFLEXIVE ANAPHOR If an English reflexive construction is translated into Riau Indonesian, the resulting construction will more often than not contain an instance of the form sendiri. Following are two examples of spontaneous speech specimens containing sendiri.2 (4) Saya tembak sendiri 1:SG shoot sendiri [Playing Nintendo game, with lots of shooting] 'I shot myself.' (5) Pakai kayu sendiri, ya? use stick sendiri yes [Playing billiards; after squabble over who is using whose cue, speaker confirms] 'I'm using my own cue, right?' In (4), sendiri looks as though it is a local reflexive taking saya as its antecedent, while in (5) sendiri appears to have a covert antecedent—though if
Reflexive Anaphor or Conjunctive Operator
89
it were present, it would presumably be similar in form and position to that in (4). Although quite large, my corpus of spontaneous speech specimens contains no examples of sendiri occurring in constructions where it would appear to have a long-distance interpretation.3 This suggests that such interpretations may be unavailable; however, it is not conclusive, since long-distance reflexives do not occur frequently even in languages that have them. In order to confirm the unavailability of long-distance interpretations, it was thus necessary to elicit judgments from native speakers—the only way to obtain explicitly negative evidence. As noted in the previous section, it is often difficult to obtain reliable judgments from native speakers of Riau Indonesian. However, three of the many speakers whom I consulted seemed to be able to provide relatively robust judgments for sentences containing sendiri. And in fact, their judgments turned out to be consistent with each other, as indicated in the following examples: (6) Alij tahu Fatimahi suka sendirii/*j/*k Ali know Fatimah like sendiri 'Ali knows that Fatimah likes herself (7) Tikusj mimpi monyeti pegang gambar sendirii/*j/*k mouse dream monkey hold picture sendiri 'The mouse is dreaming that the monkey is holding its own picture' In both (6) and (7), sendiri appears to behave like a local, clause-bound reflexive. According to all three speakers, in sentence (6), Fatimah can only be understood to like herself (i), not Ali (j) or somebody else (k); similarly, in sentence (7), the picture can only be taken to portray the monkey (i), not the mouse (j) or somebody else (k). Thus, based on these facts, sendiri appears to behave rather like a local reflexive anaphor.4 Appearances, however, can be deceiving. In fact, there are three good reasons to believe that, in spite of the preceding, sendiri is not a reflexive anaphor. These are: (a) sendiri may asymmetrically c-command its would-be antecedent; (b) sendiri may stand by itself; and (c) sendiri has a variety of clearly nonreflexive interpretations. We shall consider each of these reasons in turn.
3.1. Sendiri May C-Command Its Would-be Antecedent As anaphors, reflexives are generally c-commanded by their antecedents. However, sendiri appears, in some cases, to asymmetrically c-command its would-be antecedent.
90
David Gil
One example of this is given in (8) below: (8) Kena hidung[-nya]i sendirii undergo nose-ASSOC sendiri [Describing laptop picture of man holding gun to his head] 'He got his own nose' In (8), sendiri appears to select the possessive enclitic -nya 'his' as its antecedent. However, sendiri seems to asymmetrically c-command -nya, which is phonologically attached to hidung nose'.5 Example (9) presents a tabulation of some of the data obtained from the laptop-in-the-coffee-shop experiment in which speakers describe a drawing of a man holding a gun to his head. Of the many spontaneous utterances produced and recorded, a total of 9 consisted of exactly the same set of 3 words, tembak 'shoot,' dia 'he,' and sendiri. Remarkably, even though the context was identical, these 9 utterances exhibited 5 out of the 6 logically possible linear orders of these three words, thereby providing a vivid reflection of the freedom of word order in Riau Indonesian. The number of times each order occurred is indicated to the right of each example: (9) a. Tembak diai sendirii shoot 3 sendiri b. Tembak sendiri^ diai shoot sendiri 3 c. Diai tembak sendirii 3 shoot sendiri d. Sendirii tembak diai sendiri shoot 3 e. Ditti sending tembak 3 sendiri shoot f. Sendirii diai tembak sendiri 3 shoot [Describing laptop picture of man holding gun to his head] 'He's shooting himself
3 1 3 0 I 1
Although the distribution is not statistically significant, a tendency is evident for sendiri to follow its would-be antecedent dia; this happens in 7 out of the 9 utterances. But what of c-command? Under a conventional assignment of constituent structure, example (9f)—accounting for just 1 out of the 9 utterances—would be the only one in which sendiri asymmetrically c-commands dia. However, in actual fact, there is good reason to believe that sendiri also asymmetrically c-commands dia in examples (9a) and (9c). In general, in Riau Indonesian, syntactic constituency provides a direct iconic reflection of semantic groupings. Whereas in English, focused expressions are often
Reflexive Anaphor or Conjunctive Operator
91
marked with contrastive stress, in Riau Indonesian the most common strategy for marking focus is simply to position the focused element in a c-commanding position, typically at the beginning or the end of the sentence or the expression over which it has scope. Now in the given context, that of the man with the gun to his head, it is clearly sendiri that is in focus. Accordingly, in examples (9a) and (9c), tembak and dia form a constituent, which is the sister of sendiri; a more perspicuous paraphrase of these examples might be 'His shooting is of himself.' Thus, in (9a) and (9c), sendiri asymmetrically c-commands dia. And in total,sendiri asymmetrically c-commands its would-be antecedent dia in 7 out of the 9 utterances in (9). In summary, then, the data in (8) and (9) suggest that sendiri may readily stand in a relationship of asymmetric c-command to its would-be antecedent, behavior that is not characteristic of an anaphoric expression.
3.2. Sendiri May Stand Alone As anaphors, reflexives obtain their interpretations through coindexing with other elements, known as their antecedents. Hence, they are inherently exocentric, and therefore cannot generally stand by themselves—at least not without a very specific context, for example, as the answer to a question: Who did John shoot? Himself. In contrast, in Riau Indonesian, sendiri can readily stand on its own as a complete, nonelliptical sentence in a wide range of contexts; in this respect, it resembles almost all the other words in the language, which—as suggested in the previous section—belong to a single open syntactic category corresponding roughly to Sentence. An example of sendiri constituting a complete nonelliptical sentence is provided by the following spontaneous speech specimen: (10) Sendiri? sendiri [Checking into hotel, where, last time, I had come with a friend; this time I am alone, and receptionist asks] 'Alone?' Thus, the ability of sendiri to stand alone, as in (10), provides further reason to believe that it is not an anaphoric expression. As suggested by the preceding example, sendiri by itself means, simply, 'alone.' This brings us to the third argument against the characterization of sendiri as a reflexive anaphor, namely, that it is associated with a wide range of nonreflexive interpretations.
92
David Gil
3.3. Sendiri May Have Nonreflexive Interpretations Although sometimes, as in (4) - (9) preceding, sendiri is the translational equivalent of an English reflexive, in many other instances, as is the case in (10), sendiri is associated with meanings of seemingly quite different kinds. The range of interpretation expressed by sendiri is summarized in (11): (11) a. b. c. d. e. f.
reciprocal 'each other,' 'one another' deagentive 'by oneself,' 'without help or cause' dealiative 'only' superlative 'most,' '-est' decomitative 'by oneself,' 'alone' intensive 'oneself,' 'even'
Following are examples of spontaneous speech specimens containing sendiri illustrating each of these interpretations: (12) Nembak mati sendiri reciprocal AG-shoot die sendiri [Describing laptop picture of two men pointing guns to each other's heads] 'They're shooting each other dead' (13) Perang sendiri-sendiri fight DlSTR-sendiri [Watching movie "Village of the Damned"; climax scene in which large police force assaults group of nine possessed children, who, through supernatural powers, make the policemen turn the guns on each other, as a result of which the entire assault force is wiped out; speaker comments] 'They're fighting each other' (14) Ini putar sendiri deagentive DEM. PROX revolve sendiri [About a cassette player apparently revolving on its own] 'It's going around by itself (15) Bisa dibuka sendiri pintunya can PAT-open sendiri door-ASSOC [About the doors in a new shopping center] 'Its doors can open by themselves' (16) Aku sendiri tak ada rumah dealiative 1:SG sendiri NEG EXIST house [Speaker complaining about his life; after making this statement, he
Reflexive Anaphor or Conjunctive Operator
93
points to each of his friends in turn and says that that friend does have a house] 'Only I don't have a house' (17) Lupis sendiri ajau tang aku Lupis sendiri just debt 1:SG [Friends discussing their debts] 'Lupis is the only one I owe money' (18) Tinggi aku sendiri superlative high 1:SG sendiri [Having just done well on a laptop computer game, speaker views the high-scores table and sees his name at the top] 'I'm the highest' (19) Dia sendiri aja yang berani 3 sendiri just REL brave [Watching action movie in which woman appears to be braver than the men in her gang] 'She's the bravest one' (20) Kami tidur Siang Malam sendiri decomitative 1:EXCL sleep [restaurant name] sendiri [Street boy complaining he had to sleep alone the previous night, and not with the other gang members, as usual] 'I slept alone at the Siang Malam' (21) Anto takut sendiri? Anto afraid sendiri [Teasing younger brother about being afraid to be left alone in room] 'Are you afraid to stay alone?' (22) Mister sendiri yang kasi intensive white.person sendiri REL give [Playing cards; previously, I had complained that speaker cheats while dealing to get a good hand; this time, I dealt speaker a good hand, and speaker observes] 'You yourself gave them to me' (23) Tanya Pai sendiri ask Pai sendiri [Group of friends having an argument; interlocutor doesn't believe something speaker just said; speaker tells him to ask another friend for confirmation] 'Ask Pai'
94
David Gil
Thus, as demonstrated by the preceding data, sendiri is associated with a wide range of interpretations which are clearly not reflexive.6 But what should be made of this fact? Following is a general principle governing the structure of language: (24) A single form is associated with a single meaning. As a universal exceptionless principle, it is clearly wrong: all languages have homonyms, such as English bank, spring, and so forth. However, it is equally clear that such homonyms are the exceptions that prove the rule. Thus, as a default for linguistic analysis, a null-hypothesis for each case until proven otherwise, the above principle is indispensable. Just imagine what would happen without it. A Yagua linguist, wishing to describe English verbal tense marking, would posit five homonymous suffixes -ed, corresponding to their -jdsiy 'within a few hours,' -jay 'one day ago,' -siy 'within a few weeks,' -tiy 'within a few months,' and -jadd 'distant or legendary past' (Comrie, 1985: 99, citing Doris Payne). An Inuktitut linguist, wishing to describe a small fragment of the English lexicon, would posit several homonymous words ice, corresponding to their nilak 'clear ice from fresh water used for making water,' siku 'ice at sea, on shore, lakes, rivers,' tuvaq 'ice of the frozen sea or lake,' ivuniit 'ice broken up by heavy weather,' quasaq 'very slippery black ice on a lake, ground, sea, or in pack ice,' and others (Schneider, 1985, and E. Nowak p.c., pace Pullum 1989). Or a Riau Indonesian linguist, wishing to describe the English pronoun system, would argue that there are a number of second person pronouns you, corresponding to their own kau, engkau, kamu, anda, and others, which differ in politeness and honorificity. An unwarranted imposition of Yagua, Inuktitut, or Riau Indonesian grammatical categories on English we would doubtlessly and justifiably object to. But by the same token, English-speaking linguists must avoid imposing English grammatical categories on other languages positing multiple homonymous forms in other languages just because they correspond to different forms in English. As is the case for Riau Indonesian sendiri. For individual cases, however, the default association of a single form with a single meaning can and should be supported with explicit argumentation. Three general kinds of arguments that may be adduced in support of the association of a single form with a single meaning are given in (25): (25) A single form is associated with a single meaning to the extent that: a. in a variety of genetically, geographically, and typologically unrelated languages, there exists a single form associated with a similar range of meanings;
Reflexive Anaphor or Conjunctive Operator
95
b. the boundaries between the putative distinct alternative meanings are ill-defined; c. the meaning in question can be defined in a unified manner, without recourse to disjunctions. As suggested in (25a), a single form will be associated with a single meaning to the extent that forms with a similar semantic range recur in a variety of genetically, geographically, and typologically unrelated languages. This principle is one that I have attempted to apply in a number of previous publications, arguing, on cross-linguistic grounds, for unified semantic representations for forms such as Georgian reduplication (Gil, 1988), English any (Gil, 1991), and conjunctive operators in a variety of languages (Gil, 1993, 1994a,b,c, 1995a,b). Regarding English bank, it would be a remarkable coincidence if some unrelated language were found in which the same word denoted the side of a river and the place where one keeps one's money—it is a safe bet to say that the number of such languages is exceedingly small, if not actually zero. In contrast, in many languages, forms expressing reflexivity also express some of the other meanings associated with Riau Indonesian sendiri. In Table 1, columns correspond to the different meanings of Riau Indonesian sendiri, while rows provide a selection of reflexive morphemes in different languages. As suggested by the frequent occurrences of the V symbol, the forms in question share many of the usages associated with sendiri:7 Although none of the forms in Table 1 exhibits the same range of interpretations as Riau Indonesian sendiri, each of the interpretations of sendiri is shared by at least one of the forms in question. Thus, the recurrence, across genetically, geographically, and typologically diverse languages, of forms with a range of meanings spanning the seven listed in Table 1, provides yet further support for a unified analysis of Riau Indonesian sendiri associating it with a single common meaning.8 Patterns similar to those in Table 1 have been observed in the literature— see, for example, Moravcsik (1972), Geniusiene (1987), and Konig (1991) for a wide range of languages. However, for cases such as these, the majority view would seem to be that reflexive and nonreflexive interpretations should be kept apart, and that the forms in question are associated with several distinct meanings. Such is the position that has been argued for by, among others, Salkie (1987) for English, Moyne (1971) for Persian, and Tang (1989), Battistella and Xu (1990), and Larsen (1997) for Mandarin. In contrast, this chapter following Gil (1987), argues that the cross-linguistic facts provide additional evidence in favor of a single unified meaning for the forms in question. As proposed in (25b), a single form will be associated with a single meaning to the extent that there are no discrete and clear-cut boundaries be-
TABLE 1 MEANINGS OF REFLEXIVE FORMS ACROSS LANGUAGES reflexive English: -self French: se Bulgarian: samRussian: samLezgian: wic/ ceb Hebrew: hit-a-eMandarin: ziji
reciprocal
deagentive
dealiative
superlative
decomitative
intensive
Reflexive Anaphor or Conjunctive Operator
97
tween supposed alternative meanings. This is the truth-functional approach to homonymy proposed by Quine (1960:129). For English bank, the boundaries are as distinct as can be: it is impossible to imagine an object that is definitely a bank, but which is indeterminate between the side of a river and the place where one keeps one's money. However, for Riau Indonesian sendiri, the opposite is the case: an utterance containing sendiri is often associated with an interpretation which is difficult or impossible to classify unambivalently as either reflexive or nonreflexive. In each of the following examples, a construction containing sendiri can be translated into either a reflexive or a nonreflexive construction in English. Crucially, these examples are not ambiguous between reflexive and nonreflexive interpretations; it is difficult to imagine that the speakers who uttered them had in mind one of the two alternative translations to the exclusion of the other. Rather, they are simply vague, or undifferentiated, with respect to the reflexive/nonreflexive distinction. In the following examples, the reflexive and nonreflexive translations are separated by a'/': (26) Bayar sendiri kalian reflexive / deagentive pay sendiri 2:PL [Having drinks at restaurant; after I had given speaker some money to pay, he observes that this is not enough to cover everybody's drinks, and addresses them, gloating] 'You pay for yourselves' / 'You pay by yourselves' (27) Mister nyuci sendiri? reflexive / deagentive white.person AG-launder sendiri [Discussing laundry arrangements] 'Do you do laundry for yourself?'/ 'Do you do laundry by yourself?' (28) Dia minta sendiri reflexive / deagentive / decomitative 3 order sendiri [I am buying drinks for some friends in the night market; one person goes and orders a drink without asking permission from me; speaker complains that he's being impolite] 'He's ordering for himself / 'He's ordering by himself (29) Rumah dia rumah sendiri reflexive / decomitative house 3 house sendiri [About his friend, contrasting his friend's affluence with his own poverty] 'His house is their own' / 'His house they live in by themselves' Examples (26) and (27) are vague between reflexive readings, translated with the benefactive 'for yourself / yourselves,' and deagentive readings,
98
David Gil
rendered with the adverbial 'by yourself / by yourselves.' Example (28) is indeterminate with respect to reflexive 'for himself,' deagentive 'by himself,' and decomitative 'by himself.' And example (29) is underspecified with regard to the distinction between reflexive 'their own' and decomitative 'by themselves.' Thus, if sendiri is taken to be ambiguous between reflexive and nonreflexive interpretations, then constructions such as the preceding present an analytical conundrum: are they or are they not reflexives? In contrast, if sendiri is associated with a single, unified meaning, then constructions such as these pose no problem whatsoever.9 As suggested in (25c), a single form will be associated with a single meaning to the extent that the meaning in question can be defined in a unified and coherent fashion; that is to say, if the would-be distinct alternative interpretations are related to each other. This is the traditional view on homonymy, which can be traced back to Aristotle in the Posterior Analytics II xiii. English bank clearly fails this test, since there is nothing substantive in common to the side of a river and the place where one keeps one's money. However, Riau Indonesian sendiri passes it, since all of the interpretations listed in (11) and illustrated in (12) - (23) are, in fact, related, both to each other and to the reflexive interpretation. A definition of the unified meaning shared by all of these different usages is provided in the following Section, in the characterization of sendiri as a conjunctive operator.
4. SENDIRI AS A CONJUNCTIVE OPERATOR Roughly speaking, the basic meaning of Riau Indonesian sendiri corresponds to that of the English words 'alone' or 'only.' This is the meaning that comes to the fore when sendiri occurs by itself as a complete sentence, as in example (10). However, in many other cases, discussed in the preceding section, sendiri appears to be associated with a variety of meanings not translatable as 'alone' or 'only.' Nevertheless, all of these meanings share a common semantic core, which bears an obvious relationship to the concept of exclusivity. More specifically, whenever sendiri occurs in a particular construction, it focuses in on a certain element within the construction, marking it as being in opposition to a universally quantified set of semantically parallel elements. A more formal definition of this meaning is given below. 4.1. A Unified Semantic Analysis for Sendiri The unified meaning of sendiri may be represented within the framework of the theory of conjunctive operators proposed in Gil (1993; 1994a,b,c;
Reflexive Anaphor or Conjunctive Operator
99
1995a,b). In example (30), the interpretation of sendiri is defined in terms of a set of procedures: (30) A Unified Semantic Analysis for sendiri: For any occurrence of sendiri: a. An expression in morphosyntactic relation with sendiri is indexed as target. b. A variable x distinct from but semantically parallel to the target is indexed as source. c. An expression surrounding the target is indexed as source domain (s/dom). d. An expression surrounding the source domain is indexed as target domain (t/dom). e. The following interpretation is assigned:
In accordance with this procedure, the semantic representation of a construction containing sendiri assumes the form of a conjunction. The right-hand target conjunct comprises the interpretation associated with the construction containing sendiri, namely, the target domain. The left-hand source conjunct is parallel to a subpart of the target conjunct, but with two important differences. First, instead of the target, it contains a semantically parallel variable,x, bound by a universal quantifier; second, it is preceded by a negation, which falls within the scope of the universal quantifier.10 The above formalism provides a unified analysis for the entire range of constructions containing sendiri considered in the preceding section. The way in which this works is indicated in (31) - (38), in which a selection of examples from the preceding section are analyzed in accordance with the procedures in (30): (31) [example (4)] 'I shot myself reflexive x ~ [Msource tembak]s/dom A shoot [[saya]target
1:SG (32) [example (12)] x ~ [Msource [[[(3)]target 3
tembak] s/dom ] t/dom
shoot 'They're shooting each other dead' nembak mati]s/dom A AG-shoot die nembak mati]s/dom AG-shoot die
reciprocal
100
David Gil
(33) [example (14)] 'It's going around by itself X ~ [[x]Source [[[ini]target
DEM.PROX
Putar]s/dom
revolve
Putar]s/dom]t/dom
revolve
(34) [example (16)] 'Only I don't have a house' x
~ [x]source
[[[aku]target
1:SG
deagentive
A
tak ada
rumah] s/dom
dealiative
A
NEG EXIST house tak
ada
rumah]s/dom]t/dom
NEG EXIST house
(35) [example (18)] 'I'm the highest' x ~ [tinggi [x]source]s/dom A high
superlative
[[tinggi [aku]target]s/dom]t/dom
high 1:SG
(36) [example (20)] 'I slept alone at the Siang Malam' x ~ [[x]source Siang Malam]s/dom A [restaurant name] [[[kami]target Siang Malam]s/dom tidur]t/dom I:EXCL [restaurant name] sleep (37) [example (22)] 'You yourself gave them to me' xz ~ [[x] [[x]source]]s/dom A A [[[mister]target]s/dom yang kasi]t/dom white.person REL give
decomitative
intensive
source s/dom
(38) [example (26)]
'You pay for yourselves' 'You pay by yourselves' x ~ [bayar [x]source]s/dom A pay [[bayar [kalian]target]s/dom]t/dom pay 2:PL
reflexive deagentive
In each of the preceding examples, a construction containing sendiri is assigned the interpretation of a conjunctive operator, in accordance with the formalism in (30). Let us now examine each of these interpretations in turn, and translate all of the formalism into plain English. In (31), a semantic representation is assigned to an apparent reflexive construction, example (4) Saya tembak sendiri 'I shot myself.' As noted in Section 2 in the discussion of example (3), simple constructions in Riau Indonesian are vague with regard to thematic roles. In particular, in Riau Indonesian, a construction such as Saya tembak can be understood as either 'I
Reflexive Anaphor or Conjunctive Operator
101
shot (someone)' or '(Someone) shot me.' A more perspicuous single English gloss encompassing both possibilities would thus be "There was a shooting activity with which the speaker was associated.' This, in fact, is how the right-hand target conjunct in (31) is to be read. What the left-hand source conjunct says is that nobody else was associated with the same shooting activity. Given that shootings prototypically involve both an agent and a patient, one plausible interpretation of sentence (4)—though not the only one—is that the speaker is associated with the shooting activity in both capacities, as agent and as patient. Thus, it obtains an interpretation corresponding to the English reflexive construction 'I shot myself.' However, the Riau Indonesian sentence itself, Saya tembak sendiri, is not a reflexive construction: saya is not assigned a thematic role of agent, sendiri is not assigned a thematic role of patient, and, crucially, sendiri is not bound by saya in the way a reflexive anaphor is bound by its antecedent. Rather, as suggested by its semantic representation in (31), sentence (4) provides an example of a different construction type, namely, a conjunctive operator. In (32), a semantic representation is assigned to a seemingly reciprocal construction, example (12) Nembak mati sendiri 'They're shooting each other dead.' As in the preceding example, the right-hand target conjunct says that there is a shooting activity in which an understood set of third-person participants (indicated in parentheses) is involved. What the left-hand source conjunct says is that no other participants are associated with the same shooting activity. As before, one plausible interpretation of sentence (12) is that the understood set of participants is associated with the shooting activity as agents and also as patients. This allows for a variety of combinatorial possibilities, including that of the given context, involving a picture of two men pointing guns to each other's heads. Thus, sentence (12) obtains an interpretation corresponding to the English reciprocal construction 'They're shooting each other dead.' However, just as sentence (4) is not a reflexive construction, so sentence (12), Nembak mati sendiri, is not a reciprocal construction. Rather, as shown by their parallel analyses above both instantiate the conjunctive operator construction. Moving on to (33), here a semantic representation is assigned to the deagentive example (14) Iniputar sendiri 'It's going around by itself.' Again, what the right-hand target conjunct says is vague with respect to thematic roles, namely, that there is a revolving activity with which the object being pointed to is associated. And what the left-hand source conjunct says is that nothing else is associated with the same revolving activity. However, unlike shootings, which typically involve two participants, revolvings often involve but a single participant, the theme. Accordingly, in the case at hand, the relevant interpretation of Ini putar sendiri is one which asserts that the revolving activity does, in fact, involve only a theme, and not, in addition, an agent
102
David Gil
or cause. In this way, the sentence arrives at an interpretation corresponding to the English deagentive construction 'It's turning around by itself. Note that the interpretation of (14), as given in (33), is formally isomorphic to those in (12) and (13): in all of these cases, the effect of sendiri is to assert the absence of any additional participants in the activity. The only difference between these constructions is with respect to thematic roles: whereas in (12) and (13) the single participant or set of participants takes over the two roles of agent and patient, in (14) it assumes the single role of theme. While contributing to their distinct interpretations and English glosses, this difference is of no grammatical consequence within Riau Indonesian: all three sentences are formally and semantically parallel, as exemplars of the conjunctive operator construction. In (34), now, a semantic representation is assigned to the dealiative example (16) Aku sendiri tak ada rumah 'Only I don't have a house.' Here, the right-hand target conjunct says that the speaker is not associated with the existence of a house, while the left-hand source conjunct says that nobody else is not associated with the existence of a house. Whereas in the preceding examples, the source variable x was understood as having a different thematic role than that of the target, here the source variable x is taken to have the same thematic role as that of the target. Specifically, both source and target are, broadly speaking, the experiencers of the existential form ada, or, in simpler terms, the would-be possessors of a house. In such a way, sentence (16) acquires its dealiative interpretation, with sendiri corresponding in its meaning to English 'only.' However, as indicated in (34), the semantic representation of this sentence is completely parallel to that of its predecessors, involving, again, the conjunctive operator construction. In (35), a semantic representation is assigned to the superlative example (18) Tinggi aku sendiri 'I'm the highest.' In this case, the right-hand target conjunct says that the speaker is high, while the left-hand source conjunct says that nobody else is high. As in the preceding example, but not (31) (33), the source variable x is understood to have the same thematic role as that of the target, namely, the theme of tinggi 'high.' In the case at hand, this results in a superlative interpretation, with sendiri corresponding in its meaning to the English '-est.' However, once again, this is not a superlative construction in Riau Indonesian; rather, its semantic representation is isomorphic to that of the preceding sentences, exemplifying the conjunctive operator construction. Something just a little bit different is provided by (36), in which a semantic representation is assigned to the decomitative example (20) Kami tidur Siang Malam sendiri 'I slept alone at the Siang Malam.' Whereas in (31) (35) the source and target domains are identical, here the source domain is wholly included within the target domain, with tidur 'sleep' occurring in the
Reflexive Anaphor or Conjunctive Operator
103
latter but not in the former. Thus, the right-hand target conjunct says that the speaker slept alone at the Siang Malam, while the left-hand source conjunct asserts that nobody else was associated in any way whatsoever with the Siang Malam, that is to say, nobody else was there. (If, for example, there had been other people staying up all night and drinking at the restaurant, the speaker's complaint would have been inappropriate.) Resulting from this is a decomitative interpretation, yet another instantiation of the range of usages of sendiri as a conjunctive operator. Yet another variation is provided by (37), in which a semantic representation is assigned to the intensive example (22) Mister sendiri yang kasi You yourself gave them to me.' Here, as in the preceding example, the source domain is wholly included within the target domain, with yang kasi 'gave them to me' occurring in the latter but not in the former. However, unlike in all of the examples up to now, the source domain consists entirely of the source variable x and the semantically parallel target Mister 'you' in the respective conjuncts. How should the minimal source conjunct in (37) be understood? As noted in Section 2, most lexical items in Riau Indonesian belong to the syntactic category of Sentence; thus, for example, Orang can stand on its own as a complete nonelliptical sentence meaning 'It's a person,' or 'There's a person.' Accordingly, in the case at hand, the source conjunct may be understood simply as saying 'There is nobody else.' Its function, in other words, is to mark a switch in the topic of the sentence. Accordingly, the semantic representation in (37) may be read: 'Not talking about anybody else, you gave them to me.' In this way, the intensive use of sendiri may also be accounted for in terms of the conjunctive operator construction.11 Thus, the unified semantic analysis of sendiri proposed in (30) accounts for the entire range of interpretations of constructions containing sendiri discussed in Section 3. An additional benefit of the analysis is that it provides a convenient way to represent the meanings of examples such as those in (26) - (29), which were argued to be vague, rather than ambiguous, between reflexive and non-reflexive interpretations. As an illustration of this in (38), a single semantic representation is provided encompassing both the reflexive and the deagentive interpretations of example (26) Bayar sendiri kalian 'You pay for / by yourselves.' This representation reads, roughly: 'There should be an act of paying with which you, but nobody else, are to be associated.' What this single unified representation shows is that the apparent ambiguity of this example, between reflexive and deagentive interpretations, is a mere artefact of the English translations, which instantiate these distinct construction types in English. In Riau Indonesian, examples such as these are unambiguous, their unified meanings being accounted for in accordance with the analysis proposed in (30).
104
David Gil
The unified semantic analysis proposed above characterizes sendiri as a conjunctive operator. All of the semantic representations in (31) - (38) share the common structure specified in (30e). A corollary of this is that sendiri is not a marker of any of the other construction types that are alluded to by the ad hoc names appearing in (31) - (38) in italics to the right. In particular, sendiri is not a reflexive anaphor, even though some constructions containing sendiri, such as those in (4) - (9), correspond in their meanings to reflexive constructions in English. Rather, the semantic representation for (4), given in (31), is parallel to those of the various other constructions, given in (32) - (38), thereby showing that in (4), as elsewhere, sendiri is not a reflexive anaphor, but rather a conjunctive operator. 4.2. Why Sendiri Does Not Have Long-distance Interpretations The unified analysis of sendiri as a conjunctive operator suggests an account for why, if, contrary to the above analysis, sendiri is viewed as a reflexive anaphor, it appears, in examples such as (5) and (6), to be a local rather than a long-distance one. In the unified analysis proposed in (30), item (30a) states simply that the target expression stands in some kind of "morphosyntactic relation" with sendiri. However, this claim can be strengthened. In general, whereas anaphors are typically c-commanded by their antecedents, conjunctive operators generally c-command their targets.12 Evidence for this claim is provided with regard to the English conjunctive operator only in (39), presenting a partial parallel to (6) above: (39) a. Alie onlyabcde knows [that Fatimahc likesb Ahmada]d b. Alie knows [that Fatimahc onlyabc*d*e likesb Ahmada]d Sentences (39a) and (39b) differ only with respect to the position of the conjunctive operator only; however, this difference bears important consequences with regard to the range of possible interpretations of the two sentences. In both sentences there are (at least) five potential targets for the conjunctive operator only; these five targets are indexed with the letters a to e. The five interpretations corresponding to these five choices of target are shown in (40), in plain English (with the target expression in italics, suggesting contrastive stress), and then in accordance with the analysis proposed in (30), with the indices a to e corresponding to examples (40a) to (40e), respectively: (40) a. 'Ali knows that Fatimah likes Ahmad (but nobody else)' x ~ [Ali knows that Fatimah likes [x]source]s/dom A [[Ali knows that Fatimah likes [Ahmad]^^^^
Reflexive Anaphor or Conjunctive Operator
105
b. 'Ali knows that Fatimah likes (but doesn't love) Ahmad' x ~ [Ali knows that Fatimah [x]source Ahmad]s/dom A [[Ali knows that Fatimah [likes]target Ahmad]s/dom]t/dom c. 'Ali knows that Fatimah (but nobody else) likes Ahmad' x ~ [Ali knows that [x]source likes Ahmad]s/dom A [[Ali knows that [Fatimah]target likes Ahmad]s/dom]t/dom d. 'Ali knows that Fatimah likes Ahmad (but nothing else)' x ~ [Ali knows [x]source]s/dom A [[Ali knows [that Fatimah likes Ahmad]target/s/dom]t/dom e. 'Ali (but nobody else) knows that Fatimah likes Ahmad' x ~ [[x]source knows that Fatimah likes Ahmad]s/domA [[[Ali]target knows that Fatimah likes Ahmad]s/dom]t/dom In (39a), only c-commands all five potential targets, and indeed, all five interpretations given in (40a) to (40e) are available here. In contrast, in (39b), only c-commands only three out of the five potential targets, Ahmad, likes, and Fatimah, indexed with a, b, and c. The remaining two targets, that Fatimah likes Ahmad and Ali, indexed with d and e, are not c-commanded by only, and in fact, as suggested by the stars in front of the latter. two indices in (39b), these two interpretations, (40d) and (40e), are not available for sentence (39b). As evidenced by the above example, conjunctive operators typically c-command their targets. This generalization accordingly provides an account for why, in sentences such as (6) and (7), the interpretations that would be analyzed as involving long-distance reflexivization are unavailable. Consider, once again, sentence (6), reproduced below:13 (41) Alij tahu Fatimahi suka sendirii/*j Ali know Fatimah like sendiri 'Ali knows that Fatimah likes herself The two interpretations corresponding to the indices i and j may be represented, in accordance with the analysis proposed in (30), as follows: (42) a. x ~ [Ali tahu [x]source suka]s/dom A [[Ali tahu [Fatimah]target suka] s/dom ] t/dom b. x ~ [[x]source tahu Fatimah suka]s/dom A [[[Ali]target tahu Fatimah suka]s/dom]t/dom In (42a), corresponding to the interpretation with index i, sendiri c-commands its target Fatimah. This is because, as argued in Section 3.1 with respect to the analogous construction in (9c), Fatimah and suka form a constituent which is a sister to sendiri. (Note, however, that even under a
106
David Gil
"flat" analysis in which Fatimah suka sendiri were three-way branching, sendiri would still c-command Fatimah.) In contrast, in (42b), corresponding to the interpretation with index j, sendiri does not c-command its target Ali. Thus, the generalization to the effect that conjunctive operators typically c-command their targets accounts for the fact that the "local" interpretation in (42a) is available, while the "long-distance" interpretation in (42b) is not. In conclusion, then, the analysis of sendiri as a conjunctive operator provides an answer to the question posed at the beginning of this chapter, namely, why, as an apparent reflexive anaphor, sendiri does not allow long distance interpretations. As a conjunctive operator, sendiri seeks out a target expression that is within its c-command domain; the resulting interpretations are thus those which, if sendiri were taken to be a reflexive, would be considered local rather than long distance.
5. GENERALIZING TO OTHER FORMS The moral of the story is that things are not always as they seem. Just because a certain expression translates into a reflexive construction in English does not mean that the expression is itself a reflexive construction in its own language. Rather, each apparent instance of a reflexive form must be examined on its own terms, in order to determine whether it is indeed a reflexive anaphor or perhaps some completely different kind of construction. This is true both for different forms within the same language and for different forms in different languages.
5.1. Other Forms in Riau Indonesian Although, as noted in the beginning of Section 3, the most common way of translating an English reflexive into Riau Indonesian involves the form sendiri, there are a variety of other morphosyntactic strategies for expressing a similar range of meanings. First, it is worth pointing out that sendiri is itself polymorphemic, consisting of the proclitic se- 'one', the agentive prenasalizing prefix N- (which assimilates in place of articulation to the following consonant and is therefore realized here as n-), and the stem diri, which by itself means 'stand.' Although the meaning of sendiri is related to that of its constituent parts, the relationship is not straightforwardly predictable, and therefore it is necessary to consider the entire form as a single unit.
Reflexive Anaphor or Conjunctive Operator
107
Nevertheless, alongside sendiri, other forms based on the stem din also occur, with meanings that correspond to English reflexive constructions. Some examples are given in (43) - (45) below: (43) Bunuh diri aja la aku kill diri just CONTR 1:SG [Playing Nintendo; after losing, speaker exclaims in mock despair] 'I'll just kill myself (44) Bodoh dia nembak diri sendiri stupid 3 AG-shoot diri sendiri [Describing laptop picture of man holding gun to his head] 'Stupid, he's shooting himself (45) Dia sendirian ditembak 3 sendiri-AUG PAT-shoot [Describing laptop picture of man holding gun to his head] 'He's shooting himself Sentence (43) illustrates the use of the bare stem diri. In addition to its basic use with the meaning "stand," diri also occurs in constructions with a very limited number of words, with a meaning corresponding to an English reflexive. In this construction, diri invariably occurs immediately after the relevant word and has no other meanings other than the reflexive. Thus, in this construction, diri is probably most appropriately analyzed as a reflexive particle or enclitic.14 Sentence (44) shows what appears to be a straightforward collocation of diri with sendiri. However, whereas diri occurs only with a very small number of words, and immediately following them, diri sendiri can occur freely with an open class of words and not necessarily adjacent to them. In fact, the two components of the construction, diri and sendiri, must be adjacent to each other, suggesting that they constitute a single, complex unit. Like diri, diri sendiri has no meanings other than the reflexive. In general, the diri sendiri construction occurs infrequently, and in somewhat higher registers than do simple diri and sendiri. To the extent that it occurs in basilectal Riau Indonesian, it is probably best viewed as the product of the grammaticalization of the collocation of diri and sendiri, forming a new complex reflexive anaphor.15 Lastly, sentence (45) provides an instance of the form sendirian, derived from sendiri by affixation of the word-formation suffix -an, associated with a variety of usages broadly subsumed under the general rubric of augmentative. Unlike diri and diri sendiri, sendirian has a wide range of non-reflexive interpretations similar to that of sendiri, though, in contrast to sendiri, its use in constructions corresponding to English reflexives, as in
108
David Gil
(45) above, is relatively infrequent. Like sendiri, sendirian is most appropriately characterized as a conjunctive operator. In addition to forms based on the stem diri, Riau Indonesian offers several other ways of expressing meanings corresponding to English reflexive constructions. A sample of these is given in (46) - (48) below: (46) Saya buat juga la kopi, haos I:SG make also CONTR coffee thirsty [Speaker having just made the interlocutor a cup of coffee] 'I'm going to make myself a cup of coffee too, I'm thirsty' (47) Tembak seorang shoot one-person [Describing laptop picture of man holding gun to his head] 'He's shooting himself (48) Mister usir mister? white.person send.away white.person [Speaker teasing me about the large number of friends who have come to see me; going through each of their names, he asks me, in turn, whether I would send each one away; then, finally, as a joke, he asks] 'Would you send yourself away?' Sentence (46) presents an example of the positive conjunctive operator juga, which, in many contexts, is best translated into English as "too" or "also." However, in this particular instance, it ends up with an interpretation corresponding to an English reflexive construction. In the given context, the speaker had just made the interlocutor a cup of coffee; he then says that he will make himself one too, that is to say, himself in addition to the interlocutor. Thus, in the preceding example, juga selects saya 'I' as target, and the understood interlocutor as its semantically parallel source. This example shows how a conjunctive operator of a different kind may occasionally also result in an interpretation corresponding to an English reflexive construction. Sentence (47) illustrates the use of the proclitic se- 'one,' also forming part of sendiri, in construction with orang 'person,' Unlike sendiri, seorang is completely compositional: it is a common way of saying "one person." In addition, though, seorang often bears the meaning "alone," which it shares with sendiri. Accordingly, in (47), the reflexive translation associated with seorang may be viewed as a particular instantiation of its basic meaning of aloneness, as is the case with respect to sendiri. However, whereas sendiri, as a conjunctive operator, explicitly predicates aloneness, seorang does so merely by implicature: if the speaker asserts the involvement of one person
Reflexive Anaphor or Conjunctive Operator
109
in the activity, then by the Gricean maxim of informativeness, the hearer is entitled to infer that no other person is involved. Thus, in spite of its reflexive translation in (47), seorang is neither a reflexive anaphor nor a conjunctive operator. Finally, sentence (48) reflects the fact that in Riau Indonesian, there is no class of words which are subject to a condition of disjoint reference. Indeed, there is no well-defined category of pronoun in Riau Indonesian: a variety of kinship terms and assorted other titles, such as the English loan-word mister, glossed as "white person," are capable of being used, in argument position, with second-person reference, as is the case in (48). In the preceding example, then, the two occurrences of mister are coreferential, though there would seem to be little reason to characterize one of them as an anaphor and the other as its antecedent; rather, both refer to the addressee independently of each other. In summary, then, although examples (43) - (48) can all be translated into reflexive constructions in English, they represent very different construction types within Riau Indonesian. Whereas diri and diri sendiri may perhaps be analyzable as reflexive anaphors, sendirian and juga are probably more appropriately considered as conjunctive operators, though of quite different types, while seorang and mister would appear to be neither reflexive anaphors nor conjunctive operators. The preceding constructions thus underscore the fact that a single language may have at its disposal a variety of grammatically distinct morphosyntactic strategies to convey essentially similar messages.
5.2. Other Forms in Other Languages Many of the observations made in this paper with regard to Riau Indonesian sendiri obtain also, to varying degrees, with respect to other apparently reflexive forms in other languages. In Standard Malay and Indonesian, translations of English reflexives typically contain a form based on diri: diri, sendiri, diri sendiri, diri-PRO, or diri-PRO sendiri, where -PRO stands for a personal enclitic. While some of these forms are probably most appropriately characterized as reflexive anaphors, at least one of them, diri-PRO, has been argued, by Cole and Hermon (1997,1988), to be ambiguous between a reflexive anaphor, in which case it is clause-bound, and a free pronoun, with usages which may, in some cases, result in interpretations resembling those of a would-be long-distance reflexive.16 As noted in Section 3.3 and Table 1, many languages have forms which, like the Riau Indonesian sendiri, occur not only in reflexive constructions, but also in constructions with various other kinds of interpretations. Such
110
David Gil
forms accordingly meet two of the three conditions put forward in (25) as justifying a unified analysis: in accordance with (25a), their range of interpretations recurs crosslinguistically, and in accordance with (25c), these interpretations can be defined in a unified manner. Prima facie, then, there is reason to suspect that much of what is said in this chapter about Riau Indonesian sendiri may also be valid for these other forms. Nevertheless, each and every case needs to be judged on its own individual merits. In particular, it is necessary to test such forms against condition (25b), to see to what extent the boundaries between the putative distinct alternative meanings are ill-defined. English forms based on -self provide an interesting case. As noted in Table 1, such forms allow reflexive, deagentive, decomitative, and intensive interpretations. Indeed, with some effort, it is possible to construct examples containing -self which, like (26) - (29) in Riau Indonesian, allow either reflexive or non-reflexive interpretations: (49) John is shooting himself This sentence may be understood either as a reflexive construction, involving a person holding a gun to his head, or, less obviously, as an intensive construction, describing a person who usually delegates shooting to other people but in the particular case at hand is doing the job on his own. However, unlike (26) - (29), the preceding example would appear to be not vague but rather ambiguous with respect to the contrast between reflexive and intensive interpretations. In particular, whereas Riau Indonesian constructions with sendiri may occur in situations where the speaker is indifferent between reflexive and nonreflexive interpretations, it is hard to imagine a situation in which a speaker of English could make use of a construction such as (49) without having in mind one of the two interpretations, reflexive or intensive, to the exclusion of the other. The difference between Riau Indonesian (26) - (29) and English (49) is reinforced by the observation that whereas the former have a single syntactic structure underlying their entire range of interpretations, the latter is also structurally ambiguous, himself functioning as a direct object in the case of the reflexive interpretation but as an adverbial in the case of the intensive interpretation.17 In general, then, constructions containing forms based on -self fail to meet condition (25b): the boundaries between the alternative interpretations are, in fact, quite clearly demarcated in all cases. In order to accommodate facts such as these, it is necessary to acknowledge that vagueness and ambiguity are not discrete all-or-nothing categories but rather two opposite ends of a broad spectrum of intermediate possibilities involving various degrees of polysemy—see Zwicky and Sadock (1975) and others for discussion. In accordance with a polysemy
Reflexive Anaphor or Conjunctive Operator
111
analysis, forms containing -self would be associated with a set of distinct meanings expressed through different grammatical constructions; in particular, the same form might function in some cases as a reflexive anaphor but in other cases as a marker of a different kind. At the same time, however, these distinct meanings would be related to each other, either as particular instances of a single unified meaning, such as that defined in Section 4.1, or perhaps through a kind of Wittgensteinian network of resemblances. Thus, the analysis proposed in this chapter for Riau Indonesian sendiri may perhaps be only partially extendable to the corresponding English forms based on -self. A corollary of this is that an explanation must be sought elsewhere for the absence of long-distance reflexive interpretations associated with forms based on -self and their counterparts in other languages. The arguments put forward in this chapter provide just one of several possible reasons why an apparent reflexive anaphor may fail to possess long-distance interpretations. Nevertheless, in doing so, they underscore the need for detailed analyses of apparent reflexive constructions in different languages, ever bearing in mind the possibility that what seems to be a reflexive anaphor may, in fact, turn out to be a very different kind of object, such as a conjunctive operator.
ACKNOWLEDGMENTS This paper would not have been possible without the unwitting assistance of numerous speakers of Riau Indonesian, who, in the course of their daily lives, contributed to my database of spontaneous speech specimens: for making it into the present paper, I am grateful to Adi, Anton, Anton, Arip, Dedi, Imansaputra, Joni, Junjun, Jumbro, Kuduak, Mardona, Nano, Rudi Candra, Sapri, Suriyadi, and Zuhardi. For their judgments on examples (6) and (7), I am indebted to Nasar, Rudi Candra, and Saiful. For their assistance with the crosslinguistic data in Table 1,I am grateful to Zlatka Guentcheva, Koyka Stoyanova, and Ljuba Veselinova (Bulgarian), Martin Haspelmath (Lezgian), Xue Nianwen and Siew Im Yeoh (Mandarin), and Elena Maslova and Vladimir Tourovski (Russian). Versions of this chapter were presented at the Thirteenth Annual Conference of The Israel Association for Theoretical Linguistics, Ramat-Gan, Israel, 3 June 97; at the Workshop on Long Distance Reflexives, LSA Linguistic Institute, Ithaca, NY, USA, 5 July 1997; and at the Institut fur Englische Philologie, Freie Universitat Berlin, Berlin, Germany, 10 May 1999.I am grateful to participants at all three events for helpful comments and suggestions. In the course of my work on Riau Indonesian, I have benefited considerably from ongoing discussions with Peter Cole, Gabriella Hermon, and Uri Tadmor, to whom my thanks are due. The work described in this paper was supported in part by National
112
Science Foundation INT-9423291.
David Gil
Grants
Numbers SBR-9121167,
SBR-9729519,
and
NOTES 1
One apparent example of such an account is the suggestion by Manzini and Wexler (1987) and Wexler and Manzini (1987) to the effect that in languages with long-distance reflexives, the binding domain, or Governing Category, is the root clause. However, since no other logically independent observable properties are proposed to correlate with this parameter, the characterization of the binding domain is not a general property of a language of the kind suggested in (1b), but rather a mere recapitulation of whether or not the language in question allows long-distance reflexives. 2 The interlinear glosses make use of the following abbreviations: AG 'agent'; ASSOC 'associative'; AUG 'augmentative'; CONTR 'contrastive'; DEM 'demonstrative'; DISTR 'distributive'; EXCL 'exclusive'; NEG 'negation'; PAT 'patient'; PL 'plural'; PROX 'proximal'; REL 'relative'; SG 'singular'; 1 'first person'; 2 'second person'; 3 'third person.' In addition, because of the nature of spontaneous speech specimens, it is necessary, for each example, to include, in addition to the customary three lines (text, interlinear gloss, and free translation), an additional line describing the context in which the example was uttered, thereby justifying the translation that is provided, as opposed to any number of other translations potentially available for the same sentence had it been uttered in a different context. This additional line is enclosed in square brackets. 3 At the time of writing, the corpus contained 1833 spontaneous speech specimens in Riau Indonesian. Of these, 95 contained an occurrence of sendiri, and of these, 49 had an apparent reflexive interpretation. Of these 49, all appeared to be bound by a local antecedent. 4 Indeed, in more standard varieties of Malay and Indonesian, sendiri and other morphologically related forms are typically characterized as reflexives, in reference grammars, such as Sneddon (1996), and in theoretical discussions, such as Huang (1996) and Cole and Hermon (1998). However, in Standard Malay/Indonesian, the actual form sendiri appears to function as a reflexive only in possessive constructions resembling (5), where, as in Riau Indonesian, it would seem to be local. See Alsagoff (1992:40-42) for discussion. 5 Constructions similar to (8), with a reflexive apparently asymmetrically c-commanding its NP-internal antecedent, occur also in Icelandic (Maling, 1984) and in Mandarin (Tang, 1989; Wang, 1990; Cole, Hermon, and Sung, 1993; Pollard and Xue, 1998), thereby posing a problem for the treatment of reflexives in those languages as well. 6 Impressionistically, these different interpretations are not all of equal frequency. Specifically, whereas the deagentive, dealiative, and decomitative interpretations occur very often, the other three interpretations are substantially less frequent. For reciprocal meanings, the preferred constructions involve a reduplicated expression
Reflexive Anaphor or Conjunctive Operator
113
with sama 'same' / 'with' / 'and,' e.g., sama-sama tembak or tembak sama tembak for 'shoot each other.' For superlative meanings, the most common strategies involve prefixation of ter-, as in tertinggi tallest,' a periphrastic construction with paling, as in paling tinggi 'tallest,' or marking the property word as definite by putting it in topic position. As for the intensive, the most frequent way of expressing this meaning is probably via intonation. 7 Table 1 contains both free forms (or the bases of free forms) as in English, Bulgarian, Russian, Lezgian, and Mandarin, and verbal proclitics or affixes as in French and Hebrew. For Bulgarian and Russian, the parentheses enclosing the "A" symbol in the reflexive column reflect the fact that the form in question does not express a reflexive construction on its own, but only in construction with another form. Finally, it should be noted that Table 1 presents only those interpretations associated with Riau Indonesian sendiri; some of the other forms may have additional interpretations not represented in the table. 8 Inspection of Table 1 suggests certain non-random distributional patterns. Thus, for example, the deagentive interpretation is shared by all of the forms, while other interpretations appear to be less common. In addition, the two verbal reflexives, in French and Hebrew, exhibit a similar semantic range. Although suggestive, the sample of languages in Table 1 is too limited to enable any firm generalizations to be made regarding these patterns. 9 One commonly invoked diagnostic for distinguishing between vagueness and ambiguity involves the phenomenon of VP ellipsis—see Sag (1976), Williams (1977), and others. Unfortunately, such tests cannot be invoked here for the following two reasons: (a) speakers find it difficult or impossible to even comprehend the putative distinction between reflexive and nonreflexive interpretations, without which it is impossible to conduct any tests; and (b) there is no construction corresponding to VP ellipsis in Riau Indonesian. 10 The above formulation differs from that proposed in Gil (1993; 1994a,b,c; 1995a,b) primarily in the presence of the negative operator in the source conjunct. Whereas those works deal with positive conjunctive operators such as English also and too, Riau Indonesian sendiri is thus a negative conjunctive operator bearing a greater resemblance to English only and just. Detailed discussion of negative conjunctive operators in a variety of languages can be found in Konig (1991), who refers to them as "exclusive" or "restrictive focus particles." 11 The analysis of intensifier sendiri as a conjunctive operator of the kind represented in (30) thus runs counter to Konig's (1991:92) claim that intensifiers are not analyzable as exclusive focus particles. Konig's claim is based upon examples such as 7 cannot lend you any money. I am a little short of cash myself. Konig observes, correctly, that in an example such as this, the intensifier myself cannot entail that the speaker is the only person who is short of money, since, in the given context, the interlocutor is apparently also in need of a loan. However, Konig's conclusion that intensifier myself should therefore be analyzed as an additive focus particle, such as too or also, is unwarranted. It is true that in the preceding example, myself may be replaced by the additive too, salvae veritatis: 7 am a little short of cash too. However, there is a subtle difference in meaning between the two. While too does indeed presuppose that the interlocutor is short of cash, myself, in the preceding example, actu-
114
David Gil
ally does not. Rather, what it does is to signal a shift in topic from the interlocutor, in the first sentence, / cannot lend you any money, toward the speaker in the second sentence,I am a little short of cash myself. Or, to paraphrase the latter sentence: "Not talking about you or anybody else, I am a little short of cash." Thus, the intensive interpretations of forms such as English myself can, pace Konig, be characterized as exclusive focus particles. The crucial step in the analysis is allowing the source conjunct to consist wholly of the source variable x, as done for the intensive interpretation of sendiri in (37). 12 Similar claims have been made by, among others, Jackendoff (1972:251) for English even, though they have been disputed by other scholars, such as Bayer (1985, 1988), for various forms in German. 13 For ease of exposition, the (unavailable) interpretation involving would-be discourse anaphora, represented in (6) with the index k, is omitted. 14 The construction exemplified in (43) thus bears a resemblance to various verbal reflexive forms in other languages, such as English self-destruct, French se tuer 'kill oneself,' and so forth. 15 The doubling (as distinct from reduplication) of reflexive markers is quite common cross-linguistically, occurring in, among others, Turkish (Kornfilt, 1997), Hindi/Urdu (Davison, 1997), Malayalam, and the Munda language Juang (Gair, 1997). 16 Cole and Hermon (1997,1988) refer to the dialect that they are describing as a mesolectal variety of "Singapore Malay," but it is actually a Singaporean approximation to Standard Malay, which is quite different from the basilectal variety of Malay spoken in Singapore. 17
An abundance of evidence can be cited in support of the structural ambiguity in (49). Two arguments: (a) Whereas adverbial himself can be moved in front of the auxiliary, direct-object himself cannot; hence, John himself is shooting has only the intensive interpretation; (b) If a particle such as up is added to the verb, its position varies, occurring before the adverbial himself, in John is shooting up himself with intensive interpretation only (unless himself is assigned contrastive stress), but after the reflexive himself, in John is shooting himself up with reflexive interpretation only. It should be noted that none of these arguments carries over to the corresponding constructions with sendiri in Riau Indonesian.
REFERENCES Alsagoff, L. (1992). Topic in Malay: The Other Subject. Ph.D. thesis, Stanford University. Battistella, E., and Y. Xu (1990). Remarks on the Reflexive in Chinese. Linguistics 28,205-240. Bayer, J. (1985). Adjazenz und Kettenbildung, Bemerkungen zur Syntax der
Reflexive Anaphor or Conjunctive Operator
115
deutschen Gradpartikeln. Unpublished manuscript, Max Planck Institute for Psycholinguistics, Nijmegen. Bayer, J. (1988). Rightward Movement and the Syntax of Quantificational Particles in German. In V. Rosen (ed.)., Papers from the Tenth Scandinavian Conference of Linguistics I, Bergen. Cole, P., and G. Hermon (1997). VP Ellipsis and Malay Reflexives. Paper presented at the Thirteenth Annual Conference of The Israel Association for Theoretical Linguistics, Ramat-Gan, Israel, 3, June, 1997. Cole, P., and G. Hermon (1998). Long-Distance Reflexives in Singapore Malay: An Apparent Typological Anomaly. Linguistic Typology 2,57-77. Cole, P., Hermon G., and Sung L.-M. (1993). Feature Percolation. Journal of East Asian Linguistics 2,91-118. Cole, P., and Sung L.-M. (1994). Head Movement and Long-Distance Reflexives. Linguistic Inquiry 25,355-406. Cole, P., and C. Wang (1996). Antecedents and Blockers of Long-Distance Reflexives. Linguistic Inquiry 27,357-390. Comrie, B. (1985). Tense. Cambridge Textbooks in Linguistics, Cambridge University Press, Cambridge. Davison, A. (1997). Anti-Antilocality. Paper presented at the Workshop on Long-Distance Reflexives, LSA Linguistics Institute, Ithaca, NY, USA, 5, July, 1997. Davison, A. (this volume). Long-distance Anaphors in Hindi/Urdu. Faltz, L. M. (1985). Reflexivization: A Study in Universal Syntax, Garland, New York. Gair, J. W. (1997). Some Aspects of (Long Distance) Anaphors in South Asian Languages. Paper presented at the Workshop on Long-Distance Reflexives, LSA Linguistic Institute, Ithaca, NY, USA, 5, July, 1997. GeniuSiene, E. (1987). The Typology of Reflexives. Mouton de Gruyter, Berlin, New York and Amsterdam. Gil, D. (1987). Gil Replies to Salkie. In S. Modgil and C. Modgil (eds.)., Noam Chomsky, Consensus and Controversy, Falmer International Master-Minds Challenged: 3, Falmer Press, Barcombe, 143-145. Gil, D. (1988). Georgian Reduplication and the Domain of Distributivity. Linguistics 26,1039-1065. Gil, D. (1991). Universal Quantifiers: A Typological Study. EUROTYPWorking Papers, Series 7, Number 12, The European Science Foundation, EUROTYP Programme, Berlin. Gil, D. (1993). Conjunctive Operators: Areal Phenomenon or Semantic Universal. In D. Gil (ed.)., Studies in Number and Quantification, EUROTYP Working Papers, Series 7, Number 19, The European Science Foundation, EUROTYP Programme, Berlin, 18-47. Gil, D. (1994a). Conjunctive Operators: A Unified Semantic Analysis. In P.Bosch and R. van der Sandt (eds.)., Focus and Natural Language Processing, Volume 2, Semantics, Proceedings of a Conference in Celebration of the 10th Anniversary of the Journal of Semantics, 12th-15th June, 1994, Hotel SchloB Wolfsbrunnen, Meinhard-Schwebda, Germany, Working Papers of the Institute for Logic and Linguistics, Working Paper 7, ISSN 0946-7521, IBMTR-80.94-007,311-322.
116
David Gil
Gil, D. (1994b). Conjunctive Operators in South Asian-Languages. In A.Davison and F.M. Smith (eds.)., Papers from the Fifteenth South Asian Language Analysis Roundtable Conference, South Asian Studies Program, University of Iowa, Iowa City, 82-105. Gil. D. (1994c). 'Everything Also Must Grab: A Unified Semantic Analysis of Singlish 'Also'. In F. Plank (ed.)., Agreement Gender Number Genitive &, EUROTYP Working Papers, Series 7, Number 23, The European Science Foundation, EUROTYP Programme, Berlin, 133-145. Gil, D. (1994d). The Structure of Riau Indonesian. Nordic Journal of Linguistics 17, 179-200. Gil, D. (1995a). Conjunctive Operators: A Cross-Linguistic Study. IATL— Proceedings of the Tenth Annual Conference of The Israel Association for Theoretical Linguistics, 72-90. Gil, D. (1995b). Singlish also: A Unified Semantic Representation. Proceedings, INTELEC '94, International English Language EducationConference, National and International Challenges and Responses, Language Centre, Universiti Kebangsaan Malaysia, Kuala Lumpur, Malaysia, 197-210. Gil, D. (1999). Riau Indonesian as a Pivotless Language. In E.V. Raxilina and Y.G. Testelec (eds.)., Tipologija i Teorija Jazyka, Ot Opisanija k Objasneniju, K 60-Letiju Aleksandra Evgen'evicha Kibrika (Typology and Linguistic Theory, From Description to Explanation, For the 60th Birthday of Aleksandr E. Kibrik)., Jazyki Russkoj KuFtury, Moscow, 187-211. Gil, D. (2000). Syntactic Categories, Cross-Linguistic Variation and Universal Grammar. In P. Vogel and B. Comrie (eds.)., Approaches to the Typology of Word Classes, Empirical Approaches to Language Typology, Mouton, Berlin and New York, 173-216. Gil, D. (to appear a). Escaping Eurocentrism: Fieldwork as a Process of Unlearning. In P. Newman and M. Ratliff (eds.)., Linguistic Fieldwork, Cambridge University Press, Cambridge. Gil, D. (to appear b). The Prefixes di- and N- in Malay/Indonesian Dialects. In F. Wouk and M. Ross (eds.)., Papers From the 8-ICAL Workshop on Focus, Pacific Linguistics, Canberra. Gil, D., and U. Tadmor (to appear). Towards a Typology of Malay/Indonesian Dialects. In D. Gil (ed.)., Malay / Indonesian Linguistics, Curzon Press, London. Huang, C.-T. J., and C.-S. Liu (this volume). Logophoricity, Attitudes, and Ziji at the Interface. Huang, Y. (1996). A Note on the Head-Movement Analysis of Long-Distance Reflexives. Linguistics 34,833-840. Jackendoff, R. S. (1972). Semantic Interpretation in Generative Grammar. MIT Press, Cambridge. Konig, E. (1991). The Meaning of Focus Particles, A Comparative Perspective. Routledge, London and New York. Kornfilt, J. (1997). Long-Distance (and other) Reflexives in Turkish. Paper presented at the Workshop on Long-Distance Reflexives, LSA Linguistic Institute, Ithaca, NY, USA, 6 July 1997. Larsen, U. B. (1997). Adverbial and Adnominal Uses of ziji. Paper presented at the
Reflexive Anaphor or Conjunctive Operator
117
Ninth North American Conference on Chinese Linguistics, Victoria, BC, Canada, 3 May 1997, to appear in the proceedings. Maling, J. (1984). Non-Clause-Bounded Reflexives in Modern Icelandic. Linguistics and Philosophy 7,211-241. Manzini, R., and K. Wexler (1987). Parameters, Binding Theory and Learnability. Linguistic Inquiry 18,413-444. Moseley, C. and R.E. Asher (eds.). (1994). Atlas of the World's Languages. Routledge, London. Moravcsik, E. A. (1972). Some Crosslinguistic Generalizations about Intensifier Constructions, In P.M. Peranteau, J.N. Levi and G.C. Phares(eds.)., Papers from the Eighth Regional Meeting, Chicago Linguistic Society, Chicago Linguistic Society, Chicago, 271-277. Moyne, J.A. (1971). Reflexive and Emphatic. Language 47,141-163. Nichols, J. (this volume). Canonical Long-Distance Reflexivization in ChechenIngush. Pica, P. (1984). Subject, Tense and Truth: Toward a Modular Approach to Binding. In J. Gueron, H.-G. Obenauer and J.-Y. Pollock (eds.)., Grammatical Representation, Foris, Dordrecht, 259-291. Pollard, C. and P. Xue (1998). Chinese Reflexive Ziji: Syntactic Reflexives vs. Nonsyntactic Reflexives. Journal of East Asian Linguistics, 7,287-318. Pullum, G. K. (1989). Topic... Comment: The Great Eskimo Vocabulary Hoax. Natural Language and Linguistic Theory 7,275-281. Quine, W.V.O. (1960). Word and Object. MIT Press, Cambridge. Reinhart, T, and E. Reuland (1991). Anaphors and Logophors: An Argument Structure Perspective. In J. Koster and E. Reuland (eds.)., Long-Distance Anaphora, Cambridge University Press, Cambridge, 283-321. Sag, I. (1976). Deletion and Logical Form. Ph.D. thesis, Massachusetts Institute of Technology, Cambridge. Salkie, R. (1987). Core Grammar and Periphery. In S. Modgil and C. Modgil (eds.)., Noam Chomsky, Consensus and Controversy, Falmer International Master-Minds Challenged: 3, Falmer Press, Barcombe, 109-117. Schneider, L. (1985). Ulirnaisigutiit, An Inuktitut-English Dictionary of Northern Quebec, Labrador and Eastern Arctic Dialects. translated from French by D.R.F. Collis, Les Presses de 1' Universite Laval, Quebec. Sneddon, J. N. (1996). Indonesian, A Comprehensive Grammar. Routledge, London and New York. Wang, J.-H. (1990). Ziji—A Chinese Long-Distance Anaphor. Ms., Carnegie Mellon University. Wexler, K., and R. Manzini (1987). Parameter and Learnability in Binding Theory. In T. Roeper and E. Williams (eds.)., Parameter Setting, Reidel, Dordrecht, 41-89. Williams, E. (1977). Discourse and Logical Form. Linguistic Inquiry 8,101-139. Tang, C.-C. Jane (1989). Chinese Reflexives. Natural Language and Linguistic Theory 7,93-121. Zwicky, A., and J. M. Sadock (1975). Ambiguity Tests and How to Fail Them. In J.P. Kimball (ed.)., Syntax and Semantics 4, Academic Press, New York, 1-36.
This page intentionally left blank
SYNTACTIC VS. LOGOPHORIC BINDING: EVIDENCE FROM NORWEGIAN CHILD LANGUAGE ARILD HESTVIK* WILLIAM PHILIP+ *Department of Linguistics and Comparative Literature University of Bergen, Norway +
Utrecht Institute of Linguistics OTS Utrecht University, The Netherlands
1. INTRODUCTION Work on Binding Theory during the last decade has shifted from theories positing a single theoretical module, designed to capture all occurrences of reflexives, to theories that account for the distribution of reflexives in terms of the interaction of multiple modules and that hypothesize the existence of different types of reflexives. The single-module approach is represented by Chomsky (1981,1986), whereas the multi-module approach has been advanced most prominently by Pollard and Sag (1992) and Reinhart and Reuland (1993) (although the need for a distinction between local vs. nonlocal binding was recognized already by Bouchard (1984) and Lebeaux (1985)). The two central questions running through these works are: (i) What is the core syntactic domain in which the binding principles determine the selection of a reflexive or a pronoun?—i.e., How is "core binding" defined?— and (ii) What are the properties of noncore anaphoric relations involving reflexives—i.e., What are the properties of "logophoric binding"? Syntax and Semantics, Volume 33 Long-Distance Reflexives
119
Copyright © 2001 by Academic Press All rights of reproduction in any form reserved. 0092-4563/00 $35.00
120
Arild Hestvik and William Philip
In this chapter, we present experimental evidence from Norwegian child language that binding into locative PPs and of possessive reflexives constitute cases of core binding and not logophoric binding, whereas binding into picture NPs is indeed logophoric binding. The results match the predictions of a theory in which the relation between a reflexive and its antecedent may involve LF-movement, such that the range of possible LF-movements partially determines what counts as local binding. Any reflexive that cannot move into a local relation with its antecedent is ungrammatical unless it can be reinterpreted via logophoric binding. The chapter is organized as follows: First, we review what are widely assumed to be the distinctive semantic and syntactic properties of logophoric reflexives. Then we consider how two alternative binding frameworks deal with such reflexives. After this theoretical discussion, we review a recent psycholinguistic study which has bearing on the central theoretical questions in that it sheds light on the logophoric status in child English of a reflexive in the object position of a locative PP. Finally, we present our own experimental evidence from Norwegian child language and discuss its implications.
2. LOCAL VS. NONLOCAL (LOGOPHORIC) BINDING It is widely assumed that logophoric binding generally exhibits the following properties: • Logophoric anaphors tend to have antecedents that occupy the "center of consciousness" of the discourse in the sense that the propositional content of the sentence containing the logophor is viewed from the perspective of the person denoted by the antecedent. • Logophoric reflexives exhibit pronounlike properties: They may have non-c-commanding antecedents or split antecedents; they may allow discourse reference, and they may allow strict identity in VP-ellipsis. The best theory should predict which occurrences of reflexives have logophoric properties and differentiate them from locally bound reflexive occurrences and ordinary pronouns. We will now consider the predictions of two such alternative theories. 2.1. The Non-Coargument Theory The theory developed by Reinhart and Reuland (1993) proposes that only reflexives bound by coarguments are subject to core binding condi-
Syntactic vs. Logophoric Binding
121
tions, whereas reflexives with non-coargument antecedents either are ungrammatical or are in principle acceptable as logophors. To illustrate with a clear-cut case, consider the following pair: (1) a. [[Oscar1's father]2's picture of himself2] was on display. b. [Oscar1's father]2 commented [that a picture of himself1/2 was on display]. c. Oscar felt a wave of panic. He had believed that his identity was unknown to the police. At breakfast he learned otherwise. His father commented that a picture of himself was on display at the post office on an FBI poster. In (la), himself can only refer to Oscar's father, whereas in (1b), himself can refer to Oscar, even though it is not c-commanded by any nominal expression denoting Oscar. Embedding (1b) in an appropriate context improves the acceptability of a logophoric reading of the reflexive, as shown in (1c). In short, Reinhart and Reuland's theory—henceforth, the "non-coargument theory"—holds that if the antecedent and the reflexive are not coarguments, the possibility of logophoric interpretation is simply left open by the syntax. (In Reinhart and Reuland's theory, the distribution of reflexives is also constrained by the A-Chain Condition; however, since we will not be concerned with cases in which this constraint would apply—i.e., a reflexive c-commanding its own antecedent—we will ignore this aspect of the theory as well as the general question of how, within this theory, the A-Chain Condition interacts with logophoric binding.) The non-coargument theory leads to the expectation that, nothing else being said, any occurrence of a reflexive which does not have a coargument antecedent (and which does not c-command its own antecedent) should, in principle, be acceptable under a logophoric reading and should display at least some properties of logophoricity. In particular, binding into locative PPs in English should exhibit logophoric properties, as should the binding of possessive reflexives in languages that have them. However, consider the contrast between (2a) and (2b). (2) a. Clinton1's car carried a picture of himself1 on the roof. b. * Clinton1's car backfired/collapsed/exploded behind himself1. According to the non-coargument theory, the reflexive in an English locative PP must be a logophor since it is not a coargument of its antecedent. However, this reflexive does not appear to have the expected logophoric property of always admitting a non-c-commanding antecedent, as seen by (2b). This raises the question of whether the non-coargument theory makes the right predictions for English. We return briefly to this issue in the discussion of Avrutin and Cunningham's experimental work.
122
Arild Hestvik and William Philip
Let us now consider which occurrences of reflexives in Norwegian should be logophoric according to the non-coargument theory. First, possessive reflexives should be logophors, since they are not bound by their coarguments. (The NP that contains the possessive reflexive is a coargument of its antecedent, but the possessive reflexive itself is not.) However, consider the Norwegian possessive reflexives in (3). Sentence (3a) illustrates a grammatical instance, but as (3b-3d) show, these reflexives do require c-commanding antecedents, do not alow discourse reference, and do not allow split antecedents. (3) a. John1 holdt i [hatten sin1]. 'John held onto his(self 's) hat.' b. *John1s mor2 holdt i [hatten sin1]. 'John's mother held his(self's) hat.' c. John1 var forskrekket. *Marit2 hadde lagt vekk [hatten sin1]. 'John was terrified. Mary had put away his(self's) hat.' d. *John1 ba Marit2 ringe [slektningene sine1+2]. 'John asked Mary to call their(selves') relatives.' In general, when a non-c-commanding NP is a contextually plausible antecedent, it cannot be the grammatical antecedent of a possessive reflexive in this language. For example, the LF represented in (3c) remains unavailable when immediately preceded by a context such as "John was terrified." In sum, these possessives exhibit none of the expected properties of logophoric reflexives. The same is true of Norwegian long-distance reflexives in locative PPs. They should be logophors according to the non-coargument theory, but exhibit none of the typical properties of logophors, as seen in (4b-4d). (4) a. John1 kikket [bak seg1]. 'John looked behind self.' b. *John1s mor2 satte stolen [bak seg1], 'John's mother put the chair behind self.' c. John1 var forskrekket. *Marit2 kikket [bak seg1]. 'John was terrified. Mary looked behind self.' d. *John1 ba Marit2 sette stolen [bak seg1+2]. 'John asked Mary to put the chair behind self.' Finally, consider the case of Norwegian seg binding into a picture NP. As shown in (5), these occurrences of a reflexive also do not appear to behave like logophoric reflexives.
Syntactic vs. Logophoric Binding
123
(5) a. John1 kj0pte [et bilde av seg1]. 'John bought a picture of self'. b. *John1s mor2 kj pte [et bilde av seg1]. 'John's mother bought a picture of self.' c. John1 var forskrekket. *Marit2 kj pte [et bilde av seg1]. 'John was terrified. Mary bought a picture of self.' d. *John1 ba Marit2 kj0pe [et bilde av seg1+2]. 'John asked Mary to buy a picture of selves.' The facts presented in (2)-(5) might simply be taken as evidence that Reinhart and Reuland's non-coargument theory fails to provide an empirically adequate definition of core binding and miscategorizes a subclass of core binding as logophoric binding. On the other hand, it is also possible that the discourse-related rules of logophoric binding and the syntactic rules of core binding partially overlap in their effect on reflexive antecedent selection. Specifically, in the case of Norwegian, it may be that logophoric occurrences of seg conform to a language-specific discourse rule of subject-orientation that "mimics" the c-command restriction of core binding. This hypothesis would entail that absence of a c-commanding antecedent is a possible but not a necessary correlate of logophoric antecedent selection. If it is possible for discourse constraints on logophoric binding to "mimic" syntactic constraints on core binding, what evidence could differentiate between core binding and logophoric binding in a language like Norwegian? As we will discuss, child language may be the only source of such evidence. But first let us consider how an alternative binding theory deals with logophoric reflexives. 2.2. The LF-Movement Theory As an alternative to Reinhart and Reuland's approach, we might define core binding in the standard manner—i.e., with reference to nominal features [+/— anaphor, +/— pronominal] and with reference to governing category—but posit additionally that the binding requirement for a reflexive must be satisfied in the smallest domain in which it could be satisfied (e.g., Huang (1983), Chomsky (1986), and Hestvik (1991)). Under this approach, a possessive reflexive extends its domain to the containing matrix clause, as represented in (6a), since it already occupies the highest position in the nominal that a potential binder could occupy. Here, it finds a potential binder, namely the matrix subject. Similarly, a reflexive in a locative PP can, and therefore must, extend its binding domain to the containing clause, as represented in (6b).
124
Arild Hestvik and William Philip
(6) a. [D(sin) John1 holdt i [hatten sin 1 ]. 'John held onto his(self s) hat.' b. [D(seg) John1 satte botten [bak seg1]]. 'John put the bucket behind self.' In contrast, a reflexive that is the object of a picture NP does have a position for a potential binder, namely, the specifier position of that NP. This makes the picture NP the binding domain, as represented in (7) (specifier position indicated by underlining). (7) Jon kjopte [D(seg) bilder av seg]. 'John bought pictures of self.' In the case of picture NPs, therefore, unless the NP has a subject that can bind it, the reflexive will not be bound in the smallest domain in which it could potentially be bound and will have to resort to logophoric binding to receive a semantic value, following whatever discourse-related rules of logophoricity exist in the language. This approach fails, however, to predict the antisubject orientation of pronouns in languages such as Norwegian. Huang's theory, for example, would incorrectly predict that a pronoun in a locative PP or a possessive pronoun could satisfy its binding requirements internal to those phrases. In fact, in these constructions a pronoun must be non-coreferent with the higher subject. This problem is solved, however, in the LF-movement approach developed in Hestvik (1992). Under this approach, LF-movement applies both to pronouns and reflexives, yielding antisubject orientation in the case of pronouns and subject orientation in the case of long-distance reflexives. In other words, Norwegian simplex reflexives like seg or sin move at LF toward their antecedents in a way that mirrors reflexive clitic movement at surface structure in the Romance languages. For concreteness, let us assume that these reflexives must connect with their antecedents by adjoining at LF to the head of the Agr-projection of which the antecedent is a specifier. This explains the c-command requirement, since movement is only to positions that c-command the trace-position. The movement itself could follow from a general requirement that reflexives must get their features from a linguistic antecedent via a head-specifier relationship. Logophor binding can then be viewed as a strategy available to a reflexive which has not been able to establish the requisite locality relation via movement. On this account, after movement of the reflexive, the LF for a Norwegian sentence with a possessive reflexive such as Gutten Ioftet hatten sin 'The boy
Syntactic vs. Logophoric Binding
125
lifted his(self's) hat' would be as shown in (8), abstracting away from irrelevant details.
Likewise, the LF for a Norwegian sentence with a locative PP such as Gutten satte stolen bak seg "The boy put the chair behind self would be (9).
This analysis requires that the DP and PP boundaries do not block LF movement of the reflexive. (Specifically, the DP does not block movement of its head D.) On the other hand, DP must block movement of elements contained in a lower NP. This is in accord with general constraints on movement, where DP and CP most often act as the categories that block movement of elements they properly contain. In the case of picture NPs, therefore, or reflexives internal to any NP, the hypothesized LF-movement of the reflexive is blocked; there is no escape hatch. Thus, the LF for a Norwegian sentence with a picture NP such as Gutten viste hesten et bilde av seg 'The boy showed the horse a picture of self would be as shown in (10).
126
Arild Hestvik and William Philip
Since movement is blocked by the DP, the reflexive must stay in situ and can only resolve its referential requirements via logophoric binding. In sum, the LF-movement theory makes the claim that in adult Norwegian, possessive reflexives and reflexives in locative PPs enter into core binding (at LF), but this is not possible for reflexives in picture NPs, which are true logophors. The same question arises, however, as with the non-coargument theory: If the difference between core binding and logophoric binding is not manifested in adult intuitions, how can we distinguish the two? Specifically, in the case of seg in a picture NP, what theory-independent evidence do we have that this reflexive is governed by logophoric binding rather than core binding? We now turn to the study of child language, where we believe such evidence can be found.
3. USING CHILD LANGUAGE AS EVIDENCE: AVRUTIN AND CUNNINGHAM (1997) One well-established and widely accepted generalization about first language acquisition and development is that children acquire mastery of the discourse-related constraints of their target grammar long after they have fully acquired its syntactic constraints. If syntactic principles alone constrain the interpretation of a sentence in adult grammar, the young child will show perfectly adultlike comprehension performance with this sentence. On the other hand, if discourse-related constraints also contribute to the meaning of
Syntactic vs. Logophoric Binding
127
a sentence in the adult grammar, then the young child may well show significant levels of non-adultlike comprehension performance with this sentence. A well-known example of this is observed in the acquisition of the adult grammatical constraints on intersentential pronominal anaphora. When pronoun interpretation is only constrained by syntax, as in the case of a bound variable reading, children show virtually no divergence from adultlike performance. However, when a discourse-related rule also participates in determining the meaning of a pronoun, as in the case of a referential reading, children show the well-known (misleadingly labeled) "delay of Principle B" effect (e.g., Chien and Wexler (1990) and Grodzinsky and Reinhart (1993)). Another example of the delayed acquisition of discourse-related rules can be seen in children's use of the grammatical features of pronouns. In the case of the intersentential phi feature agreement of a pronoun with its antecedent, which is purely a matter of syntax, children show virtually no errors (Chien and Wexler, 1990). However, in the case of the extrasentential phi feature agreement of a pronoun with its discourse antecedent, which is not a matter of syntax, on-line psycholinguistic studies have shown that young children differ significantly from adults, often failing to use features such as number or gender to disambiguate pronominal discourse anaphora (Tyler, 1983). Given the general developmental dissociation of syntactic and discourse-related rules, an observation of substantial child "errors" with a given sentence-type or construction becomes evidence that the "correct" adult performance with this sentence-type or construction is the effect of a discourse-related rule of grammar, not the effect of a syntactic rule. Applying this reasoning, Avrutin and Cunningham (1997) provide an argument from English child language that binding into locative PPs is logophoric in adult English. Avrutin and Cunningham tested pre-school children's interpretation of reflexives in locative PPs, a construction which clearly differentiates Reinhart and Reuland's theory from other accounts. Building on work by Huang (1983), Hestvik (1991) showed that binding into locative PPs can be considered a case of core binding if the binding domain for reflexives is defined as requiring a subject. If so, children should not have problems with the type of construction. On the other hand, according to Reinhart and Reuland's theory, reflexives in locative PPs are logophorically governed. If so, one might expect children to have problems with such reflexives insofar as the discourse-related rules of logophoricity had not been fully acquired. Avrutin and Cunningham found that, indeed, young children acquiring English as a first language accepted non-adultlike (#) readings such as in (11) on average 53% of the time. In each sentence in (11), the antecedent of the reflexive is embedded in the subject and therefore is not a coargument of the reflexive (nor does it c-command it).
128
Arild Hestvik and William Philip
(11) a. #[The man near the boy1]2 hid a book behind himself1. b. #[The woman near the girl1]2 drew a circle around herself c. #[The man near the boy1]2 put a hat on himself1. In stark contrast, the same children accepted the ungrammatical readings represented in (12) on average only 13% of the time. In each sentence in (12), the antecedent is a coargument of the reflexive (and c-commands it). (Presumably, the 13% non-adultlike responses observed here should not be attributed to the occasional assignment of an ungrammatical LF but rather to the occasional confounding effect of nonlinguistic factors.) (12) a. *[The man near the boy1]2 is washing himself1. b. *[The woman near the girl1]2 is tickling herself1. c. *[The woman near the girl1]2 covered herself1. Avrutin and Cunningham concluded that children have trouble finding the adult-grammatical antecedent of a reflexive in a locative PP because antecedent selection in this case is determined by logophoric binding. Since the rules of logophoric antecedent selection require access to discourse-related knowledge, and since this is missing, incomplete, or hard to access for preschool children, logophoric reflexives are effectively liberated from such constraints in the children's grammar, with the result that significant deviance from adultlike performance is observed. (Note that the incomplete acquisition of a discourse-related rule constraining logophoric binding does not necessarily result in 0% adultlike performance.)
4. EXPERIMENTAL EVIDENCE FROM NORWEGIAN CHILD LANGUAGE We turn now to our own experimental work on child Norwegian, which applied the same research strategy as that of Avrutin and Cunningham. The three reflexive constructions we examined were possessive reflexives, simplex reflexives in locative PPs, and simplex reflexives in picture NPs. The prediction of the non-coargument theory was that the Norwegian children would treat all three reflexive constructions in the same way. Either they would show fully adultlike performance with each or they would make the same kind of coreference assignment "error" with each. In the latter case, this would be due to the circumstance that (i) in each construction the reflexive was a logophor and (ii) the children had not fully mastered the discourse-related rules of logophoric binding. In contrast, the LF-movement theory predicted that if the children showed any errors, it would only be
Syntactic vs. Logophoric Binding
129
with reflexives in picture NPs, since, according to this theory, the possessive reflexive and the reflexive in a locative PP involve core binding, not logophoric binding. We carried out two experiments with two overlapping groups of Norwegian preschool children of roughly the same age range. One group was tested on possessive reflexives and reflexives in locative PPs, and the other group was tested on reflexives in picture NPs. Eleven children participated in both experiments, i.e., were tested on all three reflexive constructions.
4.1. Experiment 1: Possessive Reflexives and Reflexives in Locative PPs The stimuli sentences testing possessive reflexives were the three questions in (13). Sin is the third person singular form of the Norwegian possessive reflexive, unspecified for gender. The three sentences testing reflexives in locative PPs are shown in (14). Seg is the Norwegian third person simplex reflexive, unspecified for number or gender. (For brevity, we only give a morphological analysis of (13a) and (14a).) (13) a. Knytt-er dame-n sko-en sin? Knit-Pres lady-def shoe-def PosRefl 'Is the lady tying her(self's) shoe?' b. L fter damen hatten sin? 'Is the lady lifting her(self's) hat?' c. Holder gutten i buksen sin? 'Is the body holding his(self's) pants?' (14) a. Ha-r dame-n lagt en harb rste bak seg? Aux-Pres lady-def put a hairbrush behind Refl 'Has the lady put a hairbrush behind self?' b. Har jenten satt en stol bak seg? 'Has the girl put a chair behind self?' c. Tegner gutten en sirkel rundt seg? 'Is the boy drawing a circle around self?'
4.1.1. SUBJECTS Thirty-two monolingual children acquiring Norwegian as a first language participated in Experiment 1. The age range was from 3:11 to 6:8. The mean age was 5:10. There were approximately as many boys as girls in this group of children. In addition, 23 adult Norwegian native speakers participated as a control group.
130
Arild Hestvik and William Philip
4.1.2. MATERIALS, DESIGN, AND PROCEDURE The experiment was presented to the child as guessing game with pictures. For each trial of an experimental condition, one experimenter, who sat opposite the child, would make a guess about a picture that he could not see. The child, who could see the picture, determined whether or not the guess was correct. The guess, which was the target input sentence, was always presented in the form of a yes/no question. This allowed the child to respond by simply saying "Yes" or "No." A second experimenter sat with the child and manipulated the pictures. There were three different trials of each test condition, varying as to the types of objects depicted in the picture and as to the predicate used in the target input sentence. The predicates were knytte 'tie,' se pa 'look at,' and I fte 'lift.' Trials eliciting "Yes" responses from adults were perfectly counterbalanced by trials eliciting adult "No" responses because each sentence was tested once with a "correct guess" picture and once with a "wrong guess" picture, as exemplified in Fig. 1. (Figure 1 represents two separate experimental items presented at different points in the experiment.) In each trial of each test conditions, both an inter- and an extrasentential potential antecedent for the reflexive was explicitly introduced into the discourse context in a thermatically neutral fashion immediately prior to presentation of the guess. For example, for each of the trials shown in Fig. 1, just
Figure 1. Possessive Reflexive Sin Conditions. The target input sentence presented with the correct guess picture was a trial of the test condition eliciting an adult affirmative response; the same sentence presented with the wrong guess picture was a trial of the test condition eliciting an adult negative response.
Syntactic vs. Logophoric Binding
131
before delivering the guess, the guessing experimenter would say something like "Hmm. Let's see. A girl and a lady...," as if musing out loud about how to guess. The grammatical antecedent was always mentioned last in this explicit definition of the discourse context. If the child's grammar treated sin as a logophor—i.e., as a reflexive not subject to a grammatical constraint but rather to a discourse-related constraint—and if the child had not fully acquired these discourse-related constraints, then we might expect the child to assign reference to the reflexive in a non-adultlike manner significantly often and to answer quite differently from adults much of the time. For example, such a child should significantly often take the girl to be the referent of sin in the trials of the two test conditions represented in Fig. 1, and therefore often respond "No" with the correct guess picture and "Yes" with the wrong guess picture. On the other hand, if the grammatical system of the child analyzes these cases as grammatically conditioned, we would expect adultlike performance virtually all of the time. The two conditions testing reflexives in locative PPs are exemplified in Fig. 2. These conditions were presented in the same manner and with as many repeated measures as the conditions testing possessive reflexives. For the three different trials of each condition, the predicates sette 'place' and legge 'put' were used. For each trial, the experimenter manipulating the pictures always gave the guessing experimenter a "hint." For example, for each of the trials represented in Fig. 2, the hint was "The lady put the hairbrush somewhere." This was done to make the target input sentences maximally felicitous as guesses.
Figure 2. Seg in Locative PP Conditions.
132
Arild Hestvik and William Philip
The experiment also included filler items using the same picture types and sentence types as the test items, but with pronouns instead of reflexives in the target input sentences. This was part of another experiment (Hestvik and Philip, 1997, 2000), but may serve here as an independent measure of the children's linguistic development. The filler and experimental items were presented to all subjects in a single, pseudo-random order, in two 15-minute sessions spaced approximately one or two weeks apart. Note that, in our experiment, it was not the child, but rather the experimenter making the guesses who appeared to be under examination. The experimenter's guesses were judged by the children for their truth value, and all the children immediately grasped that these guesses could just as often be wrong as right. (Thanks to the counterbalancing, they were wrong for an adult half the time.) For this and other related reasons, we believe that our adaptation of the Truth Value Judgment Task is formally and functionally equivalent to that used by Avrutin and Cunningham, and that their experimental results may be validly compared to ours despite superficial differences between the two experimental paradigms. 4.1.3. RESULTS OF EXPERIMENT 1 An ANOVA showed there to be no significant age effects when comparing 5-year-olds (n = 16) with 6-year-olds (n = 16). The average percentages of adultlike responses of the 32 children under each test condition are shown in Table 1 (standard error in parentheses). The average percentages of expected responses under these experimental conditions for the 23 adults in the control group are given in Table 2. 4.1.4. DISCUSSION OF EXPERIMENT 1 Even though it was possible to choose an antecedent other than the local subject, the 32 children of Experiment 1 performed virtually adultlike under all the test conditions. If antecedent selection for the reflexives were subject to discourse-related rules, and if children at this age do not master these TABLE 1 PERCENT ADULTLIKE PERFORMANCE OF CHILDREN Picture Tye (Adult Response) Correct Guess (Yes) Wrong Guess (No)
Possessive Reflexive Sin
Seg in Locative PP
99% (1) 95% (2)
98% (1) 96% (2)
Syntactic vs. Logophoric Binding
133
TABLE 2 PERCENT EXPECTED PERFORMANCE OF ADULTS Picture Type (Adult Response) Correct Guess (Yes) Wrong Guess (No)
Possessive Reflexive Sin
Seg in Locative PP
97% (2) 100%
100% 99% (1)
rules, or due to general processing limitations cannot apply them easily, this highly adultlike performance is not expected. On the other hand, such performance is indeed expected if antecedent selection for possessive reflexive sin and seg in a locative PP is determined purely by syntactic rule. Furthermore, if we consider how these children performed with nonreflexive pronouns in the same sentence types and with the same picture types, we find an interesting contrast with their performance on the test items. Consider Table 3, which shows the average percentages of adultlike responses of the children under the same experimental conditions but with pronouns substituted for reflexives. (Hans is the possessive form of the third person masculine singular pronoun; ham is the accusative form. In a sentence such as John pekte pa hatten Hans, "John pointed to his hat," coreference between John and hans is ungrammatical in adult Norwegian. Likewise, in a sentence such as John satte stolen bak ham, "John put the chair behind him," John is not a grammatical antecedent of ham.) The children's comparatively poor performance with pronouns may be accounted for in the standard manner, i.e., by the hypothesis that they have not fully acquired some set of discourse-related rules or lack the computational power to apply these rules successfully most of the time. But why does this incomplete acquisition or general immaturity of computational resources not affect the same children's performance with reflexives? The simplest answer to this question is that discourse rules play no role whatsoever in the antecedent selection of Norwegian possessive reflexive sin and seg in a locative PP, neither in the child grammar nor in the adult grammar. We conclude, therefore, that reflexive antecedent selection in the case of TABLE 3 PERCENT ADULTLIKE PERFORMANCE WITH PRONOUNS Picture Type (Adult Response) Correct Guess (Yes) Wrong Guess (No)
.
Possessive Pronoun Hans
Ham in Locative PP
78% (4) 46% (6)
60% (5) 34% (6)
134
Arild Hestvik and William Philip
these two reflexive constructions is determined by purely by syntactic mechanisms, which children of the age tested have fully acquired. 4.2. Experiment 2: Reflexives in Picture NPs We now turn to a second experiment in which we examined children's comprehension of reflexive seg in picture NPs. The test sentences are shown in (15). (15) a. Driv-er hund-en og vis-er gutt-en et bilde av seg? Do-Pres dog-def and show-Pres boy-def a picture of Refl 'Is the dog showing the boy a picture of self?' b. Driver gutten og viser hesten et bilde av seg? 'Is the boy showing the horse a picture of self?' c. Driver politimannen og viser gutten et bilde av seg? 'Is the policeman showing the boy a picture of self?'
4.2.1. SUBJECTS Forty-four children acquiring Norwegian as a first language participated in Experiment 2. The age range was from 4:5 to 7:4. The mean age was 6:3. The same 23 adults from Experiment 1 also participated as a control group for Experiment 2.
4.2.2. MATERIALS, DESIGN, AND PROCEDURE. As in Experiment 1, tokens of the target input sentence type were matched with correct guess and wrong guess pictures, as exemplified in Fig. 3. For the trials represented in Fig. 3, the context-setting linguistic input delivered by the guesser immediately prior to making his guess was a statement such as "Hmm. Let's see. A horse, a boy, and two pictures." Experiment 2 used exactly the same experimental procedure as that in Experiment 1, and the design was the same as well. Aside from having different test conditions, the two experiments differed only in that they contained different filler items and in that for Experiment 2, a single predicate, vise 'show,' was used for the three different trials of each test condition. The filler items of Experiment 2 consisted of sentences like that shown in Fig. 3 but with the reflexive seg replaced by the pronoun ham 'him'—e.g., Driver gutten og viser hesten et bilde av ham? 'Is the boy showing the horse a picture of him?'—and also simple sentences containing a locally bound instance of the Norwegian complex reflexive segselv—i.e., sentences such as Driver gutten og torker segselv? 'Is the boy drying himself off?' Again, these
Syntactic vs. Logophoric Binding
135
Figure 3. Seg in Picture NP conditions.
may serve as independent measures of the linguistic development of the children who participated in Experiment 2. 4.2.3. RESULTS OF EXPERIMENT 2 Again, there were no age significant effects (ANOVA). The average percentages of adultlike responses of the 44 children under the test conditions exemplified in Fig. 3 and under the control condition with segselv are given in Table 4 (standard error in parentheses). The adult control group performed 100% as expected for each of these experimental conditions, i.e., always selecting a subject antecedent. 4.2.4. DISCUSSION OF EXPERIMENT 2 As can be seen from Table 4, the children performed significantly worse with seg in picture noun phrases. Under the test condition for which the adult answer was "No," i.e., with the wrong guess picture, the children anTABLE 4 PERCENT ADULTLIKE PERFORMANCE Picture Type (Adult Response) Correct Guess (Yes) Wrong Guess (No)
Seg in Picture NP
Segselv Locally Bound
84% (3) 52% (6)
98% (1) 99% (1)
136
Arild Hestvik and William Philip TABLE 5 PERCENT ADULTLIKE PERFORMANCE WITH PRONOUNS Picture Type (Adult Response) Correct Guess (Yes) Wrong Guess (No)
Ham in Picture NP 81% (4) 39% (6)
swered "Yes" roughly half of the time, suggesting that they do not have the adult rules for selection of antecedent for seg when it occurs in picture NPs. (We attribute the fact that they performed slightly better under the correct guess condition to the general confounding effect of a yes-response bias.) This finding is expected if antecedent selection for the reflexive in this context requires mastery of a discourse-related rule and children at this age generally lack such mastery. It is highly unexpected, on the other hand, if syntax governs antecedent selection for the reflexive in this context, since children at this age generally show complete mastery of the syntactic rules of the adult grammar they are acquiring. Interestingly, the results for seg in picture noun phrases closely parallel those for pronominals in picture noun phrases. Consider Table 5, which shows the average percentages of adultlike responses of these children for the filler conditions with pronouns. This suggests that in practice, the children did not distinguish between pronouns and reflexives in this particular construction, supporting the hypothesis that they treat the picture NP as a context for logophoric binding. In other words, they recognize this sentence type as a logophoric context, but not knowing the rules for logophoric antecedent selection, they fail to differentiate between pronouns and reflexives in this context. 4.3. Within-Subjects Comparison The results of Experiments 1 and 2 suggest that, in child Norwegian, possessive reflexives and reflexives in locative PPs are governed by the syntactic principles of core binding, while reflexive seg in picture NPs are governed by the discourse-related principles of logophoric binding. It might be objected, however, that the contrasts in performance observed between the test conditions of Experiment 1 and those of Experiment 2 may simply indicate a developmental difference between the two groups of children that participated in each experiment. In other words, perhaps the children of Experiment 1 simply happened to have been further along in their acquisition of the target grammar than those of Experiment 2—despite the fact
Syntactic vs. Logophoric Binding
137
TABLE 6 PERCENT ADULTLIKE PERFORMANCE OF 11 CHILDREN IN BOTH EXPERIMENTS Picture Type
Possessive Reflexive Sin
Seg in Locative PP
Seg in Picture NP
Correct Guess Wrong Guess
100% 94% (6)
98% (1) 97% (3)
79% (7) 61% (12)
that they were slightly younger—and that is why they showed more adultlike performance with possessive reflexives and reflexives in locative PPs than the children of Experiment 2 showed with reflexives in picture NPs. This objection can be directly addressed by considering the performance of the 11 children who participated in both experiments. Consider, then, Table 6, which shows the percentages of adultlike responses under the three test conditions for the 11 children who participated in both experiments, and Table 7, where the performance of these 11 children under the wrong-guess conditions is compared with that of each larger group they were part of. As can be seen at a glance, the response pattern for the 11 children who participated in both experiments is statistically nondistinct from that of each larger group they were part of. The ability of these 11 children to detect grammatically determined falsity with sentences containing a possessive reflexive—i.e., to recognize that such sentences falsely described wrong-guess pictures—differed significantly (t-test, p < 0.0189) from their ability to detect grammatically determined falsity with sentences containing a reflexive in a picture NP. (By transitivity, the contrast is even greater between false sentences with seg in a locative PP and false sentences with seg in a picture NP.) This within-subjects comparison provides strong evidence that the contrast in performance observed between reflexives in picture NPs, on the one hand, and possessive reflexives and reflexives in locative PPs, on the other, is due to linguistic factors, not to developmental differences between the two groups of children. TABLE 7 PERCENT ADULTLIKE PERFORMANCE UNDER WRONG GUESS CONDITIONS Group (n) Experiment 1 (32) Experiment 2 (44) Experiment 1+2 (11)
Possessive Reflexive Sin
Seg in Locative PP
95% (1) 94% (6)
98% (1) 97% (3)
Seg in Picture 52% (6) 61% (12)
138
Arild Hestvik and William Philip
One caveat should be mentioned, however. Although we find it implausible, there is a possible nonlinguistic explanation of the asymmetry observed in the children's performance under the test conditions. Recall that under the conditions testing possessive reflexives and reflexives in locative PPs, one potential antecedent for the reflexive was extrasentential and the other was sentence-internal, while under the experimental conditions testing reflexives in picture NPs, both potential antecedents were sentence-internal. It might be speculated, then, that the children's highly adultlike performance with the former two reflexive constructions was confounded by the "Clever Hans effect," a nongrammatical strategy of picking the linearly closest NP as the antecedent for the reflexive. For the sentences testing picture NPs, this strategy would yield non-adultlike responses, since the linearly closest NP was the indirect object. On the other hand, for the sentences testing the possessive and locative constructions, this strategy would yield pseudo-adultlike performance, since the linearly closest NP, i.e., the subject, happened to be the grammatical antecedent. To conclusively rule out this alternative, strategy-based account of our findings, one would have to test sentences with possessive and locative constructions that also had an additional, sentence-internal NP linearly closer to the reflexive than the subject NP. We leave this for future research, and simply note in the meantime that the null hypothesis in assessing experimental data with children is that grammatical linguistic performance is determined by grammar.
5. CONCLUSION Our findings take quite a different direction from those of Avrutin and Cunningham. Although our results have no bearing on the grammar of anaphors in child English, they do suggest the following general observations about child Norwegian: • Possessive reflexive sin heading a DP and reflexive seg in a locative PP are not logophors in child Norwegian and, for this reason, Norwegian preschool children show fully adultlike performance with such reflexives. • Reflexive seg in a picture NP is a logophor in child Norwegian and, for this reason—given a general (and presumably universal) delay in the acquisition of logophoric binding—Norwegian preschoolers show significant levels of non-adultlike performance with such reflexives. Given the Strong Continuity Hypothesis of first language acquisition theory, these observations about child Norwegian constrain theories of adult
Syntactic vs. Logophoric Binding
139
Norwegian linguistic competence as well. They suggest that in adult Norwegian, possessive reflexives and reflexives in locative PPs are governed by core binding, not by logophoric binding, and that only reflexives in picture NPs are true logophors. If this is the correct descriptive generalization, then the coargument theory of Reinhart and Reuland (1993) is in need of revision. The LF-movement theory, on the other hand, predicts precisely this state of affairs.
REFERENCES Avrutin, S., and Cunningham, J. (1997). Children and reflexivity. In Proceedings of the 21st BU Conference on Language Development, 13-23. Cascadilla Press, Somerville, MA. Bouchard, D. (1984). On the Content of Empty Categories. Foris. Chien, Y.-C. and Wexler, K. (1990), Children's knowledge of locality conditions in binding as evidence for the modularity of syntax and pragmatics. Language Acquisition 1,225-295. Chomsky, N. (1981) Lectures on government and binding. Dordrecht: Foris. Chomsky, N. (1986). Knowledge of Language. New York: Praeger. Grodzinsky, Y, and Reinhart, T. (1993). The innateness of binding and the development of coreference. Linguistic Inquiry 24,69-103. Hestvik, A. (1991). Subjectless binding domains. Natural Language and Linguistic Theory 9,455-96. Hestvik, A. (1992). LF-movement of pronouns and anti-subject orientation. Linguistic Inquiry 23,557-594. Hestvik, A., and Philip, W. (1997). Reflexivity, anti-subject orientation and language acquisition. In Proceedings of NELS 27. GLSA, Univ. of Massachusetts/Amherst. Hestvik, A. and Philip, W. (2000). Binding and coreference in Norwegian child language. Language Acquisition 8,171-235. Huang, J. (1983). A note on binding theory. Linguistic Inquiry 14,554-561. Karmiloff-Smith, A. (1980). Psychological processes underlying pronominalization and non-pronominalization in children's connected discourse. In J. Kreiman and A.E. Ojeda (eds). Papers from the Parasession on Pronouns and Anaphors. Chicago, IL: Chicago Linguistic Society. Lebeaux, D. (1985). Locality and anaphoric binding. Linguistic Review 4,343-363. Pollard, G, Sag, I. (1992). Anaphors in English and the scope of the binding theory. Linguistic Inquiry 23,261-303. Reinhart, T, and Reuland, E. (1993). Reflexivity. Linguistic Inquiry 24,657-720. Tyler, L. K. (1983). The development of discourse mapping processes: The on-line interpretation of anaphoric expressions. Cognition 13,309-341.
This page intentionally left blank
LOGOPHORICITY, ATTITUDES, AND ZIJI AT THE INTERFACE C.T. JAMES HUANG* C.-S. LUTHER Liu *Department of Linguistics University of California Irvine, California Department of Foreign Languages and Literatures National Chi-Nan University, Puli, Nantou, Taiwan
1. INTRODUCTION Long-distance reflexives (LDRs) constitute an interesting phenomenon for theoretical linguistics because they pose a challenge to the standard theory of anaphor binding as put forth in Chomsky (1981) and subsequent revisions of it within the Principles-and-Parameters framework. The standard theory takes a reflexive pronoun to be an anaphor subject to the condition in (1) (henceforth, BCA), with the notion of a governing category as defined in (2): (1) Binding Condition A (Chomsky, 1981) An anaphor is bound in its governing category. (2) The Governing Category (cf. Huang, 1983, and Chomsky, 1986) a is the governing category for b if and only if a is the minimal category containing b, a governor of b, and a SUBJECT accessible to P. Long-distance reflexives are those that have their antecedents outside their governing categories. Their existence in any language would refute the BCA as a principle of UG under the assumption that these reflexives Syntax and Semantics, Volume 33 Long-Distance Reflexives
141
Copyright © 2001 by Academic Press All rights of reproduction in any form reserved. 0092-4563/00 $35.00
142
C.-T. James Huang and C.-S. Luther Liu
are anaphors and that the antecedents outside their governing categories are indeed their most immediate binders. Much research in the past 15 or so years has been devoted to the task of ascertaining whether this problem should lead one to abandon or revise BCA, or to rethink the assumption that they are indeed anaphors that are directly bound long-distance. Furthermore, since the phenomenon of long-distance reflexivization seems more widespread in some languages than in others, any attempt to deal with the phenomenon must also keep such cross-linguistic variations in mind. One of the most often talked about items in this subfield of binding theory is the reflexive pronoun ziji in Mandarin Chinese. Although Huang (1982) provided examples in which ziji could only be construed with a local antecedent (in compliance with BCA), he did not consider structurally similar examples in which it could prima facie be long-distance bound (in apparent violation of BCA). The first serious look at Mandarin reflexives was taken by Y.-H. Huang (1984) (cf. also Huang et al, 1984), where it was observed that (a) only the monomorphemic ("bare") reflexive ziji 'self, but not the polymorphemic taziji /himself/herself can be long-distance bound; (b) only subjects may qualify as antecedents; (c) long-distance binding may be blocked by certain local potential antecedents with (j-features distinct from those of the remote antecedent. Tang (1989) further noted that a "sub-commanding" subject may qualify as an antecedent under appropriate conditions. These properties—monomorphemicity, subject-orientation, sub-commanding, and blocking—as illustrated in (3a-d), respectively, constitute the basic properties of the Chinese reflexive which have been the subject of much study in recent years. (3) a. Monomorphemicity: Zhangsani renwei [Lisij hen zijii/j /taziji i/j ]. Zhangsan think Lisi hate self /himself 'Zhangsani thinks that Lisij hates himselfj/himi.' b. Subject-Orientation: Zhangsani song (gei) Lisij yi-zhang zijii/*j-de xiangpian. Zhangsan give to Lisi one-CL self's picture 'Zhangsani gives Lisij a picture of himselfyi/*j.' c. Sub-Commanding Antecedent: Zhangsani-de jiaoao hai-le zijii Zhangsan's arrogance hurt-Perf self 'Zhangsani's arrogance harmed himi.' d. Blocking Effects: Zhangsani renwei [ni j hen ziji *i/j ]. Zhangsan think you hate self
Logophoricity, Attitudes, and Ziji at the Interface
143
'Zhangsan thinks that you hate yourself.' (cf. (3a) with ni replaced by Lisi) Many scholars have attempted to explain these properties with proposed accounts that range from being purely syntactic to purely pragmatic. Our discussion will be directed to support the claims indicated in (4): (4) a. Neither the pure syntactic nor the pure pragmatic approach provides an adequate account of reflexive binding. b. A fundamental distinction must be drawn between the anaphoric and the logophoric uses of the reflexive, the former subject to syntactic conditions of anaphoricity, and the latter subject to certain pragmatic conditions of logophoricity. c. The dividing line between these two uses, we claim, is given by the traditional notion of "governing category" (GC) as given in (2). We shall start out in Section 2 with a review of major previous studies. In Section 3, we shall see that important evidence exists for treating the LDR as a logophor subject to pragmatic conditions, thus that the pure syntactic account is incorrect. We examine a clustering of properties which presented difficulties for the pure syntactic account and show that they lend themselves to a natural explanation we shall provide in logophoric terms. In Section 4 it will be shown that a pure pragmatic account is also incorrect, as there exists evidence that certain occurrences of the reflexive must be treated as syntactic anaphors subject to formal binding requirements. The conclusion that the reflexive may be a syntactic anaphor in some contexts but a pragmatic logophor in others thus supports the same claims already made by Reinhart and Reuland (1993), Pollard and Sag (1992), for English, and Xue, Pollard, and Sag (1994) for Mandarin. We argue that the traditional notion of governing category provides a satisfying "dividing line" between the anaphoric and logophoric uses of the reflexive, and compare our analysis with Reinhart and Reuland and Xue et al In Section 5, we take up the syntax and semantics of logophoricity, adopting an LF syntax involving IP-adjunction as originally proposed in Huang and Tang (1991), and a semantics of de se attitudes a la Chierchia (1989), and provide a brief comparison with Pan (1997). A summary of the results follows in Section 6.
2. PREVIOUS ANALYSES We will not be able to discuss in detail all of the worthy proposals that have been made in the literature, but mention only a few representative ac-
144
C.-T. James Huang and C.-S. Luther Liu
counts that have touched upon the Chinese LDR. Three approaches can be distinguished, the formal/syntactic, the functional/pragmatic, and a "mixed" approach that incorporates both the formal and the functional views. 2.1. The Formal Analyses The strict formal/syntactic approach treats all instances of the LDR as anaphors subject to BCA and seeks to cope with their problematic behavior by one of two strategies: (i) by revising the theory or (ii) by reanalyzing the facts. The first strategy is represented by the works of Yang (1983) and Manzini and Wexler (1987), who attempted to accommodate LDRs by parameterizing the notion of a GC across languages. According to this approach, languages differ in the size of the GC in which anaphors must be bound, which may range from the minimal clause in one language to the entire root sentence in another. LDRs exist in those languages with binding domains larger than the size of a minimal clause. The second strategy, which has enjoyed greater popularity, sought to reanalyze apparent LD binding as involving a number of successive steps of local binding, each in full satisfaction of BCA. Tang (1989) developed an early account using this strategy, with the proposal of an LF-reindexing rule. Following an idea of Lebeaux (1983), Chomsky (1986), and Pica (1987), Battistella (1989) proposed that the LDR was made possible by ziji undergoing LF head-movement across clause boundaries. This LF head-movement account was developed most fully in a number of papers by Cole, Hermon, and Sung (see Cole, Hermon, and Sung, 1990; Cole and Sung, 1994; Cole and Wang, 1996, among others), which turned out to be highly influential because of its apparent ability to explain a number of special properties of the LDR, including all of the basic properties illustrated in (3). According to this hypothesis, the monomorphemic ziji obligatorily moves in LF to I° of the minimal IP containing it, and optionally moves I°-to-I° to a higher IP, as illustrated: (5) [yZhangsan[ I ziji][VP yiwei [IP Lisi [I t'i] [VP piping-le t i ] ] ] ] Zhangsan self think Lisi criticize-Perf Apparent LD binding of the reflexive is possible because the reflexive has moved to the matrix Infl position, where it is locally bound by the matrix subject. The requirement of monomorphemicity follows from the natural assumption that only monomorphemic ziji is an X° category which can undergo head-movement, but not polymorphemic, phrasal taziji, etc. Subject orientation follows from the fact that, with ziji in I°, the subject is the only NP that c-commands the reflexive. The blocking effect also follows, under the assumption that I° agrees with its Specifier in j-features. Because the
Logophoricity, Attitudes, and Ziji at the Interface
145
Head Movement Constraint (HMC, of Travis 1984) requires ziji to move to the lower I° before it moves to the higher I°, ziji (and its trace t') must agree with the Spec's of both the lower and the higher IP, and hence the two Spec's (subjects) themselves must also agree in (j-features. Finally, a sub-commanding subject qualifies as a binder in Mandarin when it is the most prominent potential antecedent that enters into agreement with I°. Although the LF head-movement account enjoyed considerable popularity, it also encountered a number of problems—both theoretical and empirical. Among other things, Huang and Tang (1991) noted that LD binding into an island (e.g., an adjunct or relative clause) is possible: (6) Zhangsani bu xihuan [NP [CP neixie piping ziji i de ren]]. Zhangsan not like those criticize self DE person 'Zhangsanj does not like those people who criticized hinij.' (7) Zhangsani shuo [zhiyao Lisi bu zai piping ziji i jiu rang ta Zhangsan say if Lisi not again criticize self then let him canjia]. join 'Zhangsanj said that if Lisi will stop criticizing himi, he will let him participate.' Because the requisite head movement out of an island is prohibited by well-known strict locality conditions (HMC or the ECP), the head-movement approach would wrongly rule out these cases. Huang and Tang (1991) developed a different approach involving IP adjunction of the phrasal categories containing ziji which also derived most of the facts in (3) and was free from the difficulty posed by (6)-(7).1 Regardless of their differences, all these formal accounts suffer from a number of important empirical problems that have since come to light. Some of these surround the putative blocking effects first observed by Y.-H. Huang (1984). First, as pointed out by Xue et al. (1994), blocking effects may be triggered by non-subjects which, in general, are not potential antecedents of ziji(8) a. Zhangsani gaosu wo-j Lisik hen ziji *i/*j/k Zhangsan tell me Lisi hate self 'Zhangsanj told mej that Lisik hated self*i/*j/k.' b. Zhangsaniduiwoj shuo Lisik chang piping ziji *i/*j/k . Zhangsan to me say Lisi often criticize self 'Zhangsani said to mej that Lisik often criticized self*i/*j/k. Second, there is a number asymmetry with respect to the blocking effect. As Tang (1989) and Huang & Tang (1991) noted, a plural local subject does not
146
C.-T. James Huang and C.-S. Luther Liu
block a remote singular antecedent, though a singular local subject does block a remote plural antecedent, thus raising a question as to why this should be the case: (9) Lisii zhidao tamenj chang piping ziji i/j Lisi know they often criticize self 'Lisii knows that they often criticize himi/themselves.' (10) tamenj zhidao Lisii chang piping ziji i/*j they know Lisi often criticize self 'Theyi know that Lisi often criticizes himself/*themi.' Third, a person asymmetry also exists between first/second and third person NPs with respect to their ability to induce blocking effects. As noted in Xu (1993) (cf. also Pan, 1997), it appears that although a local first/second-person NP may block a remote third-person NP from being a LD antecedent, a local third-person NP does not block a remote first/second-person NP from being a LD antecedent.2 (11) a. Zhangsani danxin wo/nij hui piping ziji *i/j . Zhangsan worry I/you will criticize self 'Zhangsan is worried that I/you might criticize myself/yourself/*him.' b. woi danxin Zhangsanj hui piping ziji ? i / j I worry Zhangsan will criticize self 'I am worried that Zhangsan will criticize ?me/himself.' c. nii danxin Zhangsanj hui piping ziji ?i/j ma? you worry Zhangsan will criticize self Q 'Are you worried that Zhangsan will criticize ?you/himself?' In fact, we found that under some circumstances, even a third-person NP may induce blocking. One such circumstance is when the local third subject is presented deictically, as in (12), where the pointing finger indicates that the speaker points to someone in the audience as he utters the sentence. (12) Zhangsan shuo ta qipian-le ziji. Zhangsan say he/she cheat-Perf self 'Zhangsan said that he/she cheated himself/herself.' Another situation where a third-person NP may induce blocking is when multiple occurrences of ziji are involved. The relevance of the following example was first pointed out by Pan (1997), who attributes it to C. L. Baker. The available readings are summarized in (13), where ZS = Zhangsan, LS = Lisi, and WW = Wangwu:
Logophoricity, Attitudes, and Ziji at the Interface
147
(13) [ZS renwei [LS zhidao [WW ba zifi 1 de shu song-gei le ZS think LS know WW BA self DE book gave-to Perf ziji2 de pengyou]]]. self DE friend. a. ziji1 = ziji2 = WW b. ziji1 = ziji2 = LS c. zijij1 = ziji2 = ZS d. ziji1 = WW, ziji2 - LS e. ziji1 = WW, ziji2 = ZS f. ziji1 = ZS,ziji2 = WW g. ziji1 = LS, ziji2 = WW h. *ziji1 = ZS, ziji2 = LS i. *ziji1 = LS, ziji2 = ZS In this sentence, there are two occurrences of ziji and three c-commanding subjects. As indicated previously, the two occurrences of ziji may refer to the same antecedent, in which case any of the c-commanding subjects can be the antecedent (a, b, c). The two occurrences may also refer separately, as long as one of these occurrences is locally bound by Wangwu (d, e, f, g). Crucially, if both occurrences of ziji are to be LD bound, then they must be bound by the same LD antecedent (as in (b, c)), but not separately bound, as in (h, i). This range of possibilities indicates that a third-person NP does not induce blocking when it is itself a nonbinder or local binder of ziji, but does so when it is itself an LD binder of ziji. In the illicit cases (h, i), the intermediate subject Lisi is the LD binder of one occurrence of ziji, and it prevents the other ziji from being bound by the matrix subject Zhangsan. All of these complications are unexpected under the formal accounts discussed here. In fact, they call into serious question the very existence of a generalization concerning blocking effects as stated in connection with (3) above, and also to all versions of the formal account that were designed to derive this putative generalization. In addition to the blocking effects, certain objections have also been raised concerning other properties that have been claimed to hold, though the force of the objections is sometimes unclear.
2.2. The Functional Analyses Although the formal approach came into vogue after the problems were introduced by Y.-H. Huang (1984), the first account proposed in Huang et al (1984) was, in effect, a functional one. Essentially, the proposal was that the Chinese LD reflexives are not true anaphors in the sense of Binding Theory, but a special kind of anaphoric pronoun referring to the matrix sub-
148
C.-T. James Huang and C.-S. Luther Liu
ject as the "speaker" in an underlying representation in which the embedded clause originates as a direct quote, following Kuno's (1972) "direct discourse complementation" analysis of certain pronouns in English. According to Kuno, under one coreferential reading, the sentence (14a) is directly derived from (14b) as its underlying structure: (14) a. John said that he saw Bill, b. John said, "I saw Bill." In the terms of earlier generative studies, the claim is that the transformational process that forms an indirect complement structure from its direct discourse underlying source converts the first-person pronoun T in (14b) directly into the third-person pronoun 'he' in (14a), without going through the intermediate step (15a) or (15b): (15) a. John said that John saw Bill, b. John said, "John saw Bill." In other words, the pronoun 'he' in (14a) is not a result of "pronominalizing" a full NP under identity with another NP somewhere else in a sentence, but has as its direct source the first-person pronoun T in the direct-discourse representation of the complement clause. The pronoun refers to the matrix subject, the "speaker" of the embedded clause, identified by the first person pronoun in the underlying direct-discourse complement source. The referent of the matrix subject may be the actual speaker of the direct discourse complement as in (14b), or a "virtual speaker" (e.g., thinker, feeler, fearer, knower, experiencer) in situations like (16) below: (16) a. John was afraid that he might lose her. b. John feared in his mind: "I might lose her." Huang et al. (1984) suggested, following Kuno (1972), that the reflexive ziji in its LD use was permitted when it corresponds to T in the direct discourse representation of a sentence in which it occurs. Thus, in (17): (17) a. Zhangsan manyuan Lisi chang piping ziji. Zhangsan complain Lisi often criticize self 'Zhangsan complained that Lisi often criticized him.' b. Zhangsan manyuan, "Lisi chang piping wo." Zhangsan complained, "Lisi often criticized me." the embedded object reflexive is not the result of reflexivization on identity with its own matrix subject, but the result of converting from the speaker-referring wo 'me' in the underlying direct discourse. The idea that the LD reflexive ziji is related to wo in a direct discourse representation source would go a long way toward an explanation of some
Logophoricity, Attitudes, and Ziji at the Interface
149
of the basic properties including subject-orientation and observed blocking effects. For example, in the typical cases represented by (17b), the first person pronoun in the direct discourse typically refers to the matrix subject, the "source" of the discourse. The blocking effects, where they seem to hold, also seem to lend themselves to explanations in perceptual terms, i.e., conflicting perspectives, once the sentences are given a direct discourse analysis. Speaking of cases like (17a) where the long-distance construal would be blocked if the embedded subject Lisi were replaced by wo 'I', Huang et al (1984,208) write:3 (18) The above analysis enables us to explain why the appearance of 'I/me' in the sentence would block the LD-binding of ziji by the matrix subject. The reason is that if ziji is long-distance bound by the matrix subject, then it would be the first person pronoun 'I/me' in underlying structure. When the sentence is reported by a third party, another appearance of 'I/me' [i.e., in place of Lisi in (17a)] would refer to the reporter [i.e., the external speaker], but not the matrix subject [i.e., the 'internal speaker']. Thus two instances of 'I' occurring in the same clause would be used to refer to two separate individuals [i.e., the speaker of the entire sentence, and the 'speaker' of the embedded discourse]. Under such a situation the hearer is apt to be confused, and communication cannot be effective.... However, the idea was not developed further in these earlier studies of ziji. Part of the reason was the lack of an explicit and fully developed model that convincingly established the need to recognize a functional account in addition to a formal syntactic account of the behavior of reflexives, and partly due to the ready availability of a formal syntactic model embodying head-movement and LF interpretation, etc., which was producing attractive, though by no means conclusive, results. Since it is clear that not all uses of the reflexive fall under the direct discourse perspective, this approach entails the "ambiguity thesis," i.e., that in English and Chinese (and many other languages) the reflexive is an anaphor in some uses but something else in others. By Occam's Razor, a theory that treats all uses of the reflexive in a uniform way should be preferred over one that advocates the ambiguity thesis, in the absence of strong evidence for the latter. At the time, evidence for recognizing two reflexives ziji did not seem strong enough. The direct discourse perspective received crucial support, however, from a paper by Clements (1975), who showed that a separate series of "logophoric" pronouns exist in Ewe for the sole purpose of referring to an antecedent "whose speech, thoughts, feelings or general state of consciousness are reported" (p. 141). Such logophoric pronouns generally correspond to those cases of a reflexive or pronoun in English that, in Kuno's (1972) terms, de-
150
C.-T. James Huang and C.-S. Luther Liu
rives directly from I, me, my, etc., in the underlying direct discourse representation. The independent existence of such logophoric pronouns in natural language opened the door for the ambiguity thesis and made it seem a natural hypothesis for languages in which no such distinct pronouns exist. In the meantime, it has become increasingly clear that many uses of the reflexive simply cannot be accommodated as an anaphor falling under any version of BCA. Many authors, including Kuno (1987), Sells (1987), lida and Sells (1988), and Zribi-Hertz (1989), have since provided further evidence for logophoricity and substantial accounts of conditions on its use. Regarding the Chinese reflexive ziji, Yu (1992, 1996) gave important evidence for its logophoric use, some of which was also reproduced in Xu (1993,1994) and Chen (1992). Another recent account of the Chinese reflexive is provided in Pan (1997), who, following suggestions of Manfred Krifka, proposed to analyze the LD ziji as an element expressing beliefs "dese"in the sense of Lewis (1979). As we shall see, the treatment of LD ziji as a de se anaphor is essentially a restatement of Kuno's original insights in interpretive terms—without postulating direct discourse underlying structures and transformational mechanisms for forming indirect discourse. We shall return to Pan's analysis. It seems clear that research on logophoricity in the past few years has established, beyond doubt, its firm place in any adequate description of the reflexive. A natural question that arises is whether all uses of the reflexive can be adequately accounted for within a functional account of logophoricity without reference to the formal syntactic account of anaphor binding. While a hypothesis with a positive answer to this question may be a priori desirable, there is also abundant evidence that this cannot be correct. For one thing, if Clements' (1975) study crucially established the role of logophoricity, then it certainly also reaffirmed the role of syntactic anaphoricity since, after all, the crucial evidence comes from the existence, in languages like Ewe, of a series of logophoric pronouns in addition to a separate series of nonlogophoric anaphoric forms. In spite of certain attempts to account for all occurrences of the reflexive in functional terms (e.g., Chen 1992), it seems clear that the most promising account is one that treats some instances of reflexive binding as instances of syntactic anaphor and others as instances of logophoricity. The two most prominent recent works advocating the "mixed approach" are Reinhart and Reuland (1993) and Pollard and Sag (1992). An adaptation of the latter for an analysis of Mandarin ziji is found in Xue et al. (1994).4 We turn now to such analyses briefly. 2.3. Nonuniform Approaches Reinhart and Reuland (1993) propose a radical conceptual departure from the standard notion of anaphor binding. In contrast to familiar concep-
Logophoricity, Attitudes, and Ziji at the Interface
151
tions treating reflexivization as a property of reflexive pronouns, they argue that it is a property of predicates. Central to their account are the definition of a "syntactic predicate" and the notions of being "reflexive-marked" and being "reflexive." (19) a. The syntactic predicate formed of (a head) P is P, all its syntactic arguments, and an external argument of P (subject). (The syntactic arguments of P are the projections assigned theta-role or Case by P.) b. A syntactic predicate (formed of P) is reflexive-marked iff either P is lexically reflexive or one of P's arguments is a self anaphor. c. A predicate is reflexive iff two of its arguments are coindexed. By (19a), all argument positions receiving theta-role or Case from a given head P belong together in the same "syntactic predicate."5 By (19b), such a syntactic predicate is termed "reflexive-marked" if it involves an inherently reflexive verb like behave or one of its arguments is himself (or ziji). By (19c), this syntactic predicate is "reflexive" if two of its arguments are coindexed. The relevant condition on anaphor binding they propose is the following "Condition A":6 (20) Condition A (R&R) A reflexive-marked syntactic predicate is reflexive. According to (20), whenever a sentence contains an argument reflexive (hence, "reflexive-marked"), that reflexive must be coindexed with another argument of the same predicate (thereby making the predicate "reflexive"). This accounts for the following basic patterns: (21) a. Maryi said that Lindaj criticized herselfi/j b. Maryi believed herselfi to be honest.7 The condition (20) says nothing about adjunct reflexives or those in sentences like the following. (22) a. Maryi said that John loved no one but herselfi. b. The queeni invited both Max and herselfi to our party. Reinhart and Reuland (1993) assume that these instances of the reflexive pronoun represent their logophoric uses falling outside the domain of syntactic binding. The following cases are also attributed to logophoricity: (23) a. Theyi heard several rumors about themselvesi. b. Theyi thought that rumors about themselveSi had spread too fast. c. Theyi thought that for rumors about themselvesi to spread would be bad.
152
C.-T. James Huang and C.-S. Luther Liu
d. Theyi thought that it would be bad for rumors about themselvesi to spread. e. Theyi arrived after rumors about themselvesi had already spread. According to Reinhart and Reuland, then, syntactic binding of an anaphor is closely tied to the notion of argument structure. Argument structure (or theta and possibly Case theory) provides the dividing line between syntactic anaphoricity and discourse-pragmatic logophoricity. This view has also been independently proposed by Williams (1989,1994), Pollard and Sag (1992), and Xue et al. (1994), though these authors also differ in the specifics of their proposals. Analyzing Mandarin ziji sentences in the framework of Pollard and Sag, for example, Xue et al. (1994) argue that ziji can be syntactically bound by an argument by being its co-argument (as is the case with Reinhart and Reuland), or by being contained within a constituent that is the antecedent's co-argument. A syntactic reflexive (or what they call a "Z pronoun") is subject to the following principle: (24) Principle Z Z-pronouns must be o-bound by animate subjects. where a o-binds b iff a and P are coindexed and a is a less oblique co-argument of (i) b or (ii) some constituent g properly containing p. X is a less oblique (or more prominent) co-argument of Y iff X and Y are co-arguments and X precedes Y on the following "Obliqueness Hierarchy": Subject < Primary Object < Secondary Object < Other Complement. In other words, a syntactic reflexive ziji is o-bound by an animate subject iff the coindexed animate subject is (i) a co-argument of ziji in nonsubject position or (ii) a co-argument of some nonsubject properly containing ziji- In case (i), ziji is locally o-bound; in case (ii), ziji is long-distance o-bound. The systems of Reinhart and Reuland and of Xue et al. are very different conceptually: for Reinhart and Reuland reflexivity is a property of predicates, but for Xue et al. it is a configurational relation between two NPs. Empirically, however, they are similar in that they consider only arguments of a predicate to be relevant in defining the domain of reflexivity. The major difference between Xue et al. and Reinhart and Reuland is that Xue et al. allow a syntactic reflexive to be genuinely long-distance bound (as long as it is contained in some co-argument of its antecedent). Thus, a syntactic reflexive may be locally bound as in (25), or long-distance bound as in (26)-(27). (25) Zhangsani piping-le ziji i Zhangsan criticize-Perf self 'Zhangsan criticized himself.'
Logophoricity, Attitudes, and Ziji at the Interface
153
(26) Zhangsani shuo [List piping-le zijii]. Zhangsan say Lisi criticize-Perf self 'Zhangsarij said Lisi criticized himi.' (27) Zhangsani bu xihuan [nage [ei piping zijii/j] de ren j . Zhangsan not like that criticize self DE person 'Zhangsani; does not like the person who criticized himi himself.' In (25) the local ziji is a co-argument of its antecedent Zhangsan. In (26)-(27) the long-distance bound ziji is contained in a constituent (a complement clause in (26) and a relativized NP in (27)) which is a co-argument of Zhangsan. On the other hand, in the following sentences, ziji is not o-bound and hence, cannot be treated as a syntactic reflexive: (28) Zhangsani de jiaoao hai-le ziji i Zhangsan DE arrogance harm-Perf self 'Zhangsan's arrogance harmed him.' (29) yinwei Lisi piping-le ziji i Zhangsani gandao hen shangxin. because Lisi criticize-Perf self Zhangsan feel very sad 'Because Lisi criticized hinij, Zhangsanj felt sad.' In (28) the antecedent Zhangsan sub-commands ziji but is not a co-argument of ziji- In (29), ziji is contained in an adjunct of the matrix clause, and neither ziji nor the adjunct clause is a co-argument of Zhangsan. For Xue et al, the reflexives in these cases are treated as instances of the discourse reflexive. Note that once (29) is further embedded in a complement, as in (30), ziji may be bound by the matrix subject, as a syntactic reflexive: (30) Zhangsani shuo [yinwei Lisi piping-le ziji i ta gandao hen Zhangsan say because Lisi criticize-Perf self he feel very shangxin]. sad 'Zhangsan said that because Lisi criticized him, he felt sad.' Note that because Xue et al. allow for a broader notion of a syntactic reflexive than Reinhart and Reuland, some of the cases attributed to logophoricity by Reinhart and Reuland are admitted under Pollard and Sag's syntactic account (e.g., all of the sentences in (22) and (23a-d)). All cases of the discourse-pragmatic reflexive under Pollard and Sag's account (e.g. (23e), (28)-(29)) are necessarily attributed to logophoricity by Reinhart and Reuland. In this section, we have reviewed previous formal, functional, and nonunitary approaches to the reflexive. We indicated that the nonunitary approach holds most promise to a successful account. In Section 3, we pro-
154
C.-T. James Huang and C.-S. Luther Liu
vide crucial evidence for ziji as a logophor, thus falsifying any version of a pure formal approach that does not take logophoricity into account. In Section 4, we provide crucial evidence for ziji as a syntactic anaphor and address the question of the "dividing line" between ziji as an anaphor and ziji as a logophor.
3. ZIJI AS A LOGOPHOR 3.1. Logophoricity, Direct Discourse, and Attitudes de se As indicated in Section 2.2, Kuno (1972) was the first to demonstrate the importance of what has now come to be known as logophoricity for a full understanding of natural language (co)-reference (see also Kuno, 1987, Chapter 3). He argued that many sentences with an embedded pronoun (or reflexive) bound by a matrix constituent should be analyzed as deriving from direct-discourse complements in which the pronoun originates as the first (or second) person pronoun. Such a direct-discourse representation is obligatory when sentences with quotative or attitudinal verbs and their complements are understood as describing the thoughts of the matrix subjects. Thus, under such a reading, the "deep structure" of (31a) would be represented as in (31b), but not as in (31c): (31) a. AH claimed that he was the best boxer in the world. b. Ali claimed, "I am the best boxer in the world." c. Ali claimed, "Ali is the best boxer in the world." The sentence (3la) is obtained by a rule of the form / he/she as part of the process of indirect-discourse formation. Among other things, this analysis explains a contrast that can be observed with pairs of sentences like the following: (32) a. That hei was the best boxer in the world was claimed by Alii. b. ??That Alii was the best boxer in the world was claimed by himj. Both these sentences are well formed with coreference of Ali and the pronoun he/him if they are used to report the speaker's own knowledge of the relevant state of affairs. A contrast is observed as indicated, however, if they are intended as reports (by the speaker) on the inner or direct feelings of Ali, the underlying matrix subject. Under Kuno's theory, this contrast follows from the direct discourse analysis indicated in (31a-b), since the pronoun he in (32a) is directly derived by the I he rule but there is no similar rule that derives the name Ali in (32b).8
Logophoricity, Attitudes, and Ziji at the Interface
155
While in English the same pronoun is used in reports of both "the speaker's knowledge" and "the subject's direct feeling," Hagege (1974) and Clements (1975) report on some West African languages where the distinction is grammaticalized. In these languages, a distinct set of logophoric pronouns exists for the sole purpose of referring to an antecedent "whose speech, thoughts, feelings or general state of consciousness are reported." Another case where this distinction is grammaticalized is provided by the Italian possessive reflexive proprio, in contrast to the pronoun suo, as illustrated by the following pair (from Chierchia (1989,24)): (33) a. Pavarotti crede che i propri pantaloni siano in fiamme. 'Pavarotti believes that self's pants are on fire.' b. Pavarotti crede che i suoi pantaloni siano in fiamme. 'Pavarotti believes that his pants are on fire.' Following Lewis (1979), Chierchia distinguishes two ways in which the English sentence Pavarotti believes that his pants are on fire (and its Italian counterpart (33b)) may be true, with the pronoun his bound by Pavarotti. Suppose that Pavarotti's pants are on fire and he just caught sight of the burning pants in the mirror. Pavarotti has the belief that a certain man's pants are on fire. The speaker knows that this man is, in fact, Pavarotti himself, hence, that his and Pavarotti refer to the same person. Pavarotti himself, however, may or may not realize that it is, in fact, his own pants that are on fire. Pavarotti is said to have a de re belief. In the subcase of the de re reading where Pavarotti is himself aware that it is his own pants that are on fire, we say that Pavarotti has a de se belief. In the de se situation, Pavarotti is disposed to say, "My pants are on fire!" (and run for the fire extinguisher). In the (non-de se) de re situation, he may be disposed to say, "His pants are on fire!" (and continue to sing "O sole mio"). Clearly, the distinction between the de re and the de se beliefs corresponds to Kuno's distinction between a report of "speaker's knowledge" and that of the "matrix subject's direct feeling." (We shall return to the syntactic and semantic representation of de se and de re readings.) Chierchia's relevant point regarding (33a) and (33b) is that, whereas (33b) with the bound pronoun suo may be acceptable under the de re scenario, (33a) with the LD bound reflexive proprio is acceptable only under the de se scenario. In other words, proprio is a logophor, while suo is a normal anaphoric pronoun. While Kuno's direct-discourse representation captures an important, perhaps the core, case of logophoricity, the underlying representation that he proposed does not quite fit into any sophisticated current theory of syntax-semantics interface. In Section 5, we adopt Chierchia's (1989) semantics of attitudes de se, which not only captures Kuno's essential insights but also accommodates a wider range of phenomena that are now believed to fall
156
C.-T. James Huang and C.-S. Luther Liu
under logophoricity. Sells (1987) provided a useful taxonomy of such phenomena in terms of three primitive roles of the antecedent of the logophor. (34) a. Source: the one who is the intentional agent of the communication. b. Self: the one whose mental state or attitude the proposition describes. c. Pivot: the one with respect to whose (time-space) location the content of the proposition is evaluated. In other words, a logophor refers to a person whose (a) speech or thought, (b) attitude or state of consciousness, and/or (c) point of view, or perspective, is being reported. This person may be the speaker (the external Source, Self, or Pivot) or an internal protagonist denoted by an argument of the sentence (e.g., the matrix subject). These roles are illustrated by the following examples, with Lisi being the internal protagonist under consideration. (35) a. Lisii shuo [Zhangsan chang piping Zijii. Lisi say Zhangsan often criticize self 'Lisii says that Zhangsan often criticizes himi.' b. [Ziji i -de xiaohaimei de jiang]-de xiaoxi shi Lisii hen shangxin. Self's child not get prize DE news make Lisi very sad 'The news that hisi child didn't win the prize made Lisii very sad.' c. ?[Zhangsan lai kan ziji i -de shihou, Lisii zheng zai kan shu. Zhangsan come see self DE moment Lisi now at read book 'Lisii was reading when Zhangsan came to visit himi.' In (35a), Lisi may be understood as the source antecedent of ziji. In (35b), Lisi is the internal Self whose mental state is being reported. And (35c) may be understood as reporting an event from the perspective of Lisi, the Pivot. Sells (1987) notes that there is an implicational relation among these three roles of a logophoric antecedent. Thus, if a sentence is interpreted as reporting on the speech or thought of an internal Source antecedent, the same antecedent must also be a Self whose mental state is described and a Pivot from whose perspective the report is made or perceived. Similarly, if a sentence simply reports on the mental state or consciousness of its antecedent (Self), it must also be the case that the sentence is evaluated from the viewpoint of the antecedent (Pivot). The reverse does not hold, however. Thus, in (35c) the speaker is simply empathizing with Lisi but does not purport to be reporting on his mental state (as in (35b)) or his speech or belief (as in (35a)). Sells suggests that these roles characterize certain cross-linguistic variations and shows that languages differ as to whether they permit one, two, or all three kinds of logophoric antecedents. We submit that, rather than primitives, these three labels express a progressive degree of liberation in the linguistic expression of logophoricity, Source being the "core," Self
Logophoricity, Attitudes, and Ziji at the Interface
157
being the "extended," and Pivot having yet further extended uses. Thus, some languages may permit logophoric reference to the Source only, others allow either Source or Self, and still others allow all three roles; but we do not expect to find languages allowing Pivot as a logophoric antecedent but specifically excluding NPs that denote the Source. As indicated previously, Huang et al. (1984) made an early attempt to analyze certain cases of Mandarin LD ziji using the direct-discourse representation of Kuno (1972). The cases that were naturally analyzable in this way are typically the "core" cases, i.e., Source in Sells' system. In spite of the fact that this attempt was abandoned for several years, we think now that there is abundant reason to go back to this position. In the rest of this section, we point out a number of cases from Mandarin where ziji cannot be fully accounted for by any reasonable version of the Binding Condition A, but must be treated as a logophor falling under one of Sells' categories. 3.2. Evidence for Ziji as a Logophor 3.2.1. SENTENCE-FREE ZIJI REFERS TO SPEAKER Yu (1992,1996) points out that ziji may be completely unbound syntactically. Furthermore, when it's sentence-free, it must refer to the speaker. Hence, in the following examples, ziji would be replaceable by a first-person pronoun 'I, me' or reflexive 'myself. (36) Zhe-ge xiangfa, chule ziji, zhiyou san-ge ren zancheng. This-CL idea, besides self only three-CL people agree 'As for this idea, besides myself, only three other people agree.' (37) Zhe-pian wenzhang shi Ann he ziji he-xie de, (qing duoduo This-CL article be Ann and self co-author DE, (please many zhijiao). advise) 'This article was co-authored by Ann and myself; please give me your advice.' It is obvious that these sentences would be a problem for any account that treats all instances of ziji as syntactic anaphors. Under the logophoric account, however, these sentences are nicely explained as each involving a logophor bound, by default, to the speaker (the external Source). These sentences are reminiscent of phrases like as for myself in English (Kuno, 1987; Ross, 1970): (38) a. As for myself, coffee will be fine. b. ??As for yourself, coffee will be fine. c. * As for himself, coffee will be fine.
158
C.-T. James Huang and C.-S. Luther Liu
3.2.2. BINDING BY INTERNAL SOURCE OR SELF There is evidence that LD binding of ziji as in (35a) is most readily acceptable only under the de se interpretation, as is expected from Huang et al.'s (1984) analysis. This point is not easily appreciated from (35a) itself, but now with Chierchia's demonstration in (33), a clearer point can be made. Suppose that Zhangsan sees a pickpocket running away with his purse without realizing it's his own purse; he may kindly report the theft to the police. The speaker, who knows that the purse actually belongs to Zhangsan, can report on Zhangsan's deed as follows: (39)
Zhangsan shuo [pashou tou-le ta-de pibao] Zhangsan shuo [pickpocket steal-Perf his purse 'Zhangsan said that the pickpocket stole his purse.'
That is, the sentence under the coreference interpretation can be true as expressing a (non-de se) de re belief of Zhangsan that his purse was stolen. In this same scenario, the reflexive ziji does not fit in place of the pronominal ta-de 'his.' The following sentence is well formed only under the de se interpretation: (40) Zhangsan shuo [pashou tou-le ziji-de pibao] Zhangsan shuo [pickpocket steal-Perf self's purse 'Zhangsan said that the pickpocket stole his purse.' The LD ziji, then, is like Italian proprio in being limited to logophoric uses, here, a case of logophoric ziji referring to the internal Source. The following examples provide evidence for ziji referring to the internal Self: (41) a. Zhangsan kuajiang ziji i xia-le Lisii yi tiao. Zhangsan praise self scare-Perf Lisi one jump 'That Zhangsan praised himl greatly surprised Lisii.' b. *Zhangsan kuajiang Lisii xia-le ziji i yi tiao. Zhangsan praise Lisi scare-Perf self one jump 'That Zhangsan praised Lisii greatly surprised himi.' These sentences show that backward reflexivization is preferred over forward reflexivization, a pattern somewhat reminiscent of Kuno's (32), though forward pronominalization (as shown below) is fine: (42)
Zhangsan kuajiang Lisi xia-le ta yi tiao. Zhangsan praise Lisi scare-Perf he one jump 'That Zhangsan praised Lisi greatly surprised him.'
The contrast in (41) may be derived along Kuno's (32) if the sentential subject is given a direct discourse representing the experiencer Lisi's thought,
Logophoricity, Attitudes, and Ziji at the Interface
159
i.e., the internal Self. Under such an analysis, the ziji in (41a) takes the internal Self, Lisi, as its antecedent. In (41b), Lisi is not the internal Self and so does not qualify as the antecedent. On the other hand, (42) is acceptable because, unlike LD ziji, use of a normal pronoun ta is allowed under normal de re conditions. 3.2.3. CONSCIOUSNESS In the preceding cases involving a Source or Self antecedent, it is a necessary property of logophoric ziji that its antecedent denotes an individual conscious of the relevant event being reported. This is so as a matter of definition, for a de se reading is one in which the antecedent is disposed to refer to the logophor by the first-person pronoun. Now we shall show that LD ziji does exhibit consciousness effects. In the following examples, LD binding of ziji is possible in a scenario in which the remote antecedent can be assumed to denote someone conscious of the event involving him/her, but is difficult to obtain if the antecedent denotes someone who cannot possibly have conscious knowledge of the event. In (43a), Zhangsan could be aware that someone was criticizing him; in (43b), Zhangsan is assumed not to be aware at all of the event, or even the plot, of his killing. (43) a. Zhangsani kuajiang-le [[changchang piping zijii de] naxie Zhangsan praise-Perf often criticize self DE those ren j . persons 'Zhangsani praised those people who criticize himi a lot.' b. HZhangsani kuajiang-le [[houlai sha si ziji i de] naxie ren j . Zhangsan praise-Perf later kill die self DE those persons 'Zhangsanj praised those persons who later killed himi.' Similarly, in the following scenario depicting the historical hero Xue Rengui who, after leaving his pregnant wife to serve in the troops for years and not knowing his son—now grown up and a brave soldier—was praising his son. Under this scenario, the sentence (44) does not allow an LD construal of ziji: (44) ?? Xue Renguii zai yudao Wang Baochuan zhi-qian, yizhi shuo Xue Rengui at meet Wang Baochuan before always say [buxia renwei [ziji i -de erzi, Xue Dingshan, zui yonggan]]. staff think self s son Xue Dingshan most brave 'Before he saw his wife Wang Baochuan, Xue Renguij always said that his staff thought that hisi own son, Xue Dingshan, was the bravest.'
160
C.-T. James Huang and C.-S. Luther Liu
These contrasts do not follow from any pure syntactic account, but illustrate the relevance of Self, or consciousness, as a property of logophoricity. An effect of the consciousness requirement can also be observed through the following contrast: (45)
(46)
Yinwei Lisi piping ziji i suoyi Zhangsani hen shengqi. Because Lisi criticize self so Zhangsan very angry 'Because Lisi criticized him, Zhangsan was very angry.' ?? (Dang) Lisi piping ziji i de shihou, Zhangsani zheng zai kan (at) Lisi criticize self DE moment Zhangsan right at read shu. book 'At the moment Lisi was criticizing him, Zhangsan was reading.'
The contrast seems to stem mainly from the fact that a causal relation between the two events described obtains in (45) but not in (46). This contrast seems to be reducible to the effect of consciousness, a property of the Self as a logophoric antecedent..9 Thus, in (45), the explicit causal relation strongly implicates Zhangsan's awareness of Lisi's criticism of him, and the use of LD ziji is licit. In (46), however, which describes two concurrent events and implicates the nonexistence of a causal relation and possible nonawareness on the part of Zhangsan, it is considerably less natural to use ziji to refer to Zhangsan. (Replacement of the reflexive with the pronoun ta makes (46) completely natural, as it does (45).) Note that, for some speakers, the status of (46) ranges from acceptable to somewhat marginal. It seems that this is possible when the speaker, rather than maintaining a neutral perspective, empathizes with the internal protagonist Zhangsan, i.e., taking Zhangsan as the Pivot antecedent of ziji. If the speaker retains a completely neutral position, (46) is unacceptable under the relevant interpretation. Thus, it seems that Mandarin logophoric ziji may take a Source or Self readily, but a pure Pivot only in environments in which the pivothood is clear, as its antecedent.10 Note, incidentally, that the coindexing in (46) becomes completely licit when (46) is embedded under 'Zhangsan said': (47) Zhangsani shuo (dang) Lisi piping ziji i de shihou, tai zheng zai Zhangsan say (at) Lisi criticize self DE moment he right at kan shu. read book 'Zhangsanj said that at the moment Lisi was criticizing himi, hei was reading.' This follows because ziji now refers to the internal Source Zhangsan, thus fully meeting the consciousness requirement.
Logophoricity, Attitudes, and Ziji at the Interface
161
3.2.4. BLOCKING EFFECTS EXPLAINED In Section 2.1., we noted that the so-called "blocking effects" are more complicated than first met the eyes of those who discovered them and attempted to account for them in pure syntactic terms. In particular: (48) a. A person asymmetry exists such that a first/second-person pronoun may block a third-person LD antecedent, but not the other way round. b. LD ziji may be blocked by nonsubjects which are not potential antecedents. c. A deictically identified third-person NP does induce blocking. d. In cases with multiple occurrences of ziji, some third-person NPs may induce blocking effects. e. Plural NPs behave differently than do singular NPs, both as potential antecedents and as blockers. We now show that all of these complications may be naturally explained, based primarily on the suggestion made in Huang et al. (1984) but drawing in part on similar suggestions made by others (Chen, 1992; Pan, 1997). Recall that the suggestion made in Huang et al. (1984)—as quoted previously in (18)—was that blocking effects are the effects of a perceptual strategy, i.e., to avoid perspective conflicts when the relevant sentences are put in the context of a direct speech act. Most of these effects can be explained by taking literally Kuno's direct discourse representation hypothesis. Take (49a), for example, which, as we saw, does not permit LD binding of ziji. Assuming that the LD ziji is underlyingly wo 'I/me', then (49a) under the long construal would have the representation (49b): (49) a. Zhangsan juede wo zai piping ziji. Zhangsan think I at criticize self 'Zhangsan thinks that I am criticizing self.' b. Zhangsan juede, "wo zai piping wo." Zhangsan think I at criticize me Zhangsan thinks, "I am criticizing me." In (49b), there are two occurrences of wo. Under the intended reading, the first wo refers to the (external) speaker of the entire sentence, and the second wo refers to Zhangsan, the (internal) "speaker" of the direct discourse complement. Because of the perspective conflict it would cause, (49b) is not acceptable under the intended reading. This, in turn, explains why the intended LD binding of ziji is impossible to get. (The local binding of ziji is, of course, licit, ziji simply being an anaphor obeying BCA.) The blocking ef-
162
C.-T. James Huang and C.-S. Luther Liu
feet induced by a second-person subject as in (50a) is similarly explained, given the representation (50b): (50) a. Zhangsan juede ni zai piping ziji. Zhangsan think you at criticize self 'Zhangsan thinks that you are criticizing self.' b. Zhangsan juede, "ni zai piping wo." Zhangsan think you at criticize me Zhangsan thinks, "You are criticizing me." In (50b), ni refers to the addressee with respect to the external speaker, but wo refers to the internal speaker Zhangsan. Again, we have a conflict between the internal Source (to whom wo is "anchored") and the external Source (to whom ni is anchored). On the other hand, unlike surface first- or second-person pronouns which are obligatorily anchored to the external speaker, a third-person NP is not obligatorily anchored to the external speaker and can always be anchored to the internal speaker, the matrix subject. Thus, third-person NPs do not induce blocking. (51) Zhangsan juede List zai piping ziji. Zhangsan think Lisi at criticize self 'Zhangsan thinks Lisi is criticizing him.' This is the case even if the matrix is first or second person: (52) wo juede Lisi zai piping ziji. I think Lisi at criticize self 'I think that Lisi is criticizing me.' (53) ni juede Lisi zai piping ziji. You think Lisi at criticize self 'You think that Lisi is criticizing you.' In all three cases (51)-(53), the direct discourse complement is 'Lisi zai piping wo' with nothing obligatorily anchored to the external Source. This allows wo to refer to the matrix subject, regardless of the latter's person feature. Thus we explain not only why blocking effects should occur in cases like (49)-(50), but also why they do not occur in cases (52)-(53), i.e., the "person asymmetry" described in (48a).11 It also follows that the "blocker" does not have to be a subject (i.e., the property (48b)), as illustrated (= (8a)): (54) Zhangsani gaosu wo Lisik hen ziji i / k . Zhangsan tell me Lisi hate self 'Zhangsan, told me that Lisi hated *himi/himself.'
Logophoricity, Attitudes, and Ziji at the Interface
163
The deep structure source of (54) would be 'Zhangsan gaosu wo, "Lisi hen wo",' again with two occurrences of wo that are anchored to different speakers.12 This account also explains why even third-person NPs may induce blocking when they are introduced deictically (i.e., property (48c)), as shown in (12), repeated here: (12) Zhangsan shuo ta qipian-le ziji. Zhangsan say he/she cheat-Perf self 'Zhangsan said that he/she cheated himself/herself.' A deictic NP involving the speaker's pointing finger is, by definition, anchored to the external speaker. Thus, a deictically identified third-person is not unlike the first- or second-person pronoun in this respect. LD binding of ziji by the internal speaker Zhangsan is therefore ruled out on a par with (49)-(50). As for property (48d), the striking pattern of blocking displayed by sentences with multiple occurrences of ziji also receives a straightforward explanation. As noted previously, (13) allows the construals (a) through (g), but disallows (h) and (i). The generalization is that an LD binder of ziji blocks a higher LD binder, but a local binder does not. (13) [ZS renwei [LS zhidao [WW ba zijii de shu song-gei le ziji2 ZS think LS know WW BA self DE book give-to Perf self de pengyou]]]. DE friend. a. ziji1 = ziji2 = WW b. ziji1 = ziji2 = LS c. ziji1 = ziji2 = ZS d. ziji1 - WW, ziji2 = LS e. ziji1 = WW, ziji2 = ZS f. ziji1 = ZS, ziji2 = WW g. ziji1 = LS, ziji2 = WW h. *ziji1 = ZS, ziji2 = LS i. *ziji1 = LS, ziji2 = ZS We assume that only LD ziji's are logophors originating as first-person pronouns in direct discourse, but local ziji's are just anaphors subject to BCA. Thus, when—and only when—there are two LD binders involved, the direct discourse representation will involve one direct quote within another. This is the case with the construals (h) and (i). The direct discourse representation for (13h-i) has the following form:
164
C.-T. James Huang and C.-S. Luther Liu
(55) ZS renwei, "LS zhidao, 'WW ba wo-de shu song-gei le wo-de pengyou.'" Zhangsan thinks, "Lisi knows, 'Wangwu gave my book to my friend.'" with two occurrences of wo, one of which is bound by Lisi, the inner "speaker" and the other bound by Zhangsan, the outer speaker. This situation is precisely the same as cases of first/second-person blocking like (49) and (50), with the direct discourse complement "wo/ni. .. wo ...." In this latter cases, the first wo (or m) is anchored to the external speaker, and the second wo to the internal speaker (the matrix subject). The unavailability of (13h) and (13i) is simply another example of blocking arising from the perceptual strategy of avoiding conflicting reference to speech act participants. Note that each of (b)-(g) involves, at most, one LD binder, and hence, only one underlying direct discourse complement. No similar conflict needs to occur here, and all these readings are readily available. Under the current hypothesis, the readings (a-g) are analyzed as follows (where 'his own' is used to translate the locally bound ziji-de): (56) a. ZS thinks that LS knows that WW gave his own book to his own friend. b. ZS thinks that LS knows, "WW gave my book to my friend." c. ZS thinks, "LS knows that WW gave my book to my friend." d. ZS thinks that LS knows, "WW gave his own book to my friend." e. ZS thinks, "LS knows that WW gave his own book to my friend." f. ZS thinks that LS knows, "WW gave my book to his own friend." g. ZS thinks, "LS knows that WW gave my book to his own friend." Finally, the peculiar asymmetry between plural and singular NPs with respect to blocking (i.e., property (48e)) can also be explained. Recall that although a plural NP does not block a singular LD antecedent, a singular NP does seem to block a plural antecedent: (57) Lisii shuo tamenj chang piping ziji i/j Lisi say they often criticize self 'Lisij said that they often criticized him/themselves.' (58) tameni shuo Lisij chang piping ziji*i/j. they say Lisi often criticize self 'Theyi said that Lisi often criticized himself/*themi.' In fact, a plural LD antecedent may also be blocked by another plural NP. The following sentence, with i j, permits only local binding:
Logophoricity, Attitudes, and Ziji at the Interface
165
(59) tameni shuo tameni chang piping ziji*i/j they say they often criticize self 'Theyi said that theyjoften criticized *themi/themselveSj.' On the other hand, if a plural NP is accompanied by the universal quantifier dou, then neither singular nor plural NPs will block LD binding. Compare (58)-(59) with the following: (60) tameni dou shuo Lisij chang piping ziji i/j they all say Lisi often criticize self 'Eachiof them said that Lisi often criticized himi/himself.' (61) tameni dou shuo tamenj chang piping zijii/j. they all say they often criticize self 'Eachiof them said that theyj often criticized himi/themselvesj.' The generalization is that a LD antecedent of ziji must be semantically singular, so when ziji takes a plural antecedent, the antecedent must be understood in the distributive sense. In fact, these sentences require a strict distributive interpretation of both the antecedent and the reflexive, so that each person must have the de se belief "Lisi/They criticized me" but not "Lisi/They criticized us." It has been observed that, unlike English plurals, Chinese plural NPs do not distribute without the aid of a distributor such as the universal quantifier dou (see S. Huang, 1996; Li, 1997; Lin, 1996).13 The impossibility of LD-binding in (58)-(59) arises because the matrix subjects cannot distribute on their own. Why should LD ziji disallow group-denoting antecedents? We submit that this arises because de se attitudes typically represent the inner feelings of individuals, and groups are not individuals with inner feelings (unless they are treated collectively as an individual). In Section 5, we shall represent de se readings by structures of operator-variable binding. The distributive reading comes from treating the reflexive as a variable ranging over singular entities.14 To summarize, we have seen that the cluster of blocking effects indicated in (48) receive a natural or plausible explanation in logophoric terms, though they have posed nontrivial problems for previous pure syntactic accounts. More generally, we have seen extensive evidence for treating Mandarin ziji as a logophor whose reference is not determined by Binding Condition A but is identified with the Source or Self (or, possibly, Pivot) of a reported event or proposition.15 There are many other instances of ziji binding, however, which cannot be treated in logophoric terms, but which seem to fall under BCA rather nicely. Such cases constitute evidence for also treating ziji as a syntactic anaphor. We turn to such cases now.
166
C.-T. James Huang and C.-S. Luther Liu
4. ZIJI AS A (SYNTACTIC) ANAPHOR 4.1. Local Binding One clear case of a syntactic anaphor is when ziji is bound by a co-argument, i.e., a subject of the same clause: (62) a. Zhangsan piping-le ziji. Zhangsan criticize-Perf self 'Zhangsan criticized himself.' b. Zhangsan gen ziji guo-bu-qu. Zhangsan with self pass-not-go 'Zhangsan gave himself a hard time.' c. Zhangsan ji-le yi-ben shu gei ziji. Zhangsan send-Perf one-CL book to self 'Zhangsan sent a book to himself.' Another case is when ziji is contained in an NP and is bound by a co-argument of the containing NP. Thus, instead of ziji as an object in (62), we have ziji de pengyou 'self's friend', etc., in (63): (63) a. Zhangsan piping-le ziji de pengyou. Zhangsan criticize-Perf self DE friend 'Zhangsan criticized his own friend.' b. Zhangsan gen ziji de didi guo-bu-qu. Zhangsan with self DE brother pass-not-go 'Zhangsan gave his own brother a hard time.' c. Zhangsan ji-le yi-ben shu gei ziji de erzi. Zhangsan send-Perf one-CL book to self DE son 'Zhangsan sent a book to his own son.' In these sentences, binding is possible even when no logophoric requirements hold. In each case, the local binder is not, or need not be, a Source, Self, or Pivot. For example, since these sentences are reports on an action performed by Zhangsan but not of his speech or thought, the notion Source is irrelevant. Second, these sentences do not require the speaker or hearer to take the empathy focus of Zhangsan, but can be uttered entirely from the speaker's own viewpoint. Third, consciousness, which we see as a common property of logophoricity, clearly also does not obtain. Thus, (62a) and (63a) are entirely licit even though Zhangsan may not be aware that the person he was criticizing was actually himself or his own friend. It is also easy to imagine a scenario in which the following holds true with Zhangsan, even at the time of his death, still not knowing whom he was victimized by:
Logophoricity, Attitudes, and Ziji at the Interface
167
(64) Zhangsani bei ziji i (de pengyou) hai-si le. Zhangsan by self DE friend wrong-death Perf 'Zhangsan was wronged to death by himself / his own friend.' That local binding does not exhibit consciousness effects is also evident from the contrast between the grammatical (65) and the ungrammatical (44) we saw above. (65) Xue Renguii zheng zai chengzan zijii-de erzi Xue Dingshan Xue Rengui right at praise self-DE son Xue Dingshan 'Xue Rengui is praising his own son Xue Dingshan right now.' In the event that is being reported at present, both the narrator and the reader—but not Xue Rengui himself in the historical past—are aware of the father-son relationship. Given what we know in the plot, in other words, Xue Rengui only had a de re belief about the virtues of his son, but not a de se belief. As we saw in (44), where ziji is the object of an embedded clause that excludes Xue Rengui, LD binding is hard to obtain as an expression of Xue's de se belief about his son. Finally, such locally bound ziji's also do not exhibit any blocking effects. In the following examples, the intervening first- and second-person pronouns do not induce blocking: (66) Zhangsani gaosu wo ziji i de fenshu. Zhangsan tell me self DE grade 'Zhangsan told me about his own grade.' (67) tai xiang ni tidao ziji i de quedian le ma? he to you mention self DE shortcoming Perf Q 'Did he mention his own shortcoming to you?' (68) ta zheng-tian dui-zhe wo chuipeng ziji. he whole-day to-DUR me boast self 'He boasted about himself in front of me all day long.' Summarizing, we have identified two cases where ziji is locally bound: when it is bound by a co-argument ((62), (64), and (68)) and when it is contained in an NP and bound by a co-argument of the containing NP ((63)(67)). We saw that these cases of binding do not require any de se beliefs on the part of the referent of the antecedent, nor do they exhibit any perspectivity, consciousness, or blocking effect. The same is true where a locally bound ziji is an adjunct or contained in an adjunct: (69) Zhangsan zai ziji jiali bei da-hun le. Zhangsan at self home by hit-unconscious Perf 'Zhangsan was knocked unconscious at his own home.'
168
C.-T. James Huang and C.-S. Luther Liu
(70) Zhangsan cong ziji de jiaodu pinggu zhe-jian shi de Zhangsan from self DE angle assess this-CL matter DE libi. pro-and-con 'Zhangsan assessed the pros and cons of this matter from his own point of view.' All of these cases therefore do not involve logophoricity, and it is our claim that in each case, the locally bound reflexive is an anaphor. 4.2. The Dividing Line: Governing Category What is the "dividing line" between the local reflexive which we take to be an anaphor and the LD reflexive which we take to be a logophor? It turns out that the traditional notion of a governing category (GC), as given in (2), defines roughly the correct environments for a syntactic anaphor to be bound. Thus, in (62)-(70), each reflexive is bound within the minimal category that contains it, its governor, and an accessible subject. The GC also seems to make a correct distinction between embedded subject and object reflexives: (71) a. Zhangsan yiwei ziji de erzi zui congming. Zhangsan think self DE son most clever 'Zhangsan thought that his son was the cleverest.' b. Zhangsan yiwei Lisi zui xihuan ziji de erzi. Zhangsan think Lisi most like self DE son 'Zhangsan thought that Lisi liked his son most.' (72) a. Zhangsan shuo ziji kanjian-le Lisi. Zhangsan say self see-Perf Lisi 'Zhangsan said that he saw Lisi.' b. Zhangsan shuo Lisi kanjian-le ziji. Zhangsan say Lisi see-Perf self 'Zhangsan said that Lisi saw him.' In all these examples, ziji may be bound by the matrix subject Zhangsan under appropriate logophoric conditions (e.g., under a de se scenario). In addition, under a non-de se scenario (when the coreference is reported purely as the speaker's knowledge from the speaker's own perspective), ziji may also be bound by the matrix subject Zhangsan in (71a) and (72a), though not in (71b) or (72b). In the (b) sentences, under the non-de se scenario, the reflexive is only naturally bound by the embedded subject Lisi. This is consistent with the view that, when logophoric conditions do not hold, a reflexive is admitted if locally bound and that "local binding" means binding in its GC: the
Logophoricity, Attitudes, and Ziji at the Interface
169
GC for the reflexive in (a) is the main clause, and the GC for the reflexive in (b) is the embedded clause.16 The subject-object asymmetry displayed in (71)-(72) is corroborated by the distribution of blocking effects. As shown in the following (b) and (c) examples, a first-person pronoun blocks the LD binding of ziji in the embedded object. However, as shown in the (a) examples, a first-person pronoun does not induce blocking when ziji occurs in or within an embedded subject position. (73) a. Zhangsani gaosu wo ziji i de erzi zui congming. Zhangsan tell me self DE son most clever 'Zhangsanj told me that his, son was the cleverest.' b. ??Zhangsani gaosu wo Lisi zui xihuan ziji i de erzi. Zhangsan tell me Lisi most like self DE son 'Zhangsanj told me that Lisi liked his, son most.' c. * Zhangsani yiwei wo zui xihuan zijii de erzi. Zhangsan think I most like self DE son 'Zhangsanj thought that I liked hisj son most.' (74) a. Zhangsani dui wo shuo ziji i piping-le Lisi. Zhangsan to me say self criticize-Perf Lisi 'Zhangsan; said to me that hCj criticized Lisi.' b. ??'Zhangsani dui wo shuo Lisi piping-le zijii Zhangsan to me say Lisi criticize-Perf self 'Zhangsanj said to me that Lisi criticized himi.' c. * Zhangsani shuo wo piping-le ziji i . Zhangsan say I criticize-Perf self 'Zhangsan; said that I criticized himi.' This asymmetry is observed not only with complement constructions, but also with adjuncts, such as relative clauses: (75) a. Zhangsani xiang wo tuijian [ziji i piping-le Lisi de Zhangsan to me recommend self criticize-Perf Lisi DE nei-ben shu]. that-CL book 'Zhangsani recommended to me the book in which hei criticized Lisi.' b. 11 Zhangsani xiang wo tuijian [Lisi piping-le ziji i de Zhangsan to me recommend Lisi criticize-Perf self DE nei-ben shu] that-CL book 'Zhangsanj recommended to me the book in which Lisi criticized him,.'
170
C.-T. James Huang and C.-S. Luther Liu
c. *Zhangsani zui xihuan [wo piping-le ziji i de nei-ben shu] Zhangsan most like I criticize-Perf self DE that-CL book 'Zhangsarij likes the book in which I criticized him;.' We therefore submit that the binding domain of ziji as a syntactic anaphor is the traditional GC. If ziji is not bound in its GC, it will have to be admitted as a logophor, and in that case, we expect it to exhibit logophoricity effects.17 4.3. Subcommanding Antecedents We have considered various constructions in which a given occurrence of ziji can be identified as either a syntactic anaphor or a logophor, each with its expected properties. What about cases where ziji takes a subcommanding NP as its antecedent? An example is repeated in (76): (76)
Zhangsan-fde jiaoao hai-le ziji i Zhangsan's arrogance hurt-Perf self 'Zhangsanj's arrogance harmed himj.'
Is ziji an anaphor or a logophor in (76)? The fact that it is not c-commanded by its antecedent might suggest that it is a logophor (see, e.g., Xue et al, 1994). From this, we should expect it to display some logophoricity properties that are by now familiar. We now argue that it should be treated as an anaphor, rather than a logophor. This view is supported by the following facts. First, the antecedent Zhangsan cannot be appropriately characterized as a logophoric antecedent, as it is not the Source, nor necessarily the Self conscious of the event described, nor does it necessarily represent the speaker's empathy focus. Example (76) can be naturally read as a speaker's report of his own observation. Second, sentences with subcommanding antecedents do not exhibit any blocking effect. (77) Zhangsani de biaoqing gaosu woj [ziji i/*j shi wugude]. Zhangsan DE expression tell me self is innocent 'Zhangsanis [facial] expression tells me that hei is innocent.' (78) Zhangsani de xin gen wo tandao-le ziji i Zhangsan DE letter to me discuss-Perf self 'Zhangsanis letter discussed himi with me.' Third, note that long-distance binding by a subcommander is unacceptable in cases like (79):18 (79) *Zhangsani de shibai biaoshi tamen dui ziji i mei xinxin. Zhangsan DE failure indicate they to self no confidence 'Zhangsan's failure indicates that they have no confidence in him.'
Logophoricity, Attitudes, and Ziji at the Interface
171
This fact would be unexpected if a subcommander is a possible logophoric antecedent, since a logophor is typically long-distance bound. The absence of logophoricity effects on subcommand binding thus leads us to the conclusion that it must be a case of anaphoric binding. The question that remains is why in cases like (76)-(78) strict c-command is not required for syntactic binding to be possible. The answer has already been provided by Tang (1989), who attributes it to the fact that Chinese reflexive ziji is inherently [+animate]. A subcommander qualifies as an antecedent precisely because, by definition, it is contained only in inanimate c-commanders that are themselves nonpotential antecedents. In fact, under Kayne's (1994) conception of phrase structure, there is no need for the notion of subcommand, as subcommand is just a case of c-command. Specifically, assuming that specifiers are introduced by adjunction, and that c-command is as defined here: (80) X c-commands Y iff X and Y are categories and X excludes Y and every category that dominates X dominates Y (Kayne, 1994) then any specifier of X c-commands everything that X c-commands. In the case of (76), both Zhangsan (DPI) and Zhangsan de jiaoao (DP2) c-command ziji:
But since the inanimate Zhangsan de jiaoao is not a potential antecedent, Zhangsan minimally binds ziji and, hence, is the syntactic antecedent of ziji as a syntactic anaphor.19 We therefore conclude that binding of ziji by a subcommander is a case of anaphor binding. From here, it follows that (a) subcommand binding exhibits no logophoricity effects and (b) subcommand binding is always local. One may still ask why a logophor cannot take a subcommander as its antecedent. The answer seems simple: a subcommander, being a modifier, does not represent the Source, Self, or Pivot of speech or reported event, and hence, plays no role in logophorocity. Summarizing, we have seen that in the environments described in
172
C.-T. James Huang and C.-S. Luther Liu
(82a-d), ziji is a syntactic anaphor obeying BCA, and in (82e), it may be used as a syntactic anaphor or as a logophoric pronoun. (82) a. When it is bound by a co-argument subject. b. When it is contained in an argument NP and bound by a co-argument of that NP. c. When it is contained in an adjunct and locally bound by an argument outside. d. When it is locally bound by a subcommanding NP. e. When it is the subject of an embedded clause, or contained in the subject of an embedded clause, and is locally bound in the matrix clause. 4.4. Comparison with Other Mixed Approaches We thus share with major proponents of the "mixed" approach, including Reinhart and Reuland (1993) Pollard and Sag (1992) and Xue, Pollard, and Sag (1994), in assuming that some instances of reflexive binding are governed by syntactic conditions of anaphor binding while others are governed by logophoric considerations. We differ from these proponents, however, in seeing where the dividing line lies between these two uses of ziji. According to Reinhart and Reuland (and also Williams 1989,1994), for example, the syntactic notion of reflexivity is defined in terms of the argument structure of a predicate. Put in somewhat simplified terms, a predicate is said to be reflexive iff two of its arguments are coindexed. A sentence that contains the reflexive pronoun as one of its arguments is said to be "reflexive-marked." As far as Binding Theory is concerned, it provides that if a predicate is "reflexive-marked," then it must be "reflexive." That is, if a sentence contains a reflexive pronoun as an argument, that reflexive argument must be bound by a co-argument. All other occurrences of the reflexive fall outside of the core theory of anaphor binding. Reinhart and Reuland's conception of the syntactic anaphor is thus considerably narrower than ours, covering only cases of co-argument binding (i.e., (82a)). If we were to adopt their approach regarding Mandarin ziji, then all other cases in (82) would have to be attributed to logophoricity, along with those cases in which ziji is truly long-distance bound. As we have seen, the cases represented in (82) represent those that do not exhibit logophoricity effects associated with truly LD-bound ziji, and that the traditional GC gives us a convenient dividing line to categorize two (partially overlapping) groups of environments, each with a set of distinct, predictable properties. Adopting Reinhart and Reuland's view amounts to adopting a wider definition of the notion of logophoricity. However, to the extent that they did not offer a theory of
Logophoricity, Attitudes, and Ziji at the Interface
173
logophoricity that characterizes (82b-e) and other LD ziji environments to the exclusion of (82a), such a liberal notion of logophoricity appears to be little more than a catchall term for everything that violates their co-argument requirement for anaphor binding. A similar objection may be raised against the approach developed in Xue et al. (1994). Ignoring the technical details, Xue et al. propose that ziji, their "Z-pronoun," occurring in or within some nonsubject argument position, is a "syntactic reflexive" subject to syntactic binding by an animate subject. A reflexive that does not meet this description is a "pragmatic reflexive" subject to certain pragmatic-discourse conditions. The cases for a syntactic reflexive include: (83) a. When ziji is bound by its co-argument subject, and b. When ziji is contained in an argument NP or clause and is coindexed with a subject that is a co-argument of the containing NP or clause. Condition (83a) corresponds to our case (82a). The second case includes our (82b) and also includes some instances of (82e), where ziji occurs in some argument embedded clause or other. In these cases, then, Xue et al.'s "syntactic reflexive" is empirically more inclusive than is Reinhart and Reuland's in correctly recognizing more environments in which a reflexive is syntactically bound. However, like the latter, they also incorrectly exclude adjunct and adjunct-contained ziji's that are locally bound (82c), as well as ziji's with locally subcommanding antecedents (82d) from the class of syntactic anaphors. Xue et al, in fact, explicitly consider subcommanding binding as a case of a pragmatic reflexive, but as we have seen, subcommand binding exhibits strict locality effects but neither logophoricity nor blocking effects, and is more reasonably regarded as a case of syntactic binding. In other words, Xue et al.'s notion of a syntactic ziji is narrower than the notion of a syntactic anaphor adopted here. Furthermore, because Xue et al.'s notion of a syntactic reflexive does not impose locality on the binding of ziji, their syntactic ziji is also broader than our notion of a syntactic anaphor. Thus, according to Xue et al., a syntactic ziji may occur as the object of an embedded complement clause and LD bound by the subject of a higher clause, as a subcase of (83b). We consider such binding configurations to be logophoric, however, as evidenced by the various logophoric (including blocking) effects they display.20 In other words, Xue et al.'s conception of a syntactic reflexive includes cases of LD binding that exhibit certain logophoricity effects (as well as locally bound ziji with no such effects) and their conception of a pragmatic reflexive includes cases of local binding with no logophoricity effects (as well as LD binding with such effects). On the other hand, using the GC as the dividing
174
C.-T. James Huang and C.-S. Luther Liu
line, we capture the two uses of ziji in terms of two sets of largely exclusive properties: the syntactic anaphor exhibits BCA effects (e.g., locality but no blocking), and the logophor exhibits logophoricity effects (e.g., blocking). It seems that our approach gives a more satisfactory partitioning of the relevant matters.21
5. THE LONG DISTANCE ZIJI: SYNTAX AND SEMANTICS In the preceding pages, we have argued for distinguishing between the anaphoric and the logophoric uses of ziji, each with its distinct properties. We assume that, syntactically, the anaphoric ziji is subject to Condition A of the Binding Theory and, semantically, it is referentially dependent on the antecedent that locally binds it. We shall have nothing further to say about the syntax and semantics of the anaphor ziji, but assume that they would be in line with general conceptions concerning the syntax and semantics of A-binding.22 As for the logophoric ziji, although we have isolated a number of its properties (that it is bound by a Source, a Self, or a Pivot, and that it may display blocking effects), these properties have been presented and discussed at an informal and intuitive level. We have discussed the pragmatic properties of the logophoric ziji, but have not addressed its syntax or semantics. We turn to these issues now.
5.1. The Semantics of Logophoricity Although Kuno's (1972,1987) early observations and discussion of the discourse-pragmatic effects of anaphora provide valuable insight into the nature of logophoricity in an intuitively satisfactory way, his account of them in terms of direct-discourse representations did not tie itself with a general theory of semantics and of the syntax-semantics interface. In particular, his proposal that logophoric sentences are literally derived from deep structures with direct discourse complements was framed in the early Standard Theory or generative semantics framework, but the most sophisticated semantic theory developed to date in generative grammar is not generative, but interpretive. The direct discourse representative itself is too crude to be a true semantic representation of logophoric sentences in general. The hypothesis that logophoric sentences are literally derived from underlying structures with direct quotes may be reasonable for some sentences—especially those with matrix quotative verbs like say, claim, exclaim, ask, etc. (or their equivalents in other languages), it is not entirely suitable for a large
Logophoricity, Attitudes, and Ziji at the Interface
175
number of other embedding constructions (e.g., dream, know, forgot, etc.), and would be highly contrived for constructions like the following: (84) Zhangsan kan-wan-le [Lisi piping ziji de nei-pian wenzhang] Zhangsan read-finish-Perf Lisi criticize self DE that-CL article 'Zhangsan finished reading the article in which Lisi criticized him.' (85) [Lisi bu xihuan ziji de shi] shide Zhangsan hen shangxin. Lisi not like self DE thing make Zhangsan very sad 'The fact that Lisi dislikes him caused Zhangsan to be very sad.' Sells (1987) argues that the notion of a logophoric antecedent should be analyzed in terms of the three roles Source, Self, and Pivot, and develops a semantics within Discourse Representation Theory (DRT) in which these roles are directly postulated as primitives. Chierchia (1989) argues, however, that the notion of logophoricity can be integrated into an interpretive theory of semantics by independently needed notions in semantic theory (but without recourse to newly postulated "primitives" like Source or Self). Chierchia adopts Lewis' (1979) distinction between de re and de se beliefs and capitalizes on his insight that while the de re readings of attitudinal sentences express a relation between a believer and a proposition, the de se readings express a relation between a believer and a property. The former is a normal believe relation, where the believer holds a certain proposition to be true, and the latter a believe-to-have relation, where a believer (knowingly) ascribes a certain property to himself/herself. Chierchia proposes that this distinction can be directly captured in semantic representation by representing the complement clause either as a propositional argument or as a (secondary) X-predicate. Thus for the sentence (86), the de re and de se readings are respectively represented in (87): (86) Pavarotti believes that his pants are on fire. (87) a. (l,x (believe (x, x's pants are on fire))) (P) b. believe (P, A,x (x's pants are on fire))23 In other words, a surface sentence like (86) is analyzed as having a normal structure of complementation or a structure of (secondary) predication. The structure of predication is appropriate for the de se reading, because it is already a fact of other known structures of secondary predication that they permit only de se interpretations. For Chierchia (1984,1989) (and Williams, 1980, etc.), structures of secondary predication include structures of obligatory control. Note that the obligatory control sentences in (a) must be interpreted as expressing de se beliefs only, whereas the (b) sentences, with pronominal binding, may express general de re beliefs.
176
C.-T. James Huang and C.-S. Luther Liu
(88) a. John claims [PRO to be innocent], b. John claims that he is innocent. (89) a. John expects [PRO to win], b. John expects that he will win. The (b) sentences could be true of scenarios in which John points to someone in the mirror (who the speaker knows is John's own image), but the (a) sentences are true only of scenarios in which John actually says, or is disposed to say, "I am innocent" or "I will win." Taking obligatory control to be predication, (88a) and (89a) have the following representations: (90) a. claims (J, X,x (innocent (x))) b. expects (J, X,x (will-win (x))) These are identical in relevant respects to (87b),24 each representing a belief-to-have relation between an individual and a property. To ensure that these structures are interpreted as desired, Chierchia (1989) proposes the following definition of de se attitudes: (91) x stands in a belief relation with the property Q (i.e., x self-ascribes Q in Lewis' terms) iff x believes (de re) that x has Q, and furthermore, K(x, x), where K is the cognitive access that we have to ourselves. (92) K (x,x) = def x is disposed to describe the relevant belief by referring to x by the first-person pronoun. Assuming Chierchia's semantics, we treat the LD ziji as a variable bound by a X-operator, in the semantic representation. 5.2. The LF Syntax of Logophoricity How would such a semantic representation as (90) be related to, or derived from, the syntactic structure of LDR sentences, in a proper theory of syntax-semantics interface? The answer, we suggest, is readily provided by the syntax of LF proposed in Huang and Tang (1991). According to them, the LD reflexive adjoins in LF to an IP, and from the IP-adjoined position, it is coindexed with—rather, predicated on—its antecedent. Under the LD construal of the reflexive, the sentence (93) has the LF representation (94): (93) Zhangsan shuo Lisi you zai piping ziji le. Zhangsan say Lisi again at criticize self Prt 'Zhangsan said that Lisi was again criticizing him.' (94) Zhangsan shuo [ziji i [Lisi you zai piping ti le]].
Logophoricity, Attitudes, and Ziji at the Interface
177
As an alternative, and in the spirit of Huang (1984) and especially Rizzi's (1997) recent work on the "fine structure of the left periphery," we can also replace the IP-adjunction hypothesis with the hypothesis that ziji moves to the Spec of a CP-type functional category in LF which identifies itself as the property of a self-ascriber (i.e., Source or Self in the core cases, to be expanded to include the Pivot). Rizzi (1997) argues for a "split CP" hypothesis, according to which the traditional CP consists of a complex of functional categories (including a Force Phrase, a Focus Phrase, a Finiteness Phrase, and potentially several Topic Phrases). These are elements that play an important role in mediating between syntax and discourse. Let's hypothesize that one such "Topic" Phrase is reserved for marking the Source, i.e., that it is, in fact, a SourceP. The LF representation of (93) then has ziji as the Spec of the SourceP:25
Note that this LF representation is directly convertible to the semantic representation (96). l-abstraction consists of simply translating the operator ziji in Spec of "Source Phrase" into the A,-operator and its trace into a bound variable. (96) shuo [Zhangsan, lx [Lisi you zai piping x le]] Coindexing the LF-raised ziji with the antecedent Zhangsan is part and parcel of what Chomsky (1982,1986) terms predication or "strong binding." In semantic terms, the i-predicate as a whole is taken as denoting a property of Zhangsan.26 Note that the postulation of a Source Phrase as in (95) provides a direct way to obtain the semantics of de se attitudes that Chierchia's (stipulated) definition in (91)-(92) is intended to bring out, and makes it possible to eliminate the latter. According to (95), the LF-raised ziji is identified as Source.
178
C.-T. James Huang and C.-S. Luther Liu
Having this identity as Source (i.e., 'me') included in the belief of an internal protagonist amounts to the claim that the protagonist has a belief about himself/herself in the first person, which is the same as saying that the protagonist has a de se belief about the property described in the Source Phrase. Summarizing, although it is clear that some important properties (e.g., blocking effects) of the LD reflexive fall outside of syntax in the traditional sense, we claim that the LF syntax proposed in Huang and Tang still captures the syntax of logophoricity in important ways. For one thing, the LF-raised structure provides a direct input for translation into semantic representation. Furthermore, the idea that ziji LF-raises into [Spec, SourceP] also provides the syntactic correlate of the semantics of self-ascription. A third important aspect of the Huang and Tang approach that is preserved here is the claim that so-called LD binding is, in fact, local. In particular, the predication or strong-binding relation between the LF-raised ziji and its antecedent is strictly local, though the operator-variable chain consisting of ziji and its trace may be long distance.27,28 In other words, we treat a logophoric ziji not as a normal "pronoun in coreference" but as a variable A'-bound by an operator which is itself anaphoric, on a par with a null operator (OP) or an obligatory control PRO in constructions like the following:29 (97) John is easy [OP [to please t]]. (98) John left [PRO tired]. (99) John wanted [PRO to win]. In addition to strict locality, another well-known property of such constructions is that they exhibit obligatory sloppy identity under VP ellipsis: (100) John is easy [OP [to please t]], and Bill is, too. (101) John left [PRO tired], and Bill did, too. (102) John wanted [PRO to win], but Bill did not. These sentences cannot have a "strict" reading according to which Bill is easy for John to be pleased, or Bill left with John being tired, or Bill did not want John to win. Our treatment of the logophor along the lines of (94)-(96) predicts that an LD ziji allows only sloppy identity under VP ellipsis. As pointed out by Cole et al. (this volume), this prediction is correct. In the following sentence, (103) Zhangsan juede List chang qipian ziji, Wangwu ye shi. Zhangsan feel Lisi often cheat self Wangwu also be 'Zhangsan feels that Lisi often cheated him, and so does Wangwu.'
Logophoricity, Attitudes, and Ziji at the Interface
179
there is a reading according to which Wangwu feels that Lisi often cheated Wangwu (the sloppy reading), but no reading according to which he feels that Lisi often cheated Zhangsan (the strict reading). According to the analysis depicted in (94)-(96), (103) has the following relevant representation: (104) . . ., Wangwu ye shi [lx [x juede [ly [Lisi change qipian y]]]]. Wangwu also be feel Lisi often cheat where the property denoted by [Ay [Lisi change qipian y]] is predicated of the subject x of juede 'feel' (i.e., Wangwu). Such a representation gives rise only to the sloppy reading, because neither the operator ly, nor the variable y it binds, can be bound by anything other than Wangwu.30 Summarizing, we assume, in much the same spirit as Huang and Tang (1991), that the LF syntax of logophoricity involves operator-movement of ziji to Spec of SourceP.31 The logophor ziji is an anaphoric element (akin to null OP or obligatory control PRO) subject to predication or strong binding by an appropriate local antecedent, or is interpreted in discourse by default to denote the speaker. The operator-variable structure is translated into a A-predicate denoting a property that ziji's antecedent self-ascribes. 5.3. Extensions: Consciousness and Perspectivity The syntax and semantics of the LDR described previously works straightforwardly for the core cases—sentences in which the logophoric ziji is bound by a matrix subject of an attitudinal or communicative verb. In such a case, the antecedent is directly represented, in semantic representation as in the syntax, as the self-ascriber of de se beliefs. However, there are many cases of logophoric ziji where this relation cannot be directly obtained from syntactic structure. These include sentences with experiencer or causative predicates, like (84)-(85), repeated below: (84) Zhangsan kan-wan-le [Lisi piping ziji de nei-pian wenzhang]. Zhangsan read-finish-Perf Lisi criticize self DE that-CL article 'Zhangsan finished reading the article in which Lisi criticized him.' (85) [Lisi bu xihuan ziji de shi] shide Zhangsan hen shangxin. Lisi not like self DE thing make Zhangsan very sad 'The fact that Lisi dislikes him caused Zhangsan to be very sad.' These sentences exhibit both blocking and consciousness effects, and the reflexive is therefore logophoric. In particular, LD binding becomes unacceptable if Lisi is replaced by wo I. Furthermore, in both cases the natural assumption is that Zhangsan is aware that Lisi criticized him or does not like him in some context and therefore is disposed, if asked, to describe such
180
C.-T. James Huang and C.-S. Luther Liu
states of affairs using the first-person pronoun. In other words, Zhangsan is disposed to self-ascribe some property all right, but neither (84) nor (85) is a description of Zhangsan's self-ascription of that property. These cases present a problem for the LF syntax and interpretive semantics of attitudes de se we have adopted here, as much as they did for Kuno's generative approach involving direct discourse complementation. One way to account for these cases would be to postulate an additional subtype of logophoricity. This is essentially what Sells (1987) has done, according to whom there are three subtypes of logophoric antecedents (Source, Self, and Pivot), and (84)-(85) are cases with Self antecedents. These subtypes are not entirely independent of each other, but are related by an implicational relation: Source implies Self, which, in turn, implies Pivot, but not the other way round. We have seen that the semantics of Source logophoricity fits well under the semantics of attitudes de se. What about cases with Self logophors? One way to accommodate both Source and Self in the semantics of de se beliefs while preserving the implicational relation between them is to say that, in the former cases, the de se belief is asserted, but in the latter cases, it is simply implied by the truth of a given sentence. For example, under the de se reading, the sentence John claimed that he was the smartest asserts that the Source self-ascribes the property of being smart. For a sentence like John is saddened by Bill's criticism of him, no similar assertion is being made of John, but for the sentence to be true under the de se reading, it must be true also that John implicitly self-ascribes the property of being the object of Bill's criticism. For the examples (84) or (85) at hand, we can say that their truth implies the following at the level of semantic representation:32 (105) believe (NP,lx (. .. x ... )) Although stating (105) as an implication allows us to unify Source and Self under the semantics of de se attitudes, this move is not sufficient to account for those cases of logophoricity that involve Pivot antecedents. As mentioned earlier, some speakers find such sentences as (46)—repeated below—to be acceptable, even under the assumption that Zhangsan was unaware that Lisi was criticizing him, i.e., even if (105) does not hold of the antecedent NP. (46) ?? (Dang) Lisi piping ziji i de shihou, Zhangsani zheng zai kan (at) Lisi criticize self DE moment Zhangsan right at read shu. book 'At the moment Lisi was criticizing him, Zhangsan was reading.' A crucial requirement for this to be possible, however, is that the speaker and the antecedent must share the same perspective. Example (46) is unacceptable if the speaker does not empathize with Zhangsan. This require-
Logophoricity, Attitudes, and Ziji at the Interface
181
ment can also be illustrated by the following contrast, which obtains for some speakers.33 (106) ?*Zhangsani kuajiang-le [[houlai sha si ziji i de] naxie ren i .
Zhangsan praised-Perf later kill die self DE those persons 'Zhangsanj praised those persons who later killed himj.' (107) ?Zhangsani kuajiang-guo [[houlai sha si zijiii de] naxie renj].
Zhangsan praised-Exp later kill die self DE those persons 'Zhangsan; has praised those persons who later killed himi.' The contrast arises from the fact that the main verb ends with the perfective aspect -le in (106) but with the experiential aspect -guo in (107). Sentence (106) is a report of a past event, whereas (107) is an evaluation from the standpoint of the speaker's present. A clearer contrast also obtains between the somewhat marginal (106)-(107) and the following which are fully unacceptable: (108)
*Zhangsani san-nian qian zanmei-le [zuotian
Zhangsan 3-year nei-ge
shasi-le ziji i de
ago praise-Perf yesterday kill-Perf self DE
ren].
that-CL person 'Zhangsan praised 3 years ago the person who killed self yesterday.' (109) *Zhangsani shang xingqi zanmei-le [jin zao
Zhangsan last nei-ge
piping ziji i de
week praise-Perf this morning criticize self DE
ren].
that-CL person 'Zhangsan praised last week the person who criticized self this morning.' These sentences are worse because explicit time expressions are provided to indicate the sequence of events. Thus, these sentences are more naturally regarded as objective descriptions of the relevant sequence of events, rather than the subjective evaluation from the speaker's present perspective. When the events are evaluated from the speaker's present perspective and when the speaker empathizes with the internal protagonist, binding by the internal protagonist becomes possible. Obviously, these sentences cannot be accommodated by the condition (105) because we are looking at scenarios where the internal protagonist was actually unaware of the event that would occur later which would be ascribed to him or her. The question is why an exception is allowed only when a shift in perspectives occurs that puts together the internal protagonist with the speaker. It seems to us that this shift in perspective allows the speaker to
182
C.-T. James Huang and C.-S. Luther Liu
"go proxy" for the internal protagonist in viewing the relevant event or state. In this way, the internal protagonist has virtual consciousness given the benefit of the speaker's actual consciousness, and hence, had virtual de se attitudes about the relevant event or state.34 In other words, Pivot logophors represent an extension of core logophoricity, which is limited to cases where the LD ziji is part of the description of a property which the relevant protagonist is disposed to self-ascribe in some way. The Source may be said to explicitly self-ascribe that property by using the first-person pronoun, but the Self is simply disposed to do so given its own state of consciousness, while the Pivot is enabled to do so with the help of the speaker's empathy. Put in another way, Kuno's (1972,1987) various functional principles of anaphora, including direct discourse, consciousness, and empathy, are not independent of each other, but represent the different degrees of directness in which attitudes de se manifest themselves. How are these various cases of logophoricity to be integrated to the syntax we have proposed? One solution that suggests itself is to postulate a Self Phrase and a Pivot Phrase in addition to the Source Phrase, though given the implicational relation among these subtypes, the Pivot Phrase is all that we need, as one of the positions in the "fine structure of the left periphery" in Rizzi's (1997) system. Thus, (46) has the following structure:
An appropriate predication relationship can obtain for this structure if the Spec of Pivot Phrase is bound by an NP that, minimally, is the Point-of-View holder of the relevant discourse. When no internal protagonist is identified as a Point-of-View holder, the Speaker is the default antecedent of the Spec of Pivot Phrase. 5.4. Comparison with Pan As indicated previously, our treatment of LD ziji in terms of attitudes de se a la Chierchia (1989) is a descendant of Huang et al. (1984) a la Kuno
Logophoricity, Attitudes, and Ziji at the Interface
183
(1972). The idea that LD ziji is a "de se anaphor" was also proposed by Pan (1997), based on a suggestion of Manfred Krifka.35 In spite of the similarities, however, we differ from Pan in several important respects.36 First of all, we differ in how we situate the account of de se attitudes in a theory of logophoricity. Pan argues that LD ziji is not a logophor, as existing theories of logophoricity (e.g., Clements, 1975; Sells, 1987) cannot account for all the properties that ziji exhibits. This has the implication that while Chinese ziji is a "de se anaphor," LD-reflexives (and logophoric pronouns) in other languages belong to a distinct category called logophors. On the other hand, we consider logophoricity to be a descriptive cover term for a number of related phenomena whose content has been enriched by the properties of Chinese LD ziji, and we take it that the syntax and semantics de se beliefs that we have assumed here constitute (at least the beginnings of) a theory of logophoricity Second, we differ from Pan in how we treat the syntax and semantics of de se beliefs. Following Chierchia (1989) and Lewis (1979), we assume that de se beliefs denote relations between the believer and a property, represented by a l.-predicate in semantic representation. Such l,-predicates, in our system, correspond to LF structures created by A'-movement of the LD ziji. Pan does not assume an LF-syntax of the kind we do, nor a semantics that can make the proper distinction between de re and de se beliefs. For the sentence John thinks that he is smart, for example, Pan gives the following for the de se and de re readings: (111) a. A,x [x thinks that x is smart] (j) (de se) b. lx [x thinks that j is smart] (j) (de re) This amounts to the claim that the de se vs. de re distinction is simply one of binding vs. accidental coreference. But de se interpretation cannot be reduced to variable binding alone. In the following sentence, where the pronoun is bound by everyone, we still have both the de re and the de se readings: (112) Everyone says that he is smart. The sentence can readily have the de se reading, but it is also possible that everyone unknowingly praises himself. This latter situation is possible if everyone, all of whom have never seen mirrors before, respectively points at his own image in a mirror and says "he is smart" without realizing he was praising himself. This latter reading is de re, but it is a bound reading. The semantics Pan gives in (Ilia) is not adequate for the de se reading.37 Finally, we differ from Pan in our explanation for the blocking effects. We follow Huang et al. (1984) and claim that the blocking effects reflect perceptual difficulties that arise when elements within the same discourse domain
184
C.-T. James Huang and C.-S. Luther Liu
are "anchored" to different "speakers." Thus, an embedded clause containing ziji bound by a third-person matrix antecedent cannot contain first- or second-person pronouns: (113) *Zhangsani shuo [wo/nibu yinggai qipian ziji i . Zhangsan say I/you not should cheat self That is, the intended reading requires the protagonist Zhangsan to be able to describe the content of the embedded clause in the first person, but this is prevented by the presence of wo and ni in (113), which are anchored to the external speaker. Pan's account for the blocking effect is based on the claim that the firstand second-person pronouns are "obligatory self-ascribers" whereas third-person NPs denote "optional self-ascribers." According to him, in (113) the presence of the obligatory self-ascribers wo and ni in the c-domain of the remote third-person Zhangsan prevents the latter from being the LD antecedent of ziji. That is, a closer obligatory self-ascriber blocks a remote optional ascriber.38 There are two flaws of Pan's account that make it an unlikely candidate as an explanation of blocking. For one thing, to call the embedded subject in (113) a self-ascriber is not appropriate for what the term self-ascription means. An ascriber is one who has certain mental attitude over some property expressed by the complement of an attitudinal predicate (at whatever level of representation). In (113), the embedded subject is simply the subject of some event-denoting predicate; its referent does not ascribe any property. Furthermore, Pan's theory also fails to account for blocking effects induced by indexical/deictic elements. Recall that a speaker may introduce a third-person embedded subject by pointing a finger to its referent in the audience (see (12)). A third-person indexical NP in the embedded subject position is no ascriber, let alone obligatory self-ascriber, of any property. But such a deictic/indexical pronoun does induce blocking. And this, again, cannot follow from Pan's account.39
6. SUMMARY In this chapter, we have re-examined the bare reflexive ziji in Mandarin and showed that its properties are best accounted for if a distinction is made between local ziji and long-distance bound ziji- The local ziji is an anaphor whose reference is governed by Condition A of binding theory. The LD ziji is a logophor, whose occurrence is limited by the condition that it be part of a description of some property that its antecedent's referent either explic-
Logophoricity, Attitudes, and Ziji at the Interface
185
itly self-ascribes, or is implicitly disposed to self-ascribe on its own state of consciousness, or virtually do so with the help of the speaker's perspective. We showed that the blocking effects are best accounted for as perceptual effects resulting from conflicting "anchors" of discourse-sensitive NPs (including first- and second-person pronouns, indexical terms, etc.).40 In so doing, we claim that the earlier account of Huang and Tang (1991) in terms of LF movement is inadequate as an account of the blocking effects. At the same time, we have also shown that the LF movement process they proposed is preserved, as a means to create a functional structure that mediates between grammar and discourse and that captures the syntax, semantics, and pragmatics of logophoricity.41
ACKNOWLEDGMENTS Earlier versions of this chapter were presented at the NSF-supported Workshop on Long-distance Reflexive at Cornell University in summer 1997, and at the 7th IACL Conference at Stanford University and the 6th IsCLL Conference at Academia Sinica, Taipei, in summer 1998. We are indebted to the many people who discussed some of the materials here with us and provided us with very useful comments, including Francesca Del Gobbo, Naomi Harada, Hide Hoshi, Shizhe Huang, Jaeshil Kim, Audrey Li, Yafei Li, Jonah Lin, Dylan Tsai, Matthew Whelpton, Di Wu, and especially Gennaro Chierchia, Peter Cole, Gaby Hermon, Ken Safir, and Ping Xue. The usual disclaimers apply. The project on the theory of long-distance reflexives of which this research is a part is supported by NSF Grant #SBR-9729519. Part of the research was carried out when one of the authors (Huang) was a Fellow of the Center for Advanced Study in the Behavioral Sciences (1997-1998). The fellowship support was made possible, in part, by NSF Grant #SBR-9601236.
NOTES 1
See also Progovac (1992) for further discussion of problems of the LF-movement account and an account in terms of the notion of a "Relativized Subject." Huang and Tang's account makes crucial use of the fact that ziji lacks f-features, but dissociates itself from the property of monomorphemicity. This has both desirable and undesirable consequences. 2 Although there is still a clear preference for local binding over LD binding in (b-c), we believe that there is a real contrast between these and (a), where the LD reading seems quite impossible. translated from the Chinese text. 4 Pollard and Xue (1998, and this volume) now assume a somewhat different approach. While arguing for the need to distinguish syntactic and nonsyntactic uses of
186
C.-T. James Huang and C.-S. Luther Liu
reflexives, they do not treat reflexives as being ambiguous in having pronominal or anaphoric properties. They treat reflexives as items that can be, but do not need to be, bound syntactically but that, if not syntactically bound, must be interpreted in accordance with appropriate pragmatic conditions. 5 This includes positions that make up the Complete Functional Complex (Chomsky, 1986) of a verb plus the embedded subject in an ECM construction. 6 Reinhart and Reuland assume a two-module theory of anaphora. In addition to Conditions A and B defined over reflexivity and reflexive marking, they also invoke a condition defined over A-chains which, among other things, requires its head to contain only nonreferential elements. Thus, the sentence *Himselfsaw John satisfies Condition A, but is ruled out by the Chain Condition. 7 In (21b), herself is Case-marked by believe, hence, a "syntactic co-argument" of Mary. 8 As Kuno argues, this contrast would not follow from any existing formal version of Binding Theory, according to which both (32a) and (32b) would be well-formed. The standard theory correctly characterizes the sentences as grammatical, however, under the interpretation that they report on the speaker's knowledge. 9 This pair bears great similarity to one discussed in Sells (1987), though Sells argues that the relevant contrast between the Japanese examples he provided was illustrative of a property of the Pivot or perspectivity. We consider this an effect of consciousness, for the reasons stated in the text. 10 The status of a Pivot antecedent seems not only marginal but also variable among dialects or speakers. Cf. Cole, Hermon, and Lee (this volume), where it is shown that whereas Singapore Mandarin speakers are quite permissive with Pivot antecedents, Teochew speakers are quite reluctant to accept Pivot antecedents. We view Source, Self, and Pivot as constituting a hierarchy representing progressive grammaticalization of these pragmatic concepts. n This explanation in terms of conflicting sources is reminiscent of a principle proposed by Kuno (1987), the Ban on Conflicting Empathy Foci. Since Source implies Pivot, conflicting sources lead to conflicting perspectives. 12 The fact that nonsubjects may induce blocking was first due to Xue et al. (1994). Pan claims that they do not. Our own judgment agrees with that of Xue et al.'s. We do share the feeling that the blocking effect observed in (54) is not as strong as is observed in (49a), where the blocker is the embedded subject. We submit that this arises from the fact that in (54) the first person pronoun originates in the matrix clause but is not part of the direct discourse complement. This presumably makes it easier to tell apart the wo that refers to the external speaker from the wo that refers to the internal speaker, and therefore reduces the perceptual difficulty involved. 13 In English, John and Bill bought a car can mean either that they bought a car together or that they each bought a car (two cars total), but the Chinese counterpart in (i) only has a collective reading. To obtain the distributive reading, either the distributor ge 'respectively' or the universal quantifier dou 'all' is required: (i)
Zhangsan gen Lisi mai-le yi-liang che. Zhangsan and Lisi buy-Perf one-CL car
(Collective)
Logophoricity, Attitudes, and Ziji at the Interface
(ii)
Zhangsan gen Lisi ge mai-le yi-liang che.
(iii) Zhangsan gen Lisi dou mai-le yi-liang che.
187
(Distributive) (Distributive)
14
The local ziji also has only a distributive reading, but it does not require its local subject to be distributed by dou or ge: (i)
tamen zai piping ziji. they at criticize self 'They are criticizing themselves.'
Huang (1998b) argues that this arises from LF head-movement of ziji to the governing verb, forming an LF-compound meaning 'self-criticize'. The obligatory distributive reading thus parallels that of They are self-criticizing. Since head-movement cannot go long-distance, a long-distance plural antecedent must be distributed by some other means. 15 If a logophor is not subject to Condition A, the question arises as to what a logophor is in the general typology of noun phrases: is it a pronoun, an r-expression, or something new? It is tempting to suggest that the logophoric ziji is simply a pronoun, obeying only Condition B that requires it to be free in its governing category. As just noted in the preceding paragraph, and for reasons that will become clear in Section 5, however, we shall take the logophor to be a variable bound by an operator which is itself predicated on an appropriate antecedent, i.e., akin to the null operator. 16 This means that in or within the embedded subject position, the reflexive bound by the immediately higher subject may have ambiguous status, much as the pronoun his in Pavarotti believes that his pants are on fire. 17 We exclude from consideration cases where ziji is used to express focus and/or contrast in discourse. They don't seem to be reducible to logophoricity and will not be covered in this work. 18 In contrast to (79), the following is acceptable: (i)
Zhangsan^ de baogao biaoshi tamen dui ziji i mei xinxin. Zhangsan DE report indicate they to ziji no confidence 'Zhangsan's report indicates that they had no confidence in self.'
This is because (i) implies that Zhangsan himself indicates that they had no confidence in him. (If his report indicates P, then he indicates P.) No similar implication holds of the unacceptable (79). 19 The illicit *John's arrogance harmed himself in English is ruled out by minimality, on a par with *John's mother loves himself, because in English reflexives are not inherently [+animate]. 20 In all fairness, we should mention that Xue et al. (and Ping Xue, personal communication) do cite examples with LD binding which exhibit no clear de se effects, but we found them to exhibit blocking effects quite generally. Since Mandarin LD ziji may take Pivot as an antecedent (i.e., be a Perspective logophor), the requirement of actual de se attitudes (sourcehood and consciousness) is not absolute. See discussion in Section 5.3. 21 So far, our argument for the GC as the dividing line has been based largely on
188
C.-T. James Huang and C.-S. Luther Liu
the distribution of certain logophoricity effects (especially blocking effects). Two other independent arguments for the dividing line are offered here. The first has to do with the distribution of reciprocals. As is well known, unlike reflexives, reciprocals are typically locally bound cross-linguistically. This is not surprising if true long-distance binding means logophoricity. Since reciprocals by definition do not refer to protagonists who ascribe a relevant property by the first-person pronoun, they cannot be logophors and hence, cannot be LD bound. However, reciprocals are admitted in many positions where they are locally bound in their GC, but not by a co-argument: (i) a. b. c. d. e.
Theyj heard several rumors about each other;. Theyi thought that rumors about each otheri spread too fast. Theyi thought that for rumors about each otheri to spread would be bad. They; thought that it would be bad for rumors about each other; to spread. Theyi arrived after rumors about each otheri had already spread.
Since reciprocals cannot be logophors, the reciprocals in these examples must be anaphors. This conclusion is, of course, compatible with our view that the GC defines the domain of anaphor binding. A second independent argument comes from a difference between local and LD ziji with respect to the distributive interpretation of their plural antecedents. As observed in Huang (1998b), both the local reflexive and the LD reflexive require their plural antecedents to be interpreted distributively. An important difference between local and LD ziji, however, is that whereas the LD antecedent requires the distributor dou 'all' or ge 'each', the local plural antecedent may distribute on its own without the aid of dou or ge. (ii) a. tameni (dou) zai piping ziji i they all at criticize self "They are all involved in the act of self-criticism.' b. tamen-i *(dou) yiwei Lisi hui piping ziji i they all think Lisi will criticize self 'Each of them thinks that Lisi will criticize him.' When ziji is embedded in a possessor position and is bound in its GC, note that no distributor is required: (iii) a. tameni taiqi-le ziji i -de gangqin. they lift-Perf self s piano 'They each lifted their (own) pianos.' b. tameni zhi gen zijii-de mama shuo-hua. they only with self's mother talk 'They only talked to their (respective) mothers.' This shows that the possessive reflexive in (iii) behaves on a par with a local anaphor. The facts shown in (i-iii) thus provide further evidence for taking the GC as the domain in anaphor binding, against Reinhart and Reuland's claim that only co-argument counts as local binding.
Logophoricity, Attitudes, and Ziji at the Interface
189
22
Some interesting properties of the local reflexive are treated in Huang (1998b). A fuller representation of the de se reading that parallels (87a) is (i):
23
(i)
Ax (believe (x, Ay (y's pants are on fire))) (P)
where ly (y's pants are on fire) is predicated on jc, hence, co-indexed with x. 24 Unlike claim and expect, the verb believe does not permit an obligatory control structure. Thus, there is no grammatical sentence *John believes PRO to be innocent on a par with (88a) representing the ambiguous reading. It is a grammatical, not a semantic, property peculiar to English believe (and some other verbs) that they permit Exceptional Case Marking but exclude PRO. Claim disallows ECM but permits PRO, whereas expect allows both. The ungrammaticality of *John believes PRO to be innocent does not preclude believe ( J , l x (innocent (x))) as an appropriate semantic representation for the de se reading of John believes he is innocent. 25 For the view that the bare reflexive ziji and its Japanese counterpart zibun are operators, see Katada (1991) and Huang (1993). In very much the same spirit, one may take Chierchia's assumption that a null operator is directly inserted to an IP-adjoined position, where it binds the reflexive as its variable: (i)
Zhangsan xiangxin [OP i [Lisi you zai piping ziji i le]]
It is generally agreed that null operator is the closest syntactic correlate of a l-operator. The reason why this is the case is that the null operator is itself anaphoric, itself needing to be bound (i.e., predicated on some term). From this point of view, it is entirely natural that the IP-adjoined ziji is also an appropriate syntactic source of the l-operator in semantic representation. One consideration that led Chierchia to the assumption of OP-insertion rather than LF-movement is that the binding relation is unconstrained by familiar island conditions. However, at least at a descriptive level, it has also been well known that LF-movement affecting arguments is exempt from island conditions (see Huang, 1982, etc.). Whatever the ultimate explanation of why this is the case (cf. Tsai, 1994, among others), it should be noted that normal familiar null-operator constructions (e.g., towg/i-constructions, some relatives) do exhibit island effects. One must, on a principled basis, ensure that such constructions are not generated by the insertion of a null operator. The idea represented by (95) is also in the spirit of Nishigauchi's (1999) proposal of a Point-of-View (POV) projection for constructions headed by the auxiliary te simaw in Japanese. Nishigauchi explicitly assimilates this proposal to the "Empty Topic" Phrase proposed in Huang (1984). In order to accommodate those cases involving Pivot antecedents, we shall, in fact, have to assume that the relevant phrase is the Pivot Phrase (or the POV Phrase). Another point of convergence that has come to our attention is Jayaseelan (1998). 26 The IP-adjunction assumed in (94) is entirely in the same spirit as the null-operator movement proposed in Huang (1984) for null pronouns that are discourse bound. 27 Part of Huang and Tang's claim concerning strict locality that is not preserved here is the idea of successive-cyclic IP adjunction. In their study, it is assumed that ziji is IP-adjoined successive-cyclically. This assumption was made for the purpose of
190
C.-T. James Huang and C.-S. Luther Liu
deriving the blocking effects. However, as they also note (see also Cole and Sung, 1994), the result was achieved only at additional technical costs. As we have seen, the blocking effects should be explained on pragmatic but not syntactic considerations. We therefore do not require the IP-adjunction of ziji to operate successive-cyclically. 28 Yet another aspect of the Huang and Tang analysis that is preserved here is the assumption that the relevant LF-movement that accounts for the LDR is a case of XP-movement, rather than head-movement of the sort advocated most forcefully by Cole, Hermon, and Sung (1990) and subsequent works. As we have seen, an XP-moved ziji is an A'-phrase that corresponds readily to the l.-operator in the semantic representation. According to the head-movement hypothesis, however, ziji does not occur at the A'-position of a proposition, but is moved to a position above the matrix verb: (i)
Zhangsan ziji i xiangxln [Lisi you zai piping ti le]] Zhangsan self believe Lisi again at criticize Perf
In such an LF representation, the movement chain does not identify the bracketed clause as the property of a self-ascriber Source or Self. Such an LF representation does not provide us with the desired input for translation into the semantic representation (96). One way to avoid this problem under the head-movement approach is to assume, as in Cole et al. (this volume), that ziji stops at the C of the CP under 'believe' but does not move up to the matrix I. The head-movement analysis (with movement to Infl above 'believe') was proposed originally to account for blocking effects by reducing them to agreement effects and, in part, to derive subject orientation. But if these considerations are not relevant, movement-to-C is a possible alternative. 29 Williams (1980) and Chierchia (1984) take obligatory control to be a case of predication. 30 Cole et al. (this volume) also point out that an LD ziji requires a c-commanding antecedent (generally), and they take this, together with the sloppy identity requirement, to mean that Mandarin LD reflexives are anaphors rather than pronouns. In our treatment, the anaphoric nature of the LD ziji follows from our taking the LF operator-variable construction (as in (94) and (96)) to be a structure of predication on a par with obligatory control and null-operator constructions. There is, of course, the case where the ziji operator is unbound, in which case the speaker is taken to be the "default" antecedent. (This could be reduced to a special case of local control under the performative analysis of declarative sentences.) 31 On the other hand, in line with Cole et al (1990) (also Lebeaux, 1983, Chomsky, 1986), we assume that head-movement is involved in the LF of locally bound ziji. See Huang (1998b) for considerations that led to this conclusion. Among other things, the LF head-movement analysis explains why the (locally bound) bare ziji permits only of distributive readings, unlike himself in English. It also explains subject-orientation, as has been pointed out by Cole et al. and others. Note that only the local ziji is truly subject-oriented in the syntactic sense, and this fact is explained by head-movement and the requirement of c-command (or subcommand). According to our logophoric analysis, the LD ziji is bound by an NP denoting the holder of a de
Logophoricity, Attitudes, and Ziji at the Interface
191
se belief. In the most typical cases, the holder of the de se belief is also the subject of a higher verb. But there are cases where this is not true. One of them is when the logophor is sentence-free, in which case the antecedent is the speaker, but not a syntactic subject. Another case is when the antecedent is the object of an experiencer verb that represents the Self, or an NP in a preceding discourse representing the Pivot. In these latter cases, there is no subject-orientation for the LDR. 32 An alternative to accommodate cases like (84) in syntax is to postulate a higher intentional predicate above the verb xihuan. This (light) predicate would denote the relevant mental attitude of its subject (cf. also note 25). This will not quite work with cases like (85), however, where the antecedent is an embedded subject. We are indebted to Gennaro Chierchia (personal communication) for discussion of related issues. 33 We thank Shizhe Huang, Audrey Li, and Yafei Li (personal communication) for the observation and useful discussion concerning (106)-(109). 34 Yafei Li (personal communication) noted that the following sentence is acceptable even if Lisi has not actually expressed a de se belief. (i)
ni kan, Lisi juran gan shuo Laoshi mei zhaogu ziji. you see, Lisi incredibly dare say teacher not care-for self 'Look, Lisi even has the nerve to say that the teacher did not care for him.'
The crucial part of the sentence that makes this acceptable is the speaker-oriented adverb 'incredibly'. The speaker of (i) takes it for granted that other listeners know that Lisi is talking about himself (even if unknowingly) and expresses his disbelief that Lisi makes such an unfair statement about the teacher. The speaker's point is that, given what we all know, Lisi is, in effect, virtually expressing a de se belief. We take this to be also a perspectivity effect, similar to the cases illustrated in the text. 35 Pan (1997) did not refer to these three earlier sources. He provided some critical comments on the perspectivity account presented in Kuno (1987), but did not discuss an earlier chapter of the same book devoted to direct discourse complementation. 36 Pan's contribution to this volume contains ideas adopted or revised from Pan (1997). Our discussion will be concerned with Pan (1997) primarily; we shall not discuss any point specific to Pan's current contribution. 37 See Chierchia (1989) for further discussion. This problem for Pan was pointed out in Huang (1998a). 38 An obligatory self-ascriber that occurs of the c-domain of an optional self-ascriber will not induce blocking. See Pan (1997) and his contribution in this volume on this point. 39 Pan (1997, p. 42) claimed that ziji could be used as a verb meaning 'self control', citing examples like (i): (i)
yi kandao nage qifen, jiu zaiye buneng ziji [sic] le. once see that atmosphere, then any-more cannot self Perf 'Once [he] sensed that atmosphere, [he] could no more control himself anymore.'
192
C.-T. James Huang and C.-S. Luther Liu
This claim clearly results from a misreading of the relevant Chinese orthography. The relevant character that means 'control' (actually, 'cease') is yi, not ji of ziji. The two characters are very similar in shape, but represent different morphemes. 40 Because we analyze all three subtypes of logophoricity in terms of self-ascription (explicit, implicit, and virtual) and attribute blocking effects to conflicts that arise in contexts of self-ascription, we also explain why all three subtypes of logophoricity exhibit blocking effects. 41 There are many questions that have remained unanswered. One of these is why the blocking effects are found only in Chinese, but not, or not as clearly observed, in other languages. In fact, we have been told that some other languages also exhibit similar, but milder, blocking effects, including Korean, Japanese, and Italian (for the use oiproprio). (Also, Malayalam, per Jayaseelan, 1998.) The fact that the blocking effects are mild and may be over ridden under special conditions goes along with the view that these are pragmatic, not grammatical, effects. There are, in fact, contexts in Chinese where the blocking effects may be suppressed. Two well-known examples are given below (from Yu (1992) and Xu (1993)): (i)
niiwang qing wo zuo zai ziji de shenbian. queen ask I sit at self DE side 'The queen asked me to sit by her side.'
(ii)
Zhangsan shen-pa wo chaoguo ziji. Zhangsan deep-fear I surpass self 'Zhangsan very much fears that I might surpass him.'
The question still remains why the blocking effects seem, in general, to be stronger in Mandarin than in most other languages. We tentatively surmise that this is due to the fact that Mandarin (and other Chinese dialects) do not have a real complementizer than can distinguish direct from indirect speech. Whether and how precisely this idea may work remains to be seen.
REFERENCES Baker, C. L. (1995). Contrast, discourse prominence, and intensification, with special reference to locally free reflexives in British English. Language 71,63-102. Battistella, E. (1989). Chinese reflexivization: A movement to INFL approach. Linguistics 27,987-1012. Bickerton, D. (1987). He himself: An anaphor, pronoun, or Linguistic Inquiry 18, 345-348. Chen, P. (1992). The reflexive ziji in Chinese: Functional vs. formalist approach. In T. H.-T. Lee (ed.), Research on Chinese Linguistics in Hong Kong: 1-36. Hong Kong: The Linguistics Society of Hong Kong. Chierchia, G. (1984). Topics in the Syntax and Semantics of Infinitives and Gerunds. Doctoral dissertation. University of Massachusetts, Amherst.
Logophoricity, Attitudes, and Ziji at the Interface
193
Chierchia, G. (1989). Anaphora and attitudes de se. In R. Bartsch, J van Benthera, and P. van Emde Boas (eds.), Semantics and Contextual Expression: 1-31. Dordrecht: Foris. Chomsky, N. (1981). Lectures on Government and Binding. Dordrecht: Foris. Chomsky, N. (1982). Some Concepts and Consequences of the Theory of Government and Binding. Cambridge, MA: The MIT Press. Chomsky, N. (1986). Knowledge of Language: Its Nature, Origin, and Use. New York: Praeger. Clements, G. N. (1975). The logophoric pronoun in Ewe: Its role in discourse. Journal of West African Languages 2,141-177. Cole, P. and Sung, L.-M. (1994). Head movement and long-distance reflexives. Linguistics Inquiry 25,355-406. Cole, P., and Wang, C. (1996). Antecedents and blockers of long-distance reflexives: The case of Chinese ziji. Liguistic Inquiry 27,357-390. Cole, P., Hermon, G., and Sung, L.-M. (1990). Principles and parameters of long-distance reflexives. Linguistics Inquiry 21,1-22. Hagege, C. (1974) Les pronouns logophoriques, Bulletin de la Societe de Linguistique de Paris 69,287-310. Huang, C.-T. J. (1982). Logical Relations in Chinese and the Theory of Grammar, MIT dissertation; also published by Garland Publishing, New York (1998) Huang, C.-T. J. (1983). A note on binding theory. Linguistic Inquiry 14,554-561. Huang, C.-T. J. (1984). On the distribution and reference of empty pronouns. Linguistic Inquiry 15,531-574. Huang, C.-T. J. (1993). Reconstruction and the structure of VP: some theoretical consequences. Linguistic Inquiry 24,103-138. Huang, C.-T. J. (1998a). The bare anaphor. Paper presented at IACL-7, Stanford University. Huang, C.-T. J. (1998b). Distributivity and Reflexivity. Unpublished manuscript, University of California, Irvine; to appear in C.-S. Liu and S.-W. Tang (eds.), On the Formal Way to Chinese Languages, Stanford: CSLI. Huang, C.-T. J. and Tang, C.-C. J. (1991). The local nature of the long-distance reflexives in Chinese. In J. Koster and E. Reuland, (eds.), Long-Distance Anaphor: 263-282. Cambridge University Press. Also in Proceedings ofNELS 19 (1989). Huang, C.-T. J., Huang, Y.-H., Teng, T.-H., and Tiedeman, R. (1984). Reflexives in Chinese and the teaching of Chinese. Proceedings of the First World Conference on Chinese Language: 205-215. Huang, S. (1996). Quantification and Predication in Mandarin Chinese: A Case Study o/Dou. Doctoral dissertation. University of Pennsylvania. Huang, Y.-H. (1984). Chinese reflexives. Studies in English Literature and Linguistics 10,163-188. National Taiwan Normal University, Taipei. Hyman, L., and Comrie, B. (1981). Logophoric reference in Gokana.Journal of African Languages and Linguistics 31,19-37. lida, M. and Sells, P. (1988). Discourse factors in the binding of Zibun. In William J. Poser(ed.), Papers from the Second International Workshop on Japanese Syntax, 23–46. Stanford: CSLI. Jayaseelan, K. A. (1998). Blocking effects and the syntax of Malayalam Taan. In
194
C.-T. James Huang and C.-S. Luther Liu
Rajendra Singh (ed.), The Yearbook of South Asian Languages and Linguistics 1998,11-27. Katada, F. (1991). The LF representation of anaphors. Linguistic Inquiry 22, 287-315. Kayne, R. (1994). The Antisymmetry of Syntax, Cambridge, MA: The MIT Press. Kuno, S. (1972). Pronominalization, reflexivization, and direct discourse. Linguistic Inquiry 3,161-195. Kuno, S. (1987). Functional Syntax. Chicago: University of Chicago Press. Lebeaux, D. (1983). A distributional difference between reciprocals and reflexives. Linguistic Inquiry 14,723-730. Lewis, D. (1979). Attitudes de dicto and de se. The Philosophical Review 88,513-543. Li, X. (1997). Deriving Distributivity in Mandarin Chinese. Doctoral dissertation. University of California, Irvine. Lin, J.-W. (1996). Polarity Licensing and Wh-Phrase Quantification in Chinese. Doctoral dissertation. University of Massachusetts, Amherst. Maling, J. (1984). Non-clause-bounded reflexives in modern Icelandic. Linguistics and Philosophy 7,211-241. Manzini, M. R., and Wexler, K. (1987). Parameters, binding theory, and learnability. Linguistic Inquiry 18,413–444. Nishigauchi, T. (1999). "Point of View" and Phrase Structure. Theoretical and Applied Linguistics at Kobe Shoin 2,49–60. Pan, H.-H. (1997). Constraints on Reflexivization in Mandarin Chinese. New York: Garland Publishing Inc. Pica, P. (1987). On the nature of the reflexivization cycle. In J. McDonough and B. Plunkett (eds.), Proceeding of NELS 17,483–499, GLSA, University of Massachusetts, Amherst. Pollard C. J., and Sag, I. A. (1992). Anaphors in English and the scope of binding theory. Linguistic Inquiry 23,261-303. Pollard C. J., and Xue, P. (1998). Chinese reflexive ziji: Syntactic reflexives vs. nonsyntactic reflexives. Journal of East Asian Linguistics 7,287-318. Progovac, L. (1992). Relativized SUBJECT: Long-distance reflexives without movement. Linguistic Inquiry 23,671–680. Reinhart, T, and Reuland, E. (1993). Reflexivity. Linguistic Inquiry 24,657–720. Rizzi, L. (1997). The fine structure of the left periphery. In L. Haegeman (ed.), Elements of Grammar, 281-338. Dordrecht, Netherlands: Kluwer Academic Publishers. Ross, J. (1970). On declarative sentences. In R. A. Jacobs and P. S. Rosenbaum (eds.), Readings in English Transformational Grammar. Waltham, MA: Ginn & Co. Sells, P. (1987). Aspects of logophoricity. Linguistic Inquiry 18,445–479. Tang, C.-C. J. (1989). Chinese reflexives. Natural Language and Linguistic Theory 7, 93-121. Travis, L. (1984). Parameters and Effects of Word Order Variation. Doctoral dissertation. MIT. Tsai, W.-T. D. (1994). On Economizing the Theory ofA-Bar Dependencies. Doctoral dissertation. MIT.
Logophoricity, Attitudes, and Ziji at the Interface
195
Wang, J.-L., and Stillings, J. T. (1984). Chinese reflexives. In C.-Y. Ning et al. (eds.), Proceedings of the First Harbin Conference on Generative Grammar. Heilongjiang University, Harbin, China. Williams, E. (1980). Predication. Linguistic Inquiry 11,203–238. Williams, E. (1989). The anaphoric nature of theta-roles. Linguistic Inquiry 20, 425-456. Williams, E. (1994). Thematic Structure in Syntax. Cambridge, MA: The MIT Press. Xu, L. (1993). The long-distance binding of ziji. Journal of Chinese Linguistics 21, 123-141. Xu, L. (1994). The antecedent of ziji. Journal of Chinese Linguistics 22,115–137. Xue, P., Pollard, C. J., and Sag, I. A. (1994). A new perspective on Chinese reflexive ziji. In Proceedings of the 13th West Coast Conference on Formal Linguistics, 432–447. CSLI Publications, Stanford University. Yang, D.-W. (1983). The extended binding theory of anaphors. Language Research 19,169–192. Yu, X.-F. W. (1992). Challenging Chinese reflexive data. The Linguistic Review 9, 285-294. Yu, X.-F. W. (1996). A Study of Chinese Reflexives. Doctoral dissertation, University of London. Zribi-Hertz, A. (1989). Anaphor binding and narrative point of view: English reflexive pronouns in sentence and discourse. Language 65,695-727.
This page intentionally left blank
LOCAL AND LONG-DISTANCE REFLEXIVES IN TURKISH JAKLIN KORNFILT Department of Languages, Literatures, and Linguistics Syracuse University Syracuse, New York
1. INTRODUCTION: TYPES OF MORPHOLOGICALLY FREE REFLEXIVES Turkish has two types of reflexives: a bound morpheme, which is a verbal suffix, and a free morpheme. I will be concerned here with the free morpheme. Its shape is kendi, and it is inflected for person and number, agreeing in these features with its antecedent. This is illustrated in the following examples: (1) Sen niye kendi -n -den nefret ed -iyor -sun? you (sg.) why self -2.sg.-Abl. hate do-Pres.Pr. -2.sg. 'Why do you hate yourself?' (2) Biz kendi -miz -e kitap at -a -mi -yor -uz we self -l.pl. -Dat. book buy -Abil. -Neg. -Pres.Pr.-l.pl. 'We are unable to buy books for ourselves.' As we shall see later on, however, the third-person singular can also show up in its bare form; the bare and the inflected forms of the third-person sinSyntax and Semantics, Volume 33 Long-Distance Reflexives
197
Copyright © 2001 by Academic Press All rights of reproduction in any form reserved. 0092-4563/00 $35.00
198
Jaklin Kornfilt
gular reflexive behave differently with respect to binding, as illustrated by the following examples. (3) a. Ahmeti kendin-ii cok begen -iyor -mus Ahmet self -Acc. very admire -Progr. -Rep.Past '(They say that) Ahmet admires himself very much.' b. Ahmeti kendi -sin -ii/j qok begen -iyor -mus Ahmet self -3.sg. -Acc. very admire -Progr. -Rep.Past '(They say that) Ahmet admires himself/him very much.' Both the bare and the inflected third-person singular reflexive differ in their local binding properties from the regular pronoun, an element which cannot have a local antecedent: c. Ahmet{ on -ui/j cok begen -iyor -mus Ahmet he -Acc. very admire -Progr. -Rep.Past '(They say that) Ahmet admires him very much.' Nonlocally, the bare reflexive and the pronominal differ, since the bare reflexive cannot be bound nonlocally, while the pronominal can; in turn, the pronominal cannot be locally bound: (4) a. Fatmai [Ahmed -ini kendin -ii/*j/*k cok Fatma Ahmet -Gen. self -Acc. very begen -dig -in] -i bil -iyor admire -Ger. -3.sg. -Acc. know -Pres.Progr. 'Fatmaj knows that Ahmetj admires selfi/j/k very much.' b. Fatmai [Ahmed -ini on -u*i/j/k cok Fatma Ahmet -Gen. (s)he-Acc. very begen -dig -in] -i bil -iyor admire -Ger. -3.sg. -Ace. know -Pres.Progr. 'Fatmaj knows that Ahmeti admires him/her*i/j/k very much.' On the other hand, the inflected reflexive and the pronominal appear to pattern together in their nonlocal binding properties (not, however, in their local binding properties): c. Fatmai [Ahmed-ini kendi -sin -ii/j/k cok Fatma Ahmet -Gen. self -3.sg. -Acc. very begen -dig -in] -i bil -iyor admire -Ger. -3.sg.-Acc. know -Pres.Progr. 'Fatmaj knows that Ahmet admires selfi/j/k very much.' While the uninflected reflexive (3a) in can only refer to 'Ahmet,' the inflected reflexive in (3b) can refer to either 'Ahmet' or to someone else mentioned in the discourse.
Local and Long-distance Reflexives in Turkish
199
For the purpose of this chapter, it is the behavior of these two types of morphologically free reflexives that I shall mainly be interested in, and, in particular, in the properties of the inflected reflexive. The syntactic properties of non-third-person reflexive forms will be touched upon only in passing. I claim that the inflected third-person reflexive in Turkish is actually a phrase in disguise—more specifically, it is an Agreement Phrase (AgrP) whose specifier is pro. That AgrP serves as the binding domain for both the reflexive and the pro. The reflexive is locally bound by pro, while pro is free within this domain and can therefore "pick-up" antecedents within the clause as well as outside of it, or even outside the utterance, i.e., in the discourse. I further claim that the bare reflexive is a genuine syntactic anaphor with local binding requirements, but that a stylistic level (which I call narrative style) exists, where the bare reflexive is used as a pronoun of empathy or point of view.1 Furthermore, I briefly illustrate and discuss complex, reduplicated reflexives which are strictly local for all speakers and in all stylistic levels, and I claim that this strict locality is due to the analysis of such reflexives, previously proposed in the literature for similar elements in other languages, as phrases which cannot undergo head movement at LF. The chapter is organized as follows: The first section introduces the two main types of morphologically free reflexives. Section 2 discusses the differences between these two types. In the third section, the question is raised about the nature of the inflected reflexive. Two possibilities are considered: (1) that this element might be a LD-reflexive and (2) that this element might be a pronominal. The properties of the (at first glance) hybrid nature of the inflected reflexive are presented and discussed in separate subsections, and it is concluded that neither an extension of the binding domain for the inflected reflexive nor an analysis of that element as a LD-reflexive is an appropriate solution. Section 4 discusses various approaches to account for that hybrid nature. I discuss my own account in some detail, arguing for a phrasal analysis of the inflected reflexive. Section 5 discusses the properties of bare reflexives and distinguishes between a narrative stylistic level, where these elements are pronouns of empathy, versus a standard stylistic level, where the same elements are genuine syntactic, local anaphors. This characterization is further confirmed in subsection 5.2, where the bare and inflected reflexives and the overt pronoun are contrasted once again. In 5.3, some apparent counterexamples to the characterization of the bare reflexive as a locally bound anaphor are discussed, and the problem is resolved by proposing that possessor raising applies in the context of certain static predicates and certain unaccusatives. Section 6 illustrates complex, reduplicated reflexives and shows that they are strictly local. Section 7 offers the conclusions of the chapter.
200
Jaklin Kornfilt
2. DIFFERENCES BETWEEN THE BARE AND THE INFLECTED REFLEXIVE For a majority dialect, the bare reflexive is locally bound in regular, nonnarrative style, as illustrated in (3). We shall talk about narrative style and some of its properties later on. By being locally2 bound, the bare reflexive in (3) is well-behaved syntactically: it obeys Binding Condition A. However, the inflected reflexive in (4) is problematic; we saw that while it can be understood as co-referential with a local antecedent, it does not have to be, since it can also refer to a nonlocal antecedent, indeed even to a discourse antecedent which is not part of the utterance. Thus, (3b) is well-formed in a discourse of the following sort: (5)A. Alii hakkmda Ahmet ne dusun-uyor? Ali about Ahmet what think -Progr. 'What does Ahmet think of Ali?' B.
Ahmet kendi -sin -ij cok begen -iyor -mus Ahmet self -3.sg. -Acc. very admire -Progr. -Rep.Past. '(They say that) Ahmet admires him (i.e., Ali) very much.'
Not too surprisingly, the uninflected third-person reflexive would be ill-formed here, since it requires a local antecedent; therefore, B' is ill-formed as a reply to A, in contrast to the well-formed B; in B', the assignment of index i to the bare reflexive is grammatical but ill-formed in the discourse, since A is a question about Ali, not about Ahmet. Assignment of index;, on the other hand, is ungrammatical, due to the requirement that the bare reflexive be bound locally: B'. Ahmeti kendin-i*j/i cok begen -iyor -mus Ahmet self -Acc. very admire -Progr. -Rep.Past. Intended reading: '(They say that) Ahmet admires him (i.e., Ali) very much.' The reflexive can, of course, refer to the local antecedent 'Ahmet'; this reading, however, is non-sensical in this particular discourse. It is clear, then, that the reflexive which is inflected for third person is not a local anaphor. On the other hand, it is not a regular syntactic pronominal either, as I argue in the next section.
Local and Long-distance Reflexives in Turkish
201
3. THE NATURE OF THE INFLECTED REFLEXIVE: A LD-REFLEXIVE OR A PRONOUN? Offhand, the idea that a special form of the reflexive might actually be a syntactic pronoun in disguise is a reasonable one. As a matter of fact, it is claimed in some traditional grammars of Turkish that this form of the reflexive is a pronoun. Thus, the following quotation is typical of such claims: kendisi and its plural kendileri are commonly employed as simple third-person pronouns with no reflexive or emphatic sense: kendisi evde 'he is at home'; kendilerini gordiiniiz mil 'have you seen them?' (Lewis, 1985:71) However, if these elements were syntactic pronominals, they should be disjoint from a local antecedent. Yet, as the coindexation in (3b) shows, local binding indeed provides one of the possible interpretations for inflected third-person reflexives. Clearly, local binding is ruled out for genuine, syntactic pronominals, due to Condition B, as evidenced by the "regular" personal pronouns; (3c) illustrates this point and is repeated here as (6): (6) Ahmeti on -u*i/j cok begen -iyor -mus Ahmet he -Acc. very admire -Progr. -Rep.Past. '(They say that) Ahmet; admires him*i/j very much.' Not only do personal pronouns behave in predictable fashion locally, their syntactic behavior is predictable nonlocally, as well. In other words, these elements must be free locally, but they can be bound nonlocally. The nonlocal binder may be part of the utterance, as illustrated in (4b); (7) makes the same point: (7) Ahmeti [AH -nini on -ui/*j/k cok begen -dig -in] -i Ahmet Ali -Gen. he -Acc. very admire -Ger. -3.sg. -Acc. bil -iyor know -Progr. 'Ahmeti, knows that Alii admires himi/*j/k very much.' The nonlocal antecedent of the personal pronoun can also be a discourse antecedent. Thus, in the answer segment of the discourse in (5), a personal pronoun can be used instead of the inflected reflexive: (5)'A. Alii hakkmda Ahmet ne dusun-tiyor? Ali about Ahmet what think -Progr. 'What does Ahmet think of Ali?'
202
B.
Jaklin Kornfilt
Ahmet on -ui cok begen -iyor -mus Ahmet he -Acc. very admire -Progr. -Rep.Past. '(They say that) Ahmet admires him (i.e. Ali) very much.'
In turn, an inflected reflexive can be used in the place of a personal pronoun in examples like (7), where a non-local antecedent is present in the utterance; this was illustrated in (4) and is further illustrated in (7)': (7)' Ahmeti [Ali -ninj kendi -sin -ii/j/k cok Ahmet Ali-Gen. self -3.sg.-Ace. very begen -dig -in] -i bil -iyor admire -Ger. 3.sg. -Ace. know -Progr. 'Ahmetj knows that Alij admires himi/j/k very much.' Descriptively speaking, we see that the inflected third-person reflexive is in complementary distribution with the corresponding personal pronoun in local contexts, while it appears to be in free variation with such a pronoun in nonlocal contexts. This descriptive fact is very difficult to capture in formal terms, for obvious reasons: If a given element is a pronominal, it has to obey Binding Condition B, which requires that the element be free in a specified domain. If that element is an anaphor, on the other hand, it has to obey Condition A, which requires that element to be bound in its binding domain (which is, in most accounts, the same domain which is relevant for Condition B, as well). The distribution of the inflected reflexive as just stated is obviously paradoxical, as long as we make the reasonable (and generally accepted) assumption that any given element has a fixed feature composition which is the same in all syntactic contexts. The inflected third-person reflexive, however, appears to have a dual nature in this respect: locally, it seems to act as an anaphor, while nonlocally, it seems to act as a pronominal. 3.1. The Binding Domain of Inflected Reflexives One initial way out of this paradox might be to claim that the locality domain for binding is different for pronominals and for anaphors—or, more specifically, for certain anaphors. We saw that the uninflected as well as the inflected reflexive are, as predicted, in complementary distribution with personal pronouns in a local binding domain consisting of the minimal maximal projection that contains them and which is headed by strong AGR. Therefore, it would be misguided to give up our definition of this binding domain, at least for the uninflected reflexive and for the personal pronouns. On the other hand, it might be the case that the inflected third-person reflexive requires a different, more extensive binding domain. Such an approach is conceptually not very appealing, because it would weaken the
Local and Long-distance Reflexives in Turkish
203
explanatory power of a domain-based notion of binding that is valid for both anaphors and pronominals, while this extended binding domain would be relevant for these inflected reflexives only and would not have anything to say about any pronominal counterparts. Nevertheless, it would be useful to see if such a move introduces any insights into the syntactic nature of these inflected third-person reflexives.
3.2. Properties of LD-reflexives One way out of this paradox might be to claim that the inflected third-person reflexive is a Long Distance (LD) reflexive, i.e., an anaphor which must be bound, but whose antecedent need not be contained within a limited binding domain. This would explain why these elements do behave like local anaphors, while being able to be bound long-distance, as well. However, for this characterization not to be empty in the sense of simply giving a name to a puzzling fact, we must have clear differences between bound pronouns and LD-reflexives. I turn therefore to a discussion of LD-reflexives and their properties. Two widely mentioned properties of LD-reflexives in the languages which have been claimed to have such elements are linked to each other in at least one account, namely, one based on LF-movement of anaphors: (1) LD-reflexives are subject-oriented (i.e. their binders must be subjects in their respective clauses); (2) LD-reflexives are monomorphemic—an observation due to Pica (1987). This latter property is seen clearly in languages which have both local and LD-reflexives. In such languages, the local anaphors are polymorphemic, while the LD-reflexives are monomorphemic. The account in question (cf. Cole, Hermon, and Sung, 1990; Cole and Sung, 1994; Cole and Wang, 1996; Hermon, 1994) posits head movement of the LD-reflexive in LF; this movement would first target the INFL or AGR of the reflexive's own clause and would go from there to the higher INFL/AGR nodes, until it reaches the INFL/AGR of the matrix clause. The Binding Conditions are applied at LF, at which point the reflexive is locally bound. Since, in most instances, the only constituent which c-commands INFL/AGR (and thus the LF-moved reflexive) is the subject, the property of subject orientation observed with LD-reflexives becomes a consequence of this account. Another welcome consequence is the monomorphemic nature of LD-reflexives, since heads are monomorphemic. Polymorphemic reflexives, on the other hand, are viewed as nonheads, i.e., as phrasal elements. As such, they cannot undergo head movement, and therefore they cannot function as LD-reflexives; rather, they are local anaphors—a dichotomy first posited by Pica (1987).
204
Jaklin Kornfilt
3.3. The Turkish Inflected Reflexive is Not a LD-reflexive Are the inflected third-person reflexives in Turkish LD-reflexives, then? If they are, they would pose a challenge to the generalization in Pica (1987), since they are clearly polymorphemic. Perhaps we can get around this problem by hypothesizing that the relevant notion of being polymorphemic is not one of inflection, but one of compounding. As a matter of fact, most of the polymorphemic and locally bound anaphors mentioned in the literature consist of a sequence of free morphemes, forming compound-like entities, rather than consisting of inflected single morphemes. It is more difficult, however, to get around the fact that the Turkish inflected third-person reflexives are not subject-oriented. Examples like (5), where, in the discourse portion of B, the inflected reflexive refers to an adjunct in A, illustrate the lack of subject-orientation very clearly. To maintain that these elements are LD-reflexives without exhibiting one of the most typical properties of such elements found in other languages would come close to emptying this characterization of any empirical content. Note, furthermore, that in the exchange of (5), the antecedent of the inflected reflexive is not even part of the same sentence. In other words, while the binder of the inflected reflexive is a discourse-antecedent, it is not a genuine syntactic antecedent of that inflected reflexive in the strict sense. Thus, the strongest condition for binding, namely c-command of the bound element by its antecedent, would obviously not hold here either. (Note also that the discourse-antecedent in this example is an adjunct.) Moreover, the lack of subject-orientation and of c-command can also be seen in examples where the antecedent of the inflected reflexive is syntactic: (8)
[Oya -nm kendi -sin -ii begen -me -si] Ahmed-ini Oya -Gen. self -3.sg. -Ace. admire -Ger. -3.sg. Ahmet-Gen. hos -un -a git-ti liking -3.sg. -Dat. go-Past 'Oya's admiring him was to Ahmet's liking.'
(9) a. [Oya -nm kendi -sin -ii begen -dig -i] Ahmet -cei Oya -Gen. self -3.sg. -Acc. admire-Ger.-3.sg. Ahmet-by bil -in -iyor -du know -Pass. -Progr. -Past 'Oya's admiring him was known to Ahmet.' The fact that these examples involve a psych-verb and a passive, respectively, might be interpreted to mean that the antecedent does c-command the inflected reflexive at some level of representation. Other examples where the antecedent does not c-command the inflected reflexive might be
Local and Long-distance Reflexives in Turkish
205
explained away by appealing to scrambling. But such a claim would be more difficult to maintain in examples like (9b) with an object orientation, where the matrix oblique (dative) object is a possible antecedent of the inflected relfexive (and, in certain discourse contexts, might even be the only possible one): b. Alii Ahmed -ei [Selim -ink kendi -sin -ii/j/k cok Ali Ahmet -Dat. Selim -Gen. self -3.sg. -Acc.very begen -dig -in] -i soyle -di admire-Ger.-3.sg.-Acc. say -Past 'Alij told Ahmetj that Selimk admires him(self)i/j/k very much.' Summing up, lack of subject-orientation remains a problem. Together with examples like (5) where the antecedent is provided by the discourse alone, all of these examples conspire to show that the inflected third-person reflexive in Turkish is not subject-oriented and that it does not have to be c-commanded by its antecedent. We said previously that the inflected reflexive behaves like a pronominal in its nonlocal domain. Yet another property otherwise found with pronominals and not with reflexives is the ability of allowing for split antecedents; this, too, is a property we find with the Turkish inflected third-person reflexives: (10) Alii Oya-yai [Ankara-ya kendi -lerin -ini+j Ali Oya-Dat. Ankara-Dat. self -3.pl. -Gen. gonder-il -eceg -in ] -i soyle -di send -Pass. -fut. -3.sg. -Acc. tell -Past 'Ali told Oya that THEY3 would get sent to Ankara.' Summing up, then, we see that the inflected third-person reflexives in Turkish are neither genuine anaphors nor LD-reflexives.4 While they do appear to have anaphoric properties, this is confined to local domains. Outside of that domain, they exhibit pronominal properties. We are back, then, to our initial dilemma of how to best give a formal account of one and the same element's different binding properties according to its binding domain.
4. ELEMENTS OF A THIRD (OR FOURTH?) KIND 4.1. Underdetermined Binding Features Before offering my own solution to this dilemma, I would like to discuss one proposal for analyzing an element with similar binding properties in an-
206
Jaklin Kornfilt
other language, namely, Singapore Malay. Cole and Hermon (1998) discuss the element dirinya. This, too, is an element which behaves like an anaphor locally and as a pronominal nonlocally. Cole and Hermon treat this as an element whose features with respect to Binding Theory are underdetermined. The element dirinya takes on the values of an anaphor, i.e., [+ anaphoric, -pronominal], in a local domain, i.e., if its antecedent is contained within its local domain. If its antecedent is outside of the local binding domain of dirinya, the element takes on the values of a pronominal, i.e., it becomes [—anaphoric, + pronominal]. This proposal would, by and large, account for the Turkish facts. However, I would like to propose a different account—not for dirinya, which is, incidentally, also inflected (diri + nya 'self + 3rd poss.'), but for the Turkish counterpart. (The question of whether my account for the Turkish inflected reflexive can carry over to dirinya is left open here, but future research on this topic should be conducted.) This alternative account is discussed in the following section.
4.2. The Inflected Reflexive: A Phrase in Disguise My proposal for solving the problem posed by the different properties of the Turkish inflected reflexive in local versus nonlocal domains is tied to the overt and strong nature of the inflectional element on the reflexive. By strong, I mean the fact that this agreement inflection is part of a paradigm in which every combination of the relevant j-features (which are, for Turkish, person and number) has a distinct overt expression, i.e., the same notion of strength which is implicated in the identification of pro. I assume here that the overt agreement element is the head of an Agreement Phrase (AgrP) whose specifier is the phonologically empty pronominal pro. The reflexive is locally bound within the AgrP, and its binder is the pro. This is in parallel to possessive DPs which are also headed by an overt Agr inflection and can have a variety of specifiers, with pro being one of the possibilities. A few representative examples for such possessive phrases, all of which I analyze as AgrPs,5 follow: (11) Ali -nin araba -si Ali -Gen. car -3.sg. 'Ali's car' (12) on -un araba -si s/he -Gen. car -3.sg. 'His/her car'
Local and Long-distance Reflexives in Turkish
207
(13) pro araba -si car -3.sg. '[His/her] car' I claim, then, that the structure of the inflected third-person reflexive is similar to that of the possessive phrase in (13): (14) pro kendi -si self -3.sg. '[His/her] self (i.e. himself/herself)' The binding domain relevant for both the reflexive and the pronominal in (14) is the AgrP itself, since it is headed by a strong Agr element. In making this claim, I follow Kornfilt (1984,1987,1988) and related work, in which this claim included embedded clauses and possessive phrases. Here, I am extending it to the inflected reflexive element. As mentioned, I assume that the reflexive is locally bound, in conformity with Condition A, within the AgrP. The pronominal specifier is obviously free within that same domain; thus, Condition B is not violated.6 The antecedent of the pronominal specifier (i.e., of pro), if that pronominal is bound, can be in the same clause, but it can also be outside the clause, or even in the discourse. It is in the nature of a pronominal to be able to pick up binders in a variety of domains. What looks, then, like local binding of the inflected reflexive in examples like (4) (under the coreferential reading with a "clause-mate"), is actually one of the various possibilities for the pronominal to acquire a binder outside of its own local binding domain, i.e., outside of the AgrP that includes the reflexive.7 In such a constellation, i.e., one where the inflected reflexive appears to be locally bound within the clause (I would like to call this apparent binding a sort of "indirect binding," since it would be a consequence of the fact that it is actually the pronominal that is bound by the "clause-mate"—but not "domain-mate"—antecedent), it appears that the inflected third-person reflexive and the bare, uninflected reflexive are in free variation: Direct binding of the bare reflexive: (15) a. Ahmeti bu kitab -i kendin-ei al -di Ahmet this book -Acc. self -Dat. buy -Past Ahmet bought this book for himself.' (16) a. Ahmeti ayna -da kendin-ii gor-dii Ahmet mirror -Dat. self -Acc. see-Past 'Ahmet saw himself in the mirror.' Indirect binding of the inflected reflexive:8
208
Jaklin Kornfilt
(15) b. Ahmeti bu kitab-i [pro, kendi -sin -ej al -di Ahmet this book-Acc, self -3.sg. -Dat. buy -Past Ahmet bought this book for himself.' (16) b. Ahmeti ayna -da [proi kendi -sin -ij] gor -du Ahmet mirror -Dat. self -3.sg. -Acc. see -Past Ahmet saw himself in the mirror.' However, while the binding facts are the same for the a.-sentences and the b.- sentences, it is not fully correct to say that these examples are in free variation. In such situations, the bare reflexive is preferred by most speakers. This might be due to a "squishy" preference of the pronominal to find its binder not "too close" in the syntactic structure. What is important for our purposes at this point is that the coreferential reading with the "local" antecedent is possible for the inflected reflexive as well as for the bare reflexive. However, in the account here, this similarity is superficial. The coreference of the bare reflexive with the clause-mate binder is due to Condition A, while the well-formed binding of the inflected reflexive is an artifact; it is its pronominal "possessor" which happens to be bound by that same antecedent. As we said before, this does not lead to a violation of Condition B, since the relevant binding domain for this pronominal is its AgrP, which "protects" it from such a violation. Note that this analysis predicts straightforwardly and without taking recourse to any ad hoc stipulations that the bare reflexive cannot be bound by an antecedent outside its binding domain, namely, the sentence, while the inflected reflexive can be so bound: Bare reflexives: (15) c. *Ahmeti bu kitab -i kendin-ej al -di Ahmet this book-Acc. self -Dat. buy -Past Intended reading: Ahmeti bought this book for himj (i.e., someone mentioned in the discourse).' (16) c. *Ahmeti ayna -da kendin-ij gor-du Ahmet mirror-Dat. self -Acc. see-Past Intended reading: 'Ahmeti saw himj (i.e., someone mentioned in the discourse) in the mirror.' Inflected reflexives: (15) d. Ahmeti bu kitab-i [proj kendi -sin -ej al -di Ahmet this book-Acc. self -3.sg. -Dat. buy -Past 'Ahmetj bought this book for hinij (i.e., someone mentioned in the discourse).'
Local and Long-distance Reflexives in Turkish
209
(16) d. Ahmeti ayna -da [pro kendi -sin -ij gor-du Ahmet mirror-Dat. self -3.sg. -Acc. see-Past 'Ahmetj saw himj (i.e., someone mentioned in the discourse) in the mirror.' The ungrammatically of the examples in (15c) and (16c) is due to violations of Condition A, while the grammaticality of (15d) and (16d) is due to the fact that pro, the pronominal specifier of the AgrP, can pick up an antecedent anywhere, as long as that antecedent is outside of the AgrP. Thus, it is that pronominal that determines the distribution of the inflected reflexive, and all the pronominal properties of the inflected reflexive are thus predicted.9 4.3. Empirical Support for the Phrasal Analysis: Other Possessive Phrases Additional empirical support for preferring this analysis of the Turkish inflected reflexive over one that posits underdetermined binding features is due to the fact that the inflected reflexive can also show up with an overt "possessor": (17) [Ahmed -ini kendi -sin -ii gor-du -m Ahmet -Gen. self -3.sg. -Acc. see-Past-l.sg. 'I saw Ahmet himself.' (18) kitab-i [Ahmed-in i kendi -sin -ei ver -di -m book-Ace. Ahmet -Gen. self -3.sg. -Dat. give-Past -l.sg. 'I gave the book to Ahmet himself.' The binding facts concerning the inflected reflexive here are just as I posited them before: the reflexive is bound locally, by the specifier of the AgrP. The only difference between these examples and all the previous examples involving inflected reflexives is that here, the specifier of the AgrP is an R-expression, while in the previous examples, that specifier was a pronominal, i.e.,pro. What is important, however, is that the phrasal structure I posited for the inflected reflexive does, in fact, exist overtly. One interesting question that arises at this point in the discussion is whether the specifier of such AgrPs can be an overt pronoun. Such examples are possible, but they are marginal: (19) ??kitab -i [on -uni kendi -sin -ei ver -di -m book -Acc. he -Gen. self -3.sg. -Dat. give-Past -l.sg. 'I gave the book to him himself.' The reason the overt pronoun is awkward and marginal here (while not being totally ungrammatical) is due to something like an "Avoid Pronoun"
210
Jaklin Kornfilt
principle of the kind first proposed in Chomsky (1981) and later adapted to Turkish in Kornfilt (1984,1987,1988, and 1991). This is not a fully formal principle, but rather a functional one; it says that in positions where a phonologically empty pronominal, i.e.,pro, is licensed, an overt pronominal will either be ungrammatical (or disjoint in reference from a potential antecedent) or, at best, marginal. As a matter of fact, this marginality of the overt pronominal specifier deteriorates to full ungrammaticality in smaller domains: (20) *Ahmeti kitab-i [on -uni kendi -sin -ei al -di Ahmet book-Acc, he -Gen. self -3.sg. -Dat. buy -Past Intended reading: 'Ahmeti bought the book for himi himselfi.' The domain that includes the inflected reflexive, its pronominal specifier, and the antecedent of the pronominal is the sentence in (20); thus, it is smaller than the corresponding domain in (19), where the antecedent of the pronominal specifier is in the discourse. This deterioration in acceptability, i.e., the difference between the marginality of (19) and the full ungrammaticality of (20), may be interpreted as a problem for the analysis proposed here, since in both examples, the pronominal should be bound from outside its local binding domain which, I claim, is the AgrP and not the sentence. Therefore, whatever the explanation is for the ungrammaticality of (20), that same factor should also render (19) ungrammatical. The fact that the two differ might suggest that the ungrammaticality of (20) is due to a violation of Binding Condition B, under the assumption that the AgrP does not protect the pronoun and that the relevant binding domain is the sentence rather than the AgrP. I would like to suggest here that the problem is only apparent. First of all, the difference in acceptability between (19) and (20) is small. Second, and more importantly, the distinctions we see here between overt pronouns and pro in terms of their binding possibilities in various domains when either type of pronominal is "in construction with" a reflexive can also be found when the two pronominal types are "in construction with" other elements. Note the following contrasts and compare them to those just discussed: (21) *Ahmeti kitab -i [on -uni baba -sin -ai al -di Ahmet book-Acc. he -Gen. father -3.sg. -Dat. buy-Past Intended reading: 'Ahmetj bought the book for hisi father.' This example would be acceptable under a reading where the overt pronoun and Ahmet are disjoint and where the pronoun has a discourse antecedent. This, then, is similar to the contrast between the examples involving overt pronouns with "agreeing" reflexives. Note also that the reading intended in (21) is expressed by the corre-
Local and Long-distance Reflexives in Turkish
211
spending construction where the specifier of the AgrP, i.e., the "possessor," is nonovert, i.e., is pro: (22) Ahmeti kitab-i [proi baba -sin -a] al -di Ahmet book-Acc. father -3.sg. -Dat. buy -Past 'Ahmeti bought the book for hisi father.' This example, then, would completely correspond to those with inflected reflexives. Furthermore, just like those examples, (22) is also open to an interpretation where the empty pronominal can be bound by an antecedent further removed or nonsyntactic, i.e., in the discourse. This, too, is a reading found with the inflected reflexive, as we saw previously. What, then, explains the contrast between (21) and (22)? It cannot be Condition B, since the examples are completely parallel; they both have benefactive Dative AgrPs with pronominal specifiers. The only difference is that in (21), that specifier is overt, while in (22), it is nonovert, i.e.,pro. If we make the assumption, as I have been making here, that it is pro which is the representative pronominal for the purposes of Binding Theory whenever it is licensed, rather than the overt pronoun (while the latter is the representative pronominal when the former is not licensed), we arrive at the result that the grammaticality of (22) is expected, if we make the additional assumption—again made throughout this chapter—that the smallest AgrP is the relevant binding domain for both pronouns and anaphors contained in the AgrP. If this is correct, we have to explain the ungrammaticality of (21) under the intended co-referential reading, since it cannot be explained by Binding Condition B. Again, I claim that this is due to the Avoid Pronoun Principle. Thus, the contrast between (21) and (22) is entirely parallel to the one between (20) and (15b), repeated here for the reader's convenience: (20) *Ahmeti kitab-i [on -uni kendi -sin -ei al -di Ahmet book-Acc. he -Gen. self -3.sg. -Dat. buy -Past Intended reading: Ahmet; bought the book for him; himselfi.' (15) b. Ahmeti bu kitab-i [proi kendi -sin -ei] al -di Ahmet this book-Acc, self -3.sg. -Dat. buy -Past 'Ahmet bought this book for himself.' The issue of how best to account for the differences in distribution between overt pronouns and pro in the language at large is a complex one; in addition to my work mentioned earlier (Kornfilt 1984,1987,1988,1991), where this issue is addressed from the point of view of Binding Theory and binding domains, there is work that addresses this question in more general terms, e.g., Erguvanli-Taylan (1986) and Enc (1986). To include a full discussion of
212
Jaklin Kornfilt
this issue here is impossible, due to space restrictions, and it is tangential to our main concerns. The important point here is that differences exist between overt third-person pronouns and pro in the language at large and they are not confined to the position of the "possessor" of the inflected reflexive, and that these differences are similar across the board. My conclusion at this point, then, is as follows: In languages like Turkish which have a reflexive which appears to be locally bound yet nonlocally free and which displays properties different from those of genuine LD-reflexives, and where this element is inflected, the best analysis is one which analyzes these elements to be "in construction with" a pronominal specifier, where this pronominal is pro (provided, of course, that the inflection on the reflexive is of the kind which generally licenses pro in that language). This analysis enables us to treat the inflected reflexive as just a special instantiation of AgrPs which are DPs with pronominal specifiers. The analysis captures the pronominal properties of the inflected reflexive as well as the similarities between that element's binding properties and those of pro, in contrast to those of the overt regular personal pronouns. Furthermore, this analysis also enables us to continue viewing the reflexive itself as a locally bound anaphor. I leave the question open here of whether in languages like Singapore Malay, where the apparent (but not genuine) LD-reflexive is inflected as the Turkish element under discussion here, an analysis such as the one in Cole and Hermon's (1998) account, which posits an underlying undeterminacy of binding features is the more appropriate one.10
5. THE BARE REFLEXIVE: ANAPHOR OR EMPHATIC PRONOUN? 5.1. Differences between Narrative and Standard Styles If this is so, let us turn now to the bare reflexive in Turkish and discuss to what extent it is indeed a locally bound anaphor, since I just claimed that preservation of this property for it is an advantage of my analysis for the inflected reflexive (which, under that analysis, crucially consists of the bare reflexive, together with the inflection viewed as the head of the AgrP, i.e., of the binding domain relevant for both the reflexive itself and the pro it is in construction with). It has been claimed in some of the literature on this element that it is not a genuine syntactic anaphor. For example, in Sezer (1979-1980), it is claimed that the bare reflexive is a special element which expresses "the speaker's
Local and Long-distance Reflexives in Turkish
213
attitute towards the person(s) he is describing, rather than by the syntactic configurations that contain coreferential Noun Phrases" (Sezer, 1979-1980: 748) Some of the evidence he presents to strengthen his claim comes from literature (in the sense of belles lettres); a representative passage follows: (23) Lamiai yuzbasinm elini, kolunu sallayarak birgeyler anlattigmi, binbasmm arasira basini cevirip vagona baktigini goruyordu. Genc kiZi onlarm ne konustuklarmi isitmedigi halde kendindeni bahsettiklerini zannediyor, Makbule'ye soyleyecegi sozleri sasmyordu. (R. N. Giintekin, Dudaktan kalbe [From the lip to the heart], as cited in Sezer, 1979-1980:750, without page reference to the novel) 'Lamiaj saw that the captain was telling something while gesticulating, and that the major was turning his head from time to time and was looking at the train carriage. The young girlj, even though shei, did not hear what they were saying, was thinking that they were talking about her(self)i and was mixing up the words that she was saying to Makbule.' The bare "reflexive" here seems to express a certain point of view, namely Lamia's own. The inflected reflexive is always fine in these same contexts, but appears to reflect a more objective point of view, i.e., a point of view "from the outside," that of the speaker, narrator, writer, when s/he does not put herself or himself into the place of one of the participants in the narration. In narrative style, then, it might be true that Turkish has no form which is confined to genuine reflexives exclusively. We find the extension of this narrative style in some idiolects to regular, colloquial discourse. However, the inflected reflexive is always preferred to the bare reflexive in the function of a logophor in colloquial style. As for the "regular," standard, nonliterary, written style, i.e., the style which is, in general, the subject of formal syntactic analysis, the bare reflexive does have the properties imputed to it in general, i.e., it is indeed a locally bound anaphor. 5.2. Second Round: Contrasting the Bare and the Inflected Reflexives and the Overt Pronoun The characterization of the bare reflexive as a local anaphor is shared (albeit implicitly) by some Turkish grammarians; the following example and discussion in Sinanoglu (1967) is revealing in this respect. The following example is taken by Sinanoglu from Deny's (1921) grammar; what's interesting is the author's interpretation of the example and his
214
Jaklin Kornfilt
explanation of the usage of the inflected reflexive in this construction, since he contrasts this usage with that of the bare anaphor, on the one hand, and that of the regular personal pronoun, on the other: (24) [proi] Hiss -ed-er -di [ki [bu kadin kendi -sin -ii he feeling -do-Aor. -Past that this woman self -3.sg. -Acc. sev -mi -yor ]] love - Neg. -Progr. 'Hei felt that this woman does not love himi(self).' (Sinanoglu, 1967:207, without number, and without page attribution to Deny, 1921) "Bu ciimlede baska bir zamirin kullamlamayacagi apaqiktir; kendini dense, yan cumlecigin oznesi yani kadin anlasilacak, onu dense, hissedenden ve kadindan baska bir kimse kastedilmis olacaktir." 'It is totally clear that no other anaphoric form can be used in this sentence; if kendini (i.e., the bare reflexive—J.K.) were used, one would understand it to refer to the subject of the subordinate clause, i.e., the woman, and if onu (i.e., the personal pronoun—J.K.) were used, the person referred to would have been different from the one who feels as well as from the woman.' This, then, shows that the inflected reflexive is, in fact, different from the bare reflexive, and that the latter must be locally bound here, while it cannot be bound by the subject of the matrix clause. Yet, if the bare reflexive were uniquely ruled by discourse considerations such as empathy and point of view, it should also have been possible to interpret it as coreferential with the subject of the matrix clause. At the same time, the writer makes clear that the inflected reflexive is different from the overt personal pronoun. In this chapter, I have attributed this difference to the more general issue of the differences between the overt pronoun and pro, the nonovert pronoun. The pro which is the specifier of the inflected reflexive would be bound by the matrix subject—a perfectly expected reading. This pro could actually also be bound by a discourse antecedent, if such an antecedent were provided (thus resulting, indirectly, in the interpretation of the inflected reflexive as discourse-bound—a possibility not mentioned by Sinanoglu). But Sinanoglu is right that if a personal pronoun replaces the inflected reflexive, there is a strong tendency to interpret it as bound by a discourse antecedent rather than by a syntactic (albeit nonlocal) antecedent. The reason for this is probably a pragmatic principle that dictates use of the overt pronoun for switches of discourse topics in general (cf. Eng, 1986). Summarizing the facts considered here, the differences in use between the overt pronoun and the inflected reflexive with respect to their use in nonlocal domains are really differences in preference, as to be expected, if
Local and Long-distance Reflexives in Turkish
215
they are interpreted as resulting from pragmatic and functional principles like the topic-switch principle and the Avoid Pronoun Principle. On the other hand, the differences between the bare reflexive, the inflected reflexive, and the overt pronoun in local domains are much more clear-cut, which is, in turn, not surprising, if those differences result from formal, syntactic principles like the binding conditions. I had mentioned previously that for many speakers, only the bare reflexive is well-formed in local domains, and the inflected reflexive is either rejected or is judged to be less well-formed. Some additional examples in Sinanoglu (1967) illustrate this difference: (25) Evldd -in -i dov -me -yen (kisii) child -3.sg. -Acc. spank -Neg. -Rel.Part. person kendin-ii dov-er self -Acc. hit -Aor. 'He who doesn't hit his child hits himself.' (Sinanoglu, 1967:205, without number; taken from Deny, 1921, without page reference) (26) Epikurosfelsefe -si [insan-ini kendin-ei Epicure philosophy-Cmp.M. person-Gen, self -Dat. kiy -ma -sin ] -i yasak et-me -z kill -Subj.Ger. -3.sg. -Acc. forbid -Neg. -Aor. "The Epicurean philosophy does not forbid for a person to kill himself.' (Sinanoglu, 1967:205, without number) In both instances, the bare reflexive is used. Using the inflected reflexive would be rather strange for many speakers under the coreferential reading; the primary reading would be the nonsensical (or pragmatically infelicitous) readings where people outside the domain of the utterance are referred to. For such speakers, two accounts come to mind: For the more tolerant of such speakers, for whom the inflected reflexive bound (indirectly, via pro) within the clause is possible, even if less preferred, I assume that there is a pragmatic, functional principle that dictates choosing a simpler form over a more complex form, when the simpler form is available with the same syntactic function. For the less tolerant speakers (who would form, as far as I can observe, a minority), for whom the inflected reflexive cannot function as an anaphor bound within the clause, the restriction must be more strictly formal. For those speakers, I propose that the AgrP projected by the agreement inflection is defective and is therefore "transparent" for binding. Therefore, the pro is no longer "protected." The relevant binding domain becomes larger
216
Jaklin Kornfilt
than the possessive AgrP (in most instances, as large as the smallest IP) for such speakers, and the pro specifier of the inflected reflexive would be bound by a co-indexed, c-commanding DP, i.e., it would be no longer free. This would result in a violation of Binding Condition B, thus ruling out the inflected reflexive from functioning as an anaphor bound locally (i.e., within the IP). Returning to the bare reflexive, we had said that in narrative style (and certain colloquial uses), it does not act as a genuine, locally bound anaphor. There are, however, certain instances where even in a more standard, written, or careful stylistic level, the bare anaphor is felicitous where it does not appear to display the main anaphoric property of being bound, i.e., c-commanded, by a co-indexed antecedent; I now turn to such instances. 5.3. Possessor Raising and the Bare Reflexive: Possessor Raising Examples in Unaccusative Constructions It has been noticed (e.g., Underhill, 1976; Sezer, 1979-1980) that the bare reflexive is felicitous in a clause when it is bound by the possessor of a c-commanding DP; thus, the binder itself does not c-command the reflexive: (27) [Orhan -ini ogl -u]i kendin-deni/j buyuk Orhan -Gen. son-3.sg. self -Abl. tall '[Orhan'Sj son]j is taller than hei/himselfj.' (Underhill, 1976: 356; no example number) This example is ambigous between a nonsensical interpretation, under which Orhan's son is taller than himself, i.e., the son, and a sensible reading, under which his son is taller than Orhan. It is under the sensible reading that the bare reflexive is not c-commanded by its antecedent. Interestingly, this kind of example is well-formed even for less permissive speakers, and this is true even for "regular," nonnarrative styles. Note, however, that this initially disturbing fact holds for certain static predicates only. Thus, in the following examples, the bare reflexive must indeed be c-commanded by its antecedent, and co-indexations parallel to the "sensible" one in the previous example are ill-formed: (28) [Orhan-ini ogl-u]j kendin-deni/j nefret ed-iyor Orhan-Gen son-3.sg. self -Abl. hatred do-Progr. '[Orhan'Si son]j hates himselfi/*himi.' (29)
[Orhan-ini ogl-u]j kendin-i*i/j ol -dur -du Orhan-Gen son-3.sg. self -Acc. die -Caus.-Past '[Orhan'Si son]j killed himselfi/*himi.'
Local and Long-distance Reflexives in Turkish
217
My proposal for explaining the well-formedness of examples like (27) are, at this point, sketchy and only suggestive. There is some evidence for static predicates like the one in (27) showing that a particular instance of NP-movement takes place that has the effect of "possessor raising." (In a nutshell, these arguments are based on the fact that, in general, a possessor within a DP in subject position of an embedded clause cannot undergo Exceptional Case Marking, but this is possible in certain constructions with Unaccusatives and static predicates; not all Unaccusatives allow this, but in order for this to be possible, the embedded predicate must be an Unaccusative or static verb). If this suggestion is on the right track, then possessor raising would take place in (27), resulting in the bare reflexive's being c-commanded by the raised possessor.
6. COMPLEX REFLEXIVES I turn now to another complex form involving the reflexive, namely, a reduplicated form. This form is very strictly local: (30) Akrep kendi kendin -i sok -ar scorpion self self -Acc. sting -Aor. 'The scorpion stings itself.' (31) Kabahat-i (siz) kendi kendi -niz -de ara -yin! fault -Acc. you(pl) self self -2.pl. -Loc. search -Imp.2.pl. 'Look for the fault in yourselves!' Thus, even for permissive speakers who extend the relevant properties of the narrative style to their colloquial speaking style, i.e., speakers that treat the bare third-person reflexive as an element of empathy11, reduplicated reflexives are strictly local: (32) [Oya-ninj kendin-den(??)i/j hoslan -ma -si] Ahmed-ii Oya-Gen. self -Abl. like -Ger. -3.sg. Ahmet-Acc. sevin -dir -di happy -Caus.-Past 'That Oya likes herself/him made Ahmet happy.' (33) [Oya-nmi kendi kendin-den*i/j hoslan-ma -si] Ahmed-ii Oya-Gen. self self -Abl. like -Ger.-3.sg. Ahmet-Ace. sevin -dir -di happy -Caus.-Past Only available reading: 'That Oya likes herself/*him made Ahmet happy.' (Despite the nonsensical/implausible nature of this reading)
218
Jaklin Kornfilt
I would like to suggest that the strict locality of the reduplicated reflexive is, in fact, due to its genuine complexity (and, thus, phrasal nature). The locality of this element for the nonpermissive speakers does not need any special explanation, since the single bare reflexive is local for these speakers, as well; thus, Binding Condition A is sufficient to account for the distribution of both the single (bare) and the reduplicated forms. However, for the permissive speakers (especially given that the bare reflexive is treated by them as a LD-reflexive and not just as an emphatic pronoun), the phrasal nature of the reduplicated reflexive would preclude LF-movement of this element in the fashion sketched by Cole et al. (1990), Cole and Wang (1996), and related work, while the bare anaphor would, undeed, undergo such head movement at LF. For the less permissive speakers, this movement of the bare reflexive would be strictly confined to narrative, literary style.12 The difference between the reduplicated and the inflected reflexives is a structural one. The reduplicated reflexives are compoundlike, with the second reflexive as the head. The first (noninflected) part of the compound (within the phrasal reflexive) is bound by (and, in some sense, incorporated into) the second part, i.e., the head. The head, in turn, is bound by a local (i.e., essentially clause-mate) antecedent. While the head of the compound can be inflected, it is inflected for first and second persons—and we said previously that these forms behave like the bare reflexive for third-persons, i.e., they are genuine local anaphors. As a matter of fact, inflected third-person reflexives cannot be used here. Contrast the grammatical (30) with the ungrammatical (34): (34) *Akrep kendi kendi -sin -i sok -ar scorpion self self -3.sg. -Acc. sting -Aor. Intended reading: 'The scorpion stings itself.' In contrast, the inflected third-person reflexive, the topic of the largest part of this chapter has, I claimed, as its first part not an "incorporated" reflexive as part of a compound, but an independent pronominal specifier which determines the overall binding properties of the phrase. The conclusion of this section, then, is that Turkish does obey the cross-linguistic generalization about morphologically complex reflexives being strictly local. Neither the reduplicated nor the inflected reflexives can undergo head movement at LF. However, because the inflected reflexives are associated with a.pro specifier (and the reduplicated reflexives are not), it appears that the inflected reflexives can, impressionistically speaking, extend their binding domain. But as I hope to have shown in this chapter, this is not a real extension of the binding domain; rather, pronominal rather than anaphoric binding takes place in these "large domains."
Local and Long-distance Reflexives in Turkish
219
7. CONCLUSION In this chapter, I have looked at inflected third-person reflexives in Turkish and have characterized them as AgrPs, headed by a strong agreement element, and with an internal structure that includes a pronominal specifier, pro, and a bare reflexive. The AgrP is the binding domain for both elements, within which they obey Conditions B and A, respectively. The bare reflexive is a locally bound anaphor—here as well as in the language in general. The phonologically empty pronominal specifier can pick up antecedents anywhere in the syntactic structure (as long as they are outside the AgrP in question), as well as in the discourse. This analysis explains the initially puzzling fact that the inflected reflexive behaves like an anaphor in local domains but like a pronominal in nonlocal domains. I have also shown that the inflected reflexive does not have the syntactic properties of LD-reflexives observed in other languages. In a rather sketchy fashion, I have also discussed the properties of the uninflected reflexive and have claimed that a narrative level must be distinguished from the standard style. While the bare reflexive does have the expected properties of a locally bound anaphor in the standard style, it has the syntactic properties of a LD-reflexive and the pragmatic properties of an empathic element in narrative style. Furthermore, I attributed the existence of some apparent counterexamples to this characterization as a syntactic anaphor to a phenomenon of possessor raising, restricted to certain predicates. Another possible analysis for the inflected reflexives, based on underdetermined feature specification, namely, the one of Cole and Hermon (this volume) for a Singaporean Malay element with similar syntactic properties, was considered but rejected for the Turkish inflected reflexive. I claimed that while underdetermined feature specification is appropriate for such elements if they are not inflected, my analysis captures the more general properties of the inflected reflexives better, because it explicitly draws parallels between them and other possessive AgrPs in the language. Finally, I briefly looked at complex, reduplicated reflexives and showed that these are strictly locally bound for all speakers and in all syntactic levels. This, I claimed, is due to previously proposed ideas about head movement of simple reflexives at LF in languages that have genuine LD-reflexives, whereby it is proposed that complex reflexives, due to their phrasal nature, cannot undergo such head movement and therefore cannot function as LD-reflexives and must consequently be bound locally. There are a number of issues concerning various reflexive forms in Turkish which I have left undiscussed (e.g., interaction with scrambling, with argument structure and with case marking) or which I have only skimmed.
220
Jaklin Kornfilt
One additional issue of relevance to inflected reflexives, namely, differences between overt pronominals and pro, was discussed in some detail, but more work does need to be done in the context of a more general binding theory. This chapter has been an attempt to give an overview of some of the reflexive forms in Turkish, with an analysis that attempts to integrate them into other aspects of the language.
ACKNOWLEDGMENTS I would like to thank the audience at the presentation of this paper at the LD-reflexives workshop, held during the Linguistic Institute of 1997 at Cornell University, for helpful questions and comments, especially Peter Cole and Wynn Chao. I am particularly grateful to Gaby Hermon for interesting discussion before and after that presentation, as well as for constructive comments and criticism after a close reading of a preliminary version. I am furthermore indebted to Mehmet Yamlmaz and Cigdem Bahm for their native judgments (which differ in interesting and, mercifully, predictable, ways from mine). All shortcomings are my responsibility.
NOTES 1
There is a rich literature on the topic of empathy and anaphora (see, among others, Kuno (1987) and Kuno and Kaburaki (1977), and the references cited in these works) and on logophors (see, for example, Reinhart and Reuland, 1991; Thrainsson, 1991, and references cited in these articles). 2 The local domain relevant for A-binding (i.e., for Binding Conditions A and B) of a constituent X is the smallest maximal projection which contains X and a SUBJECT accessible to X. The accessible SUBJECT is either the structural subject of that smallest maximal projection or a "strong AGR" element heading that projection, whereby "strong AGR" is understood to mean that the AGR element is part of a paradigm that fully and overtly distinguishes for all relevant j-features. This approach to locality is, by and large, the one adopted in Chomsky (1981) and related work. The specific proposals for Turkish are discussed in Kornfilt (1984,1988, and 1991). 3 As with other instances in the language where pro could be used (because the position is licensed and identified by overt agreement) in the same position, an overt element has some emphatic or contrastive overtones. The example with pro is as follows: (i) Alii Oya-ya-. [pro-l+j Ankara-ya Ali Oya-Dat. Ankara-Dat. gonder -il -ecek -ler + in ] -i soyle -di send -Pass. -Put. -pi. + 3. -Acc. tell -Past 'Ali told Oya that they would get sent to Ankara.'
Local and Long-distance Reflexives in Turkish
221
Here, there is no emphasis on the embedded subject. It is suggestive that the element which is in almost free variation with the inflected reflexive in such instances is a pronominal, and this fact goes along well with the analysis defended in this contribution that the inflected reflexive is in construction with a pro specifier, and that it is that pro which is bound nonlocally (points mentioned in the introduction and to be elaborated in Section 4.2.). Incidentally, the attentive reader will have noticed that the agreement element in (10) is for third-person singular, while we find the form for third-person plural in (i). I have discussed elsewhere (e.g., in Kornfilt, 1984,1988,1991) the fact that the plural portion of the third-person plural agreement in both the verbal and nominal paradigms is optional and, in fact, is often elided when an overt subject is present, as is the case in (10). Thus, in (i), the third-person plural agreement form -leri(n) is actually -ler + i(n) 'pl. + 3.' Where the overt subject is a possessive (as I claim is the case with the third-person inflected reflexive), this elision is often preferred. However, where the subject is pro, as in (i), such elision is not possible, since the j-feature content of pro must be identified by fully overt agreement morphology. This explains the difference in the agreement morphology between (10) and (i). 4 Another property commonly observed with LD-reflexives in certain languages is blocking effects; when there is a type of reflexive which can undergo LD binding in general, all the subjects between it and all potential antecedents must share (p-features; if one potential antecedent does not share those features, then additional potential antecedents which are further removed from the reflexive cannot bind it, even if they do share its j-features. (Discussion, overview and examples can be found in Cole and Hermon (1998), in Cole, Hermon, and Lee [this volume], in Hermon (1994), and in references cited in these works.) This property is not observed in Turkish. The following examples illustrate the lack of blocking effects: (i) Hasarii [Zeyneb -inj [Ali-nink kendi -sin -ii/j/(??)k Hasan Zeynep -Gen. Ali-Gen. self -3.sg.-Acc. sev -dig -in ] -i bil -dig -in -i] son -tyor love-Ger.-3.sg. -Acc. know -Ger. -3.sg.-Acc. think -Progr. 'Hasan thinks that Zeynep knows that Ali loves him/her/(??)himself.' (ii) Hasani [ben-im [Zeyneb-in-. kendi -sin -i\jm\\ Hasan I -Gen. Zeynep-Gen. self -3.sg. -Acc. sev -dig -in ] -i bil -dig -im -i] san -lyor love-Ger.-3.sg. -Acc. know -Ger. -l.sg. -Acc. think -Progr. 'Hasan thinks that I know that Zeynep loves him/(??)herself.' The lack of blocking effects observed here is, by itself, not damaging to a classification of the third-person inflected reflexive as a LD-reflexive, since the blocking effect (whether analyzed as a strictly grammatical phenomenon, as in Cole and Hermon (1998) and Hermon (1994) [among others] or as a combination of a logophoric and grammatical phenomenon, as in Cole, Hermon, and Lee [this volume]) is otherwise observed cross-linguistically with LD-reflexives that are not inflected. However, it is nevertheless important to show that inflected reflexives indeed do not exhibit blocking effects. The arguments given in the text against ana-
222
Jaklin Kornfilt
lyzing the Turkish inflected reflexives as LD-reflexives should suffice to make the point. 5 These AgrPs are inserted in positions otherwise appropriate for DPs. It is not immediately obvious that this can be done, given that sentences are a type of AgrP, as well, yet do not necessarily appear exclusively in positions reserved for DPs. I follow a proposal advanced in Kornfilt (1984), where it was claimed that there is a distinction between verbal and nominal AgrPs; it is only the latter that are of the category DP. (For a somewhat different analysis of Agr in DPs and based on Hungarian data, see Szabolcsi, 1987.) The independent motivation for this claim rests on the fact that there are two main agreement paradigms: verbal and nominal. The former is found in fully finite clauses, while the latter is found with possessive DPs and embedded nominalized clauses. The inflected reflexive would be subsumed under possessive DPs. The nominal Agr element determines the nominal character of the phrases it heads. In the spirit of Grimshaw (1991), the extended projection of the DP to the level of the nominal AgrP enables this type of AgrP to appear in DP-positions. 6 A similar analysis, involving internal structure for LD-reflexives in Chinese, was proposed in Chao and Yu (1997). In their analysis, there is a pro as part of that structure, as well, but it is analyzed as the D-head of a DP, rather than as the specifier of an AgrP, as in my analysis for the Turkish element. It would be interesting to analyze all (apparent) LD-reflexives cross-linguistically and to see if one and the same analysis might be able to account for all such elements, or if these are indeed different elements with different properties, requiring different analyses. The latter is more probable, since the Chinese elements are genuine LD-reflexives, while the Turkish inflected reflexives are not. Similar remarks would carry over to analyses like the one advanced in Reinhart and Reuland (1991 and 1993), where the pronominal determiner is overt, but where otherwise the analysis of the complex reflexive in languages like Dutch and German is rather similar to that of Chao and Yu (1997) for Chinese. 7 The pro, as the specifier of the AgrP, is only c-commanded by the head of the phrase, i.e., by Agr. Since it is not c-commanded by the reflexive, the pro is A-free. On the other hand, as the specifier of the AgrP, pro does c-command the reflexive. Thus, both Conditions A and B are respected. 8 In both of the examples that follow in the text, the inflected reflexive can also be coindexed with a discourse antecedent—of course, under the present analysis, this, too, would be via the pro, which would itself be bound by such a discourse antecedent. At this point in the discussion, however, we are interested in a "clause-mate" antecedent. 9 The same analysis would not carry over to first- and second-person reflexives, since those, as remarked previously, cannot be nonlocally bound. Those reflexives, then, must be treated as nonphrasal. I surmise here that this is due to the fact that for third persons, there is a choice between two forms—bare and inflected—while for first and second persons there is no such choice. Thus, the paradigm of genuine reflexives consists of the inflected first- and second-person forms (whereby the inflection is purely morphological and does not project its own phrase in the syntax) and the bare third-person form.
Local and Long-distance Reflexives in Turkish 10
223
There is yet another approach that could be adopted to account for the properties of the inflected third-person reflexive in Turkish, namely, that advanced in Enc (1989). I am indebted to Jim Huang for (indirectly) bringing this article to my attention. The Turkish inflected reflexive is characterized in this very interesting cross-linguistic article as an element which is, in a sense, the least restricted element among the three that we have seen, namely, bare reflexive, inflected reflexive, and pronoun: the bare anaphor must be locally bound by a qualifying antecedent, while the pronoun must be locally free. The inflected reflexive is, indeed, in a sense less restricted, since it can be locally bound as well as free, and it can be nonlocally bound. This (among other considerations) leads the author to posit a three-way distinction for these three elements among three distinct binary binding features—a system where the inflected reflexive has minus values throughout. In a sense, this account is diametrically opposed to Cole and Hermon's (1998) account, since it adds a feature with its own binary specifications, while Cole and Hermon retain a two-feature system and allow for these features to be underdetermined, with the values to be filled in by syntactic context (very much in parallel to underdetermined segments in phonology). My own account also does not add to the two features (i.e., anaphoric and pronominal). All else being equal, accounts with fewer features are obviously more explanatory than those with a larger number of features, provided all facts are explained. While Enc (1989) does achieve some economy by proposing to derive the third feature from certain values of the other two, it is still true that hers is a three-featured system. Furthermore, such an account cannot capture the parallels between the inflected reflexives and other possessive phrases in the language, nor does it address the issue of overt pronouns versus pro, especially in specifier position of such phrases as well as in the language at large, the way my analysis can, since the latter crucially imputes internal structure to the inflected reflexive. It is possible, of course, that the languages considered in Enc (1989) do ultimately necessitate a three-featured system; but it would be desirable to first attempt a cross-linguistic account that is more economical and that also seeks to integrate morphological properties into the syntactic explanation. Note that neither Enc's three-featured system nor Cole and Hermon's underspecification account has an explanation for why it is that, at least for Turkish and Malay, it should be an inflected reflexive that is the least restricted form. Mine, in contrast, addresses this property directly. 11 From the limited field work I have conducted so far, it appears that this usage of the bare reflexive by certain permissive speakers as an element of empathy is actually a usage of that element as a genuine LD-reflexive with relevant properties like subject-orientation. In this respect, the "nonlocal" binding of the bare reflexive differs from the instances of nonlocal binding of the inflected third person reflexive. Thus, we have here, for this stylistic level and for these speakers, a syntactically well-defined element, whose syntactic properties have received satisfactory accounts in the literature (e.g., head movement at LF). These elements have, in addition to their well-defined syntactic properties, pragmatic functions. However, this fact does not challenge the syntactic nature of these elements as has sometimes been claimed in the literature (e.g., Sezer, 1979-1980), let alone the syntactic nature of reflexives in general.
224
Jaklin Kornfilt
12
There are a variety of other interesting properties worth investigating concerning complex reflexives, for example, restrictions concerning Case, or thematic roles. Discussion of such properties would go far beyond the concerns of this chapter and I leave such investigation to future research. Another question left open here is why the uninflected reflexive does not function as a LD reflexive, especially in the nonpermissive dialect. This is particularly surprising in an approach based on head-movement of reflexives (at LF), since the bare reflexive should undergo such movement and thus exhibit properties of LD-reflexives. Claiming that the bare reflexive is not an X° and that it projects an XP (I am grateful to Gabriella Hermon for pointing out this possibility to me) would not help solve this problem: while this analysis would, as required, preclude the bare reflexive from undergoing head-movement, it would analyze the element in a fashion similar to the inflected reflexive, since the specifier of the posited XP would need to be a phonologically null element, and thus, presumably, pro. I see two ways out of this dilemma: (1) One could claim that such a pro is indeed there, but that it is an expletive and therefore not able to be bound. Such an analysis has some independent motivation: pro has to be licensed and identified by overt as well as paradigmatic agreement, and the bare reflexive has no such agreement. However, the bare reflexive does have an alternating final n (which does not show up in the Nominative, nor does it appear before the third-person agreement form). This n might be sufficient to license an expletive pro without being able to identify a thematic pro fully. Therefore, an expletive pro is all that can be licensed (cf. Jaeggli and Safir, 1989); (2) Alternatively, one could claim that, while the "bare" reflexive does not project an XP, it is not a genuine X° either, due to the aforementioned n. In other words, it carries defective inflection and can therefore not undergo X°-movement. These sketchy remarks have to suffice at this point in time.
REFERENCES Chao, W. and W. Yu. (1997). Long-distance Reflexives and Logophoric Interpretations. Paper presented at the Long-distance Reflexives Workshop. LSA Linguistic Institute, Cornell University, Ithaca, July, 1997. Chomsky, N. (1981). Lectures on Government and Binding. Dordrecht, Netherlands: Foris. Cole, P., and Hermon, G. (1998). Long distance reflexives in Singapore Malay: An apparent typological anomaly. Linguistic Typology 2(1), 57-77. Cole, P., Hermon, G., and Lee, C. L. (1999). Grammatical and discourse conditions on long distance reflexives in two Chinese dialects. This volume. Cole, P., Hermon, G., and Sung, L.-M. (1990). Principles and parameters of long-distance reflexives. Linguistic Inquiry 21,1-22. Cole, P., and Wang, C. (1996). Antecedents and blockers of long distance reflexives. Linguistic Inquiry 27,357-390.
Local and Long-distance Reflexives in Turkish
225
Deny, J. (1921). Grammaire de la langue Turque (dialecte Osmanli). Paris: Leroux. [Reprinted in Wiesbaden: Dr. Martin Sa'ndig, 1971.] Enc, M. (1986). Topic switching and pronominal subjects in Turkish. In Studies in Turkish Linguistics (D.I. Slobin and K. Zimmer, eds.), pp. 195-208. Amsterdam: John Benjamins. Enc, M. (1989). Pronouns, licensing, and binding. Natural Language and Linguistic Theory 7,51-92. Erguvanh-Taylan, E. (1986). Pronominal versus zero representation of anaphora in Turkish. In Studies in Turkish Linguistics (D.I. Slobin and K. Zimmer, eds.), pp. 209-231. Amsterdam: John Benjamins. Grimshaw, J. (1991). Extended Projections. Unpublished manuscript. Waltham: Brandeis University. Hermon, G. (1984). Long-distance reflexives in UG: Theoretical approaches and predictions for acquisition. In Syntactic Theory and First Language Acquisition: Cross-Linguistic Perspectives, Vol 2: Binding, Dependencies, and Learnability (B. Lust, G. Hermon, and J. Kornfilt, eds.), pp. 91-111. Hillsdale, NJ: Lawrence Erlbaum. Jaeggli, O., and Safir, K. J. (1989). The null subject parameter and parametric theory. In The Null Subject Parameter (O. Jaeggli and K. J. Safir, eds.), pp. 1-44. Dordrecht: Kluwer. Kornfilt,! (1984). Case Marking, Agreement, and Empty Categories in Turkish. Ph.D. thesis, Harvard University. Kornfilt, J. (1987). Beyond binding conditions: The case of Turkish. In Studies in Modern Turkish, (R. Boeschoten and L. Verhoeven, eds.), pp. 105-120. Tilburg: Tilburg University Press. Kornfilt, J. (1988). A typology of morphological agreement and its syntactic consequences. In Proceedings of the Chicago Linguistic Society 24,117-134. Kornfilt, J. (1991). Some current issues in Turkish syntax. In Turkish Linguistics Today (H. Boeschoten and L. Verhoeven, eds.), pp. 60-92. Leiden: Brill Publishers. Kuno, S. (1987). Functional Syntax: Anaphora, Discourse and Empathy. Chicago: The University of Chicago Press. Kuno, S., and Kaburaki, E. (1977). Empathy and syntax. Linguistic Inquiry 8, 627-672. Lewis, G. L. (1985). Turkish Grammar. Oxford: Oxford University Press. [Corrected reprint of 1967 original.] Pica, P. (1987). On the nature of the reflexivization cycle. In Proceedings O/NELS17, Vol. 2, 483—499. Graduate Linguistic Student Association, Department of Linguistics, University of Massachusetts, Amherst. Reinhart, T, and Reuland, E. (1991). Anaphors and logophors: An argument structure perspective. In Long-Distance Anaphora (J. Koster and E. Reuland, eds.), pp. 283-321. Cambridge: Cambridge University Press Reinhart, T, and Reuland, E. (1993). Reflexivity. Linguistic Inquiry 24,657-720. Sezer, E. (1979-1980). On reflexivization in Turkish. In Harvard Ukrainian Studies Eucharisterion: Essays Presented to Omeljan Pritsak on his Sixtieth Birthday by his Colleagues and Students. III/IV(2), 748-759. Cambridge, MA: Harvard University.
226
Jaklin Kornfflt
Sinanoglu, S. (1967). Kendi kelimesinin kullamhslan. (The usages of the word kendi.) In Dilbilgisi Somnlan (Issues of Grammar); TDK Yaymlari: 259. Dil Konulan Dizisi: 10; Ankara Universitesi Yaymevi (Publications of the Turkish Language Academy: 259. Series of Language Topics: 10; Publication House of Ankara University), 204-207. [First published in Turk Dili (The Turkish Language), vol. 79,1959.] Slobin, D. I., and Zimmer, K. (eds.) (1986). Studies in Turkish Linguistics. Amsterdam: John Benjamins. Szabolcsi, A. (1987). Functional categories in the noun phrase. In Approaches to Hungarian, vol. 2: Theories and Analyses (I. Kenesei,ed.),pp. 167-189. Szeged. Thrainsson, H. (1991). Long-distance reflexives and the typology of NPs. In Long-Distance Anaphora (J. Koster and E. Reuland, eds.), pp. 49-75. Cambridge: Cambridge University Press. Underbill, R. (1976). Turkish Grammar. Cambridge, MA: MIT Press.
ANTI-ANTILOCALITY JEFFREY LIDZ Institute for Research in Cognitive Science University of Pennsylvania Philadelphia, Pennsylvania
1. INTRODUCTION As the range of data that is potentially relevant to Binding Theory has grown, a great deal of research has focused on whether this theory applies to long-distance reflexives, the central question being whether such forms are anaphors, pronominals, or something outside the purview of the Binding Theory. One problem which has been relatively unexamined in this research is the character of anaphors which cannot be locally bound. We know that such elements are anaphors because they require a c-commanding antecedent. Unlike most anaphors, however, they do not permit their antecedents to occur within the local domain. The contrast between an antilocal anaphor and the typical long-distance reflexive is illustrated in (1-2) below: (1) Kannada1 a. *Hari tann-annu hogaL-id-a Hari self-ACC praise-PST-3SM 'Hari praised himself.' b. Raamiti Shyaamuj tann-annuyi/*jk/*k priitis-utt-aane anta Raam-NOM Shyam-NOM self-ACC love-NPST-3SM that namb-utt-aane believe-NPST-3SM 'Raam believes that Shyam loves him.' Syntax and Semantics, Volume 33 Long-Distance Reflexives
227
Copyright © 2001 by Academic Press All rights of reproduction in any form reserved. 0092-4563/00 $35.00
228
Jeffrey Lidz
(2) Chinese (Cole, Hermon, and Sung, 1990) a. Lisii piping zijii Lisi criticized self 'Lisi criticized himself.' b. Zhangsani renwei Lisii piping ziji i/j Zhangsan thinks Lisi criticized self 'Zhangsan thinks Lisi criticized him/himself.' In Kannada, we see that local binding of the morphologically simplex anaphor tannu is not possible while long-distance binding is possible. In Chinese, the morphologically simplex anaphor ziji may be locally bound as well as long-distance bound. In both languages, the morphologically complex anaphor permits only local binding: (3) a.
Raamui Shyaamui tann-annu-taane*i/j Raam-NOM Shyam-NOM self-ACC-self namb-utt-aane believe-NPST-3SM 'Raam believes that Shyam loves himself.' b. Zhangsani renwei Lisii piping ta-zijii/j Zhangsan thinks Lisi criticized himself 'Zhangsan thinks Lisi criticized himself.'
priitis-utt-aane anta love-NPST-3SM that
An important observation with respect to the form of reflexive pronouns was made by Faltz (1977) who observed that polymorphemic reflexives may not be long-distance bound while long-distance reflexives are always monomorphemic. Capitalizing on this observation, Pica (1987) proposed that the determination of binding domain is related to the morphological form of the reflexive. In particular, Pica proposed that morphologically simplex reflexives are LF clitics, which move to INFL at LF (cf. Battistella, 1989; Cole, Hermon, and Sung, 1990; Cole and Sung, 1994). Because they are heads, these reflexives may move through INFL to Comp and then into the next clause, explaining their possible long-distance interpretation.2 A sentence like (4) has the two possible LF representations in (5) (Cole et al, 1990):3 (4) Lisii zhidao Wangwuj xihuan zijii/j/*k Lisi think Wangwu like self 'Lisi thinks Wangwu likes him/himself.' (5) a. [Lisi [zhidao [Wangwu iNFL-zijii [xihuanti,]]]] b. [Lisi NFL-zijii [Wangwu iNFL-ti [xihuan i]]] In (5a), ziji is locally bound only by Wangwu, while in (5b), ziji is locally bound only by Lisi. The long-distance interpretation in (5b) is really an in-
Anti-Antilocality
229
stance of local binding at LF. Under this account, the binding theory remains constant in enforcing the antecedent to be locally related to the reflexive. What varies from language to language is whether the reflexive moves. Whether the reflexive moves is, in turn, explained by the morphological form of the reflexive. The details of how the morphological form of the reflexive relates to the possibility of movement are irrelevant to the task at hand, however. What is relevant is that the antilocality of anaphors like Kannada tannu is not explained by this approach. The head-movement theory correctly predicts that tannu is a long-distance reflexive but fails to predict its antilocality. This chapter argues that the antilocality of anaphors like Kannada tannu is epiphenomenal. These elements may, in fact, be locally bound. In many cases where local binding obtains, however, independent principles relating to the interface of lexical and semantic representations rule the sentence out. Such restrictions are predicted to hold in all languages. What I will show is that the semantic representation of the antilocal reflexives differs from that of the reflexives which can be locally bound. The antilocal reflexives require complete identity with their antecedents, while those anaphors which allow local binding can be loosely related to the referents of their antecedents by certain kinds of similarity to be made more precise in the following. Accordingly, the existence of two types of long-distance anaphor, local and antilocal, is not a syntactic fact and can be explained by independent principles of grammar relating to the semantic representation of reflexive predicates. Thus, no changes to the binding theory are required in order to explain the behavior of antilocal anaphors. This chapter is organized as follows. In section 2,1 describe the distribution of pronouns in Kannada. In sections 3.1 and 3.2,1 explore semantic properties of the various kinds of anaphor in Kannada, drawing a distinction between Pure-reflexivity and Near-reflexivity. In section 3.3,1 introduce Condition R, a principle of grammar which forces Pure-reflexivity to be lexically expressed. In section 4,1 explore some predictions made by Condition R. In section 5,1 outline some previous proposals to account for antilocality and show why they are inferior to the account proposed here. Finally, I conclude in section 6.
2. ANAPHORA IN KANNADA The Kannada anaphor, tannu, has been characterized in the past as an antilocal reflexive because it appears to require its binder to be outside of the local domain (Bhat, 1981; Yadurajan, 1987; Amritavalli, 1991; Lidz,
230
Jeffrey Lidz
1995). In this section, I will show that this characterization is inaccurate. As a consequence, any theory based on this characterization inherits the inaccuracy. The anaphor, tannu, behaves like the pronominal but unlike the morphologically complex anaphor in resisting binding by a coargument: (6) a. Harii awan-annu*i/j hoDe-d-a Hari he-ACC hit-PST-3SM 'Hari hit him.' b. Harii tann-annu*i/*j hoDe-d-a Hair self-ACC hit-PST-3SM 'Hari hit himself.' c. 4Harii tann-annu-taanei/*j hoDe-d-a Hari self-ACC-self hit-PST-3sM 'Hari hit himself.' (7) a. Harii Raaju i awcn-annui/*j/k hoDe-d-a anta heeL-id-a Hari Raaju he-ACC hit-PST-3SM that say-PST-3SM 'Hari said that Raaju hit him.' b. Harii Raaju j tann-annui/*j/*k hoDe-d-a anta heeL-id-a Hari Raaju self-ACC hit-PST-3SM that say-PST-3SM 'Hari said that Raaju hit him.' c. Harii Raaju i tann-annu-taane*i/j/*k hoDe-d-a anta heeL-id-a Hari Raaju self-ACC-self hit-PST-3sM that say-PST-3SM 'Hari said that Raaju hit himself.' However, in clausebounded contexts where tannu is not a coargument of its antecedent, it patterns with the morphologically complex anaphor and against the pronoun in allowing binding. (8) a. Harii tann-annui/*j mattu tann-ai/*j hendati-yannu hoDe-d-a Hari self-ACC and self-GEN wife-ACC hit-PST-3SM 'Hari hit himself and his wife.' b. Harii pustaka-vannu tann-ai/*j hindee nooD-id-a Hari book-ACC self-GEN behind see-PST-3sM 'Hari saw the book behind him.' (9) a. Harii avan-annu*i/j mattu avan-ai/j hendati-yannu hoDe-d-a Hari he-ACC and he-GEN wife-ACC hit-PST-3SM 'Hari hit him and his wife.' b. Harii pustaka-vannu avan-a*i/j hindee nooD-id-a Hari book-ACC he-GEN behind see-PST-3sM 'Hari saw the book behind him.'
Anti-Antilocality
231
(10) a. Harii tann-annu-taanei/*j mattu tann-ai/*j hendati-yannu Hari self-ACC-self and self-GEN wife-ACC hoDe-d-a hit-PST-3SM 'Hari hit himself and his wife.' b. Harii pustaka-vannu tann-a-taan-eei/*j hindee nooD-id-a Hari book-ACC self-GEN-self-EMPH behind see-PST-3SM 'Hari saw the book behind him.' Finally, the restriction on coargument binding is obviated if the predicate is morphologically reflexive. In such cases, coargument binding is required: (11) Raaju j Hari i tann-annu i/*j/*k hode-du-koND-a anta heeL-id-a Raaju Hari self-ACC hit-PP-REFL.PST-3SM that say-PST-3SM 'Raaju said that Hari hit himself.' The pattern of facts presented so far is summarized in Table 1. There are two lessons to take from these data. First, the standard binding theory (Chomsky, 1981/1986) will be insufficient to characterize these facts by itself because, in some environments, tannu patterns with the anaphor and in other environments, it patterns with the pronominal. We cannot characterize the simplex anaphor tannu as a pronominal because in the contexts with a clausemate noncoargument antecedent such as those in (8-10) it does not pattern like a pronominal. Both the simplex and complex anaphors allow such an antecedent, but the pronominal does not. Similarly, the pronominal allows a discourse antecedent but neither of the anaphors does. We also cannot characterize the simplex anaphor as an anaphor since in the coargument context, it patterns like a pronominal but unlike the complex anaphor in resisting a coargument antecedent. This pattern is unexpected in a theory in which the only distinction between pronoun types is whether they are proTABLE 1 Antecedent Pronoun Awanu (pronominal) Tannu (simplex anaphor) Tannu-taane (complex anaphor) Tannu . . . V+koLLu (simplex anaphor + verbal marking)
ClausemateCoargument noncoargument
Longdistance
(=6a)
(=9)
(=7a)
(=6b)
(=8)
(=7b)
(=6c)
(=10)
(=7c)
(=11)
(=11)
Discourse
232
Jeffrey Lidz
nominal or anaphoric. It would appear that we need three types of pronoun rather than two (cf. Reinhart and Reuland, 1993). More generally, the characterization of tannu as an antilocal anaphor is inaccurate since such an element would be expected to pattern with the pronominal in all contexts with a c-commanding antecedent, and we have seen that this expectation is not met. The second conclusion we can draw is that properties of predicates will be an important ingredient of any explanation of the distribution of tannu because changing the verbal morphology also changes the distribution of the anaphor (cf. Reinhart and Reuland, 1993; Lidz, 1995). That is, although tannu resists binding by a coargument, there is no absolute prohibition on coargument binding since the addition of reflexive morphology to the verb will license such binding. In what follows, I will show that we do not, in fact, need three types of pronouns to explain the pattern of facts observed above (contra Reinhart and Reuland, 1993). Instead, we can maintain the division of pronouns into two syntactic types (pronominal and anaphoric). The anaphoric pronouns are further divided into two semantic types. This semantic division has no direct syntactic consequence, but rather interacts with lexical and semantic properties of reflexive predicates to yield the observed distribution.
3. WHY APPARENT ANTILOCALITY IS NOT SYNTACTIC The mystery of this chapter is that the Kannada simplex anaphor behaves in the coargument context like a pronominal while in other local contexts, it behaves like an anaphor. To make matters more complicated, in the coargument context if the verb is morphologically reflexive, then the simplex anaphor behaves like an anaphor and unlike a pronominal. Assuming that the noncoargument context reveals the true categorial status of this pronoun, the question we are left with is why the form of the verb changes the binding possibilities for the anaphor. In other words, the noncoargument context tells us that the simplex anaphor is, in fact, an anaphor and not a pronominal (contra Amritavalli, 1984; Jayaseelan, 1996). But if it is an anaphor, then we are left with the question of why binding by a coargument is prohibited in the absence of verbal morphology. In this section, I will show that an independent principle of the lexicon-semantics interface explains the odd behavior of the simplex anaphor in the coargument context. 3.1. Semantic Distinctions between Anaphors Assuming that both the simplex and complex anaphors in Kannada are anaphoric, we can ask why the simplex anaphor is blocked in (12a), but al-
Anti-Antilocality
233
lowed in (12b). We can also ask why the complex anaphor is not blocked in (12c): (12) a. *Hari tann-annu hoDe-d-a Hari self-ACC hit-PST-3SM 'Hari hit himself.' b. Hari tann-annu hoDe-du-koND-a Hari self-ACC hit-PP-REFL.PST-3SM 'Hari hit himself.' c. Hari tann-annu-taane hoDe-d-a Hari self-ACC-self hit-PST-3SM 'Hari hit himself.' That is, the crucial question here is what property distinguishes the two anaphors from each other such that one can take a coargument as antecedent in the absence of the verbal reflexive and the other cannot. Given that the syntactic relationship between the anaphor and its antecedent is the same in (12a-c), we will explore the possibility that the anaphors have different semantic content and that this difference is responsible for the binding differences. Let us first consider the Madame Tussaud context discussed in Jackendoff (1992). Here, imagine that Ringo Starr goes into Madame Tussaud's wax museum, which contains a statue depicting him. Upon seeing the statue, Ringo is appalled because the museum has chosen to portray him with a beard, though Ringo himself prefers a cleaner-cut image. So, Ringo pulls out his razor and begins to shave the statue that portrays him. In this situation, it is felicitous in Kannada to say (13b) but not (13a). (13) a. Ringo tann-annu boolis-i-koND-a Ringo self-ACC shave-PP-REFL.PST-3SM 'Ringo shaves.' (=Real-Ringo shaves Real-Ringo; Real-Ringo shaves Statue-Ringo) b. Ringo tann-annu-taane boolis-id-a Ringo self-ACC-self shave-PST-3SM 'Ringo shaves himself.' (=Real-Ringo shaves Real-Ringo; =Real-Ringo shaves Statue-Ringo) Conversely, if the statue in the museum is one depicting Ringo without a beard but he currently does have a beard and realizes that he was much cuter without the beard and begins to shave his own face, it is possible to say either (13a) or (13b).
234
Jeffrey Lidz
A second place where the two anaphors give divergent interpretations is in comparative deletion constructions. For those speakers who allow comparative deletion in Kannada,5 when the verbal reflexive is present (with the simplex anaphor), only the sloppy reading is available. When the complex anaphor is present in the absence of the verbal reflexive, both strict and sloppy readings are available: (14) a. Rashmi Siita-ginta cheenage tann-annu rakshisi-koLL-utt-aaLe Rashmi Sita-COMP better self-ACC defend-REFL-NPST-3SF 'Rashmi defends herself better than Sita defends herself (=Sita).' '*Rashmi defends herself better than Sita defends Rashmi.' b. Rashmi Siita-ginta cheenage tann-annu-taane rakshis-utt-aaLe Rashmi Sita-COMP better self-ACC-self defend-NPST-3SF 'Rashmi defends herself better than Sita defends herself (=Sita).' 'Rashmi defends herself better than Sita defends Rashmi.' Given that sentences with the complex anaphor are interpreted differently from those with the simplex anaphor, it is possible that the distributional differences between the anaphors are due to the semantic differences between them and not to any syntactic difference. In order to see how this could be, we must first characterize the difference between the interpretations allowed in the sentences with the local anaphor and those with the antilocal anaphor. 3.2. Pure-reflexivity vs Near-reflexivity What the Madame Tussaud examples illustrate is that the morphologically complex anaphor allows an interpretation in which the anaphor is referentially dependent on its antecedent but not necessarily identical with it. The anaphor can pick out an entity from the domain of discourse which is related to and similar to its antecedent. We will call such anaphors "Near-reflexives." The difference between Near-reflexive predicates (i.e., predicates which have a Near-reflexive as an argument) and semantically reflexive predicates is given in (15): (15) a. lx [P(x,x)] (Semantic/Pure-reflexive) b. lx [P(x, f(x))] (Near-reflexive) Example (15a) indicates that the two arguments of the predicate P are identical (or that there is one entity which realizes both roles of the predicate). In (15b), the second argument is a function taking the first argument as input and returning an entity which is representationally related to that argument (cf. Reuland, 1995).6 The representations of these predicates are formally distinct, although, in most cases, they are extensionally equivalent. That is, the Near-reflexive function does not prohibit the antecedent and the
Anti-Antilocality
235
anaphor from being the same entity in the world, although in the semantic representation they are distinguished. On the other hand, in a Pure-reflexive sentence, the anaphor and its antecedent must be completely identical, both in the world and in the semantic representation. Since the sentences with the simplex anaphor and the verbal reflexive do not allow a nonidentity interpretation, it follows that they are semantically reflexive, i.e., that they have the Pure-reflexive representation (15a). Predicates with a morphologically complex anaphor as an argument do permit the nonidentity interpretation, and so implicate the Near-reflexive representation (15b). It is important to note that the interpretation of Pure-reflexive and Near-reflexive predicates depends on inherent properties of the reflexive pronouns themselves. The simplex anaphors are pure variables and do not contribute anything to interpretation other than that the argument that they represent is identical to some syntactically determined antecedent. The complex anaphors, on the other hand, do contribute something to interpretation. They introduce the Near-reflexive function discussed previously, which establishes a mapping from the syntactic antecedent to things which are able to be construed as representations of that antecedent. This point can be illustrated in examples in which the anaphor is not a coargument of its antecedent. Even here, the Near-reflexive interpretation is only allowed with those anaphors that license the Near-reflexive interpretation with coarguments. Consider the situation in which Ringo is in the museum and someone bumps into the statue that portrays him. If Ringo sees his statue fall, we can say (16b) but not (16a): (16) a. Ringo tann-u bidd-uvudu nooD-id-a Ringo self-NOM fall-NOMNL see-PST-3sM 'Ringo saw himself fall.' (Real-Ringo saw real-Ringo fall) (*Real-Ringo saw statue-Ringo fall) b. Ringo tann-u-taane bidd-uvudu nooD-id-a Ringo self-NOM-self fall-NOMNL see-PST-3SM 'Ringo saw himself fall.' (Real-Ringo saw real-Ringo fall) (Real-Ringo saw statue-Ringo fall) The anaphor in these sentences is not an argument of the same predicate as its antecedent. The anaphor is an argument of the embedded (nominalized) verb while the antecedent is an argument of the matrix verb. Even when the simplex anaphor is not a coargument of its antecedent, it requires a pure-identity interpretation. The complex anaphor similarly always allows the Near-reflexive interpretation, independent of where its antecedent lies in
236
Jeffrey Lidz
the structure. Thus, we can conclude that individual anaphors are lexically specified as introducing the Near-reflexive function or not. The interpretation of a given anaphor is independent of properties of the predicate that it is an argument of. However, when its antecedent is a coargument, the interpretation of that anaphor does have morphological consequences for the verb that it is an argument of. In particular, when a pure-reflexive anaphor takes a coargument as its antecedent, the predicate must be marked with the verbal reflexive. The question we are left with is why. 3.3. Condition R Given that the coargument restriction on tannu is obviated in the presence of morphological reflexivity and that tannu with the verbal reflexive always involves a Pure-reflexive interpretation, the following principle suggests itself: (17) Condition R: l[P(x,x)] « ( q 1=q2) semantics theta-grid The left side of this formula depicts the semantic representation of Pure-reflexivity. The right side depicts the representation of the theta-grid of a lexically reflexive predicate. The " = " symbol is intended to convey that the two theta-roles are lexically specified as being borne by the same individual. The condition states that if a predicate is semantically reflexive, then it must be lexically reflexive. Similarly, if a predicate is lexically reflexive, then it must be semantically reflexive. I assume that the morphological reflexive marker occurs on the verb when an operation of lexical reflexivization has taken place. Thus, we expect that the verbal reflexive will be required when the simplex anaphor is a coargument of its antecedent because that anaphor requires the pure-reflexive interpretation. The pure-reflexive interpretation, given coargumenthood, requires lexical reflexivity, in accordance with Condition R. Lexical reflexivity, in turn, requires the verbal reflexive to occur.7 To see how this works, consider the following: (18) a. *Hari tann-annu nooD-id-a Hari self-ACC see-PST-3SM 'Hari saw himself.' b. Hari tann-annu nooDi-du-koND-a Hari self-ACC see-PP-REFL.PST-3SM 'Hari saw himself.' (= reflection, statue) c. Hari tann-annu-taane nooD-id-a Hari self-ACC-self see-PST-3SM 'Hari saw himself.' (= reflection or statue)
Anti-Antilocality
237
Example (18a) is ruled out in violation of the left-to-right implication of Condition R; the sentence is semantically reflexive (since tannu does not introduce the Near-reflexive function) but not lexically reflexive. Sentence (18b) is both semantically and lexically reflexive and so satisfies Condition R. The Near-reflexive interpretation is ruled out both because the simplex anaphor does not introduce the Near-reflexive function and because the predicate is lexically reflexive and so must be semantically reflexive, by the right-to-left implication of Condition R. Finally, in (18c), the anaphor introduces the Near-reflexive function and so the sentence is not semantically reflexive. The predicate is not lexically reflexive either and so Condition R is not operative. Another place where Condition R is inoperative is when the anaphor is not a coargument of its antecedent. Because such cases do not involve reflexive predication, Condition R does not apply. Hence, we expect that, in such cases, the Binding Theory will determine whether binding is licit and that all anaphors will pattern alike. This is precisely the case, as was seen above in (8-10). When there is no reflexive predicate, binding theory treats the two anaphors alike but distinct from pronominals. Similarly, in the coargument domain, the Binding Theory allows both anaphors but Condition R filters out one of these in the absence of lexical reflexivity. We should note here that a theory with both Condition R and the standard binding conditions redundantly rules out sentences with reflexive predicates and a bound pronominal, like (19): (19) *Hari awan-annu hoDe-d-a Hari he-ACC hit-PST-3SM Here, the pronominal is illicitly bound within its governing category, IP. The sentence is also semantically reflexive and so is blocked by Condition R.8 This redundancy is necessary, however, given that the binding theory can correctly account for all data outside of the coargument domain. Adding Condition R to the theory correctly captures the fact that the coargument domain is unique in distinguishing types of anaphors from each other.
4. PREDICTIONS OF CONDITION R 4.1. All Coargument-bound Anaphors are Near-reflexives Condition R makes an interesting prediction: if an anaphor can be bound by a coargument (in the absence of lexical reflexivity), then that anaphor is a Near-reflexive. This prediction follows from Condition R because if a predicate is not lexically reflexive, then there is no way for it to be semanti-
238
Jeffrey Lidz
cally reflexive. This prediction is borne out in languages as diverse as English, Chinese, and Russian. We consider these in turn.
4.1.1. ENGLISH It is well known that the English anaphor himself can be bound by a coargument: (20) John hit himself That this anaphor allows the Near-reflexive interpretation is clear (see Jackendoff, 1992). Consider the situation in which there is a carnival game in which the player throws a football at some statues depicting various presidents of the United States and that Ronald Reagan plays this game. Here, it is natural to say (21) and mean that Reagan hit the statue that portrays him. (21) Ronald Reagan hit himself with the football Of course, (21) does not force a statue-interpretation; it can be interpreted as real-Reagan hitting real-Reagan with the football. What this illustrates is that the entities picked out by the domain and the range of the Near-reflexive function can be identical even though they are distinguished in the semantic representation. Note also that sentences with inherently reflexive predicates in English do not allow a Near-reflexive interpretation, in accordance with Condition R. We cannot say (22a) to mean that Reagan dressed a statue that portrayed him, although (22b) will suffice in that situation: (22) a. Reagan dressed in the museum b. Reagan dressed himself in the museum We also saw above that Pure-reflexive sentences allow only a sloppy reading under comparative deletion and that Near-reflexive sentences allow either a strict or sloppy reading. These facts are mirrored in English: (23) a. Reagan dresses faster than his nurse does b. Reagan dresses himself faster than his nurse does
(sloppy only) (strict/sloppy)
4.1.2. CHINESE9
Chinese has a morphologically simplex anaphor, ziji, which can be bound by a coargument. Condition R correctly predicts that this anaphor is a Near-reflexive. Here, imagine the situation where Mao Tse Tsung goes into the wax museum and becomes enraged at seeing the statue depicting him.
Anti-Antilocality
239
Here, we can say (24) to mean either that Mao committed suicide (shooting himself) or vandalism (shooting the statue): (24) Mao Tse Tung ba ziji qiangbi le Mao Tse Tung BA self shoot ASP 'Mao Tse Tung shot himself (=statue or Mao).' Under comparative deletion, we expect Near-reflexives to allow strict or sloppy interpretations but Pure-reflexives to allow only sloppy interpretations. So, if ziji is a Near-reflective, then it follows that it should allow both a strict and a sloppy interpretation. The prediction is also borne out: (25) Zhangsan bi Lisi wei ziji bianhu de hao Zhangsan than Lisi for self defend DE Well 'Zhangsan defended himself better than Lisi defended himself.' 'Zhangsan defended himself better than Lisi defended him.' 4.1.3. RUSSIAN10
Russian is a further interesting example from this perspective because, like Kannada, it has both an affixal reflexive and a morphologically simplex anaphor. Unlike Kannada, however, the morphologically simplex anaphor can be bound by a coargument in the absence of the affixal reflexive. Since the affixal reflexive is taken to indicate lexical reflexivity, it follows that when it is absent, the predicate cannot be lexically reflexive and thus not semantically reflexive. So, sentences with the anaphor bound by a coargument must have a Near-reflexive interpretation. This prediction is borne out. Imagine the situation described above for Mao Tse Tung, but with Boris Yeltsin as the protagonist. When the verbal reflexive is present, the Near-reflexive interpretation is blocked (as predicted by Condition R). When the anaphor is present, the Near-reflexive interpretation is available: (26) a. Yeltsin zastrelil-sja Yeltsin shot-REFL 'Yeltsin shot himself (= Yeltsin, statue).' b. Yeltsin zastrelil sebja Yeltsin shot self 'Yeltsin shot himself (= Yeltsin or statue).' Also as predicted by Condition R, in comparative deletion constructions when the verbal reflexive is present, we get only the sloppy interpretation and when the anaphor is present, we get both strict and sloppy interpretations: (27) a. Ivan zashchischal-sja lachshe chem Petr Ivan defended-REFL better than Peter
240
Jeffrey Lidz
'Ivan defended himself better than Peter defended himself.' '*Ivan defended himself better than Peter defended him.' b. Ivan zashchischal sebja lachshe chem Petr Ivan defended self better than Peter 'Ivan defended himself better than Peter defended himself.' 'Ivan defended himself better than Peter defended him.' 4.2. Variation in Lexical Reflexivity A second prediction made by Condition R has to do with the extent of lexical reflexivity found in a given language. In Kannada, any verb can be made lexically reflexive through the affixation of the verbal reflexive. However, lexical reflexivization is not as free in other languages. Dutch allows only a small set of lexically specified verbs to be lexically reflexive. (28) a. *Max haat zich Max hates self 'Max hates himself.' b. Max wast zich Max washes self 'Max washes himself.' We can tell which verbs can be lexically reflexive by seeing which verbs allow coargument binding of the simplex anaphor zich, which is a Pure-reflexive by the criteria previously identified. It never allows Near-reflexive interpretations and allows only sloppy identity under comparative deletion (Sells, Zaenen, and Zec, 1987): (29) a. Ringo scheert zich Ringo shaves self 'Ringo shaves.' (*Near-reflexive) b. Ringo scheert zich-zelf Ringo shaves self-self 'Ringo shaves himself.' (ok Near-reflexive) (30) a. Zij verdedigde zich beter dan Peter She defended self better than Peter 'She defended herself better than Peter defended himself.' '*She defended herself better than Peter defended her.' b. Zij verdedigde zichzelf beter dan Peter She defended self-self better than Peter 'She defended herself better than Peter defended himself.' 'She defended herself better than Peter defended her.'
Anti-Antilocality
241
The fact that (28a) is ungrammatical tells us that only some verbs can be lexically reflexive in Dutch, as opposed to Kannada which allows all verbs to be lexically reflexive. Given this variability in the extent of lexical reflexivization allowed in a language, we might expect to find a language that allowed no predicates to be lexically reflexive. Such a language, if it had an anaphor which did not permit a Near-reflexive interpretation, might block binding by a coargument across the board. That is, in a language where there is no lexical reflexivity, we predict the existence of an anaphor which can never be bound by a coargument. Malayalam is such is language. The Malayalam simplex anaphor, tan, cannot be bound by a coargument whereas the complex anaphor tan-ne-tanne can (Mohanan, 1982; Jayaseelan, 1996): (31) a. *Moohan tan-ne aaraaDhik'k'uNNu Mohan self-ACC worships 'Mohan worships himself.' b. Moohan tan-ne-tanne aaraDhik'k'uNNu Mohan self-ACC-self worships 'Mohan worships himself.' This generalization holds even with verbs which are lexically reflexive in other languages (K.A. Jayaseelan, personal communication): (32) a. *Raaman tan-ne kshauram ceytu Raaman self-ACC shaving did 'Raaman shaved.' b. Raaman tan-ne-tanne kshauram ceytu Raaman self-ACC-self shaving did 'Raaman shaved himself.' Note also that (32b) allows the Near-reflexive interpretation that Raaman shaved a portrayal of himself, as predicted by Condition R. We can conclude that Malayalam has no lexical reflexivity. The fact that tan can never be bound by a coargument indicates that this anaphor does not express Near-Reflexivity and follows from the combination of lexical properties of Malayalam with Condition R. Under this conception, all variation in the behavior of reflexive forms found among the world's languages reduces to lexical variation. The syntactic binding theory is constant, as is Condition R. What varies is the extent of lexical reflexivity permitted by a language and the range of anaphors found in that language. The remaining question is what properties of the lexicon or of lexical items determines the extent of lexical reflexivity and whether a given anaphor allows a Near-reflexive interpretation. Note that although it is not clear at this point what linguistic property determines whether an
242
Jeffrey Lidz
anaphor introduces the Near-reflexive function, it should be trivial for a child learning the language to determine if an anaphor that he encounters does so. Assuming that the child knows which verbs are inherently reflexive and which verbs are not, then if the child sees an anaphor bound by a coargument in the absence of lexical reflexivity, he can conclude that the anaphor is a Near-reflexive. Thus, the child needs have no exposure to sentences involving wax museums and heads of state to determine whether a given form is a Near-reflexive.
5. ALTERNATIVE ANALYSES In this section, I examine some previous accounts of antilocality and show how they are inferior to the Condition R approach advocated here. 5.1. The Hybrid Pronoun Gambit One kind of solution to the problem presented by the Kannada anaphor is to postulate a new kind of pronoun which is partially anaphoric and partially pronominal. This is the tack taken by Mohanan (1982) and Hestvik (1990) for Malayalam and Norwegian, respectively. Mohanan's (1982) solution to the Malayalam antilocality is to allow the simplex anaphor to be both anaphoric and pronominal. Further, he creates a parameterized binding theory in which "governing category" can be left out of Condition A in some languages. Thus, in Malayalam, Condition A states only that an anaphor must be bound while principle B states that a pronominal must be free in its governing category. The Malayalam simplex anaphor tan is given the features [+anaphor] and [+pronominal] and so requires a binder outside of the local domain. This solution will be inadequate for Kannada, however, since the domain in which the simplex anaphor, tannu, must be free is distinct from the domain in which the pronominal, awanu, must be free, as we have seen in examples (8-10). The simplex anaphor allows a clausemate noncoargument antecedent in places where the pronominal does not. Hestvik (1990) also attempts to explain antilocality by giving anaphors a mixed set of features. He develops a binding theory in which anaphors may be marked with a feature indicating whether they are to be bound or free at S-structure or at LF. If an element has the feature [+bs], then it may be locally bound at S-structure. If it has the feature [—bs], then it must be locally free at S-structure. The additional feature for LF binding allows Hestvik to explain the behavior of long-distance reflexives. If an anaphor has the fea-
Anti-Antilocality
243
ture [+bL], then it must be locally bound at LF, while if it has the feature [-bL],it must be locally free at LF. Hestvik follows Pica (1987) and others in assuming that X° reflexives move to INFL at LF. So, if an X° reflexive moves out of its own clause at LF and has the feature [+bL], then it will have the appearance of being long-distance bound, exactly as we saw above in Chinese in (5). To illustrate, consider the following: (33) a. Harii Raajuj tann-annui/*j hoDe-d-a anta heeL-id-a Hari Raaju self-ACC hit-PST-3SM that say-PST-3SM 'Hari said that Raaju hit him.' b. Zhangsani renwei Lisij piping ziji i/j Zhangsan thinks Lisi criticized self 'Zhangsan thinks Lisi criticized him/himself.' The Kannada sentence in (33a) allows only long-distance binding. This follows from Hestvik's system because the anaphor, tannu, is marked with the features [-b s ] and [+bL]. Because it is marked [-b s ], it cannot be bound by Raaju, the only NP which locally c-commands the base position at S-structure. The anaphor moves at LF to the matrix INFL, however, where it can be locally bound by the matrix subject, as illustrated in (34): (34) [ip Hari [cp[ip Raaju t-i hoDeda] anta] heeLida-tannannui The Chinese sentence in (33b) enables us to illustrate how the system handles an anaphor with the features [+bs] and [+bL]. Because ziji has the feature [+bs], it may be locally bound at S-structure. Thus, the embedded subject (Lisi) may be the antecedent for the anaphor. Because ziji also has the feature [+bL], it may optionally move to the matrix INFL at LF where it can be locally bound by the matrix subject, as illustrated in (35): (35) [Lisi iNFL-zyiji [Wangwu INFL-ti xihuan ti] As we have seen, Hestvik must give antilocal reflexives like Kannada11 tannu the features [—bs] and [+bL]. As a consequence, these elements must not be locally bound at S-structure and must be locally bound at LF. Since the feature [—bs] disallows S-structure binding, the anaphor must move out of its own clause at LF, where it can be locally bound, in accordance with the feature [+bL] There are three problems with this approach to distinguishing antilocal reflexives from other reflexives. First, it is purely classificatory and has little explanatory value. By positing the existence of a feature distinguishing long-distance reflexives which can be locally bound from those which cannot, Hestvik merely stipulates and does not derive the distinction between the two types of long-distance anaphor. A satisfactory account of these facts
244
Jeffrey Lidz
will derive the difference between Chinese and Kannada from something other than a stipulation meant only to capture that difference. The Condition R approach meets this criterion by linking the semantic properties of the anaphor to its inability to be bound by a coargument in the absence of lexical reflexivity. Second, Hestvik's approach does not make the right generalization. It is not possible to state the binding conditions for "antilocal" reflexives only in terms of the level of representation at which the binding principles apply. The examples in (8-10) make this second point clear. Kannada tannu can be locally bound at S-structure in certain cases. The claim that these anaphors must be free at S-structure and bound at LF is false. There are both grammatical and ungrammatical examples of these anaphors being bound at S-structure. The appropriate descriptive generalization of this type of anaphor is that it may be locally bound unless its antecedent is an argument of the same predicate as the anaphor itself (cf. Reinhart and Reuland, 1993; Lidz, 1995). Hestvik was aware of the partially overlapping distribution of the simplex and complex anaphors. To account for these facts, repeated here in Kannada and Norwegian, he argues that the subject condition on binding domains depends on the character of the element being bound. (36) a. Hari pustaka-vannu tann-a hindee nooD-id-a Hari book-ACC self-GEN behind see-PST-3SM 'Hari saw the book behind him.' b. Hari pustaka-vannu tanna-taan-ee hindee nooD-id-a Hari book-ACC self-GEN-self-EMPH behind see-PST-3SM 'Hari saw the book behind him.' c. *Hari pustaka-vannu avan-a hindee nooD-id-a Hari book-ACC he-GEN behind see-PST-3SM (37) a. Hari tann-annu mattu tann-a hendati-yannu hoDe-d-a Hari self-ACC and self-GEN wife-ACC hit-PST-3SM 'Hari hit himself and his wife.' b. Hari tann-annu-taane mattu tann-a hendati-yannu hoDe-d-a Hari self-ACC-self and self-GEN wife-ACC hit-PST-3SM 'Hari hit himself and his wife.' c. *Hari avan-annu mattu avan-a hendati-yannu hoDe-d-a Hari he-ACC and he-GEN wife-ACC hit-PST-3SM (38) a. Jon legger boken bak seg Jon puts book behind self 'Jon puts the book behind himself.'
Anti-Antilocality
245
b. Jon legger boken bak seg selv Jon puts book behind self self 'Jon puts the book behind himself.' c. *Jon legger boken bak ham Jon puts book behind him In particular, he argues that only elements that are marked [+b] need to have a subject in their binding domains (cf. Huang, 1983; Chomsky, 1986). Hence, the S-structure binding domains for the simplex and complex anaphors are different. For the simplex anaphor, which is marked [—bs], the PP in (36/38) is a binding domain since it is the smallest category containing the pronoun and a governor for it. A subject is not required in the binding domain because a [—b] element is necessarily free (at the specified level) and so looking for an antecedent is not required. The simplex anaphor is correctly free in the PP at S-structure. Subsequent LF movement to INFL allows the simplex anaphor to be bound at LF by the matrix subject. For the complex anaphor, which is marked [+bs], the PP in (36/38) is not a binding domain because it does not contain a subject. The smallest possible governing category is the entire IP because this is the smallest governing category in which the anaphor could possibly be bound. The complex anaphor is correctly bound in this domain at S-structure. The conclusion we reach with respect to these examples is that cases where the antilocal and local anaphors are in overlapping distribution arise from a conspiracy of principles which make them look alike even though they are different. In Hestvik's system, the antilocal anaphor can be bound by the subject in (36a/38a) because the subject is outside of the S-structure binding domain of the anaphor. The local anaphor can be bound by the subject in (36b/38b) because the subject is inside the S-structure binding domain of the anaphor. the resulting system is problematic precisely because of this conspiracy. In the place where the two anaphors are alike, the theory does not give a unified explanation. The Condition R approach, on the other hand, treats all anaphors equally as far as the calculation of binding domains is concerned. What distinguishes them is simply their semantics. The third problem for the Hestvik solution comes from the licit cases of coargument binding, as in (39): (39) a. Hari tann-annu hoDe-du-koND-a Hari self-ACC hit-PP-REFL.PST-3SM 'Hari hit himself.' b. Jon vasket seg Jon washes self 'Jon washes.'
246
Jeffrey Lidz
Hestvik argues that (39b) does not involve binding at all, but rather that seg is a nonargument element that indicates that lexical reflexivization has occurred. Hence, the binding theory does not apply to (39b) (cf. Grimshaw, 1982; Lidz, 1996). This solution does not work for (39a), however. While it is true that a morphological operation of reflexivization has applied to the predicate in (39a), it is not the case that the morphological realization of this operation is the anaphor. Looking at the two languages side by side tells us that properties of the predicate can alter binding possibilities for an anaphor bound by another argument of that predicate. In both languages, the anaphor in (39) is actually an anaphor and reflexivizing the predicate allows this anaphor to be locally bound despite the general prohibition on coargument antecedents. It is important to emphasize, however, that in this chapter, the prohibition on coargument binding is not stated as a prohibition anywhere in the grammar. As far as the binding theory is concerned, coargument antecedents are perfectly acceptable. It is only when such configurations yield interpretations incompatible with Condition R that coargument binding is blocked. 5.2. Predicate-centered Binding Theory Reinhart and Reuland (1993), take a different tack in explaining antilocality, arguing against a nominal approach to binding. Instead, they claim that reflexivity is a property of predicates. On this view, the role of reflexive pronouns in language is not to express coreference but rather to reflexivize predicates. The distribution of anaphors and pronominals is not determined by reference, a property of NPs, but by reflexivity, a property of predicates. From this predicate-centered perspective, an anaphor is still defined as an NP which is referentially deficient in the relevant respect, but the anaphors are broken into two types, those that reflexivize the predicates that they are arguments of and those that do not. This division is intended to capture the apparent antilocality of anaphors like the Dutch anaphor zich, which, like tannu, must be bound, but not by a coargument. The behavior of zich contrasts with that of zichzelf, which may be bound by a coargument. (40) a. *Max haat zich Max hates self 'Max hates himself.' b. Max haat zichzelf Max hates selfself 'Max hates himself.' c. Max hoorde mij over zich praten Max heard me about self talk 'Max heard me talk about him.'
Anti-Antilocality
247
d. *Max hoorde mij over zichzelf praten Max heard me about selfself talk 'Max heard me talk about him.' While both zich and zichzelf are anaphors in the sense that they require a c-commanding antecedent, only zichzelf can reflexivize a predicate and so only zichzelf can be bound by a coargument. Technically, this is implemented through the following binding conditions: (41)12 Condition A: A reflexive-marked (syntactic) predicate is reflexive Condition B: A reflexive (semantic) predicate is reflexive-marked These conditions depend on the following definitions: (42) a. A predicate is reflexive iff it has two coindexed arguments b. A predicate is reflexive-marked iff (i) it is lexically reflexive; or (ii) one of its arguments is a SELF anaphor (43) A SELF-anaphor is a morphologically complex anaphor Returning now to the examples in (40), (40a) is ungrammatical as a violation of Condition B. The predicate is reflexive (it has two coreferential arguments) but not reflexive-marked (it is not lexically reflexive and neither of its arguments is a SELF anaphor). Both conditions are satisfied in (40b); here, the predicate is both reflexive and reflexive-marked (since one of its arguments is a SELF-anaphor). In the long-distance cases, (40c) trivially satisfies both conditions since there are no reflexive predicates and no reflexive-marked predicates, while (40d) violates Condition A because the embedded predicate is reflexive-marked (via the SELF anaphor) but not reflexive. The intuition behind Reinhart and Reuland's (1993) proposal is that there is a one-to-one correspondence between a particular formal property (reflexive-marked) and a particular semantic property (reflexive). The definition of reflexive-marked, however, is a disjunction; there are two ways for a predicate to be reflexive-marked. These two modes of reflexive-marking are required to explain the contrast in (44): (44) a. *Max haat zich Max hates self 'Max hates himself.' b. Max wast zich Max washes self 'Max washes himself.'
248
Jeffrey Lidz
Here, we see that the restriction against coargument binding of zich depends on the predicate, as in the Condition R approach. Since the syntactic relation between the anaphor and its antecedent is identical in (44a) and (44b) but the binding possibilities for the two cases diverge, we can conclude that it is the predicate which determines the availability of coargument binding for the anaphor zich. Under Reinhart and Reuland's (1993) predicate-centered binding theory, the effect is achieved by saying that some predicates are inherently reflexive and, thus, inherently reflexive-marked. While (44b) would appear to violate Condition B since the predicate is reflexive but not reflexive-marked, in fact, the condition is satisfied because the predicate is lexically reflexive and thus does not need overt reflexive-marking by a SELF anaphor. Lidz (1995) extends the definition of reflexive-marked predicate to include morphological reflexivity, that is, to include those languages which have a morpheme which is required on reflexive predicates. In Kannada, the anaphor tannu cannot be bound by a coargument in the absence of the verbal reflexive morpheme -koLLu: (45) a. *Hari tann-annu hoDe-d-a Hari self-ACC hit-PST-3SM 'Hari hit himself.'
b. Hari tann-annu hoDe-du-koND-a Hari self-ACC hit-PP-REFL.PST-3SM 'Hari hit himself.' Under this approach, (45a) is ungrammatical because the predicate is reflexive but not reflexive-marked, in violation of Condition B. In (45b), however, the verbal reflexive serves to morphologically reflexive-mark the predicate and so both binding conditions are met. Thus, we see that reflexive marking can be achieved via a SELF anaphor or lexically, with some languages marking the lexically reflexive predicates morphologically. The fact that the morphologically complex anaphor can occur in the absence of the verbal reflexive is also in accord with the predictions of Reinhart and Reuland (1993): (46) Hari tann-annu-taane hoDe-d-a Hari self-ACC-self hit-PST-3SM 'Hari hit himself.' Here, the reflexive predicate is reflexive-marked by the complex anaphor. Reinhart and Reuland's (1993) proposal is similar to Condition R in linking antilocality to properties of predicates. The two approaches differ in how reflexive predication relates to the distribution of anaphors. Under
Anti-Antilocality
249
Condition R, the distribution of Near-reflexive anaphors is unaffected by the theory of reflexive predication. Under Reinhart and Reuland's (1993) approach, reflexive-marking by a SELF anaphor is equivalent to reflexive marking by lexical means. Hence, Reinhart and Reuland's (1993) approach falsely predicts that all reflexive-marked predicates are synonymous. As noted previously, reflexive-marking can be realized lexically (sometimes through a verbal affix) or syntactically (through a SELF anaphor). The binding conditions themselves, however, make no reference to the manner of reflexive marking. The binding conditions require only that reflexive-marking, no matter how it is realized, corresponds to semantic reflexivity. As a consequence, the theory predicts that reflexive-marked predicates are semantically uniform. Since the binding conditions apply at LF and only make reference to reflexive-marking and not to the manner of reflexive-marking, it follows that all reflexive predicates have the same interpretation at LF. Semantically, reflexive-marked predicates should form a natural class. This prediction is not borne out, however, as we have seen. The one-to-one mapping between the semantic property of being reflexive and the formal property of being reflexive-marked breaks down when we examine the semantic properties of these two types of reflexive-marked predicate more closely. Predicates which are reflexive-marked by anaphors allow Near-reflexive interpretations while predicates which are reflexive-marked by lexical means do not. A second failing in the Reinhart and Reuland (1993) approach concerns the morphological character of SELF anaphors. They claim that SELF anaphors are morphologically complex. Hence, the expectation is that no morphologically simplex anaphor will be able to reflexivize a predicate. This is clearly false, as demonstrated by the Chinese and Russian simplex anaphors discussed previously. In order to accommodate these anaphors, Reinhart and Reuland (1993) would have to say that these forms are ambiguous between SELF anaphors and SE anaphors. If they were not, we would expect either local binding or long-distance binding all the time. Hence, the only criterion for determining whether an anaphor is a SELF anaphor is whether it can be bound by a coargument. The Condition R theory, on the other hand, says nothing about the morphological form of an anaphor and can explain why some anaphors can be coargument bound and others cannot on the basis of the interpretive properties of the anaphors themselves. Any anaphor which can be bound by a coargument, independent of morphological structure, introduces the Near-reflexive function and thus averts Condition R. Thus, the semantic partitioning of anaphors is superior to a morphological one. Reinhart and Reuland's (1993) important observation that the behavior of antilocal anaphors is explained by properties of reflexive predicates and not by binding configurations is maintained in our
250
Jeffrey Lidz
approach. The difference between the approaches concerns the characterization of anaphor types. For Reinhart and Reuland (1993), the two kinds of anaphor (local and antilocal) are distinguised syntactically, based on whether the anaphor has the power to reflexive-mark a predicate. In the current approach, however, the two kinds of anaphor are syntactically alike but semantically distinguished.
6. CONCLUSIONS In this chapter, I have shown that antilocal anaphors do not require any syntactic analysis beyond that of anaphors in general. Instead, the coargument restriction on these anaphors falls out from their semantic properties in concert with Condition R. Typologically, there are two classes of anaphor: Pure-reflexives and Near-reflexives. A Pure-reflexive anaphor requires complete identity with its antecedent while a Near-reflexive anaphor can be related to its antecedent via the Near-reflexive function. The distinction between Pure- and Near-reflexives is independent of the distinction between local and long-distance anaphors; there are long-distance anaphors which are Pure-reflexives (Kannada tannu) and long-distance anaphors which are Near-reflexives (Chinese ziji). Further, those anaphors which have been characterized as antilocal in the past are really Pure-reflexives. Syntactically, they allow binding by a coargument; however, in the absence of lexical reflexivity these anaphors cannot be bound by a coargument without inducing Condition R, which requires semantically reflexive predicates to be lexically reflexive. Thus, we do not need a class of anaphors which are syntactically specified as being antilocal since the antilocality of such anaphors can be attributed to their semantic representations in conjunction with Condition R. The head-movement theory of long-distance binding still naturally accounts for the difference between local and long-distance reflexives. As far as this theory is concerned, there is no difference between the so-called antilocal anaphors and other long-distance anaphors, since the antilocality effect is due to Condition R, which is part of the interface between the semantic component and the lexicon. Finally, Condition R has nothing to say about the behavior of anaphors that introduce the Near-Reflexive function or about Pure-reflexives outside of the coargument domain. It seems clear that the distribution of these anaphors in Kannada is best characterized by the standard binding theory of Chomsky (1981/1986), since the complementarity between pronominals and anaphors is maintained in all noncoargument contexts.
Anti-Antilocality
251
ACKNOWLEDGMENTS This chapter has benefited from discussion with various friends and colleagues including Abdellatif Alghadi, R. Amritavalli, Rajesh Bhatt, Tonia Bleam, Luigi Burzio, Robin Clark, Peter Cole, Alice Davison, Bob Frank, Bill Frawley, Jim Gair, Arild Hestvik, Gaby Hermon, Bill Idsardi, K. A. Jayaseelan, P. Madhavan, Bill Philip, Pierre Pica, Tom Purnell, Tanya Reinhart, Eric Reuland, K.V. Tirumalesh, Christina Tortora, and Alexander Williams. Thanks also to audiences at Delhi University, Central Institute of English and Foreign Languages, Central Institute of Indian Languages, Jawaharlal Nehru University, and the Workshop on Long-distance Reflexives at the 1997 LSA Summer Institute. This work was supported in part by National Science Foundation Grant BNS-9121167 while at the University of Delaware and in part by National Institute of Health Grant DC-00282-02 and the Institute for Research in Cognitive Science while at the University of Pennsylvania.
NOTES 1 A11 Kannada data was collected between 1994 and 1998 from various speakers including Ananda Murthy, Hari Pujar, Raama Brasad, B. Srinivas, K.V. Tirumalesh, S. Chandrashekar, and R. Amritavalli. Abbreviations used in the glosses are: ACC = accusative, CAUS = causative, COMP = comparative, EMPH = emphatic, F = feminine, GEN = genitive, M = masculine, NOM = nominative, NMNL = nominalization, NPST = nonpast, PP = participle, PST = past, REFL = reflexive, S = singular, 1 = 1st person, 2 = 2nd person, 3 = 3rd person. 2 One possible objection to this approach is that "real" reflexive clitics can never be long-distance bound. However, there are two responses to this objection. First, given the theory of verbal reflexives developed in Lidz (1996,1997) in which verbal reflexives (including reflexive clitics) are not nominal anaphors, we have no reason to expect "reflexive" clitics to behave like anaphors. Second, Kayne (1989) shows that pronominal clitics in Romance can overtly undergo long head-movement of this sort, providing support for the idea that clitic pronouns can move across clause boundaries. 3 These representations are simplified so as to show only the traces in the base position and INFL. In order to satisfy the Head-Movement Constraint (Chomsky, 1986), movement must actually be through all heads between the base position and the landing site. 4 There are many speakers of Kannada for whom the morphologically complex anaphor is not allowed in the absence of the verbal reflexive. I will put aside the discussion of these varieties for the purposes of this chapter, though detailed examination of the differences is ultimately required. 5 Most Kannada speakers require the verb phrase to be repeated in comparative constructions. 6 I leave open the exact content of the Near-reflexive function for the time being,
252
Jeffrey Lidz
though most likely this function is something like "representation of (x)." See Jackendoff (1978,1992) and Fauconnier (1985) for some illuminating discussion. 7 Note that the relationship between lexical reflexivity and the "reflexive" morpheme —koLLu is indirect. In other words, it is not the case that lexical reflexivity is always marked by —koLLu and it is also not the case that the morpheme always indicates that lexical reflexivization has applied. As for the first case, we find tannu bound by a coargument in the absence of —koLLu in the dative subject construction (Sridhar, 1979): (i)
Hari-ge tann-u isht-al-illa Hari-DAT self-NOM like-INF-NEG 'Hari doesn't like himself.'
The morpheme —koLLu is incompatible with this construction for independent reasons (Lidz, 1996; 1998), but the morpheme's absence should not be taken as evidence that there is no lexical reflexivity. We do not get near-reflexive interpretations in sentences like (i), indicating that the predicate is lexically reflexive, despite its lack of morphological marking. Turning to the second case, we also find the morphene —koLLu in nonreflexive contexts such as (ii) (Amritavalli, 1984; Lidz, 1996): (ii)
baagil-u tere-du-koND-itu door-NOM open-PP-REFL.PST-3SN 'The door opened.'
This example tells us only that lexical reflexivity is not the only factor which licenses -koLLu. Analysis of the full set of conditions responsible for the surfacing of this morphene would take us beyond the scope of this chapter. See Lidz (1996; 1998) for details. 8 It is possible that Condition R is actually inoperative here also and so there is no redundancy. It may be that pronominals always introduce the Near-reflexive function and so are exempt from Condition R. For example, we might say (i) in a situation where what Ringo thought would be taller was the statue that portrays him: (i)
Ringo thought he would be taller
If this is correct, then Condition R does not apply to pronominals and there is no redundancy. 9 Chinese data comes from Nianwen (Bert) Xue, personal communication. 10 Russian data comes from Alexander Lehrmann, personal communication. 11 Hestvik does not actually discuss Kannada, but gives an extensive discussion of Norwegian seg, whose distribution is identical to Kannada tannu. 12 The parenthetical material in the Reinhart and Reuland (1993) binding conditions is irrelevant to the current chapter. It is included in their theory because their Condition A holds of LF representations while their Condition B holds of semantic representations. This extra stipulation about levels of application is required to handle facts concerning conjoined NPs such as (i-ii): (i)
The queen invited Max and herself for dinner
Anti-Antilocality
(ii)
253
*The queen invited Max and her for dinner
See Safir (1997) for criticism of this aspect of Reinhart and Reuland's (1993) theory.
REFERENCES Amritavalli, R. (1984). "Anaphorization in Dravidian," Central Institute of English and Foreign Languages Working Papers in Linguistics 1,1-31. Amritavalli, R. (1991). "Lexical anaphora in Kannada," Hyderabad: Central Institute of English and Foreign Languages. Battistella, E. (1989). "Chinese reflexivization: A Movement to INFL approach," Linguistics 27,987-1012. Bhat, D. N. S. (1981). Pronominalization. Pune, India: Deccan College Postgraduate Research Institute. Chomsky, N. (1981). Lectures on Government and Binding. Dordrecht, Netherlands: Foris. Chomsky, N. (1986). Knowledge of Language. New York: Praeger. Cole, P., Hermon, G. and Sung, L. M. (1990). "Principles and parameters of Long-distance Reflexives," Linguistic Inquiry 21,1-22. Cole, P., and Sung, L. M. (1994). "Head-movement and long-distance reflexives," Linguistic Inquiry 25,355-406. Faltz, L. (1977). Reflexivization: A Study in Universal Syntax. Ph.D. dissertation, University of California at Berkeley. Fauconnier, G. (1985). Mental Spaces. Cambridge: MIT Press. Grimshaw, J. (1982). "On the lexical representation of Romance reflexive clitics." In The Mental Representation of Grammatical Relations J. Bresnan (ed.) Cambridge: MIT Press. Hestvik, A. (1990). LF-Movement of Pronouns and the Computation of Binding Domains. Ph.D. dissertation, Brandeis University. Huang, J. (1983). "A note on the binding theory," Linguistic Inquiry 14,554-561. Jackendoff, R. (1978). "Grammar as evidence for conceptual structure." In Linguistic Theory and Psychological Reality (M. Halle, J. Bresnan, and G. Miller, eds.) Cambridge: MIT Press. Jackendoff, R. (1992). "Madame Tussaud meets the binding theory," Natural Language and Linguistic Theory 10,1-31. Jayaseelan, K. A. (1996). "Anaphors as Pronouns," Studia Linguistica 50,207-255. Kayne, R. (1989). "Null subjects and clitic climbing." In The Null Subject Parameter (O. Jaeggli and K. Safir, eds.) Dordrecht: Kluwer. Lidz, J. (1995). "Morphological reflexive-marking: Evidence from Kannada," Linguistic Inquiry 26,705-710. Lidz, J. (1996). Dimensions of Reflexivity, Ph.D. dissertation. Univesity of Delaware. Lidz, J. (1997). "The Argument Structure of Verbal Reflexives," University of Pennsylvania. [Unpublished Manuscript]
254
Jeffrey Lidz
Lidz, J. (1998) "Valency in Kannada: Evidence for interpretive morphology," in Penn Working Papers in Linguistics 5.2,37-64. Mohanan, K. P. (1982). "Grammatical relations and anaphora in Malayalam." In MIT Working Papers in Linguistics, Papers in Syntax (A. Marantz and T. Stowell, eds.) pp. 163-190. Cambridge: MIT Press. Pica, P. (1987). "On the nature of the reflexivization cycle," in Proceedings of NELS 17, 483-499. Reinhart, T., and Reuland, E. (1993). "Reflexivity," Linguistic Inquiry 24,657-720. Reuland, E. (1995). "Primitives of Binding." University of Utrecht. [Unpublished manuscript] Safir, K. (1997). "Symmetry and Unity in the Theory of Binding." Rutgers University. [Unpublished manuscript] Sells, P., Zaenen, A., and Zee, D. (1987). "Reflexivization variation: Relations between syntax, semantics and lexical structure." In Working Papers in Grammatical Theory and Discourse Structure (M. lida, S. Wechsler, and D. Zec, eds.). Stanford: CSLI. Sridhar, S. N. (1979). "Dative subjects and the notion of subject," Lingua 49,99-125. Yadurajan, K. S. (1987). "Binding theory in Dravidian," Cornell Working Papers in Linguistics: Ithaca.
LONG-DISTANCE REFLEXIVIZATION IN CHECHEN AND INGUSH JOHANNA NICHOLS Department of Slavic Languages University of California, Berkeley
1. INTRODUCTION Chechen and Ingush are very closely related languages belonging to the Northeast Caucasian, or Nakh-Daghestanian, family and spoken in the north central Caucasus. Typologically, they are head-final, suffixing, ergative, case-using, dependent-marking, and clause-chaining languages. They make use of long-distance reflexivization (LDR) which is so regular, so extensive, and so unambiguously reflexive and long-distance, that I propose it be taken as canonical long-distance reflexivization. Chechen-Ingush LDR makes use of dedicated reflexive pronouns. The antecedent may be in either a main or a non-main clause, and the antecedent reflexivizes any coreferent that it c-commands. (The head-final word order means that reflexives generally precede their antecedents,often at some distance.) There are virtually no blocking effects; in particular, any number of clauses may intervene between antecedent and reflexive, as may subjects different in person-number categories from the antecedent, and these intervening clauses and subjects may sometimes create ambiguity but almost never block LDR. The domain of LDR is the sentence (which consists of a tensed main clause and any number of chained or complement clauses), and LDR is almost unblockable within sentences and absolutely blocked by sentence boundaries. LDR is abundant in natural texts of all styles and genres. It is clearly Syntax and Semantics, Volume 33 Long-Distance Reflexives
255
Copyright © 2001 by Academic Press All rights of reproduction in any form reserved. 0092-4563/00 $35.00
256
Johanna Nichols
distinct from logophoric reflexivization, which also occurs in Chechen and Ingush. The purpose of this chapter is to give a full description of Chechen-Ingush LDR, with particular attention to issues of theoretical interest. Example (1), from a 19th-century folklore text, shows typical traditional narrative style in Chechen.1 San maara 'my husband' in the final clause reflexivizes all overt coreferents in preceding clauses, and the hearer begins tracking reflexives several clauses before the antecedent appears. In (2), from the same text, the first clause (which is chained) has a reflexive indirect object sajna 'myself and a subject of another person-number category, and the second (main) clause contains the antecedent of the reflexive. (1) Chechen (El'darxanov, 1900:3) (transcription of vowels uncorrected; a-f are nonfinal chained clauses (a) is the continuation of (a), and (g) is the main clause): a. ... heette 'a as still & ls.Erg b. "0 hwo ca loorush (Erg) 2s.Nom not respect. CV c. ba'ajtana hwuona hwan vashas hwaalxarin kuog" sent 2s.Dat 2s.Gen brother.Erg front leg (a) sheega aalchahw, SsRFL.All said.CV d. cunax cwa duosh 'a dina, this.Loc (Erg) one word & do.CV e. da'ajtinchu deeq'e 'a ca hwozhush, (Erg) sent.Obl part.All & not look.CV f. shen vezherashcin dov 'a dina, (Erg) 3s.RFL brothers.Instr quarrel & made.CV g. shuega joollun huma jeeq'an qajkina san maara 2p.All be.Ppl thing divide.In called ls.Gen husband.Erg 'Nonetheless, when I told him that his brother had slighted him in sending him the front leg, my husband took this as grounds (for offense) and, not even looking at the part that had been sent, started a quarrel with his brothers, and called you to divide their household.' Lit.: 'Nonetheless, when I told himself, 'Your brother, not respecting you, sent you the front leg', having taken offense at this, not even looking at the part that had been sent, having quarreled with his brothers, my husband called you to divide the possessions.' (2) Ash sajna purma lahwaar, 2p.Erg IsRFL.Dat agreement give.CV
Long-distance Reflexivization in Chechen and Ingush
257
TABLE 1 SELECTED PLAIN AND REFLEXIVE PRONOUNS IN INGUSH First person Case Nominative Genitive Dative Ergative Allative
Third person
Nonreflexive
Reflexive
Nonreflexive
Reflexive
so sy suona aaz suoga
sie sej sejna ejsa sejga
yz cyn cynna cuo cynga
shie shii shiina shie shiiga
as cwa duosh eerdar shuega ls.Erg one word say.Cnd 2p.All 'If you would permit me, I would say a word to you' lit.: 'If you would permit myself, I would say a word to you.' (El'darxanov, 1900) Chechen and Ingush have a full set of reflexive pronouns which decline in all the cases and correspond to all the person-number categories of the regular personal pronouns. Each reflexive pronoun is a monomorphemic root with lexical meaning such as "first-person singular reflexive" or "third-person plural reflexive." The first- and second-person reflexives resemble the nonreflexives in consonantism but not vocalism, but are not morphologically segmentable; the third-person reflexive bears no resemblance and no etymological connection to the nonreflexive. Table 1 shows some first- and third-person pronouns from Ingush. The reflexive pronouns have three syntactically distinct functions in Chechen and Ingush: local reflexivization, logophoric reflexivization, and long-distance reflexivization.
2. LOCAL REFLEXIVIZATION Local reflexivization is obligatory: the subject reflexivizes any coreferent in its clause. Subjects can be in various cases—nominative, ergative, dative, or genitive—as required by the verb, and reflexivization occurs regardless of the case of the subject and regardless of the case or syntactic function of the object. In the following examples, coreferents to the subject must be reflexivized. Nonreflexive third-person pronouns in (6-8) are necessarily
258
Johanna Nichols
interpreted as not coreferential to a third-person subject. (Here and following, examples are Ingush unless identified as Chechen.) (3) a. Aaz sejna kinashjka iicar Is.Erg 1sRFL.Dat book.Nom bought 'I bought myself a book.' b. *Aaz suona kinashjka iicar 1s-Erg ls.Dat book.Nom bought 'I bought me a book.' (4) a. Aaz sej bierazhta kinashjka iicaad ls.Erg 1sRFL.Gen children.Dat book.Nom bought 'I bought my children a book.' b. *Aaz sy bierazhta kinashjka iicaad Is.Erg ls.Gen children.Dat book.Nom bought 'I bought my children a book.' (5) a. So sej veshiila laqagh vy ls.Nom 1sRFL.Gen brother.Cmp taller be.Prs 'I am taller than my brother.' b. *So sy veshiila laqagh vy ls.Nom ls.Gen brother.Cmp taller be.Prs 'I am taller than my brother.' (6) a. Muusaaz shiina kinashjka iicar M.Erg 3sRFL.Dat book.Nom bought 'Musa bought himself a book.' b. Muusaaz cynna kinashjka iicar M.Erg 3s.Dat book.Nom bought 'Musaj bought himj a book.' (7) a. Muusaaz shii bierazhta kinashjka icaad M.Erg 3sRFL.Gen children.Dat book bought 'Musaj bought hisi own children a book.' b. Muusaaz cyn bierazhta kinashjka icaad M.Erg 3s.Gen children.Dat book bought 'Musaj bought hisi children a book.' (8) a. Muusaa shii veshiila laqagh vy M.Nom 3sRFL.Gen brother.Comp taller be.Prs 'Musaj is taller than hisi own brother.' b. Muusaa cyn veshiila laqagh vy M.Nom 3s.Gen brother.Comp taller be.Prs 'Musai is taller than hisj brother.'
Long-distance Reflexivization in Chechen and Ingush
259
(9) a. Shii vosha handza-'a bwarjg-v»ejna-v aac cynna 3s.RFL.Gen brother.Nom still eye-saw-Neg 3s.Dat 'Hei still hasn't seen hisj (own) brother.' b. Cyn vosha handza-'a bwarjg-v• ejna-v• aac cynna 3s. Gen brother.Nom still eye-saw-Neg 3s.Dat 'Hej still hasn't seen hisj brother.'
3. LOGOPHORIC REFLEXIVIZATION Logophoric reflexivization in Chechen-Ingush differs from other reflexivization in the way in which person categories shift and in not being sentence-bounded. Logophoric reflexivization occurs in semi-direct speech, in which the quoted matter is identical to the reported speech act except that coreferents to the speaker are reflexivized and the clause is marked with a quotative particle. Logophoric reflexives are always third person in form, despite their coreference to the speaker. In (l0a), which contains semi-direct speech with logophoric reflexivization, the quoted clause contains an imperative verb form (which is necessarily second person in reference) just as the reported speech act did, but the coreferent to the speaker is third-person reflexive rather than first-person nonreflexive as it would have been in the actual original speech, shown in (l0b). (10) a. Cuo ch'woagha diexar deadar suoga, shiina axcha 3s.Erg very request made ls.All 3sRFL.Dat money daa eanna give.Impr Quot 'He implored me to lend him money.' b. Suona axcha daa ls.Dat money give.Impr 'Give me money.' In contrast to semi-direct speech with logophoric reflexivization, direct speech, shown in (11), retains the original person of coreferents to the speaker and has no reflexivization or quotative particle. (11)
T'aaqqa "So c'a v•aallalca yz hwa ma diellalahw," so 1s home come.CV it here don't open.Impr eannad cuo cu siesag.aga said 3s.Erg Dem wife-All 'So he said to his wife, "Don't open it until I get back.'"
260
Johanna Nichols
Examples (12) and (13) show logophoric reflexivization in lengthy examples of semi-direct speech from Chechen narrative. In both, logophoric reflexivization of coreferents to the speaker continues through long sequences of finite sentences, which contain unshifted second-person pronouns, intact imperatives, and so on as in the original speech acts. In (12), the speaker is third person. In (13), the speaker is first person, yet the logophoric reflexives are third person. Fixed third person in logophoric reflexives is the only shifting that occurs in semi-direct speech. (I have not been able to elicit analogs to (13) from Ingush.) (12) Chechen Irsuo zhuop lo : Happiness.Erg answer gives Shaa 'a shun duezal-ehw duqqa 'a daexniera,... 3sRFL.Nom 2s.Gen family.Loc lots & lived shun duezalehw ban b•ara, shun duezalehw berkat 2s.Gen family.Loc agreement was 2s. Gen family.Loc abundance d•ara. was. Bart b•ol-chu mett-ehw irsa 'a mettig b•u, agreement be-Ppl place.Loc happiness.Gen & place be.Prs, cundela wiiniera shu. . .. Shie juxa-d•aan jish jaac" therefore visited 2s.Nom 3sRFL back-go.Inf reason isn't" oolii dist xylu Irs cynga. Quot speak Aux H. 3s-All 'Happiness answers: "I lived in your household for a long time; there was harmony and abundance in your household. A place where there is agreement is the place of happiness, that's why I stayed with you. [Several sentences intervene about the rise of discord in the household.] There's no reason for me to come back," says Happiness to him.' (13) Chechen . . . Cuo juxa 'a diexar d•o : 3s.Erg back & request make Irs, juxa-d•aan d•iezara hwo. H, back-go.Inf want.Impf 2s-Nom Shaa v•oqqa 'a xilliera, t'aaqqa shaa dynien-chohw joqqurg 3s.RFL old & be.Pst so 3s.RFL world-in take.Fut.Ppl
Long-distance Reflexivization in Chechen and Ingush
miel xaan ju ca how much time is not shien doqqa diexar 3s.RFL great request
261
xae'ara shiena, know.Impf 3s.RFL.Dat du hwuega 'a, hwo juxa-daan diezar, is 2s.All & 2s.Nom back-go.Inf should.Nz
oolii. Quot 'He [the old man] asks in return: "Happiness, please come back. I'm old, I don't know how much more time I have to spend in this world, and my great request to you is that you come back.'" Example (14) is from a Chechen dialect folk song (Maciev, 1965:84). Third-person singular reflexive pronouns are used throughout the song to refer to the singer (who is never referred to with first-person pronouns but only with these reflexives and nouns such as 'daughter' and 'bride'). (14) Sarahw siri dinna baj-t'e barguolash ma jettahw, in.evening gray horse.Dat meadow-on hobble Neg throw.Imp k'anat, shiex dog dillaza young man 3s.RFL.Loc heart D.cover.CV.Neg 'Don't hobble your horse on the meadow in the evening, young man, if you haven't lost hope in me' (lit. 'if you haven't lost hope in oneself) (dog dilla, lit. 'cover the heart', is an idiom meaning 'lose hope.')
4. ORDINARY LONG-DISTANCE REFLEXIVIZATION Nonlogophoric, or ordinary, long-distance reflexivization applies almost without limit within tensed sentences, reflexivizing any coreferent to any subject and producing examples like (1) preceding. The following subsections illustrate its properties in more detail. 4.1. Sentence-bounded LDR does not apply between sentences (including coordinated sentences). Examples (15)-(16) each contain two conjoined finite clauses with a coordinating conjunction heata = 'a 'nonetheless, however, anyway, still.' (15) a. Aaz kuoch xoza jaac ealar cynga, heata'a iicar cuo ls.Erg dress pretty isn't said 3s.All, still bought 3s.Erg
yz. it.Nom 'I told her the dress wasn't pretty, but she bought it anyway.'
262
Johanna Nichols
b. *Aaz kuoch xoza jaac ealar shiiga, heata'a iicar ls.Erg dress pretty isn't said 3s.RFL.All, still bought cuo yz. 3s.Erg it id. (lit. 'I told herself the dress wasn't pretty, but she bought it anyway.') (16) a. Aaz kuoch xoza jaac ealar cynga, heata'a cuo suona ls.Erg dress pretty isn't said 3s.All, still 3s.Erg 1s.Dat iicar yz bought it 'I told her the dress wasn't pretty, but she still bought it for me.' b. *Aaz kuoch xoza jaac ealar cynga, heata'a cuo I.Erg dress pretty isn't said her. All, still 3s.Erg sejna iicar yz ls.RFL.Dat bought it.Nom lit. 'I told her the dress wasn't pretty, but she still bought it for myself.'
4.2. Serialization LDR applies in the tightly linked kind of chained clause termed cosubordination by Foley and Van Valin, 1984; Van Valin, 1993; and serialization by Hale, 1991: a clause in which one argument is actually shared between two clauses. Chechen-Ingush serialization is readily identifiable fom the converb forms, VS word order in the main clause, prosody, the clitic particle '« following the pre-converb argument, and other properties (see Peterson, 2000, Nichols, 1994a:54ff.; 1994b:133ff.). The shared argument (often a subject in both clauses) takes its case marking from the main clause. As there is no separate token of the shared argument in the serialized clause, it cannot be reflexivized there. LDR affects any other coreferent to the main-clause subject in the serialized clause, however. The following examples have the serializing converb in *-na. In (17), there is only one overt token of the shared argument Muusaa; in (18), there is an (ungrammatical) overt reflexive subject in the serialized clause instead of sharing; in (19), a possessor coreferential to Muusaa and to the shared argument (subject in the serialized clause) is reflexivized. (17)
deana axcha 'a danna, aara-vealar Muusaa (=Musa) father.Dat money gave.CV out-went Musa 'Musa gave (his) father money and went out.'
Long-distance Reflexivization in Chechen and Ingush
263
(18) *Shie deana axcha 'a danna, aara-vealar Muusaa 3sRFL.Erg father.Dat money gave.CV out-went Musa id. (19)
shii deana axcha 'a danna, (=Musa) 3sRFL.Gen father-Dat money gave.CV aara-vealar Muusaa out-went Musa 'Musaj gave hisi father money and went out.'
Compare chained adverbial clauses, where coreferential deletion creates a zero that can be reflexivized. The following examples use the temporal sequenced converb in -cha. The zero in the chained clause is most easily interpreted as unspecified or as anaphoric with an antecedent outside the sentence, but it can also be understood as coreferential to the main-clause subject Musa. The reflexive in (21) is obligatorily coreferential to Musa. (20) f deana axcha dalcha, aara-vealar Muusaa father.Dat money give.CVout-went Musa 'When his father had been given money, Musa went out' 'When hei had given his father money, Musai,j went out.' (21) Shie deana axcha dalcha, aara-vealar Muusaa 3sRFL.Erg father.Dat money give.CV out-went Musa 'When hei had given his father money, Musai went out.' Serialized purpose clauses have two shared arguments: one is subject in both clauses and one is object in both. Again, neither of these can be reflexivized, but any other coreferent to the main subject can be. (22) Aaz sejna [ diesha ] kinashjka iicar Is-Erg ls.RFL.Dat (=1) (=book) read.Inf book.Nom bought 'I bought myself a book to read' (not *'I bought a book to read to myself.') (23) Aaz Wadrahwmaanaa [ sej bierazhta ls.Erg Abdurahman-Dat (=Abd.) 1sRFL.Gen children.Dat (=book) diesha] kinashjka iicaad read.Inf book bought 'I bought Abdurahman a book to read to my children.' 4.3. Chained adjunct clauses Adjunct clauses have converbs with various adverbial meanings such as 'when,' 'while,' 'if,' etc. LDR affects any coreferent to the subject. The first
264
Johanna Nichols
three examples following have the temporal sequenced converb ('when, after'). The fourth has an adjunct purpose clause. (24) Aaz shiina j•iittacha, aara-v•ealar Muusaa ls.Erg 3s.RFL.Dat hit.CV out-went Musa.Nom 'When I hit him, Musa went out' (lit. 'When I hit himself . ..') (25) Suona shie bwarg-v•ejcha, hwa-aara-v•ealar yz ls.Dat 3s.RFLNom eye-saw.CV here-out-came Ss.Nom 'When I saw him, he came out' (lit. 'When I saw himself . ..') (26) Shiina Bjersnaq'uo axcha dalcha, Oarcnaq' 3s.RFL.Dat B.Erg money gave.CV O.Nom hwa-aara-v•ealar here-out-went 'When Biersnaq' gave himi money, Oarcnaq'i came out.' (27) Aaz Wadrahwmaanaa kinashjka iicaad, f yz ls.Erg Abd.Dat book bought (f=Abd.) it sej bierazhta diesharg-dolazh 1sRFL.Gen children.Datread.Fut.CV 'I bought Abdurahman a book so he would read it to my children.' 4.4. Complement Clauses LDR affects complement clauses of all types. (28) Suona xov, ejsa shiekar iicaaljga ls.Datknow ls.RFL.Erg sugar bought.Sbj 'I know I bought sugar.' [subjunctive complement] (29) Cynna dieza aaz shii nanna novq'ostal dar 3s.Dat wants ls.Erg 3s.RFL.Gen mother.Dat help do.Nz 'He wants me to help his mother.' [nominalization] (30) Suona dieza cuo sej nanna novq'ostal dar ls.Dat want 3s.Erg ls.RFL.Gen mother.Dat help do.Nz 'I want him to help my mother.' [nominalization] (31) Biersnaq'aa diezac Oarcnaq'uo shiina lerjgaxie hwama B.Dat like.Neg O.Erg 3s.RFL.Dat ear.Loc thing tuoxar hit.Nz 'Biersnaq' doesn't like Oarcnaq' to hit him in the ear.' [nominalization]
Long-distance Reflexivization in Chechen and Ingush
(32)
265
sej nanna novq'ostal die voal so (Erg) IsRFL.Gen mother.Dat help do.Inf want ls.Nom 'I want to help my mother.' [infinitive]
(33) Suonadika xiet, f sej nanna novq'ostal ls.Dat good think (Unsp.Erg) 1sRFL.Gen mother.Dat help dycha do.CV 'I like it when people help my mother.' [chained] (34) [ [shie chy-j•iecha] f shiiga tilifon 3sRFL in-come.CV (f=she) 3sRFL. All telephone tuoxa-j•aaxazh} v•ar yz strike-say.CV V.was 3s.Nom 'He would like her to call him when she gets home.'[converb subordinate] (35) So hwaajla laqagh vy ma jaax ls.Nom 2s.RFL.Cmp taller be.Prs Neg say.Impr 'Don't say that I'm taller than you.' [asyndetic finite subordination]
4.5. Relative Clauses Chechen and Ingush use relativization by "deletion" (actually, argument sharing, but the term relativization by deletion is well established in the literature), with a participle in the relative clause. The shared argument, of course, cannot be reflexivized since there is no separate token of it in the relative clause, but any other coreferent to the main-clause subject is reflexivized. (36) Cuo hwa-hwiexar txuoga [shie jaazdea ] kinashjka 3s.Erg here-showed 1p. All 3s.RFL.Erg written book.Nom 'He showed us the book he had written.'
4.6. LDR from Non-main Clause The controller of LDR must be a subject, but not necessarily in a main clause. In (37)-(38), the controllers are subjects of complement clauses. In (37), the reflexive in the adjunct to that complement is unidiomatic (zero would be equally clear and stylistically preferable), but not incorrect. Examples (38)-(39) are fully idiomatic.
266
Johanna Nichols
(37) Cynnaxov, [ [shie c'a-qeachacha] cuo shiiga 3s.Dat knows 3sg.RFLhome-come.CV 3s.Erg 3sRFL.All tilifon tuoxarg-joljga] phone strike-Sbj 'He knows she will call him when she gets home.' (38) Suona lovr [wa cynga [hwaaj nanna ls.Dat want.Impf 2s.Erg 3s.All 2sRFL.Gen mother.Dat novq'ostal-die] aalar] help-do.Inf say.Nz 'I'd like you to tell him to help your mother.' (39) Dagadoagh-ii hwuona, [aaz hwaajga [ sej remember-Q 2s.Dat ls.Erg 2sRFL.All ( =2s.Erg)lsRFL.Gen nanna novq'uostal-die] eanna]? mother.Dat help-do.Inf Quot 'Do you remember that I asked you to help my mother?' 4.7. Intervening Different Subject LDR applies even if an intervening clause has a subject not coreferential to the LDR antecedent. In (40), LDR applies to shie 'him' in the lowest clause despite the subject suona 'I' (underlined) in the intermediate clause. (40) Cynnaxov, [ [f shie bwarjg-vejna] suona 3s.Dat know (ls.Dat) 3sRFL eye-saw ls.Dat xoza xietaljga] good seem.Sbj 'He knows I'm glad I saw him.' 4.8. No Upwards LDR Regardless of word order, topicality, empathy, and all other conditions that enhance prominence of arguments, LDR cannot affect a higher clause. (41) *Bjersnaq'uo Oarcnaq'aa axcha dalcha, shie B.Errg O.Dat money gave.CV 3s.RFL.Nom hwa-aara-vealar here-out-came 'When Bjersnaq'i gave Oarcnaq' money, hei left.' Ordinary anaphoric pronominalization can go upwards, however; an example is shown in (66).
Long-distance Reflexivization in Chechen and Ingush
267
4.9. Conditions and Blocking Effects Three conditions have been found that weaken (and, in one case, completely block) LDR. (There are some further conditions on multiple LDR, discussed following.)
4.9.1. PERSON HIERARCHY Third-person LDR is obligatory, while (at least in elicitation and for some speakers) nonreflexives are of borderline acceptability in the first person. Nonreflexive third-person pronouns are of course acceptable, as in (42b), but cannot be coreferential to the main-clause subject. (42) a. Aaz shii veshiina jiittacha, hwa-aara-vealar yz ls.Erg 3sRFL.Genbrother.Dat hit-CV here-out-went 3s.Nom 'When I hit hisi brother, hei came out.' b. Aaz cyn veshiina jiittacha, hwa-aara-vealar yz ls.Erg 3s.Gen brother.Dat hit-CV here-out-went 3s.Nom 'When I hit hisj brother, hej, *i came out.' (43) a. Cuo sej veshiina jiittacha, dwa-aara-vealar so 3s.Erg IsRFL.Gen brother.Dat hit.CV there-out-went ls.Nom 'When he hit my brother, I went out.' b. ?/* Cuo sy veshiina jiittacha, dwa-aara-vealar so 3s.Erg 1s.Gen brother.Dat hit.CV there-out-went ls.Nom 'When he hit my brother, I went out.'
4.9.2. MODALITY Reflexivization is not absolutely obligatory in purposive and future or intentional clauses. This effect too has been observed only in elicitation and only for first person. (44) a. Aaz dwa-qieta-vyr yz, sejna hwama ls.Erg convinced-Aux 3s.Nom 1sRFL.Dat thing tuoxarg-joacazh hit-Fut.Neg.CV 'I convinced him not to hit me.' b. ?/*Aaz dwa-qieta-vyr yz, suona hwama tuoxarg.joacazh ls.Erg convinced-Aux 3s.Nom ls.Dat thing hit-Fut.Neg.CV 'I convinced him not to hit me.'
268
Johanna Nichols
4.9.3. ANIMACY Inanimates cannot control reflexivization. This condition appears to be absolute. Examples (45-46) have the same main clause verb 'grow,' and reflexivization is possible only in (46) where the subject is animate. (45) a. Hwazhjk' azh dika diegha-j•oagha, f dika xii tiexacha corn well grow-Aux (Dat) well water hit.CV 'Corn grows well if you water it well','.. . when it is well watered.' b. *Hwazhjk'azh dika diegha-j•oagha, shiina dika xii corn well grow-Aux 3sRFL.Dat well water tiexacha hit.CV (46) Bier yshtta sixa diegha-d•oagha-d•aacar, shiina dika child so fast grow-Aux.Neg 3sg.RFL.Dat well hwama ca ju'a-juorie thing not feed-Aux.Cond 'The child wouldn't grow this fast if it weren't fed well.' (47) a. Bod dika sovsa, f dika hwiicha dough well rise well knead.CV 'Dough rises well if you knead it well',' ... when it is well kneaded.' b. *Bod dika sowsa, shie dika hwiicha dough well rise 3s.RFL.Nom well knead.CV lit. 'Dough rises well if you knead itself well.' (Note that, although inanimates cannot control reflexives, they can perfectly well control overt nonreflexive pronouns: see (15) and following, (51) and following.) 4.10. Conditions on Multiple LDR and Multiple Anaphora Since both main and nonmain subjects can control LDR, examples with multiple LDR can easily be constructed. In the examples following, one set of coreferents is in boldface and the other underlined. In the first two examples, the subject of 'get home' and 'call' is feminine, as shown shown by thejprefix on the verbs; 'know' does not take agreement, so its subject could be either masculine or feminine. Example (48) repeats (37) above. (48) Ingush Cynna xov, [ [shie chy-j•iecha] 3s.Dat knows 3s.RFL.Nom in-come.CV
Long-distance Reflexivization in Chechen and Ingush
269
cuo shiiga tilifon tuoxarg-j•oljga] s/3s.Erg 3s.RFL.All phone strike-J.Fut.Sbj 'He knows she will call him when she gets home.' (49) Chechen Cunna xae'a [ [shaa c'a-j•e'achii] 3s.Dat knows 3s.RFL.Nom home-come.CV cuo shiega telefon tuuxur-j•ujla] 3s.Erg 3s.RFL. All telephone call-J.Fut.Subj 'He knows she will call him when she gets home.' (50) Chechen (Superscripts to verbs number clauses; see translation.) cunna [as dyyciila2] lae'a,1 3s.Dat ls.Erg say.Sbj wants [hunda aella3 as [sajna ca xae'a4 why said IsJErg 1sRFL.Dat not know [san luulaxuosha shaa hwoshalgha hunda ca 1s.Gen neighbors.Erg 3sRFL.Nom hospitality why not qajkina5 ]]] invited 'Hej wants1 me to explain2 why I said3 I didn't know4 why my neighbors didn't invite5 himi.' We now turn to constraints applying to these cases of multiple LDR or to LDR in the context of multiple pronominalization. 4.10.1. PERSON INTERACTION The first constraint concerns LDR and nonreflexive anaphora affecting the same clause. First-person and third-person pronouns can co-occur in the same nonmain clause without disrupting reflexivization, but second person for the most part cannot co-occur with first or third. These patterns obtain no matter which of the two pronouns is reflexive. The following examples (all from Ingush) have chained adjunct (i.e., nonserialized) purpose clauses. The ones with second-person pronouns are clearly and firmly rejected by the speakers I have worked with, except that (53) is sometimes accepted though pronounced difficult to interpret. The raw facts, for various combinations of pronouns, are as follows. (51) Muusaaz suona kinashjka iicar, aaz yzshii M.Erg ls.Dat book bought ls.Ergit 3sRFL.Gen bierazhta diesharg-dolazh children.Dat read.Fut.CV 'Musa bought me a book for me to read it to his children.'
270
Johanna Nichols
(52) *Muusaaz hwuona kinashjka iicar, wa yz shii M.Erg 2s.Dat book bought 2s.Erg it 3sRFL.Gen bierazhta diesharg-dolazh children.Dat read.Fut.CV 'Musa bought you a book for you to read it to his children.' (53) ? Aaz hwuona kinashjka iicaad, wa yzsej ls.Erg 2s.Dat book bought 2s.Erg it 1sRFL.Gen bierazhta diesharg-dolazh children.Dat read.Fut.CV 'I bought you a book for you to read it to my children.' (54) *Wa suona kinashjka iicaad, aaz yzhwaaj 2s.Erg ls.Dat book bought ls.Erg it 2s.RFL bierazhta diesharg-dolazh children.Dat read-Fut.CV 'You bought me a book for me to read it to your children.' (55) Aaz Muusaajna kinashjka iicaad, cuo yz sej ls.Erg M.Dat book bought 3s.Erg it 1sRFL.Gen bierazhta diesharg-dolazh children.Dat read-Fut.CV 'I bought Musa a book for him (=M.) to read it to my children.' (56) *Wa Muusajna kinashjka iicaad, cuo yz hwaaj 2s.Erg M.Dat book bought 3s.Erg it 2s.RFL bierazhta diesharg-dolazh children.Dat read-Fut.CV 'You bought Musa a book for him (=M.) to read it to your children.' Plural pronouns behave the same way and will not be illustrated, except for the inclusive, which behaves like second-person pronouns in blocking LDR. (The inclusive reflexive and nonreflexive genitives are homophonous with each other and have nominative-ergative syncretism.) (57) *Muusaaz vajna kinashjka iicaad-ii, vaj yz shii M.Erg 12.Dat book bought-Q 12.Erg it 3sRFL.Gen bierazhta diesharg-dolazh? children.Dat read-Fut.CV 'Did Musa buy us (inclusive) a book for us to read it to his children?' (58)
*Vaj Muusaajna kinashjka iicaad-ii, cuo yz vaj 12.Erg M.Dat book bought-Q 3s.Erg it 12.Gen
Long-distance Reflexivization in Chechen and Ingush
271
bierazhta diesharg-dolazh? children.DAT read-FUT.CV 'Did we buy Musa a book to read to our children?' It is not that second and other persons cannot occur in the same main clause or the same sentence; see (38)-(39) for perfectly grammatical examples of such combinations. What is ungrammatical is second person occurring together with another person in a nonmain clause where both are syntactically controlled by higher antecedents. The co-occurrence possibilities are summarized in Table 2. Table 3 breaks such examples down into two potentially problematic control situations they exemplify: an indirect object in the main clause has a coreferential subject in the purpose clause; and the possessor in the purpose clause is reflexivized across a purpose-clause subject of a different person category. Second-person coreference does not tolerate either situation, showing that it is strongly subject-oriented. Adjunct purpose clauses appear to be the only construction in Ingush in which an object (in this case, an indirect object) syntactically controls an overt pronoun, and when this unique context has two different antecedents in the main clause with coreferents in the purpose clause, the ungrammaticality pattern exhibited by the second person shows that there is a person-based hierarchy for subject orientation of control in Ingush: 2>1>3. Again, the two control situations are problematic for second person only when they co-occur. Second-person control by an indirect object, in the absence of reflexivization of another person, is perfectly acceptable as shown in (59). Reflexivization across an intervening subject, in the absence of another controlled pronominal, is also perfectly acceptable, as shown in (39) TABLE 2 CO-OCCURRENCE OF PERSON CATEGORIES IN LDR IN INGUSH PERSON CLAUSES Person of main clause subject (chained clause possessor)
Person of main clause indirect object (chained clause subject)
3 1 1 2 3 2 3
1 3 2 1 2 3
Inclusive
3
Inclusive
Acceptability
Example no.
/ / ? * * * * *
(51) (55) (53) (54) (52) (56) (57) (58)
272
Johanna Nichols TABLE 3 SYNTACTIC POSSIBILITIES FOR COREFERENCE, BY PERSON CATEGORY
Control situation Control by indirect object Reflexivization across intervening subject
Person category
2 * *
1 / */?/a
3 / /
a
Acceptability depends on person of intervening subject: 3 acceptable, 1 variable or dubious, 2 never acceptable.
preceding (the second-person pronouns, not the boldface first-person pronouns, are at issue). (59) Muusaaz hwuona kinashjka iicaad, wa yz txuona Musa.Erg 2s.Dat book bought 2s.Erg it 1p.Dat diesharg-dolazh read-Fut.CV 'Musa bought you a book to read to us.' 4.10.2. ONE REFLEXIVE CATEGORY PER CLAUSE No two non-coreferential reflexives may surface in the same clause. In (60a), the two reflexives are of different person categories, so the sentence should be unambiguous, but it is nonetheless ungrammatical. A non-third-person pronoun can be made nonreflexive in this context, and the result is grammatical, as in (60b). Note that in (60b) the controller of LDR is in a non-main clause. (60) a. *Suona xov, shiina sie bwarg-v•ejcha Muusaajna ls.Dat know 3s.RFL ls.RFL eye-see.CV Musa.Dat xosa xietaljga good seem-Sbj 'I know Musa is glad he saw me.' (lit. 'that he himself saw myself) b. Suona xov, shiina so bwarg-v•ejcha Muusaajna xoza Is-Dat know 3s.RFL ls.Nom eye-see-CV Musa.Dat good xietaljga seem-Sbj id. (61) * Muusaajna xov sejna shii bwarjg-vejcha suona M.-Dat know 1sRFL.Dat SsRFL.Nom eye-see.CV ls.Dat
Long-distance Reflexivization in Chechen and Ingush
273
xoza xietaljga good seem.Sbj 'Musa knows I'm glad I saw him.' The following examples have two third-person reflexives with different ferents. Feminine j- vs masculine v- on 'see' makes the reference unambiguous, but the sentences are nonetheless ungrammatical. (62)
*Muusaajna xov, shiina shie bwarjg-jejcha Mariemaa Musa.dat know 3sRFL.Dat SsRFL.Nom eye-J.see.CV M.Dat xoza xietaljga. good seem.Sbj 'Musa knows Mariem is glad he saw her.'
(63) *Muusaajna xov, shiina shie bwarjg-vejcha M.Dat know 3sRFL.Dat 3sRFL.Nom eye-V.see.CV Mariemaa xoza xietaljga. M.Dat good seem.Sbj 'Musa knows Mariem is glad she saw him.' 4.10.3. MAIN-CLAUSE ANTECEDENT Only when there is potential multiple third-person LDR does a difference between main-clause and non-main-clause controllers appear: in this case, the main-clause subject is the preferred controller and coreferents to other subjects can be nonreflexive or zero. (64)
Cynna xov [f [yz chy-jiecha] shiiga tilifon 3s.Dat know (Erg) 3s.Nom in-J.come 3s.RFL.All phone tuoxarg-joljga] strike-Sbj 'He knows she will call him when she gets home.'
(65) Cynna xov [ [f chy-jiecha] cuo shiiga tilifon 3s.Da know (Nom) in-J.come 3s.Erg 3sRFL.All phone tuoxarg-joljga] strike-Sbj 'He knows she will call him when she gets home.' (66)
Muusaajna xov [ [Wajshiet chy-jiecha] cuo shiiga Musa.Dat know Aisha in-J.come.CV3s.Erg 3sRFL.All tilifon tuoxarg-joljga] phone strike-Sbj 'Musa knows that when Aisha gets home she will call him.'
274
Johanna Nichols
Even here, however, multiple reflexivization is possible: (67) Muusaajna xov [[shie chy-jiecha] Wajshietaz M.Dat know 3sRFL.Nom in-J.come.CV Aisha.Erg shiiga tilifon tuoxarg-joljga] 3sRFL. All 3sRFL. All phone strike-Sbj 'Musa knows that when she gets home Aisha will call him.' 4.10.4. LOCAL REFLEXIVIZATION IS FAVORED OVER LDR In sentences with two possible controllers and, hence, potential ambiguity, wherever a local-reflexivization interpretation is possible it is strongly preferred and in most elicitation proved to be the only interpretation that came to mind: (68) Q'amarsoltaz Muusaajna kinashjka iicar, cuo yz shii Q.Erg M.Dat book bought 3s.Ergit 3s.RFL bierazhta diesharg-dolazh children.Dat read-FUT.CV 'Q'amarsoltj bought Musaj a book for hiirij to read to hisj children.' (69) Q'amarsoltaz [f shii bierazhta diesha] Muusaajna Q.Erg. (Erg) 3s.RFL children-.Datread-Inf M.Dat kinashjka iicar book bought 'Q'amarsolti bought Musaj a book to read to hisj (M.'s) children.'
5. CONCLUSION Chechen-Ingush long-distance reflexivization is analogous in its syntactic behavior to local reflexivization, the only difference being its domain: the domain of local reflexivization is the clause while that of long-distance reflexivization is the sentence or clause chain. Both local and long-distance reflexivization are strictly subject-controlled, apply virtually without limit throughout their domains, and are not sensitive to discourse or pragmatic conditions. Generative research over the last decade (Cole, Hermon, and Sung, 1990; papers in Koster and Reuland, 1991, Progovac, 1993; Cole and Sung, 1993) has led to the general expectation that in languages with LDR, the reflexive pronouns will be monomorphemic, LDR will be subject-oriented and controlled only from the main clause, and (in languages without subject-verb
Long-distance Reflexivization in Chechen and Ingush
275
agreement) it will be blocked if subjects of different person categories occur in different higher clauses (i.e., if between the potential antecedent and the reflexive there is an intermediate subject of a different person category). Chechen-Ingush LDR does not fit this pattern: the pronouns are technically monomorphemic, although semantically they distinguish person-number categories and reflexivity; LDR is subject-oriented but controllable from any clause; there are no blocking effects (as shown by examples (42)-(43), (50)) though there is no subject-verb agreement (recall that there is gender agreement, not with the subject but on the ergative pattern, with the S/O; and it occurs in under one-third of the verb roots, the rest simply lacking agreement). Even if the person interaction discussed in section 4.10.1 is considered a blocking effect, it is quite minimal, involving only the second person in a very restricted context. From a typological viewpoint, the Chechen-Ingush system is made possible by the presence of person-number distinctions in reflexive pronouns (this makes possible the multiple LDR and absence of blocking) and by clause chaining (which sets up a morphosyntax in which long strings of clauses comprise a single-tensed sentence, and in which referential properties are regularly dependent on higher clauses). The absence of a switchreference system or any regular specialization of converbs as same-subject and different-subject must also favor LDR; at least, their presence would presumably make LDR unnecessary. To my knowledge, Chechen and Ingush are the first ergative languages to be discussed in the literature on LDR, though it seems unlikely that ergativity has anything to do with LDR. LDR of the Chechen-Ingush type is not common cross-linguistically. It is not widespread within the Northeast Caucasian language family, though cognate third-person reflexive pronouns occur throughout the family. (Haspelmath (1993:413) gives two examples of LDR and on p. 414 one of logophoric reflexivization, from the very distant sister language Lezghi.) Nonetheless, in view of its regularity, the presence of multiple LDR, and the clear contrast with logophoric reflexivization, I suggest that it be taken as canonical LDR.
ACKNOWLEDGMENTS Some of the work on Ingush reported here has been supported by NSF grant SBR-96-16448 to the University of California, Berkeley. Field work on Chechen and Ingush in Tbilisi, Georgia (1979-80,1981,1984), Ingushetia, and the Chechen Republic-Ichkeria (then Chechen-Ingush ASSR; 1989) was supported by the Interna-
276
Johanna Nichols
tional Research and Exchanges Board. Research on Jordan Chechen in Berkeley (1983) was supported by the Committee on Research of the University of California at Berkeley. I am grateful to the many speakers of Ingush and Chechen who have worked with me at one time or another, and in particular to Issa Guliev for assistance on this and other syntactic topics in recent years. I thank Thomas Gamkrelidze, Firuza Ozdoeva, and Vaxa Timaev for hospitality and assistance in their respective institutions.
NOTES 1
Examples use an all-Latin transcription without diacritics in which "h" has a diacritic function as in English and other orthographies: ch [c], sh [s], etc.; gh [8]; w = pharyngealization or (non-postconsonantal position) voiced pharyngeal [9] (and hw = voiceless pharyngeal fricative [h]). y = [~t] in Ingush, [u] in Chechen. (For fuller phonological information see, .) In all examples, antecedent and long-distance reflexive are in boldface. Other conventions and interlinear abbreviations: 0 * & • 1s, 2p, etc. 12 All Aux Cmp Cnd CV Dat Erg Gen Impr Impf Inf Instr J Loc Nom Nz Pot Ppl Prs
Zero due to anaphora or deletion (in chaining and subordination) Place occupied by shared argument (in serialization) Particle used for serialization (and other functions). Boundary between gender prefix and verb First-person singular, second-person plural, etc. Inclusive Allative case Auxiliary verb Comparative (case of noun or degree of adjective) Conditional Converb Dative case Ergative case Genitive case Imperative Imperfect tense Infinitive Instrumental case Gender class (feminine) marked by prefixed ;'- agreement. Locative case Nominative case (i.e., absolutive: case of S, O, and citation form) Nominalized form of verb Potential (verbal derivation) Participle Present tense
Long-distance Reflexivization in Chechen and Ingush
Q Quot RFL Sbj V
277
Interrogative clitic particle Quotative Reflexive Subjunctive Gender class (masculine) marked by prefixed v- agreement.
Segmentation and interlinearization is minimal. Gender prefixes (not quite one-third of the verb roots in Chechen and Ingush agree in gender with the nominative S/O) are segmented off only where gender is at issue. Verb tenses and various kinds of converbs are not always distinguished in interlinears.
REFERENCES Cole, P., Hermon, G, and L.-M. Sung, (1990). Principles and parameters of long-distance reflexives. Linguistic Inquiry 21,1.1-22. Cole, P., and L.-M. Sung, (1993). Head movement and long-distance reflexives. Linguistic Inquiry 25,2. El'darxanov, T. (1900). Chechenskie teksty. Sbornik materialov dlja opisanija mestnostej i piemen Kavkaza 22,3.1-40. Tiflis. Foley, W., and Van Valin, R. D. Jr., (1984). Functional Syntax and Universal Grammar. Cambridge: Cambridge University Press. Hale, K. (1991). Misumalpan verb sequencing constructions. Clair Lefebvre, ed., Serial Verbs: Grammatical, Comparative, and Cognitive Approaches (Studies in the Sciences of Language, 8.) (C. Lefebvre, ed.) Amsterdam-Philadelphia: Benjamins. Haspelmath, M. (1993). A Grammar of Lezgian. (Mouton Grammar Library, 9.) Mouton de Gruyter, Berlin. Koster, J, and Reuland,E. (eds.) (1991). Long-distance Anaphora. Cambridge: Cambridge University Press. Maciev, A. G. (1965). Cheberloevskij dialekt chechenskogo jazyka. I. Ju. Aliroev et al., eds., Sbornik statej i materialov po voprosam naxskogo jazykoznanija, 3-95. Izvestija Chechensko-ingushskogo nauchno-issledovatel'skogo instituta istorii, jazyka i literatury, VI.2: Jazykoznanie. Groznyj. Nichols, J. (1994a). Chechen. In The Indigenous Languages of the Caucasus, 4: Northeast Caucasian Languages, part 2. (Rieks Smeets, ed.,) pp. 1-78. Delmar, NY: Caravan Books. Nichols, J. (1994b). Ingush. In The Indigenous Languages of the Caucasus, 4: Northeast Caucasian Languages, part 2 (Rieks Smeets, ed.), pp. 79-145. Delmar, NY: Caravan Books. Peterson, D. A. (2000). Ingush a: The elusive Type 5 clitic? Submitted for publication. Progovac, L. (1993). Long-distance reflexives: Movement-to-Infl vs. relativized SUBJECT. Linguistic Inquiry 24,4.755-772.
278
Johanna Nichols
Van Valin, R. D., Jr. (1993). A synopsis of role and reference grammar. In Advances in Role and Reference Grammar, (R. Van Valin, ed.,) pp. 1-164. Amsterdam-Philadelphia: Benjamins.
WHY THE BLOCKING EFFECT? HAIHUA PAN Department of Chinese, Translation, and Linguistics City University of Hong Kong Kowloon, Hong Kong
1. INTRODUCTION This chapter discusses the blocking effect observed in long-distance (LD) bound bare reflexive ziji in Mandarin Chinese. Unlike the symmetrical unlike-person blocking claimed in the literature (Huang and Tang, 1991; Xue, Pollard, and Sag, 1994), this chapter argues that (a) the blocking effect of ziji is not symmetrical: first- and second-person pronouns can block third-person noun phrases (NP) from long distance binding ziji, though third-person NPs do not necessarily block first- or second-person pronouns from long-distance binding ziji; and (b) other grammatical functions filled by first- and second-person pronouns, not just subjects or NPs contained in the subject, can induce the blocking effect. The chapter claims that long-distance bound ziji points to the carrier of belief, and reconstructs this notion as self-ascription. The blocking effect is explained by appealing to the fact that only first- and second-person pronouns are obligatory self-ascribers and thus can block long-distance binding of ziji by third-person NPs if they intervene between the potential third-person NP and the reflexive ziji, while third-person NPs do not necessarily block ziji from being long distance bound by first/second-person pronouns. The rest of this chapter is organized as follows: after describing the basic facts concerning the blocking effect in Section 2, I argue in Section 3 that the blocking effect is not symmetrical, and the crucial factor is not the local subject or the unlike-person feature conflict, as suggested in the literature. Syntax and Semantics, Volume 33 Long-Distance Reflexives
279
Copyright © 2001 by Academic Press All rights of reproduction in any form reserved. 0092-4563/00 $35.00
280
Haihua Pan
Rather, it is the asymmetry between first/second- and third-person noun phrases that plays a crucial role in the blocking effect, and syntactic functions other than subject can also induce the blocking effect. In Section 4,1 briefly discuss limitations of previous analyses employing feature agreement checking, closeness, and potential binders, logophoricity, and perspectivity. In Section 5,1 first argue that long-distance bound ziji is a de se anaphor, using Lewis' terminology, and is thus constrained by self-ascription. Then, I propose a condition to account for the properties of long distance bound ziji, and show that my account can better explain the blocking effect than can previous analyses. Before concluding the chapter in Section 7,1 discuss some relevant issues in Section 6, especially the differences between self-ascription and logophoricity.
2. THE BLOCKING EFFECT Chinese bare reflexive ziji 'self exhibits the so-called effect: long-distance binding of ziji is possible only if all intervening subjects agree in person features; otherwise, the long-distance binding is blocked, as exemplified in (1) (Y. H. Huang, 1984, Tang, 1985,1989)1 (1) a. Woi juede nij dui ziji*i/j meixinxin. I think you to self not confidence 'I think you have no confidence in yourself/*me.' b. Nii juede woj dui ziji*i/j meixinxin ma? you think I to self not confidence Q 'Do you think I have no confidence in myself/*you?' c. Zhangsani juede wo/nij dui ziji*i/j meixinxin. Zhangsan think I/you to self not confidence 'Zhangsan thinks I/you have no confidence in myself/yourself/*him.' d. Zhangsani zhidao Lisij dui ziji i/j mei xinxin. Zhangsan how Lisi to self not confidence 'Zhangsan knows that Lisi has no confidence in him/himself.' Since they do not agree in person feature with the local subjects, the matrix subjects in (la), (1b), and (1c) cannot be the antecedents of ziji. By contrast, the matrix subject in (1d) can be the antecedent of ziji, since there is no person feature conflict between the local subject and the matrix subject. The blocking effect is the motivation for requiring successive cyclic movement of ziji at LF in the movement analysis of ziji (Battistella, 1989; Cole, Hermon, and Sung, 1990; Huang and Tang, 1991; Li, 1993; Cole and Sung, 1994; Cole and Wang, 1996, among others).
Why the Blocking Effect?
281
However, the blocking fact is more complicated than presented in Tang (1989). Huang and Tang (1991) point out that an intervening subcommanding NP with different person features can also induce the blocking effect, as shown by the following contrast: first/second-person NPs but not third-person NPs contained in a subject block third-person matrix subjects from binding ziji. (2) a. Zhangsani renwei Lisij de jiao'ao hai-le zijiyy Zhangsan think Lisi DE arrogance harm-Perf self 'Zhangsanj felt that Lisi'sj arrogance harmed himi/j.' b. Zhangsani renwei woj de jiao'ao hai-le ziji*i/j. Zhangsan think I DE arrogance harm-Perf self 'Zhangsan; felt that my arrogance harmed hinUj/me.' (3) a. Zhangsani renwei Lisij neiyang zuo dui ziji i/j buli Zhangsan think Lisi that-waydo to self not-beneficial 'Zhangsan; felt that Lisi'sj acting that way didn't do him any good.' b. Zhangsani renwei nij neiyang zuo dui ziji *i/j buli Zhangsan think you that-way do to self not-beneficial 'Zhangsanj felt that your acting that way didn't do him*i/you any good.' Furthermore, direct objects and obliques, though not potential binders which are defined as c-commanding animate subjects in Huang and Tang (1991), can induce the blocking just as subjects can (Xue et al, 1994). (4) a. Zhangsani zhidao Lisij gaosu-guo nik youguan ziji *i/j/k de gongzuo. Zhangsan know Lisi tell-Guo you about self De work 'Zhangsan knew that Lisi told you about his/*your work.' b. Zhangsani shuo Lisij gen nik tan-guo ziji *i/j/k de shi. Zhangsan say Lisi with you talk-Guo self De business 'Zhangsan said that Lisi talked about his/*your business with you.' c. Zhangsani renwei Lisij cong wok nar tingshuo-le ziji *i/j/k Zhangsan think Lisi from I there hear-say-Perf self de fenshu. De score 'Zhangsan thinks Lisi heard from me his/*my score.' d. Zhangsani zhidao Lisij zai wo/nikjia xi ziji *i/j/k Zhangsan know Lisi at I/you home develop self de zhaopian. DE photo 'Zhangsan knew that Lisi was developing his picture(s) at my/your home.' Sentence (4a) indicates that a second-person object can induce the blocking effect on the third-person matrix subject, Zhangsan, though it does not
282
Haihua Pan
do so on its co-argument, the intermediate third-person subject Lisi. Sentences like (4b) and (4c) show that first- and second-person pronouns in an adjunct (wo/ni 'I/you'), though not c-commanding or subcommanding ziji, can also induce the blocking effect on the third-person matrix subject. These sentences, as pointed out in Xue et al. (1994), pose a serious problem for analyses based on I-to-I relations and Spec-head (Infl) agreement (Battistella, 1989; Huang and Tang, 1991; Progova?, 1992, 1993; Cole and Sung, 1994), which claim that only c-commanding and subcommanding subjects can induce the blocking effect. Following Huang and Tang (1991), Xue et al. (1994) suggest that the blocking effect in subcommanding cases can be influenced by discourse factors. They claim that, while the blocking effect in the latter is subject to unlike-person blocking, in the former it is constrained by animate blocking, as exemplified by the contrast observed in (5a) and (5b). The inanimate subject NP naben shu 'that book' in (5a) does not block long-distance binding of ziji, but the animate subject NP Lisi in (5b) does. (5) a. [Zhangsani de xin]j biaoming naben shuk hai-le ziji i/*j/k . Zhangsan DE letter indicate that book harm-Perf self 'Zhangsan'si letter indicates that that book harmed himi.' b. [Zhangsani de xin]j biaoming Lisik hai-le ziji*i/*j/k. Zhangsan DE letter indicate Lisi harm-Perf self 'Zhangsan'si letter indicates that Lisi harmed himi/himself.' c. [Zhangsani de xin] j anshi Lisik hai-le zijii/*j/k. Zhangsan DE letter hint Lisi harm-Perf self 'Zhangsan'si letter hinted that Lisi harmed himi/himself.' However, contrary to their claim, the subcommander Zhangsan in (5c) can bind ziji, though there is an intervening animate subject Lisi. The difference between (5b) and (5c) is the main verb. As for the verb anshi 'hint,' we have the following: if Zhangsan's letter hints at X, then Zhangsan hints at X, but this is not necessarily true for the verb biaoming 'indicate.' So only in (5c) Zhangsan can bind ziji- This contrast suggests that different verbs influence the binding possibility of ziji. Hence subcommanding and c-commanding cases behave similarly and should not be treated separately.
3. THE BLOCKING EFFECT IS NOT SYMMETRICAL The characterization of the blocking effect so far implies that the blocking effect is symmetrical; actually, this seems to be the general consensus (Tang, 1989; Huang and Tang, 1991; Xue et al., 1994). However, closer examination reveals that it is not all that symmetrical.2 Actually, all the data pre-
Why the Blocking Effect?
283
sented so far, i.e., (l)-(5), show only that intervening first/second-person pronouns induce a blocking effect on third-person NPs, while intervening NPs with third-person features do not induce such an effect for third-person NPs. What is not shown is whether intervening third-person NPs can block first/second NPs from binding ziji. For the unlike-person blocking hypothesis to hold in its most general form, intervening third-person NPs should also induce a blocking effect on first/second-person pronouns. Unfortunately, this is not supported by Chinese data. First, although intervening first- and second-person subjects block long distance binding of ziji, third-person subjects do not necessarily block long-distance binding of ziji by first- or second-person NPs, as exemplified by the following sentences. (6) a. Woi zhidao Lisij bu xihuan ziji ?i/j. I know Lisi not like self 'I knew that Lisi did not like me/himself.' b. Nii xiang meixiang guo Lisij conglaijiu mei xihuan guo ziji ?i/j ? you think not think Guo Lisi never conj not like Guo self 'Have you ever thought about the idea that Lisi never liked you/himself?' c. Woi yizhi yiwei Zhangsanj xihuan zijii/j, keshio wo cuo le. I so-far think Zhangsan like self but I wrong Prt 'I always thought that Zhangsan liked me, but I was wrong.' (7) a. Woi bu xihuan Lisij guan ziji i/j de shi. I not like Lisi interfere self De matter 'I don't like Lisi interfering in my (own) business.' b. Nii xihuan Lisij guan ziji i/j de shi ma? you like Lisi interfere self De matter Q 'Do you like Lisi interfering in your (own) business?' c. Lisii bu xihuan wo/nij guan ziji *i/j de shi. Lisi not like I/you interfere self DE matter 'Lisi does not like me/you interfering in my/your (own) business.' d. Lisii xihuan Zhangsanj guan zijiyj de shi. Lisi like Zhangsan interfere self DE matter 'Does Lisi like Zhangsan interfering in his (own) business?' (8) a. Dangshi woi pa Lisij zai lai zhao ziji i/?j de mafan, jiu Then I afraid Lisi again come find self DE trouble Conj gei-le ta 100 kuai qian. give-Perfhim CL money 'At that time I was afraid that Lisi would come to cause trouble for me again, so I gave him 100 dollars.'
284
Haihua Pan
b. Dangshi nii pa Lisij zai lai zhao ziji i/?j de mafan, Then you afraid Lisi again come find self DE trouble jiu gei-le ta 100 kuai qian, dui-bu-dui? Conj give-Perf him CL money right-not-right 'At that time you were afraid that Lisi would come to cause trouble for me again, so you gave him 100 dollars. Is that right?' c. Dangshi Lisii pa woj zai qu zhao ziji *i/j de mafan, Then Lisi afraid I again come find self DE trouble jiu gei-le wo 100 kuai qian. Conj give-Perf I CL money 'At that time Lisi was afraid that I would go to cause trouble for him again, so he gave me 100 dollars.' Although the local third-person subjects are the preferred antecedents of ziji, the first/second-person matrix subjects in (6a) and (6b) are also possible antecedents. The first/second-person matrix subjects in (6c), (7a), and (7b) are equally acceptable antecedents. The contrast between (7a)/(7b) and (7c) is a strong piece of evidence arguing for the claim that the blocking effect is not symmetrical. The pragmatic factor involved is biased toward the matrix subject, so one would expect that the matrix subject could be the antecedent for all occurrences of ziji in (7). However, the matrix subject is not a possible antecedent only in (7c). If the symmetrical unlike-person account is correct, all the long-distance binding should be blocked except for (7d), which is not consistent with the facts. The binding patterns in (8) further support our position. Thus, the binding patterns observed in (6), (7), and (8) argue strongly against the symmetrical view. Furthermore, the asymmetry between third-person NPs and first/second-person pronouns also shows up in the subcommanding cases. Although an inanimate NP does not block long-distance binding of ziji in (9), inanimate NPs with a possessive can block the long-distance binding of ziji by third-person NPs only if the possessive in question has the first- or second-person feature, as shown in (10) and (11). (9) Zhangsani shuo naben shu hai-le zijii Zhangsan say that book hurt-Perf self 'Zhangsanj said that that book hurt himi. (10) a. Zhangsani zhidao wo/nii de baogao hai-le ziji *i/j Zhangsan know I/you DE report hurt-Perf self 'Zhangsan knew that my/your report hurt me/you.' b. Woi zhidao Lisij de baogao hai-le ziji i/j I know Lisi DE report hurt-Perf self 'I knew that Lisi's report hurt me/him.'
Why the Blocking Effect?
285
c. Zhangsani zhidao Lisij de baogaohai-le zijii/j. Zhangsan know Lisi DE report hurt-Perf self 'Zhangsan knew that Lisi's report hurt himself/him.' (11) a. [Zhangsani de hua]j anshi [Lisik de xin]1 hai-le zijii/*j/k*1. Zhangsan DE speech imply Lisi DE letter harm-Perf self 'Zhangsan'Sj words imply that Lisi's letter harmed himi/himself.' b. {Zhangsani de hua] j anshi [nik de xin]1 hai-le Zhangsan DE speech imply your DE letter harm-Perf ziji *i/j/k/*1 . self 'Zhangsan'si letter implies that you harmed himi/himself.' c. [Ni i de hua]j anshi [Lisik de xin]1 hai-le zijii/*j/k/*1. youDE speech imply Lisi DE letter harm-Perf self 'Youri letter implies that Lisi harmed himi/himself.' Thus, the animate feature of the intervening subjects is not the correct factor to differentiate the long-distance binding possibility of ziji in subcommanding cases, as claimed in Xue et al (1994). Rather, it is the person feature, especially the contrast between first/second-person pronouns and third-person NPs, just like the c-commanding cases. Hence, we think that there is no need to differentiate the so-called different types of blocking effect—unlike person blocking for c-commanding cases and animate blocking for subcommanding cases—and they should be dealt with by the same principle in the semantic and discourse modules. Sentences like (12) (from Li, 1993) further support our nonsymmetrical view of the blocking effect: although the local subject is animate, the firstand second-person pronoun possessors in (12a) block the long-distance binding of ziji by the matrix subject, while the third-person possessor in (12b) does not. (12) a. Baoyuiyiwei wode/nidej xueshengk bu xihuan ziji *i/*j/k . Baoyii think my/your student not like self 'Baoyu thinks that my/your student does not like himself.' b. Baoyui yiwei Lisij de xueshengk bu xihuan zijiyi/*j/k. Baoyu think Lisi DE student not like self 'Baoyu thinks that Lisi's student does not like him/himself.' Therefore, the correct generalization for the blocking effect is that only intervening first/second-person pronouns induce the blocking effect, while third-person NPs do not necessarily do so, and first/second-person pronouns in other grammatical functions, not just subjects or those contained in a subject, can block the long-distance binding of ziji. Thus, the blocking effect is more sensitive to different person NPs than to local subjects. This de-
286
Haihua Pan
scriptive generalization indicates that any analysis crucially depending on the properties of local subjects, e.g., the LF head movement analysis of ziji, will fail to explain the blocking effect of long-distance bound ziji.
4. PREVIOUS ANALYSES AND THEIR LIMITATIONS Most of the previous analyses of ziji employ feature compatibility checking to explain the blocking effect, while Huang and Tang (1991) use the concepts of closeness and potential binder with feature compatibility in their account of ziji. In the first two subsections following, I will discuss the problems for these two approaches.3,4 In Section 4.3,1 point out the problems for the accounts based on logophoricity and perspectivity. 4.1 Feature Compatibility Checking
In order to explain the blocking effect, most of the previous GB analyses employ feature percolation and subject-head agreement (Battistella, 1989; Cole et al, 1990; Progovac, 1992,1993; Cole et al, 1993; Cole and Sung, 1994; Cole and Wang, 1996) In these accounts, it is assumed that subjects play a crucial role. The blocking effect is explained as follows: once ziji gets its person feature from the local subject or the possessor contained in the subject if the subject is inanimate, the potential antecedent will be checked against this feature. If there is a feature conflict, then we have the blocking effect. Otherwise, long-distance binding is allowed. Although feature agreement checking can explain some of the blocking effects of ziji, the following problems remain. First, feature agreement checking is symmetrical, since neither will a third-person local subject agree in person feature with a first/second-person pronoun, nor will a first/second-person pronoun with a third-person NP. Thus, this account predicts that the blocking effect will be observed for both first/second-person pronouns and third-person NPs if the intervening NP has a different person feature. Although this prediction is true for the potential candidate with a first-person feature and the intervening NP with a second-person feature, or the other way round, it is not true for first/second-person pronouns when the intervening NP has a third-person feature. In the latter case, the blocking effect is not always observed, and long-distance binding is possible, as shown in (6), (7a), (7b), (8a), and (8b). Second, inanimate subjects without possessors could be problematic for this approach, since, unlike the claim made in Tang (1989), ziji can have inanimate antecedents (Pan, 1995,1997), as exemplified in the following.5
Why the Blocking Effect?
287
(13) [Meige gongyuan]idou you (ta)-zijii[de tedian. every-CL park all have (it)-self DE special-feature 'Every park has its (own) special feature.' Since inanimate subjects can be the antecedents of ziji, sentences like (14) below, if feature checking is applied, would be wrongly predicted to observe the blocking effect, since the local subject and the matrix subject do not agree in person features. (14) Nii shuo-guo naben shu hai-le zijii ma? yousay-GUO that-CL book hurt-Perf self Q 'Did you; say that that book hurt youi?' Third, subcommanding cases are also problematic for feature compatibility checking. According to feature percolation (under the assumption that inanimate subjects are not possible antecedents of ziji), the subcommanding possessor will percolate its person feature to the embedded subject and ziji will get this feature. Since the person feature of the possessors in (l0a), (l0b), (11b), and (11c) is not compatible with that of the matrix subject, the blocking effect should be observed for all the sentences. This prediction is only compatible with (10a) and (11b) but not with (10b) or (11c). Hence, sentences like (10b) and (11c) are not expected for the feature agreement checking approach. Furthermore, since the basic assumption for feature percolation (inanimate NPs cannot be antecedents of ziji) is in doubt, the feature compatibility checking approach is in trouble. Besides, first- and second-person nonsubjects cause a serious problem for this approach, since feature checking (Spec-Head agreement) only applies to subjects or NPs contained in a subject. Thus, the approach in question makes either no prediction or wrong predictions for the sentences in (4).6 Finally, sentences like (12a) are even more problematic for the approach in question. Since the local subject is animate, feature percolation cannot apply. As the local subject and the matrix subject have compatible person feature, (12a) is predicted to allow long-distance binding, which contradicts the fact. 4.2. Closeness and Potential Binders Huang and Tang (1991) differentiate subcommanding cases from the normal cases of ziji binding. Since their treatment of the latter also uses feature compatibility checking to explain the blocking effect, and thus will have similar problems that the feature compatibility checking approach has, as pointed out in the previous subsection, I will focus on their treatment of the former which is dealt with by the following condition.
288
Haihua Pan
(15) A reflexive a may take a NP b as its binder if (a) b subcommands a; and (b) There is no NP g,g a potential binder for a, such that 7 is closer to a than b is. (16) b subcommands a iff b is contained in an NP that c-commands a or that subcommands a, and any argument containing b is in subject position. (17) A potential binder is any c-commanding animate subject in the governing category of a (whether or not it is co-indexed with a).7 Notice that the proposed condition requires that the antecedents of ziji subcommand ziji, and long-distance binding is possible only if there is no closer potential binder. Since a potential binder must be a c-commanding animate subject, the analysis of ziji by Huang and Tang (1991) can avoid the problems caused by intervening inanimate subjects without possessors, as shown in (18). (18) Ni/Lisii de xin biaoming naben shu hai-le ziji i You/Lisi DE letter indicate that-CL book hurt-Perf self 'Your/Lisi'si letter indicates that book hurt you/himi?' However, their analysis has the problems listed in the following. First, the condition defined in (15) above is an alternative way of extending the governing category of the reflexive, though it is not incorporated into the definition given in Binding Condition A. This suggests that there exists some redundancy in Huang and Tang's account, since, with LF movement, there is no need to extend the governing category with additional rules, and LF movement provides it for free. Second, the account of Huang and Tang (1991) is too strong in ruling out a remote human subject from binding the reflexive if closer subjects are animate but nonhuman NPs, as shown in (19) below. For these cases, Huang and Tang (1991) would predict that long-distance binding of ziji is not possible, since the closer animate nonhuman subjects are potential binders, and thus should block the human subcommander in the matrix subjects from binding ziji- However, this prediction is not consistent with the facts. In (19) following, the subjects in the embedded clause have the feature [+animate, -human], but they do not block long-distance binding of ziji by the possessors in the matrix subjects. (19) a. Zhangsarii de xin biaoming [NP Lisij de gou]k zai zijii/?j/k Zhangsan DE letter indicate Lisi DE dog at self de fangjian jiao-guo. DE room bark-GUO 'Zhangsan's letter indicates that Lisi's dog barked in his room.'
Why the Blocking Effect?
289
b. Zhangsani de hua biaoming [NP yitiao gou] j zai ziji i/?j Zhangsan DE speech indicate one-CL dog at self de fangjian shuijiao. DE room sleep 'Zhangsan's words indicate that a dog was sleeping in his room.' Similarly, other subcommanding cases are also problematic for Huang and Tang's analysis. Although they can handle the problematic cases for the feature agreement checking approach such as (10b) and (11c), they have problems with (l0a) and (lib) that are not problematic for the feature agreement checking approach. Since the inanimate local subjects (with or without possessors) are not potential binders, no blocking is expected under Huang and Tang's analysis, which is at odds with examples like (10a) and (11b). Although we can change the potential binder definition to c-commanding human subject to cover sentences like (19), the following sentences are still problematic for the new definition of potential binders. This is because, under the new definition, the local human subject should block the NPs farther away from binding the reflexive ziji, which contradicts the fact, as exemplified in (20). (20) a. Nii de xin anshi Lisik hai-le ziji i/k . you DE letter hint Lisi harm-Perf self 'Yourj letter hinted that Lisi harmed you/himself.' b. Zhangsani de xin anshi Lisik hai-le ziji i/k . Zhangsan DE letter hint Lisi harm-Perf self 'Zhangsan'Sj letter hinted that Lisi harmed hinii/himself.' Finally, examples like (12) given earlier are also problematic for Huang and Tang (1991). Neither the feature compatibility checking nor the condition defined in (15) can correctly predict the binding patterns in (12). Since Huang and Tang only consider the closer c-commanding animate subject NPs and do not differentiate the person feature of the potential candidate and the competing closer NPs, their account of ziji fails to deal with the blocking effect, though they employ concepts like closeness and potential binders. 4.3. Logophoricity and Perspectivity Some researchers argue that long-distance binding should be accounted for by discourse factors such as logophoricity (Maling, 1984; Zribi-Hertz, 1989; Reinhart and Reuland, 1991,1993; Yu, 1991; Huang, 1994, and others), perspectivity (Kuno, 1987; Sells, 1987; Zubin et al, 1990; Li, 1991; Iida, 1992,
290
Haihua Pan
and others), emphasis, and intensive pronouns (Baker, 1995). Others argue that thematic prominence or a combination of it with subject is the right condition to explain the long-distance binding property of reflexives (Chou, 1992; Xu, 1993,1994). Clements (1975) uses the notion of logophoricity, which was originally introduced by Hagege (1974), to differentiate morphologically different pronouns in the studies of African languages. A logophor refers to an entity "whose speech, thoughts, feeling, or general state of consciousness are reported" (Clements, 1975:141) and typically appears with predicates of communication and mental experience. Maling (1984) argues that Icelandic long-distance bound reflexive sig is logophoric. Sells (1987) tries to reduce the notion of logophoricity to three more primitive ones: the SOURCE of the report, the person whose mental state or attitude is described by the content of the report (SELF), and the person from whose point of view the report is made (PIVOT). He claims that Japanese zibun is constrained by PIVOT. Zribi-Hertz (1989) argues that an English reflexive can be long-distance bound if it refers to the minimal subject of consciousness (SC), which is very similar to the logophoricity account. Reinhart and Reuland (1991) also treat the exceptional cases of long-distance binding of reflexives as something like logophoricity, claiming that the deictic center plays a crucial role in the interpretation of long-distance bound reflexives. Based on Kuno's (1987) empathy concept, Iida (1992) argues that it is the speaker's perspective that makes long-distance binding possible for Japanese zibun. After discussing Sells' approach to logophoricity, Stirling (1993) argues that it is possible to unify the three notions of Sells' by the notion of validator, which has the feature [-first person] and is responsible for the truth of the sentence involved. Following Sells (1987), Yu (1991) and Huang (1994) claim that Chinese ziji is logophoric and thus must be constrained by logophoricity. Besides, Chen (1992) uses topicality and Sells' PIVOT notion to account for the long-distance binding properties of ziji, while Li (1991) argues that perspective-taking is the crucial factor for the interpretation of unbound and long-distance bound ziji. In this subsection, I will concentrate only on logophoricity and Li's account, a representative of the perspectivity approach, and briefly discuss their limitations. 4.3.1. LONG-DISTANCE BOUND ZIJI IS NOT A LOGOPHORIC PRONOUN According to Clements (1975), Sells (1987), and Stirling (1993), a logophoric pronoun has the following three properties: (a) it does not exhibit the blocking effect; (b) it can always have the source or the subject of consciousness as its antecedent, though it is not compatible with perception verbs; and (c) it cannot have the first-person pronoun as its antecedent.
Why the Blocking Effect?
291
However, ziji does not have any of the three properties. First, ziji exhibits the blocking effect, as pointed out in Section 1, namely, that an intervening first/second-person pronoun blocks all the remote third-person non-co-argument subjects from binding ziji, but a third person NP does not necessarily block first/second-person pronouns from binding ziji long distance. Second, ziji cannot have the NP carrying the role source as its antecedent, as shown in (21). (21) Zhangsani cong Lisij nar tingshuo naben shu hai-le zijiy i/*j . Zhangsan from Lisi there hear that-CL book hurt-Perf self 'Zhangsan heard from Lisi that that book hurt himself.' Also, ziji cannot have the subject of consciousness as its antecedent as long as there is an intervening first/second-person pronoun, See (10a) and (12a), though this would be possible if ziji were a logophoric pronoun. Third, although the logophoricity theory can account for long distance binding in belief contexts or sentences with embedded complement clauses when there is no first/second-person pronoun intervening, it fails to account for similar sentences which contain an intervening first/second-person pronoun, e.g., (10a) and (12a). The theory also fails to explain why long-distance binding is possible in the following sentences (Carl Lee Baker's lecture notes): (22) a. Johni minglin Billj [s PRO gei zijiyj guahuzi]. John order Bill to self shave 'John ordered Bill to shave him/himself.' b. Johtii bi Billi [s PRO gei ziji i/j guahuzi]. John force Bill to self shave 'John forced Bill to shave him/himself.' In (22), the local domain is the embedded infinitive clause indicated by S. Since the matrix object Bill controls PRO, Binding Condition A can only predict that Bill is the antecedent of ziji. That is, it fails to predict that the matrix subject can also be the antecedent. Since John in (22) is hardly a source, nor are sentences in (22) reports about John's feeling, thoughts, etc., the binding pattern observed in (22) is problematic for the logophoricity account. Besides, ziji can appear in the perception verb context, and can be long-distance bound, as long as there is no intervening first/second-person pronoun, as exemplified in (23). (23) Zhangsani tingshuo [NP Lisij de gou]k zai ziji i/j/k de fangjian jiao. Zhangsan hear Lisi DE dog at self DE room bark 'Zhangsan heard that Lisi's dog is barking in his own room.'
292
Haihua Pan
Finally, logophoric pronouns are said to be in complementary distribution with first-person pronouns, so they cannot have them as their antecedents, i.e., the (c) property of logophoric pronouns. However, long-distance bound ziji does not have the (c) property of a logophoric pronoun. It can readily have the first person pronoun wo 'I' as its antecedent, as shown in sentences like (6a, c), (la), and (8a). Since none of the three properties of logophoric pronouns is possessed by ziji, it is very difficult to argue that ziji is a logophoric pronoun.8 4.3.2. THE PERSPECTIVE-TAKING ACCOUNT is NOT ADEQUATE N.-C. Li (1991) argues that Chinese long-distance bound ziji is constrained by perspective-taking, and it is compatible only with reflective mental states. She differentiates unbound ziji and long-distance bound ziji. Although both have the experiencer as antecedents, they differ in the sense that the former appears in the expressive framing which does not contain the coding of the experiencer, while the latter occurs in the reportive framing with the coding of the experiencer. She suggests that the former is related to the experiencer's private experience, and the latter must refer to the experiencer who is in a mental state of self-awareness. The prediction of her analysis of long-distance bound ziji is that, in statements about facts and pure emotion or perception, ziji cannot be long-distance bound. Li's account can provide an explanation to why first/second-person pronouns cannot always bind ziji long distance when there is a third-person human NP intervening, since according to her, only verbs that are related to the experiencer's private experience, or to the reflective consciousness of the experiencer allow the long-distance binding of ziji- Although I agree with her on the role played by verbs and on the claim that ziji can have the projected ego as its antecedent, Li's analysis is not adequate in the following aspects. First, the sentence below contains a factive verb zhidao 'know,' so it should not allow ziji to appear in it, according to Li's account. This is because it is not guaranteed that Zhangsan is always self-conscious, as the belief can be attributed to him by the speaker without his awareness. However, ziji can appear in sentences like this, which suggests that self-consciousness is not a necessary condition for ziji to have a long-distance antecedent. (24) Zhangsani zhidao naben shu dui ziji i bull. Zhangsan know that-CLbook to self not-good 'Zhangsan knows that book is not good for him.' Second, sentences containing nonreflective reports also allow long-distance bound ziji to appear in them, which is not compatible with Li's ac-
Why the Blocking Effect?
293
count, because there should be no coding of the experiencer in nonreflective reports, but long-distance bound ziji must point to the projected ego of the experiencer, which requires the coding of the experiencer. (25) Zhangsani tingshuo Lisi qiangzou-le ziji i de shu. Zhangsan hear Lisi rob-Perf self DE book 'Zhangsan heard that Lisi took away his (own) book.' Also, perception verbs do allow long-distance bound ziji, as exemplified in (23) and (25), given in the previous subsection. Finally, Li does not provide an account to explain the blocking effects. Even with the highly reflective verb yiwei 'think,' ziji cannot be long-distance bound in the relevant sentences, as exemplified in (26). (26) Zhangsani yiwei wo/ni mei nazou ziji*i de shu. Zhangsan think I/you not take self DE book 'Zhangsan thought I/you did not take away his book.' From the discussions in this section, we can see that previous approaches do not provide a satisfactory account for the blocking effect of long-distance bound ziji. Thus, an alternative account is called for.
5. SELF-ASCRIPTION AND LONG-DISTANCE BOUND ZIJI Although I agree with the view that the properties of Mandarin reflexive ziji cannot be properly accounted for by pure syntactic conditions like Binding Condition A, and discourse and pragmatic principles are necessary to account for its properties, I think that the correct condition for long-distance bound ziji is not one that utilizes concepts like logophoricity or perspectivity, but one that employs self-ascription, as will be argued in the rest of this chapter. Note that my account assumes that the locally bound ziji is dealt with by the Anaphor Condition, as proposed in Pan (1998), which roughly corresponds to the category—syntactic reflexives—in Pollard and Xue (1998). 5.1. My View of Mandarin Reflexives In Pan (1995, 1997) I divide Chinese reflexives into contrastive and noncontrastive reflexives.9 The contrastive reflexive category also includes morphological reflexives like benren 'proper-person,' benshen 'proper-body,' zishen 'self-body,' and their compound forms. The noncontrastive reflexive category includes locality and self-ascription reflexives. Figure 1 below rep-
294
Haihua Pan
Figure 1. My Current View of Mandarin Reflexives
resents my current view on Chinese reflexives. Notice that the category self-ascription is now under the new category long-distance bound reflexives that also includes other long-distance bound reflexives such as the ones that are constrained by logophoricity. For example, Mandarin compound reflexive taziji can be used logophorically. See the relevant discussions in Pan (1998) and Section 6 following for the differences between logophoricity and self-ascription. In Pan (1997), I argue that the three different categories of reflexives are subject to different conditions. The contrastive ones, besides the contrastive condition, are also subject to a prominence condition like centrality, as suggested in Baker (1995), namely, that a contrastive reflexive must have the central character in the discourse as its antecedent.10 The locality reflexive is constrained by a revised condition on the basis of Chomsky's (1981) Binding Condition A. The condition utilizes the concepts of closeness and prominence (see Pan (1998) for relevant discussion). Since only long-distance bound reflexives including self-ascription ziji observe the blocking effect, I will not discuss locality reflexives and contrastive reflexives any further in this chapter. 5.2. Beliefs de re, de se, and de dicto Before getting to my analysis of self-ascription ziji, let me first introduce some necessary concepts for subsequent discussion. There are three kinds of belief. Belief de re is a belief about an entity. Belief de se, introduced in Lewis (1979), is a belief about the believer him/herself, which corresponds to
Why the Blocking Effect?
295
self-ascription. It involves self-ascribers and implies that the believer (consciously) ascribes a property to him/herself. Belief de dicto is a belief about a proposition and quantifying-in is not allowed, i.e., the elements in the belief cannot be accessed externally, i.e., outside the scope of the believer. Lewis (1979) suggests that belief de dicto is subsumed under belief de se, since, if a person believes a proposition, then we can say s/he has a belief about her/himself that s/he lives in the world where the relevant proposition is true, and thus this person has a de se belief about him/herself, which involves the relevant proposition. Because the dese belief in the de dicto case is attributed to him/her by the speaker, the person in question is not necessarily aware of it, and thus s/he is not necessarily self-conscious; see endnote 12. According to Lewis (1979), self-ascription can be applied to any attitude related to belief, knowledge, and desire. It also applies to any attitude that amounts to lack of belief, knowledge, or desire. This latter case is possible, because, as noted before for de dicto beliefs, the speaker can attribute the relevant de se belief to the person in question without his/her awareness. Connected with self-ascription are the concepts of self-ascriber and property: a self-ascriber ascribes a property to him/herself. First/second-person pronouns are obligatory self-ascribers, while third-person human NPs are optional self-ascribers, as shown in (27). (27) a. I think I am smart. b. You think you are smart. c. John thinks he is smart. Unlike the referent of John in (27c) who can have either a de re or de se belief about himself, the referents of / and you in (27) can only have a de se belief about themselves. The preceding difference indicates that, whenever the speaker or addressee ascribes some property to him/herself, it must be a de se belief, or a self-ascription. Note that the ambiguity in (27c) comes from the third-person pronoun he. It can be interpreted as a referential pronoun or a bound variable pronoun, whereas the first occurrence of first and second persons in (27a) and (27b) is not a bound variable pronoun, and the second occurrences of them have to be bound variable pronouns.11 Furthermore, although third-person NPs, e.g., John, can be referred to as the believer/carrier of belief, or the entity named John, first/second-person pronouns can be referred to as the believer/carrier of belief only. Since firstand second-person pronouns have obligatory de se beliefs, they are obligatory self-ascribers in the context of belief de se, whereas third-person NPs are optional self-ascribers in the de se context. In the discussion following, we assume that all the sentences in the language represent a belief, knowledge, or desire of the carrier. Even for simple sentences like Zhangsan xihuan Lisi 'Zhangsan likes Lisi'—a statement
296
Haihua Pan
made by the speaker—is a belief of the speaker. Note that the sentence like Zhangsan zhidao Lisi xihuan Wangwu 'Zhangsan knows that Lisi likes Wangwu' can also be a belief of the speaker. Besides, there are two other situations (readings) for this sentence: (a) Zhangsan self-ascribes the belief 'Lisi likes Wangwu'; and (b) the speaker attributes to Zhangsan the de se belief—a belief about himself—that he lives in the world where the proposition Lisi likes Wangwu is true. In both situations, Zhangsan has the knowledge that he lives in the world where the proposition Lisi likes Wangwu is true, though in the (b) reading Zhangsan is not self-conscious, as he may not know that he was attributed a de se belief, and the speaker can stand at his place, viewing the object in question, in order to satisfy the requirement of self-ascription. (See more relevant discussion in Section 6 of this chapter.) Also note that, for the latter sentence Zhangsan zhidao Lisi xihuan Wangwu, we have one self-ascription embedded in another. The speaker self-ascribes the following belief: that Zhangsan self-ascribes the belief that Lisi likes Wangwu. 5.3. Self-ascription Ziji Points to the Carrier of Belief N.-C. Li (1991) points out that there is an unbound ziji which refers to the speaker, as exemplified here: (28) Ziji neng qu nar ma? self can go there Q 'Can self(I) go there?' She claims that this ziji is referential, which is why it can be used alone. However, I would rather say that ziji is bound in a discourse in which (the projected ego of) the speaker is its antecedent. Besides the speaker, an addressee can also be the antecedent of the so-called unbound ziji when the speaker addresses the addressee using (29). Note that (29) can also be used to talk about a third party that is salient in the discourse. (29) Ziji weishenme bu qu ne? self why not go Q 'Why didn't self (you) go?' Furthermore, a sentence with an object ziji can also have the speaker as antecedent if the speaker asks him/herself the question, as in (30a). (30) a. Zhangsan xihuan ziji ma? Zhangsan like self Q 'Does Zhangsan like (my)self?'
Why the Blocking Effect?
297
b. Zhangsan zhidao Lisi xihuan ziji ma? Zhangsan know Lisi like self Q 'Does Zhangsan know that Lisi likes myself/him/himself?' This is true even for a complex sentence like (30b). Besides the normal readings with Zhangsan and Lisi as the antecedents of ziji, the speaker can also be the antecedent of ziji in (30b), though I admit that this reading is more difficult to get than the ones with Zhangsan and Lisi. Note that this reading is highly reflective; actually, all the similar readings are presented in the form of a question, which implies that the speaker has an image of himself who has the property denoted by the sentence in question. By contrast, one of the readings in (30b), i.e., with Zhangsan as antecedent, need not be reflective, since the speaker can attribute Zhangsan a de se belief lx[Lisi xihuan x], in this case, Zhangsan is not necessarily self-conscious, as he may not know it.12 This reading suggests that Li's requirement that the antecedent of long-distance bound ziji must be the experiencer of a reflective mental state is not the correct characterization of long-distance bound ziji for third-person believers, though she may be right with respect to first/second-person pronouns. We can say that Zhangsan may self-ascribe the knowledge about the fact that Lisi likes him. Even his lack of self-ascribing the knowledge also falls in the domain of self-ascription, as pointed out by Lewis (1979), and repeated in the previous subsection. This is because the speaker can attribute Zhangsan a de se belief without his awareness, and it is the speaker who makes the connection between Zhangsan and ziji. Note that, in this case, Zhangsan may be conscious, but he is not self-conscious. What is crucial here is that Zhangsan is not necessarily self-conscious, and whether he is conscious or not is not relevant to our discussion of self-ascription. From the discussion preceding, we can see that long-distance bound ziji points to the carrier of belief, knowledge, or desire, and it is thus constrained by self-ascription (see more relevant discussions in Pan 1995,1997).
5.4 The Condition for Self-ascription Ziji Assuming the discussion preceding, I propose the following condition for self-ascription ziji. (31) The Condition for Self-Ascription Ziji Ziji can be bound to the carrier of belief, the most prominent self-ascriber, in a linguistic domain g iff there is no blocker in the believed proposition contained in g.
298
Haihua Pan
(32) The Prominence Condition a is the most prominent self-ascriber in g iff there is no b in g such that b appears higher in one of the following hierarchies than a. a. SUBJ > OBJ or OBLIQUE b. Dominating NPs > Dominated NPs (33) a is a blocker for b if a is a self-ascriber such that (a) a precedes ziji; and (b) neither a nor the NP controlled by it is an argument of an irreflexive predicate containing ziji. In a de se belief situation, self-ascribers include all the referents of the animate subjects of attitudinal verbs, and first- and second-person pronouns. The hierarchy in (32b) only compares NPs that have dominance relationships with one another. The domain g includes the clause that contains at least one attitudinal verb, and one sentence may have more than one domain if it contains more than one attitudinal verb. The believed proposition is the complement clause of the attitudinal verb under consideration. Now I will show that the condition in (31) can handle all the long-distance bound cases of ziji and can provide a natural account to the blocking effect. (34) a. Johtii yiwei billj xihuan zijiyy John think Bill like self 'Johnj thinks Billj likes hinii/himselfj.' b. Johni yiwei [naben shu]j hai-le ziji i/*j . John think that-CL book hurt-Perf self 'Johnj thinks that book hurt himi.' c. Johni yiwei wo/nij xihuan ziji*i/j. John think I/you like self 'Jhonj thinks I/youj likes himself*i/my/your-selfj.' d. Woi yizhi yiwei Billj xihuan ziji i/j , keshi wo zuo le. I always think Bill like self but I wrong Prt 'Ii always thinks Billj likes mej/himselfj, but I was wrong.' (35) a. [slJohni zhidao [S2Billj juede Mark k xihaun zijii/j/k]]. John know Bill think Mark like self 'John knows that Bill thinks that Mark likes himself/him.' b. [S1 Johni juede [S2 wo/nij, zhidao Markk xihaun ziji*i/j/k]]John think I/you know Mark like self 'John thinks that I/you know that Mark likes himself/me/you.' In (34a), the domain g is the whole sentence, and the believed proposition is the embedded clause. Since the local subject is not a first- or second-person pronoun, and neither is it the subject of an attitudinal predicate, it is
Why the Blocking Effect?
299
not self-ascriber. Because there is no self-ascriber in the believed proposition contained in g and thus no blocker, the only self-ascriber, the matrix subject, will be assigned the antecedent of ziji. Hence, (31) predicts that long-distance binding is possible in (34a), which is consistent with the fact. (The coindexation with j is due to the locality ziji, which is not the concern of this chapter). In (34b), like (34a), the domain g for long-distance bound ziji is the matrix clause, and the local subject is not a self-ascriber. Since only the matrix subject is a self-ascriber and no other NPs are more prominent, it can be the antecedent of ziji as predicted by (31). (The locality ziji in (34b) is excluded independently, because ziji carries the experiencer role, but naben shu 'that book' has the inanimate feature, and thus they are incompatible with each other.) In (34c), the first/second-person pronoun wo/ni in the embedded clause, a self-ascriber, precedes ziji and is not an argument of an irreflexive predicate containing ziji, so it is a blocker in the believed proposition, according to (33). Thus, (31) would predict that long-distance binding of ziji is not possible, which is consistent with the fact. In (34d), the matrix subject in the first conjunct can be the antecedent of ziji just as in (34a), since the local subject is not a self-ascriber and is thus not a blocker. In (35a), there are two domains for the long-distance bound ziji: the matrix clause (S1) and the intermediate clause (S2). For domain S1, since there is a self-ascriber in the believed proposition S2, i.e., the subject of the attitudinal predicate juede 'think' in S2, this self-ascriber may function as a blocker for the matrix subject if it is an obligatory self-ascriber. But as pointed out in the previous subsection, third-person NPs are optional self-ascribers, so when the intermediate subject does not function as a self-ascriber, and is thus not a blocker, the matrix subject can be the antecedent of ziji- As for the intermediate subject in domain S2, similar to (34a), (31) allows it to be the antecedent of ziji Thus, (31) correctly predicts the possible readings in (35a). In (35b), the first/second-person pronouns in the believed proposition S2 precede ziji and are not an argument of an irreflexive predicate containing ziji, so they function as blockers for the matrix subject, just as in (34c). Since they are obligatory self-ascribers, the intervening first/second-person pronouns will block the matrix subject from binding ziji. Note that, unlike the matrix subject, the first/second-person pronouns in (35b) can be the antecedents of ziji, as they are also the most prominent NPs in the domain (S2), and there are no (other) self-ascribers in the believed proposition in this domain. Since all the obligatory self-ascribers discussed so far are subjects, let us examine cases in which the obligatory self-ascribers are nonsubjects, to see if (31) also makes correct predictions.
300
Haihua Pan
(36) a. Johani shuo Billk gei wo/nij kan-guo ziji *i/j/k de shu. Jogn say Bill to I/you see-GUO self DE book 'John said that Bill showed me/you his/my/your book.' b. Johni shuo Billj gaosu-guo ni/wok Markn da-le ziji*i/?j/*k/n yixiar. John say Bill tell-Guo you/I Mark hit-Perf self once 'John said that Bill told you/me that Mark hit himself/him once.' In (36a), the obligatory self-ascriber wo/ni 'I/you' are nonsubjects; they are obliques indicated by preposition gei 'to.' Since there is no requirement of subjecthood in the definition of blockers in (33), (31) would predict that the obligatory self-ascriber in (36a), though not a subject, can block the long-distance binding of ziji by the matrix subject, because the obligatory self-ascribers wo/ni in the believed proposition—the intermediate clause—are blockers for the matrix subject, according to (33); note that they precede ziji and are not an argument of an irreflexive predicate containing ziji. This prediction is borne out by the impossibility of long-distance binding of ziji observed in (36a). In (36b), the obligatory self-ascribers wo/ni are not subjects, either; they are objects. Just as in (36a), the object wo/ni in (36b) blocks the long-distance binding of ziji by the matrix subject, as predicted by (31). This is because the obligatory self-ascriber first/second-person pronouns in the intermediate clause also function as blockers for the matrix subject. Note that there is a difference between (36a) and (36b): the first/second-person pronouns do not block the local subject from binding ziji, though they do for nonlocal subjects except for the co-arguments of firstand second-person pronouns.13 This is a difference between locality ziji and long-distance bound ziji: only the latter observe the blocking effect. Also, note that the fact that the local subject can be antecedent of ziji with or without the first/second-person pronouns intervening is accounted for by the condition on locality ziji.14 Hence, it is not a problem for our account for long-distance bound self-ascription ziji. Condition (31) also predicts that long-distance binding of ziji is possible if there is an obligatory self-ascriber that does not intervene between the potential antecedent and ziji. This prediction seems to be correct, as the contrast between the sentences in (37) exemplifies. (37) a. Johni shuo Billj ba zijii/j/*k de shu songgei-lewo/nik. John say Bill BA self DE book give-Perf I/you 'John said that Bill gave his own books to me/you.' b. Johni shuo Billj Songgie-lewo/nik yiben ziji*i/j/?k de shu John say Bill give-Perf I/you one-CL self DE book 'John said that Bill gave me/you one of his own books.'
Why the Blocking Effect?
301
The obligatory self-ascribers wo/ni are obliques in the embedded clause, but they do not intervene between the matrix subject and ziji in (37a), though they do in (37b). Since the obligatory self-ascribers wo/ni do not block long-distance binding only in sentences like (37a), we can say that obligatory self-ascribers can block long-distance binding of ziji only if they intervene between the two NPs under discussion. This is why we require the self-ascriber to precede the reflexive in the definition of blockers in (33a). Sentences like (37a) suggest that an account stipulating that ziji must take the perspective of the sentence in question is not adequate, as also argued in the previous section. Under the perspective-based account, the first- or second-person NP in the embedded clause would be the deictic center; note that word order should be irrelevant in an account based on perspectivity. Thus, the first- or second-person NP would have to be the perspective-taker. That is, an account employing perspectivity would predict that the long-distance binding of ziji in (37a) is blocked, since a first/second-person pronoun—a perspective-taker—appears in it. This account would also predict that the referent of the first- or second-person NP can always be the antecedent of ziji, a prediction that is at odds with the binding pattern observed in (37a). Hence, the facts here further show that the account proposed in this section is superior to an account based on perspectivity. Another property of (31) is that it requires the antecedent of ziji to be the most prominent self-ascriber. Let us see if this requirement is fulfilled in (38) below. (38) a. Billi cong wo/nij nar tingshuo Suek piping-le zijii/*j/k. Bill from I/you there hear Sue criticize-Perf self 'Bill heard from me/you that Sue criticized herself/him.' b. Wuqing de shishii gaosu Mark j Suek pian-le ziji*i/j/k.. cruel DE fact tell Marck Sue cheat-Perf self 'The cruel fact told Mark that Sue cheated herself/him.' In (38a), there is no intervening self-ascriber in the believed proposition, the embedded clause, so the NPs in the matrix clause are possible antecedents of ziji. If there is no requirement of the most prominent self-ascriber, then both the matrix subject and the oblique first/second-person pronouns, obligatory self-ascribers, should be possible antecedents of ziji. However, the fact that only the subject can be the antecedent suggests that some kind of prominence relation is necessary. Such a requirement is specified in (31). Note that sentences like (38b) are not counterexamples to our account. This is because the inanimate subject is not a self-ascriber in nature, and thus the matrix object can be the antecedent of ziji in (38b). Sentence (39), similar to (12a), suggests that the prominence requirement should not be imposed on the blockers (self-ascribers), and our account cor-
302
Haihua Pan
rectly predicts the long-distance binding of ziji is not possible in (39). This is because the first-person pronoun in the believed proposition is a blocker for the matrix subject, as it precedes ziji and is not an argument of an irreflexive predicate containing ziji, according to (33). (39) Baoyui yiwei wodej xueshengk bu xihuan ziji*i/*j/k. Baoyii think my student not like self 'Baoyii thinks that my student does not like himself.' Our account can also cover cases like (40) and (41), which are claimed to be evidence for the influence of pragmatic factors (world knowledge) in the determination of the antecedents of ziji.15 (40) Zongtongi quing woj [s PROj zuo zai ziji de shenbian]. president ask I sit at self DE side 'The president asked me to sit beside him.' (41) a. Zhangsan pa [s wo/ni hui chaoguo ziji]. Zhangsan fear I/you will surpass self 'Zhangsan fears that I/you will surpass him.' b. Ni bu pa[sta chaoguo ziji ma]*? you not fear he surpass self Q 'Don't you fear that he surpasses you?' Although the believed proposition, the embedded clause in (41a), contains a first/second-person pronoun, we do not take the first/second-person pronoun as a blocker for the matrix subject, since the predicate in the embedded clause is irreflexive, and thus precludes co-reference of its arguments: wo/ni and ziji. This is specified in our definition of blockers in (33). Since the first/second-person pronoun is not a blocker for it, the matrix subject in (41a) can be the antecedent of ziji, according to (31). Example (40) can be dealt with similarly. The irreflexive predicate is x zuo zai y de shenbian 'x sits beside y,' where x is PRO, and y is ziji. The only self-ascriber wo in (40) cannot be a blocker, since it controls PRO which is an argument of the irreflexive predicate in question. Thus, there is no blocker for the matrix subject Zongtong 'president,' according to (33). Hence, the matrix subject Zongtong 'president' can be the antecedent of ziji, as expected.16'17 Condition (31) also allows the co-arguments of first/second-person pronouns to be the antecedents of long-distance bound ziji, as shown in (36b) and the sentence following.18 (42) Zhangsani gaoshu woj [s ziji i/*j mei bei dahui xuanshang]. Zhangsan tell I self not by conference select 'Zhangsan told me that he was not selected by the conference.'
Why the Blocking Effect?
303
According to (31), the believed proposition, the embedded clause S, does not contain a self-ascriber, so there is no blocker for the matrix subject, Zhangsan, to bind ziji. Hence, long-distance binding is possible in (42). Similarly, the fact that the intermediate subject, Bill, in (36b) can be an antecedent of ziji is also predicted by our condition (31).19 Note that, although the existence of first- and second-person pronouns is crucial for the blocking effect, there is no need to stipulate in condition (31) the obligatoriness of self-ascribers, as shown by the following example (43) (Carl Lee Baker, personal communication).20 (43)
John renwei Bill zhidao Mark ba ziji de shu jie-gei-le John think Bill know Mark BA self DE book loan-to-Perf ziji de pengyou. self DE friend 'John thinks Bill knows that Mark has loaned his book to his friend.'
(44) a. John{ renwei Bill} zhidao Markk ba ziji i de shu jie-gei-le John think Bill know Mark BA self DE book loan-to-Perf ziji i de pengyou. self DE friend 'John thinks Bill knows that Mark has loaned his book to his friend.' b. Johni renwei Billj zhidao Markk ba zijij de shu jie-gei-le John think Bill know Mark BA self DE book loan-to-Perf zijij de pengyou. self DE friend 'John thinks Bill knows that Mark has loaned his book to his friend.' c. *John{ renwei Billj zhidao Markk ba ziji i de shu jie-gei-le John think Bill know Mark BA self DE book loan-to-Perf zijij de pengyou. self DE friend 'John thinks Bill knows that Mark has loaned his book to his friend.' d. *Johni renwei Billj zhidao Markk ba zijij de shu jie-gei-le John think Bill know Mark BA self DE book loan-to-Perf zijii de pengyou. self DE friend 'John thinks Bill knows that Mark has loaned his book to his friend.' For ease of exposition, I have represented the long-distance bound readings of sentence (43) individually in (44). The possible readings for the two oc-
304
Haihua Pan
currences of ziji are that either John or Bill can bind both occurrences of ziji. Note that the mixed readings are not possible, as shown in (44c) and (44d). Thus, sentence (43) suggests that, like first/second-person pronouns, third-person NPs, when functioning as self-ascribers, will block other NPs outside the domain from binding ziji. So we do not need to specify obligatory self-ascriber in the //-clause in condition (33). Note that sentence (43) does have mixed readings with the local subject, as exemplified here: (45) a. John{ renwei Bill} zhidao Markk ba ziji i de shu jie-gei-le John think Bill know Mark BA self DE book loan-to-Perf zijik de pengyou. self DE friend 'John thinks Bill knows that Mark has loaned his book to his friend.' b. Johrii renwei Billj zhidao Markk ba ziji j de shu jie-gei-le John think Bill know Mark BA self DE book loan-to-Perf ziji k de pengyou. self DE friend 'John thinks Bill knows that Mark has loaned his book to his friend.' c. Johrii renwei Bill} zhidao Markk ba ziji k de shu jie-gei-le John think Bill know Mark BAself DE book loan-to-Perf ziji i de pengyou. self DE friend 'John thinks Bill knows that Mark has loaned his book to his friend.' d. Johni renwei Billj zhidao Markk ba ziji k de shu jie-gei-le John think Bill know Mark BAself DE book loan-to-Perf ziji j de pengyou. self DE friend 'John thinks Bill knows that Mark has loaned his book to his friend.' In (45) ziji k is locally bound, while ziji i/j is long-distance bound. The sentences in (45) show that the local subject and the remote subjects can have mixed readings. The binding patterns in (44) and (45) indicate that local subjects function differently from long-distance ones. They also suggest that locally bound ziji and long-distance bound ziji are constrained by different conditions, as suggested in Pan (1995,1997). From the discussion preceding, we can see that our account of long-distance bound ziji fares better than do the previous analyses presented in this chapter. Hence, the condition in (31) is an appropriate condition for self-ascription ziji.
Why the Blocking Effect?
305
Now we can answer the question: why is there a blocking effect for long-distance bound ziji? It is because of the difference between the first/second-person pronouns and third-person NPs. The blocking effect is observed because of the obligatoriness of first/second-person pronouns being a self-ascriber if they do not agree in person features with the carrier of belief, knowledge, or desire involved. Since ziji points to the carrier of belief, a self-ascriber, the intervening obligatory self-ascribers will prevent it from being bound by farther-away self-ascribers. Hence, the blocking effect.
6. DISCUSSION Like Huang and Tang (1991) and Cole and Wang (1996), we recognize the need for separating blockers from antecedents. As one can see from our condition on self-ascription ziji given in (31), our concept of blockers includes only self-ascribers that are (a) first/second-person pronouns that precede the reflexive ziji and are not an arguments of an irreflexive predicate that contains ziji; and (b) animate NPs that are subjects of a belief, knowledge, or desire. There is no prominence restriction for being a blocker for first/second-person pronouns, though there is one on being the antecedent of long-distance bound ziji- Unlike Huang and Tang (1991) but in line with Cole and Wang (1996), our account also suggests that a blocker is not necessarily an antecedent of the reflexive ziji; see (12a) (See more discussion on the difference between blockers and antecedents in Cole and Wang (1996) and Pan (1998).) Another issue is the relationship between the concepts of logophoricity and self-ascription. Since these concepts are closely related, it is very easy to confuse them. In the following, I will briefly discuss their differences. Logophoricity and self-ascription differ from each other in the following aspects. First, according to Sells (1987), the SOURCE of the report is one primitive of the concept of logophoricity. However, self-ascription can never be attributed to the individual who carries the source role if it is not also the carrier of belief, which is why a source cannot be the antecedent of long-distance bound ziji, as shown in (21). Second, the subject of consciousness is another primitive in Sells' concept of logophoricity (SELF). By contrast, the concept of self-ascription does not necessarily require self-consciousness; for example, (46) would present a problem for the requirement of self-consciousness of the subject, whereas, in the self-ascription concept, self-consciousness is not necessary. This is because, in addition to the reading in which Zhangsan is self-conscious, i.e., that he self-ascribes the relevant belief, the de se belief can also be attributed to Zhangsan by the
306
Haihua Pan
speaker without his awareness. In this latter reading, Zhangsan is not aware of the fact that the speaker attributed him a de se belief. This is what I meant earlier, that in self-ascription, the person, denoted by a third-person NP, who has the de se belief, need not be self-conscious, as the speaker can attribute him the relevant de se belief. Notice that the consciousness of Zhangsan is not relevant here, and what is crucial is the self-consciousness concept; see endnote 12. Even for the positive version of (46), self-consciousness is not necessary for it to be true.21 Just like the negative version, Zhangsan can be attributed a de se belief by the speaker without his awareness. The clearer example is the question version of (46) given in (30b), since there is no guarantee that the answer to the question is positive. (46) Zhangsan bu zhidao Lisi xihuan ziji. Zhangsan not know Lisi like self 'Zhangsan does not know that Lisi likes him.' Self-ascription is also different from the third primitive in Sells' logophoricity concept, point of view (PIVOT). Self-ascription is person neutral, that is, it can be combined with all three persons, whereas point of view has separate persons: first-, second-, or third-person point of view, depending on which NP is the point of reference (Mitchell, 1986). The closest one among the three points of view to self-ascription is the first-person point of view. However, there are at least two differences between them, as far as I can see. (47) a. Zuozi zai wo de zuobian. table at I DE left 'The table is to my left.' b. Zuozi zai ziji de zuobian. table at self DE left 'The table is to self's left.' The first difference between wo T and ziji is that the interpretation for (47a) is fixed to the speaker, but that for (47b) is not. In (47b), ziji can refer to the speaker, the addressee, or a third party individual. The last reading is possible if the speaker and addressee are talking about a particular third person. This difference becomes clearer if we embed (47) in a sentence. With the different person NPs as the matrix subject, the interpretation is very different for (47b), though for (47a) it always has the reading that the point of reference is the speaker. The second difference is that ziji can refer to the projected ego, while wo cannot. It seems that wo is external in nature, but ziji is internal in nature. In other words, wo is existentially bound, i.e., identified outside the scope of the carrier of belief, while ziji is not. If we really want to name self-ascription ziji in terms of the concept of point of view,
Why the Blocking Effect?
307
we can view it as presenting the self point of view, which is the essential property of self-ascription. Another difference between logophoric pronouns and self-ascription ziji is that the former cannot have a first-person pronoun as antecedent, whereas the latter can easily do so, as pointed out in Section 4.3.1. Maybe one can see the difference between self-ascription and the concept of point of view (and logophoricity) more clearly by using the following analogy. Suppose that a viewer (the speaker) is seeing things from a point of reference, and the reference can be first-second, and third-person NPs. If the viewer takes the perspective of a third-person NP, then we call this a third-person point of view. Note that the viewer can have his/her independence from the referent of the third-person NP, namely, that they stand at different places. This is the so-called logophoricity use of reflexives. However, if the viewer becomes one with the referent of the third-person NP, namely, that they at least stand at the same place, then we have the self point of view, or it involves self-ascription. Hence, the difference between logophoricity and self-ascription is that the former allows the independence of the viewer from the point of reference, while self-ascription does not. That is, logophoricity does not necessarily require that the two (the reviewer and the point of reference) become one, though self-ascription does require it. Note that the self-ascription requirement is satisfied only if the speaker stands at Zhangsan's place and views the object in question from that place. Similarly, if the second-person pronoun is the point of reference, and the viewer takes the perspective of the referent of the second-person pronoun, but still keeping his/her independence, then we have a second-person point of view. However, if the two become one, i.e., the viewer loses his/her independence, then we have the self point of view again. Note that the independence of the viewer from the point of reference and the self-consciousness of the individual used as the point of reference are independent concepts. Although the viewer or speaker is always self-conscious, the individual as the point of reference is not necessarily so, especially when the point of reference is a third-person NP; though s/he may be conscious, that is not at issue here; see endnote 12. Self-ascription does not require the individual—the point of reference—to be self-conscious, as pointed out previously and in Section 4.3, though it does require that the viewer and the point of reference become one, i.e., at least standing at the same place.22 An anonymous reviewer asks why can't the speaker take the point of view of Zhangsan in (46). Actually, the speaker can take Zhangsan's point of view. The issue is the relationship between the speaker and Zhangsan. There are two possible ways for the speaker to take Zhangsan's point of view: (a) the speaker is independent of Zhangsan, e.g., they stand at differ-
308
Haihua Pan
ent places, and the speaker views the object from Zhangsan's perspective; and (b) the speaker and Zhangsan stand at the same place or, in my terminology, they become one. The first case is the so-called third-person point of view, while the second case is the self point of view. I think that sentences like (46) only allow the second case that corresponds to self-ascription, though logophoricity may allow both cases. Hence, only self-ascription corresponds exactly to the use of long-distance bound ziji. In summary, self-ascription and logophoricity are two different concepts, though they overlap in meaning. I think that the concept of self-ascription is more primitive and covers more cases than does logophoricity.23
7. CONCLUSION In this chapter, I have presented systematic facts on the blocking effect of long-distance bound bare reflexive ziji in Mandarin Chinese. I have argued for the nonsymmetrical nature of the blocking effect and against the symmetrical unlike-person blocking view. I have shown that not only subjects or NPs contained in subjects but also nonsubjects can block long-distance binding of ziji as long as they are first- or second-person pronouns, or contain first- or second-person pronouns. Hence, the blocking effect is more sensitive to the person feature than to the local subject. I claim that long-distance bound ziji is constrained by self-ascription and have proposed a condition in (31) to account for the binding properties of ziji. I have explained the blocking effect by appealing to the special role played by the first- and second-person pronouns in self-ascription: since first- and second-person pronouns are obligatory self-ascribers, they will block long-distance binding of ziji by third-person NPs. Since sometimes third-person NPs can indeed block the long-distance binding of ziji by first/second-person pronouns, though they are not as strong as the cases with first/second-person pronouns as blockers, we need to investigate the reason why. One possible reason is that, following Li (1991), we may say that it is required to have highly reflective contexts for long-distance bound ziji, though unlike Li (1991), we think the relevant constraint only applies to first/second-person pronouns, not to third-person NPs, when they function as long-distance antecedents. Hence, there is another asymmetry between third-person NPs and first/second-person pronouns in the sense that highly reflective contexts are required only when the latter function as the antecedents of long-distance bound ziji. We leave this issue open for future research.
Why the Blocking Effect?
309
ACKNOWLEDGMENTS Parts of this article were presented at the 1997 LSA Workshop on Long-Distance Reflexives in Ithaca, New York. The author would like to thank the participants for interesting questions. He would also like to thank Peter Cole, James C.-T. Huang, Manfred Krifka, Carlota Smith, Jonathan Webster, and Liejiong Xu for helpful comments and suggestions on an earlier version of this chapter. Thanks also go to the two anonymous reviewers for the helpful comments and suggestions. As usual, the author is responsible for all potential errors.
NOTES l
The following abbreviations are used in this chapter: CL (classifier), Perf (perfective marker), DE (possessive/modifier/complementizer marker), ZAI (progressive marker), ZHE (imperfective nonprogressive marker), GUO (the experiential marker), Q (question morpheme), Prt (sentence-final particle), Conj (sentence connective). Also, note that the Chinese names Zhangsan, Lisi, and Wangwu given in the chapter are generic names rather than real names. 2 F.-X. Li (1990) also notices the asymmetrical nature of long-distance binding in Chinese. I differ from him, though, in the sense that I group first/second-person pronouns together in contrast with third-person NPs, but he differentiates all three person forms. I think that first-person pronouns can bind ziji with a second-person pronoun's intervening only if ziji is contrastive. Since it belongs to the contrastive reflexive, and is thus subject to different conditions, this usage of ziji should not interfere with the characterization of noncontrastive reflexives (see Pan (1997) for relevant discussion). 3 Xue, Pollard, and Sag (1994) propose a new category Z-pronoun for ziji besides locality ziji subject to local o-binding and discourse ziji, and suggest that it must be o-bound (see Pollard and Sag, 1994, for the concepts of (local) o-binding). Although they suggest differentiating two types of blocking effect—unlike-person blocking for ziji in complement clauses and animate blocking for ziji in subcommanding cases—Xue, Sag, and Pollard (1994) do not provide a clear account to explain the blocking effect for ziji except pragmatic factors. Hence, I will not discuss their analysis further in the rest of this chapter. 4 The reader can find a detailed critique of all the previous analyses mentioned in this chapter and other analyses of Chinese reflexives in Pan (1995,1997). 5 Note that it is not adequate to claim that cases like (13) are metaphoric extension (or personification), as suggested in Tang (1989). One weakness of the personification claim is that there is no clear condition to determine when personification takes place, and when it does not. We think that ziji, in general, can take either an animate or inanimate NP as its antecedent, although we understand that, for a specific case, it depends on the semantics of the verb to determine the animacy feature of ziji, and whether the NP in question can be an antecedent of ziji. For instance, in the case fol-
310
Haihua Pan
lowing, the inanimate subject cannot be the antecedent of ziji because of the animate feature ziji has that is determined by the experiencer role ziji carries. i.
*Na-bian wenzhang hai -le ziji. That-CL article hurt-ASP self Lit: That article hurt self.
What we are saying, basically, is that the animacy feature of ziji is not determined lexically or inherently, but rather it is determined by context. 6 We notice that Cole and Wang (1996) claim that first/second-person pronouns do not block long-distance binding of ziji by a third-person NP if they are the BA object or an oblique introduced by the passive marker bei 'by.' It seems that there is some speaker judgment variation here. Although this possibility may be extended to adjuncts, it does not apply to objects, as in (4a). That is, even if we agree in not counting obliques and adjuncts as possible blockers, objects and possessors in a subject are still problematic to the feature agreement checking approach. The relevant examples are (4a) and (12a). Besides, all the arguments against this approach presented earlier are still valid. 7 Example (17) is a summary of Huang and Tang's (1991) original descriptions. 8 Both anonymous reviewers point out that the notion of logophoricity has been used in varying ways by different authors, and thus my arguments against the individual notions of logophoricity proposed by Sells (1987) do not refute the general notion of logophoricity. However, we notice that the general notion of logophoricity is not well defined. One possible definition is given in Reinhart and Reuland (1993): nonargument reflexives or arguments that are focused are logophors, but this notion only applies to compound reflexives like himself, not to bare reflexives like ziji. If we use the conception that all the cases that do not obey Binding Condition A belong to the so-called logophoricity, then it is very clear to see that this general notion of logophoricity is too weak to account for the properties of ziji. Although it can predict the long-distance binding possibility, the general notion fails to explain the blocking effect, and it also fails to account for the subject orientation property of ziji. One of the reviewers asks what category the logophoric pronouns belong to if logophoric pronouns are not de se anaphors. As shown in Fig. 1 in Section 5.1 following, I take the logophoric reflexives as belonging to a separate category that is dominated by long-distance bound reflexives that also dominate the self-ascription reflexives. See also endnote 23. 9 This classification does not include the adverbial usage of ziji, as discussed in Tang (1989) and Pan (1995,1997). The following is a relevant example. i.
Zhangan ye ziji zuo fan. Zhangsan also self do rice 'Zhangsan also cooks his meal himself.'
We can also group this usage of ziji with the locality reflexive, since they can only have clause-bound antecedents. 10 Note that Baker's conditions are not fine-grained enough to differentiate the four contrastive reflexives in Chinese: ziji, benren, benshen, and zishen. For the rele-
Why the Blocking Effect?
311
vant conditions to differentiate these reflexives, see the discussion in Chapter 7 of Pan (1997). n The bound variable interpretation of pronouns is a necessary condition, but not a sufficient condition, for de se beliefs. Note that, even for the bound variable interpretation of pronoun he, there are also two readings: de re and de se. See Chierchia (1989) for relevant discussion. 12 Note that I make a distinction between the concepts of consciousness and self-consciousness. The former does not entail the latter, though the latter does entail the former. A person can be conscious without being self-conscious. For example, when a person is awake, then he is conscious. He can read and speak. Nevertheless, he may not know what his name is if he has suffered brain damage and lost all of his previous memory. In this latter situation, the person in question is not self-conscious, though he is conscious. Hence, when I say a person is not self-conscious, I mean that he may not know something about himself, though he is awake and conscious. 13 Example (36b) also shows that first/second-person pronouns do not block their co-arguments from long-distance binding ziji. See more relevant discussions later. 14 The condition on locality ziji and taziji is proposed in Pan (1998), and we repeat it here for reference. (i)
The Anaphor Condition An a can be the antecedent of an anaphor X in a linguistic domain y if (a) a is a noun phrase that does not dominate X; and (b) There is no closer blocker in g.
(ii)
a is closer to X, the reflexive, than b iff the path from X to the minimal maximal projection dominating a is a subset of the path from X to the minimal maximal projection dominating p.
(iii) b is a blocker for a if (a) b is not less prominent than a; and (b) b, if not dominating a, must be a subject. (iv) a is less prominent than b if it appears lower in the animacy hierarchy than b does. (v) 15
Animacy Hierarchy (Chou, 1992) [+human] > [+animate, -human] > [-animate]
Thanks to an anonymous reviewer for pointing out that sentence (40) may be a counterexample to the claim that first-person pronouns always induce the blocking effect. Sentences like (40) are well-behaved examples showing that semantic constraints play a role in the determination of the antecedents of ziji- The involved predicates are irreflexive predicates, i.e., the meaning of the predicate precludes co-reference of its arguments. We have incorporated this constraint in our definition of blockers in (33). 16 The same reviewer points out that the following sentence may be a counterexample to the claim that third-person NPs do not block the long-distance binding of ziji by first/second-person pronouns. (Note that I only claim that third-person NPs
312
Haihua Pan
do not NECESSARILY block long-distance binding of ziji by first/second-person pronouns; see endnotes 19 and 20.) (i)
Nii weishenme meiyou shixian gaosu dajia Lisij-. bu xihuan ziji/;? you why didn't before tell all Lisi not like self 'Why didn't you tell everyone in advance that Lisi didn't like you?'
Although I agree with the reviewer that first/second-person pronouns are strong blockers and third-person NPs are weaker blockers, I think that the i reading in (i) is possible, at least marginally. It seems that for the remote first/second-person pronouns to be the long-distance antecedents of ziji, we need to make extra effort to provide appropriate context. This may be the point made in Li (1991), namely, that long-distance bound ziji requires highly reflective contexts, and it points to the projected ego. Although I don't think that her claim applies to third-person NPs as antecedents, it may be true that it applies to first/second-person pronouns as antecedents. Sentences like (i) seem to suggest that there may be additional constraints for the cases in which first/second-person pronouns are the long-distance antecedents of ziji. I leave this issue open for future research. 17 The reviewer also argues against the self-ascription account of ziji using (i) below, claiming that it is very difficult to say the subject Zhangsan is the carrier of belief, and thus a syntactic account is necessary. (i)
Cong nei yihou Zhangsani zai meiyou jian-dao guo jiu le ziji i from that after Zhangsan again didn't see ASP save Perf self mingde nage ren. life De that-CL person 'Since then Zhangsan has never seen the person again who saved his life.'
Actually, I agree with the reviewer that my self-ascription account cannot cover sentences like (i). However, these sentences are dealt with by the condition on locality ziji, as specified in endnote 14, and so are sentences involved psych-verb object antecedents. 18 Example (42) is from Luther Liu (1999), i.e., his (50c) in Chapter 1 of his Ph.D. dissertation, University of California, Irvine. 19 An anonymous reviewer suggests that the sentence following (Tang, 1989) is another counterexample to my analysis, since there is no first/second-person pronoun in the believed proposition, the embedded clause, but long-distance binding of ziji is still not possible, contrary to the prediction made by my analysis. (i) Tameni renwei taj zai piping ziji *i/j they think he at criticize self 'They think that he is criticizing himself.' However, my informants take the i reading in (i) to be possible, especially when tamen 'they' is interpreted as a group. For instance, long-distance binding of ziji is perfectly possible in (ii) below. (ii)
Like
ban de xuesheng zhidao Laoshi zai zhanyang ziji,
Why the Blocking Effect?
313
science class DE student know Teacher at praise self, gandao feichang gaoxing. feel very happy 'The students from the science class(es) knew that Teacher is praising them, and felt very happy about it.' Hence, the asymmetry between first/second-person pronouns and third-person NPs still holds, even if the long-distance binding of ziji is not possible in (i). See also endnotes 16 and 20. 20 The fact that third-person NPs can sometimes block the long-distance binding of ziji by first/second-person pronouns, as pointed out by one of the reviewers, does not falsify my claim that there is an asymmetry between first/second-person pronouns and third-person NPs in the blocking effect. This is because I only claim that third-person NPs do not NECESSARILY block long-distance binding of ziji by first/second-person pronouns, which implies that third-person NPs can, though they do not always do so. Similarly, sentences like (43) are not counter examples to the preceding claim of mine either, as long-distance binding is possible in the relevant sentences. 21 Note that here I differ from what I said in Pan (1995,1997), namely, that the antecedents of long-distance bound ziji must be self-conscious. It seems that this self-consciousness constraint should be replaced by the highly reflective context constraint and only applies to antecedents with first/second-person features, i.e., it does not apply to third-person antecedents. 22 When I say that the viewer and the referent of the point of reference become one, I mean it metaphorically. It seems that it is more appropriate to say that the viewer stands at the same place as the referent of the point of reference, viewing the object in question, when we are talking about self-ascription in the sense that the viewer attributes a de se belief to the referent of the point of reference. In this situation, the referent in question is not required to view the object, and the viewing action can be carried out by the speaker only. This is why the referent need not be self-conscious, though s/he may be conscious, which is not relevant to the concept of self-ascription, as mentioned earlier. Note that we do not deny the existence of a different reading in which the point of reference is the viewer, and thus s/he self-ascribes the relevant belief. Unlike the situation mentioned previously, the referent of the point of reference, being a self-ascriber, in this reading is self-conscious. 23 The same reviewer claims that s/he could not see the differences between logophoricity and self-ascription and suggests that, although I have refuted the individual notions (SOURCE, SELF, and PIVOT) of logophoricity for long-distance bound ziji, there is still the general notion of logophoricity that is not refuted. I believe that the general notion must be based on the individual notions. If each and every one of them is refuted, I could not see how the general notion could stand. If the general notion is a garbage-can type, as suggested in the literature, then it is still too weak to account for the properties of long-distance bound ziji, as pointed out in endnote 8. Also note that the logophoricity notion given in Reinhart and Reuland (1991), as
314
Haihua Pan
pointed out in Section 4.3, is actually perspectivity (deictic center), which is refuted in Section 4.3. Hence, it is not appropriate to claim that the general notion of logophoricity can help to account for the long-distance binding of ziji.
REFERENCES Baker, C. L. (1995). Contrast, discourse prominence, and intensification, with special reference to locally free reflexives in British English. Language 71(1), 63-101. Battistella, E. (1989). Chinese reflexivization: A movement to INFL approach. Linguistics 27,987-1012. Chen, P. (1992). The reflexive ziji in Chinese: Functional vs. formalist approaches. In T. H.-T. Lee (ed.), Research on Chinese Linguistics in Hong Kong, pp. 1-36. Linguistics Society of Hong Kong. Chierchia, G. (1989). Anaphora and attitude de se. In R. Bartsch et al. (eds.) Semantics and Contextual Expression, pp. 1-31., Dordrecht, The Netherlands: Foris Publications Chomsky, N. (1981). Lectures on Government and Binding. Foris Publications, Dordrecht, The Netherlands: Foris Publications. Chou, X.-L. (1992). An alternative approach to Chinese reflexives. MA report, the University of Texas at Austin. Clements, G. N. (1975) The logophoric pronoun in Ewe: Its role in discourse. Journal of West African Languages 10,141-177. Cole, P. Hermon G., and Sung, L.-M. (1990). Principles and parameters of long distance reflexives. Linguistic Inquiry 21,1-22. Cole, P., and Sung, L.-M. (1993). Feature percolation and Mandarin reflexives. Journal of East Asian Linguistics 2,91-118. Cole, P., and Sung, L.-M. (1994). Head movement and long distance reflexives. Linguistic Inquiry 25,239-62 Cole, P., and Wang, C.-Z. (1996). Antecedents and blockers of long distance reflexives: The case of Chinese ziji. Linguistic Inquiry 27,357-390. Hagege, C. (1974). Les pronoms logophorique. BSLP 69,287-310. Huang, C.-C. J., and Tang, C.-C. J (1991). The local nature of long distance reflexive in Chinese. In J. Koster and E. Reuland (eds.) Long-distance Anaphora. Cambridge, Cambridge University Press. Huang, Y. (1994). The Syntax and Pragmatics of Anaphora: A Study with Special Reference to Chinese. Cambridge University Press, New York. Huang, Y.-H. (1984). Reflexives in Chinese. Studies in English Literature and Linguistics 10. lida, M. (1992) Context and Binding in Japanese. Ph.D. thesis, Stanford University, Stanford, CA. Kuno, S. (1987) Functional Syntax: Anaphora, discourse, and empathy. University of Chicago Press.
Why the Blocking Effect?
315
Lewis, D. (1979) Attitudes de dicto and de se. The Philosophical Review 88,29-80. Li, F.-X. (1990). The effect of person hierarchy on the blocking of the long distance binding of the Chinese pronoun ziji, In Proceedings of WECOL 19. Li, N.-C. (1991). Perspective-Taking in Mandarin Discourse. Unpublished Ph.D. dissertation, State University of New York at Buffalo. Li, Y. (1993). What makes long distance reflexives possible? Journal of East Asian Linguistics 2,132-166. Maling, J. (1984). Non-clause-bounded reflexives in modern Icelandic. Linguistics and Philosophy 7,211-241. Mitchell, J. E. (1986) The Formal Semantics of Point of View. Unpublished Ph.D. dissertation, University of Massachusetts, Amherst, MA. Pan, H. (1991). Pro and long distance reflexive ziji in Chinese. Paper presented at the 3rd North American Conference on Chinese Linguistics, Cornell University, Ithaca, NY. Pan, H. (1995). Locality, Self-Ascription, Discourse Prominence, and Mandarin Reflexives. Ph.D. thesis, Dept. of Linguistics, the University of Texas at Austin, Austin. Pan, H. (1997) Constraints on Reflexivization in Mandarin Chinese. New York: Garland Publishing, Inc. Pan, H. (1998). Closeness, prominence and binding theory. Natural Language & Linguistic Theory 16,771-815. Pollard, C, and Sag, LA. (1994). Head-Driven Phrase Structure Grammar. The University of Chicago Press and CLSI Publication. Pollard, C., and Xue, P. (1998). Chinese reflexive Ziji: Syntactic reflexives vs. nonsyntactic reflexives. Journal of East Asian Linguistics 7,287-318. Progovag, L. (1992). Relativized SUBJECT: Long distance reflexives without movement. Linguistic Inquiry 23,671-680. Progova?, L. (1993). Long distance reflexives: Movement-to-Infl vs. relativized SUBJECT. Linguistic Inquiry 24,755-772. Reinhart, T, and Reuland, E. (1991). Anaphors and logophors: An argument structure perspective. In J. Koster and E. Reuland (eds.), Long distance Anaphora. Cambridge: Cambridge University Press. Reinhart, T. and Reuland, E. (1993). Reflexivity. Linguistic Inquiry 24,657-720 Sells, P. (1987). Aspects of logophoricity. Linguistic Inquiry 18,445-479. Stirling, L. (1993). Switch Reference and Discourse Representation. Cambridge: Cambridge University Press. Tang, C.-C. J. (1985). A Study of Reflexives in Chinese. Unpublished M. A. thesis, National Taiwan Normal University. Tang, C.-C. J. (1989). Chinese reflexives. Natural Language and Linguistic Theory 7(4), 93-121. Xu, L. (1993). The long distance binding of ziji- Journal of Chinese Linguistics 21, 123-141. Xu, L. (1994). The antecedent of ziji. Journal of Chinese Linguistics 22,115-137. Xue, P., Pollard C., and Sag, I. (1994). A new perspective on Chinese ziji. In Proceedings of the 13th West Coast Conference on Formal Linguistics, pp. 432—447. The University of Chicago Press and CSLI Publication.
316
Haihua Pan
Yu, W. (1991). Logophoricity in Chinese. Paper presented at the 3rd North-American Conference on Chinese Linguistics, Cornell University, Ithaca, NY. Zribi-Hertz, A. (1989). Anaphor binding and narrative point of view: English reflexive pronouns in sentences and discourse. Language 65(4), 695-727. Zubin, D. A., Chun, S., and Li, N. (1980). Misbehaving reflexives in Korean and Mandarin. In K. Hall and J.-P. Koenig (eds.), Proceedings of the 16th Annual Meeting of the Berkeley Linguistics Society, Vol. 16: General Session and Parasession on the Legacy of Grice. Berkeley: Berkeley Linguistics Society.
SYNTACTIC AND NONSYNTACTIC CONSTRAINTS ON LONG-DISTANCE REFLEXIVES CARL POLLARD* PINGXUEt ^Department of Linguistics, College of Humanities The Ohio State University Columbus, Ohio tApplied Research & Technology The Boeing Company Seattle, Washington
1. INTRODUCTION This chapter is part of a long-term project whose goal is to untangle the various factors involved in the interpretation of reflexive pronouns, especially in Chinese and English. Building on proposals developed in Pollard and Xue (1998), here we discuss not only syntactic conditions but also nonsyntactic conditions that determine the interpretation of the Chinese reflexive ziji and (to a lesser extent) reflexives in English (both standard American English and British and literary (Brit./Lit.) English varieties discussed in, for example, Baker, 1995 and Zribi-Hertz, 1989). The nonsyntactic conditions will involve such notions as logophoricity, contrastiveness, and discourse prominence, all of which we will attempt to clarify. It has long been observed that the Chinese ziji exhibits several interesting properties. First of all, ziji is a long-distance (henceforth, LD) reflexive in the Syntax and Semantics, Volume 33 Long-Distance Reflexives
317
Copyright © 2001 by Academic Press All rights of reproduction in any form reserved. 0092-4563/00 $35.00
318
Carl Pollard and Ping Xue
sense that it may take an antecedent across clause boundaries, as shown by the following example, where ziji may be interpreted as co-referring with any of the subjects, namely, Wangwu, List, or Zhangsan (see, e.g., Battistella, 1989; Cole, Hermon and Sung, 1990; Huang and Tang, 1991). (1) Zhangsani zhidao [Lisij renwei [Wangwuk zui xihuan zijii/j/k]]Zhangsan know Lisi think Wangwu most like self 'Zhangsanj knows that Lisij thinks that Wangwu likes himself/himi/j most.' A closely related property is that a higher subject usually can antecede ziji only if, roughly speaking, no lower subject which c-commands ziji disagrees with it (see, e.g., Tang, 1989).1 This effect, which we call nonagreement blocking, is illustrated in (2).2 (2) Zhangsani zhidao [nij renwei [Wangwuk zui xihuan ziji* ji/*j/k] ]Zhangsan know you think Wangwu most like self 'Zhangsan; knows that yoUj think that Wangwu likes himself/you*j/him*i most.' Further, ziji, in general, takes a subject, not an object, as its antecedent. This property is often called subject orientation. (3) Zhangsani songgei Lislj yizhang ziji/j de xiangpian. Zhangsa give Lisi one-CL self DE picture 'Zhangsanj gave Lisij a picture of himselfi/*j.' While these have been widely taken as the central characteristics of ziji, the relevant properties are more complex than they have appeared. In addition, as has been widely discussed in the literature, ziji may co-refer with an NP in the same sentence that does not structurally "command" it, or even with an NP in the prior discourse (see, e.g., Chen, 1992; Yu, 1992). In (5) following, for instance, ziji can refer either to the local subject na xuesheng 'that student' or to the external speaker. Examples like (4), (5), and (6) are equally important in characterizing the reflexive ziji and thus pose a serious challenge for any purely syntactic account. In fact, these examples have been cited as arguments for a purely pragmatic or discourse-based account (see, e.g., Chen, 1992; Y. Huang, 1994). (4) Zhangsan^ xiangxin [[ziji i/j de xiaohai mei de jiang de xiaoxi] Zhangsan believe self DE child not get prize DE news shi Lisij hen nanguo.] make Lisi very sad 'Zhangsan, believed that the news of hiSj/j child not getting a prize made Lisij sad.'
Syntactic and Nonsyntactic Constraints on Long-distance Reflexives
319
(5) Zheyang yilai na xueshengi jiu bu shuohua-le, yanjing cong zijii/j thus then that student then not speak-ASP, eyes from self de shen-shang yikai-le DE body-on move-away-ASP 'Then, that studentj stopped talking, and his eyes moved away from hin i/j
(6) Zhangsarii zhidao neijian shi yihou hen qifen, Lisij shuo neixie Zhangsan knows that-CL thing after very angry Lisi say those hua mingming shi zai he ziji i/j zuodui. words obviously is being with self against 'Zhangsan; was very angry after he learned that. By saying those words, Lisij was obviously acting against himselfj/himj.' In Pollard and Xue (1998), we propose that a distinction must be drawn between syntactic and nonsyntactic uses of reflexives. More importantly, we point out that this distinction should not be treated as one of lexical ambiguity between an anaphor and a logophor (or discourse pronoun, or intensive pronoun), contrary to what has usually been proposed.3 Instead, we posit just one type of referentially dependent element—reflexives—which avail themselves of two (not necessarily mutually exclusive) options for being related to their antecedents, namely, (syntactic) binding and discourse coreference; here we reserve the term "binding" for a coindexing relation between the reflexive and an NP that commands it in a certain structurally characterized way (which we will make precise in the following). Informally speaking, an NP can serve as the antecedent of ziji when it is either syntactically prominent (namely, a subject which structurally commands the reflexive) in the sentence or else pragmatically prominent in the discourse. Our account entails that, contrary to what has been commonly assumed, there is no notion of anaphor (viz., a referentially dependent element subject to an obligatory local binding condition such as Principle A) that is relevant in the grammar of Chinese (or of Brit./Lit. varieties of English). This rejection of obligatory binding is motivated by numerous examples from Chinese such as (4), (5), and (6) preceding, and by certain examples from English discussed in Zribi-Hertz (1989) and Baker (1995). On the other hand, we reject a purely pragmatic account; we believe that a syntactic notion of binding cannot be entirely dispensed with. This chapter is organized as follows. In Section 2.1, we review current analyses, demonstrating that any theory positing obligatory binding is empirically inadequate. In Section 2.2, we revise the binding theory for English developed by Pollard and Sag (1992,1994), focusing on how American and British English reflexive usages differ. In Section 2.3, we discuss similarities and differences between English and Chinese reflexives, showing that the
320
Carl Pollard and Ping Xue
notion of obligatory binding is not motivated; at the same time, we will argue that a purely pragmatic or discourse-based account cannot be maintained. In Section 3, we focus on nonsyntactic conditions relevant to Chinese ziji. In Section 3.1, we consider how logophoricity bears upon ziji, concluding that, of the three logophoric notions SELF, SOURCE, and PIVOT suggested by Sells (1987), only the first is relevant to the interpretation of ziji. Moreover, we argue that, while logophoricity is often a factor in nonsyntactic use of reflexives, it is not a necessary condition, a point also made in Baker (1995). In Section 3.2, we will identify notions of discourse prominence relevant to ziji, together with a number of related issues. And in Section 4, we summarize our findings and conclusions.
2. SYNTACTIC AND NONSYNTACTIC REFLEXIVES While differing in their views regarding the nature of reflexivization, almost all researchers seem to agree that a reflexive must find some prominent NP as its antecedent. Over the past 20 years, there have been many attempts to characterize the relevant notion of prominence. Chomsky (1981,1986) suggests that the relevant generalizations can be formulated as a single structural condition (viz., Principle A) that requires a reflexive to be bound by a syntactically prominent (i.e., c-commanding) NP in an appropriately defined local syntactic domain. That is, a reflexive (or anaphor) must be bound in the minimal phrase containing the reflexive, a subject, and an NP that c-commands the reflexive. Clearly, this theory, as it stands, is problematic with respect to the facts of ziji as illustrated previously, since it is inconsistent at least with LD antecedents, on the one hand, and with subject orientation on the other.
2.1. Limitations of Binding Principle A In the last decade or so, most analyses of ziji within generative grammar have focused on incorporating ziji into a suitable extension of Chomsky's Principle A. More precisely, the properties of ziji, particularly the possibility of LD antecedents and nonagreement blocking, have been widely considered evidence for a successive-cyclic LF association of ziji with each of the c-commanding subjects (or the highest subcommanding subjects, which will be discussed shortly), either via head movement (see, e.g., Cole and Sung, 1994), or via IP-adjunction (see, e.g., Huang and Tang, 1991), thus predicting, among other things: (i) ziji allows LD antecedents; (ii) only c-commanding
Syntactic and Nonsyntactic Constraints on Long-distance Reflexives
321
(or highest subcommanding) subjects can be the antecedents; (iii) only potential binders participate in nonagreement blocking. While having different theoretical consequences, analyses along these lines are all problematic with respect to the following facts.4 First, the antecedent of ziji does not always have to be a c-commanding subject. One class of counterexamples are sentences with so-called psychological predicates as in (4) preceding, where ziji can refer to the experiencer object List. Thus, subject orientation cannot be taken as an absolute condition for ziji in general. Another class of problematic examples are those instances of ziji which allow so-called subcommanding antecedents, that is, animate subjects which are not c-commanders themselves, but which are contained in an inanimate c-commanding subject (see, e.g., Huang and Tang, 1991). As Xue et al. (1994) and Pollard and Xue (1998) point out, the relation between a reflexive and its subcommanding antecedent is fundamentally pragmatic, not syntactic in nature. The second problem for these LF movement analyses is that nonagreement blockers need not be potential binders. For example, direct and oblique objects, in general, cannot be antecedents without bearing special discourse or pragmatic properties, such as being the individual whose point of view or state of consciousness is being reported, as in (4), but objects can trigger nonagreement blocking just as subjects can, as discussed in the literature (see, e.g,, Pollard and Xue, 1998; Pan, 1995; Xue and Sag, 1994). In fact, there is ample evidence that nonagreement blocking is actually pragmatic, not syntactic in nature. For instance, the blocking effect can even be overriden if obeying it leads to a pragmatically bizarre result. This is illustrated in (7) and (8):5 (7) Zongtongi qing wo^ zuo zai zijiy*j de shenbian. president ask I sit at self DE side 'The president^ asked mCj to sit beside him/himself*j.' (8) Lisii shengpa woj chaoguo ziji i/*j Lisi worry I surpass self 'Lisij was afraid that Ij would surpass himi/myself*j. Thus, nonagreement blocking is not an absolute grammatical constraint, and therefore cannot be accounted for in terms of such purely syntactic notions as LF movement. The third problem is that, crucially, no analysis that assumes a condition of obligatory binding, such as Principle A, however modified, can adequately explain why an instance of ziji may have a noncommanding NP as its antecendent even when a commanding subject NP is available, as in (4), (5), and (6) (which were either ignored or considered peripheral cases).
322
Carl Pollard and Ping Xue
Note that a similar phenomenon is also observed with reflexives found in BritVLit. English (after Zribi-Hertz, 1989; also see Baker, 1995, for similar examples): (9) a. Philipl was supposed to be fooling ( . . . ) , because Desiree ( . . . ) had undoubtedly explained to them the precise nature of her relationship with himself^ (Zribi-Hertz, 1989: (43b)) b. ( . . . ) his{ wife was equally incredulous of her innocence and suspected himself i , the pastor, to be the cause of her distress, ( . . . ) (Zribi-Hertz, 1989: (37)) c. But Ruperti was not unduly worried about Peter's opinion of himself. (Zribi-Hertz, 1989:(46b)) These facts show that under appropriate pragmatic or discoursal conditions—whose precise characterization remains a matter of widespread debate—in these varieties of English even a locally c-commanded reflexive pronoun does not always have to be locally bound, Zribi-Hertz assumed that logophoricity was the operative factor. Baker (1995) argued for a different notion of contrastive intensification involving contrastive reference to discourse-prominent entities. But, in any case, facts like these are inconsistent with any theory that requires reflexives to be (obligatorily) bound in accordance with Chomsky's binding Principle A. 2.2. A Revision of Binding Principle A Pollard and Sag (1992,1994) suggest that the binding theory should be formulated not in terms of c-command, but rather in terms of the relative obliqueness ordering of grammatical relations, shown in (10). (10) The Obliqueness Hierarchy: SUBJECT > PRIMARY OBJECT > SECONDARY OBJECT > OTHER COMPLEMENTS In place of Chomsky's c-command, Pollard and Sag define the relation of o-command and o-binding as in (11) and (12), respectively: (11)
O-Command: X o-commands Y just in case X is a less oblique coargument of some Z that dominates Y. In case Z = Y, X is said to locally o-command Y.
(12) O-Binding: X (locally) o-binds Y iff X and Y are coindexed and X (locally) o-commands Y; Y is (locally) o-free if it is not (locally) o-bound.
Syntactic and Nonsyntactic Constraints on Long-distance Reflexives
323
This theory relates binding to argument structure, converging with the conclusions of a number of recent studies (e.g., Everaert, 1991; Hellan, 1991; Reinhart and Reuland, 1991,1993). It explains not only the basic facts about American English reflexive (and reciprocal) binding, but also many other facts that are unexplained by Chomsky's binding theory. Unfortunately, the formulation of Principle A in Pollard and Sag (1992,1994) is also inconsistent with reflexives in Brit/Lit, varieties as shown in (9) preceding, since it assumes obligatory binding, as other analyses do. Pollard and Xue (1998) propose that referentially dependent English NPs are either R-PRONOUNS (reflexives or reciprocals) or P-PRONOUNS (pronouns other than r-pronouns), completely eliminating the notion of anaphor from the classification of NPs. Note that the term ANAPHOR has come to be widely associated with the property of being constrained by Principle A, but the term R-PRONOUN here carries no such association. We carry over from Pollard and Sag the notions of obliqueness, (local) o-command, and (local) o-binding, as stated in (10), (11), and (12) preceding. We formulate the principle for r-pronouns as follows: (13) Principle R (for English): An r-pronoun (e.g., himself) must either be locally o-bound or interpreted in accordance with certain (English-specific) pragmatic/ discourse constrains (which a complete theory must make explicit). The pragmatic/discourse contraints referenced in Principle R, involving notions such as logphoricity, contrastiveness, intensification, and discourse prominence, will be discussed in detail in Section 3. We believe that the main difference between American English and Brit./Lit. English with respect to reflexive interpretation is that American English is subject to an additional syntactic condition—a version of Principle A—which we call Principle A'. (14) Principle A' (Standard American English): A locally o-commanded r-pronoun must be locally o-bound. Because of (14), examples like those in (9) are ruled out in American English. But except for that difference, British and American English work essentially the same way as far as reflexives are concerned: they can be either locally o-bound or pragmatically interpreted (or both at the same time). While also recognizing the distinction between syntactic reflexives and nonsyntactic reflexives, Baker (1995) proposes that reflexives in Brit./Lit. English are ambiguous between anaphors and something else that he called (nonanaphoric) intensive pronouns. Baker argues that logophoricity is not a necessary condition for nonsyntactic reflexives, and that nonsyntactic reflexives can be viewed as (contrastive) intensive versions of ordinary pro-
324
Carl Pollard and Ping Xue
nouns; Baker's intensive pronouns overlap in part with what other people have sometimes called logophors or discourse reflexives. Baker's account is summarized in (15): (15) Baker's Theory of English Reflexive Pronouns a. Reflexives are lexically ambiguous between anaphors and intensive pronouns. b. Anaphors in both American and British English are subject to an obligatory local binding condition (e.g., Principle A). c. British English and American English differ in the form of intensive pronouns; in nonnominative positions, reflexives are used as intensive pronouns in British English, but not in American English, where ordinary pronoun forms are used. d. Pragmatically, intensive pronouns are used to set up a contrast between the antecedent, which must refer to a discoursally prominent entity, and some other entity. There are some problems with Baker's theory. First, it ignores facts of American English such as the following (also see, e.g., Pollardf and Sag, 1992, 1994, and references cited there for similar examples), all of which contain nonnominative reflexive pronouns that are not locally bound. (16) a. Jessiej knew full well that the local people would all feel that people like himselfj were not to be trusted, let alone hired. b. Each student; was confident that the teacher would criticize everyone but himselfj. c. The picture of himself; on page one of The Times upset Johni;. d. John's; campaign requires that pictures of himselfi be placed all over town. e. Johni; asked Maryj to send reminders to everyone except themselvesk(k=i+j). In such examples (among which some are contrastive and some not), the antecedents of the reflexives seem to be subject to discourse conditions of the same general character as those that apply to Baker's intensive pronouns in British English. A related problem is concerned with the conclusion about the difference between American and British English. Baker suggests that American English and British English differ with respect to nonsyntactic uses of reflexives in the sense that nonnominative reflexives can be used as intensive pronouns only in British English, but not in American English. While it is true that ordinary pronouns with a certain contrastive pitch accent are often used as intensive forms, there is no question that reflexives are also used to signal intensification in American English, as the following examples show:
Syntactic and Nonsyntactic Constraints on Long-distance Reflexives
325
(17) a. It suddenly dawned on Johni that it was a picture of HIMSELFi/HIMj (not of Kenneth Starr) that Mary was painting, b. John; sent flyers to everyone on the distribution list except HIMSELFi/??HIMi. In fact, in a sentence like (17b), it seems much more natural to use the reflexive form himself (in the environment introduced by the element except). Baker's theory fails to capture this similarity between American and British English, and therefore makes them appear to be more different from each other than they really are. The third problem arises from Baker's lexical ambiguity hypothesis for reflexives. The problem can be made clear by considering a very simple example like (18a): (18) a. Johni; admires himselfi;, b. John admires Mary. According to Baker's account, (18a) must be ambiguous between a reading where himself is an anaphor and a reading where himself is an intensive nonanaphoric pronoun. But there does not seem to be any evidence that the sentence is ambiguous in this way. It is true that himself can be uttered with or without contrastive pitch accent. But so can Mary in (18b), and this does not stand as evidence that Mary is lexically ambiguous! One apparent way to eliminate this spurious ambiguity would be to add to Baker's theory a further constraint that intensive nonanaphoric pronouns must be locally free, thereby eliminating the intensive reading of the reflexive in (18a). That is, intensive pronouns are pronominals (in Chomsky's terminology). This is essentially the solution adopted by Reinhart and Reuland (1993). The problem with this approach (beyond the fact that it is a further complication) is that it blurs the fact that being an intensive pronoun is a pragmatic notion entirely independent of considerations about whether the antecedent is local or not. If a reflexive is being used to signal a contrast involving the antecedent, then it seems to us that deserves to be classified as an intensive, no matter where the antecedent is. In summary, while both Baker's theory and ours recognize the distinction between syntactic and nonsyntactic reflexives, out theory differs from Baker's theory in some fundamental aspects. Our theory does not posit two different syntactic types of reflexives. In our theory, there is no notion of obligatory binding for reflexives; rather reflexives are subject to the general principle—Principle R—that there are two options for interpreting r-pronouns. In fact, these two options are nonexclusive: there is no reason why an English reflexive couldn't be intensive (or logophoric) and locally o-bound at the same time, for example, as in (18a), uttered with contrastive pitch ac-
326
Carl Pollard and Ping Xue
cent on himself. Furthermore, Baker suggests that American English and British English differ as to whether nonnominative reflexives can be used as intensifiers, while in our account, they differ primarily in the respect that Principle A' applies only to Standard American English. 2.3. Binding for Chinese and English Reflexives. Returning to Chinese, we now characterize exactly what we refer to as syntactic and nonsyntactic instances of ziji, respectively. In passing, we will also discuss the similarities and differences between Chinese and English. In Chinese, the cases in which ziji is coindexed with a (local or superordinate) animate "commanding" subject, as in (1), repeated here, have been widely considered core cases of ziji binding. (1)
Zhangsan-t zhidao [Lisij renwei [Wangwuk zui xihuan ziji i / j / k ]]. Zhangsan know Lisi think Wangwu most like self 'Zhangsanj knows that Lisij thinks that Wangwu likes himself/him^ most.'
Speaking informally, the key fact about such cases is that any commanding subject qualifies as the antecedent of ziji on the strength of its syntactic prominence alone.6 Thus, we can define an instance of ziji to be syntactic ziji or nonsyntactic ziji according to whether it is or is not coindexed with a (local or superordinate) commanding subject. Consider a sentence like (8), repeated here, where either the matrix subject or the local subject in principle can be the antecedent. Note that usually the reflexive ziji will not be interpreted as referring to the local subject wo, not because the relevant structure is ungrammatical but only because such an interpretation of the sentence is pragmatically bizarre in a normal context. (8) Lisli shengpa woj chaoguo ziji ji/*j Lisi worry I surpass self 'Lisij was afraid that Ij would surpass hinii/myselfj.' Pollard and Xue (1998) argue that argument structure is relevant to ziji binding. Thus, while the tree-configurational notion c-command is close to being descriptively correct, it is the notion of o-command defined in (11) preceding that appears to provide exactly what is required for an empirically adequate account of syntactic ziji-7 It has sometimes been suggested (e.g., by Chen, 1992) that all instances of reflexives, including those with commanding subject antecedents, can be subsumed under some unified discourse or pragmatic notion, and thus the notion of binding can be eliminated entirely. However, as has been discussed, an embedded ziji, as in (1), can take any of the superordinate sub-
Syntactic and Nonsyntactic Constraints on Long-distance Reflexives
327
jects as its antecedent, even in the absence of such conditions as logophoricity, contrast, or discourse prominence. And as far as local subjects are concerned, it seems even clearer that no logophoricity or special discourse prominence is required in order to qualify as being the antecedent of a reflexive, no matter how deeply the clause containing the subject and the reflexive may be embedded, what types of predicates may be involved, and what the sentence intonation is like (with or without pitch accent), as shown by many examples in the literature. One might argue that being a subject grants the element an automatic prominence in the discourse. Our answer is that this is not necessarily true. Consider the following example: (19) Zhangsatii zhidao Lisij bu hui rang Wangwuk [PROk zai ziji i/j/k Zhangsan know Lisi not will allow Wangwu at self de jia-li ye shou dao yingxiang.] DE home also receive ASP affect 'Zhangsanj knows that Lisij will not allow Wangwuk to be affected even at (hisi/j/k) home.' In (19), the fact that the embedded object Wangwu (or the PRO) can be the antecedent is particularly telling. It need be associated neither with any logophoric role (e.g., point of view, source of reported speech, or deictic perspective) nor with any discoursally prominent property (e.g., topicality, high external rank, thematic agency, givenness, contrastiveness) in order to be interpreted as the antecedent. Rather, we believe that it is simply its syntactic prominence (namely, being an o-commanding subject) that makes it possible for it to serve as the antecedent.8 On the basis of these considerations, we conclude that while it is necessary to recognize the status of nonsyntactic uses of reflexives, whose antecedents are determined by nonsyntactic conditions, a purely discourse or pragmatic theory cannot adequately account for the facts discussed previously, and thus a syntactic notion of binding cannot be entirely dispensed with. Given the preceding discussion, we summarize the similarities and differences between Chinese and Brit./Lit. English as follows: (20) Chinese ziji and Brit./Lit. Reflexives: Some Similarities a. They are not anaphors, in the sense that they are not subject to any obligatory binding condition. b. In spite of (a), in each language, there is a language-specific notion of (syntactic) binding that is relevant for reflexives. c. In each language, reflexives are either bound (in the sense relevant for the language in question) or interpreted according to Ian-
328
Carl Pollard and Ping Xue
guage-specific discourse/pragmatic constraints (e.g., logophoricity, contrast, intensification, and discourse prominence). This is a nonexclusive disjunction. d. In both languages, unbound reflexives include (among others) ones anteceded by subcommanders, psych-verb experiencers, NPs in preceding sentences or in the discourse (see (16) for English examples). (21) Chinese and English Have Different Notions of Binding: a. English: local o-binding b. Chinese: o-binding by a subject. Statement (21) captures the standard wisdom that Chinese reflexive binding differs from English in being subject-oriented and potentially long-distance. Of course, it is interesting to speculate but premature to conclude that (21a) and (21b) are, in some sense, settings of a parameter of cross-linguistic variation within a universal theory of binding. Now we state our binding theory for Chinese in (22): (22) Principle Z (for Chinese): ziji must either be o-bound by a subject or interpreted in accordance with certain (Chinese specific) pragmatic/discourse constraints (which a complete theory must make precise). Statement (22) is analogous to the theory for English in (13), except that (i) the English Principle R with local o-binding has been replaced with the Chinese Principle Z in (22) with o-binding by a subject, and (ii) we allow for the (strong) possibility that pragmatic/discourse factors to which antecedents of unbound reflexives are subject might differ in subtle ways between the two languages. Thus, neither Chinese nor (British) English has obligatory reflexive binding. From this point of view, the obligatory binding (as traditionally stated in Principle A) of anaphors in American English belong to a special case and should not be taken as a cross-linguistic model for the study of reflexives.
3. NONSYNTACTIC CONDITIONS ON ZIJP In the preceding sections, we have shown that there is a wide class of instances of reflexives which are not o-bound by a subject but rather are coindexed with an NP in the discourse, as exemplified by (4), (5), and (6), repeated following:
Syntactic and Nonsyntactic Constraints on Long-distance Reflexives
329
(4) Zhangsani xiangxin ziji i/j de xiaohai mei de jiang de xiaoxi Zhangsan believe self DE child not get prize DE news shi Lisij hen nanguo. make Lisi very sad 'Zhangsanj believed that the news of hiSi/j. child not getting a prize made Lisij sad.' (5)
Zheyang yilai na xueshengi jiu bu shuohua-le, yanjing cong zijiyj thus then that student then not speak-ASP, eyes from self de shen-shang yikai-le DE body-on move-away-ASP 'Then, that studentj stopped talking, and his eyes moved away from hinij/j.'
(6)
Zhangsan{ zhidao neijian shi yihou hen qifen, Lisij shuo Zhangsan knows that-CL thing after very angry Lisi say neixie hua mingming shi zai he ziji i/j zuodui. words those obviously is being with self against 'Zhangsanj was very angry after he learned that. By saying those words, Lisij was obviously acting against himselfj/himj.'
Rather than being completely free, such instances of reflexives must be interpreted subject to discourse and pragmatic conditions, as stated in Principle Z in (22). Compared with structural aspects of ziji, the relevant discourse and pragmatic aspects have been studied much less. However, there seems to be a general consensus that logophoricity, contrastiveness, intensification, and discourse prominence, in general, are among the factors involved in the distribution of nonsyntactic antecedents of reflexives. In the following sections, we focus on these aspects, examining and characterizing how discourse and pragmatic factors are relevant to nonsyntactic uses of ziji, and thus providing further justification that instances of ziji not o-bound by a subject are indeed discoursal and pragmatic in nature. We also hope to clarify a number of related issues. 3.1. Logophoricity It has been extensively noted in the literature (e.g., Sells 1987, Reinhart and Reuland, 1991) that logophoricity plays a central role in the interpretation of nonsyntactic reflexives. Roughly speaking, logophoricity involves reference to an individual whose point of view or general state of consciousness is expressed in the discourse (Clements, 1975). Sells (1987) uses the term "logophoric" in an extended sense which subsumes not only point of view and subject of consciousness (his SELF) but also source of reported
330
Carl Pollard and Ping Xue
speech (SOURCE) and deictic perspective (PIVOT). Reinhart and Reuland (1993) employ an even broader (and vaguer) usage which is, as they point out, quite widespread in the syntactic literature. Logophoricity typically arises in the presence of certain predicates, such as verbs of communication and predicates of mental experience, or even with some adverbial conjunctions, as will be discussed (see Sells, 1987). A typical and much-cited example is the sentence in (23), where the antecedent of ziji is a psych-predicate experiencer. Note that an o-commanding object normally cannot be the antecedent, as shown in (3) preceding. Moreover, an object cannot usually be the antecedent when ziji occurs within a superordinate position, as shown in (24). (23) Ziji de xiaohai mei de Jiang de xiaoxi shi Lisii hen nanguo. self DE child not get prize DE news make Lisi very sad 'The news of hiSj child not getting a prize made Lisii sad.' (24) *Ziji- de didi da le Lisii self DE younger-brother hit ASP Lisi 'His, own younger brother hit Lisij.' A psych-predicate experiencer object is logophoric in the strict sense that it reports the general state of the consciousness of the experiencer. Examples like (23) suggest that an NP can antecede ziji, even if it is not an o-commanding subject, if it refers to the discourse SELF. Certain adverbial conjunctions also seem to be logophoric. For example, when an adverbial clause is introduced by a causal conjunction like yinwei 'because,' the subject of the matrix clause can antecede ziji contained in an adverbial clause for many speakers.10 (25) Wangwui bu hui qu, [yinwei Lisij mei-you yaoqing ziji i/j ]. Wangwu not will go because Lisi have-not invite self 'WangwUj won't go because Lisij didn't invite him/himselfj.' As Sells (1987) points out, such adverbial conjunctions allow the speaker to take the point of view of the subject of the matrix clause. To put it another way, the reason expressed by the subordinate clause can be the matrix subject's reason, not necessarily the speaker's reason. By contrast with SELF, being the logophoric SOURCE does not seem to be a sufficient condition to license an antecedent which is not a subject o-commanding ziji, as shown by examples like the following: (26) Zhangsan^ cong Lisij chu tingshuo Wangwuk bu xihuan ziji i/*j/k Zhangsan from Lisi place hear Wangwu not like self 'Zhangsanj heard from Lisij that Wangwuk does not like himi/*j/himselfk.'
Syntactic and Nonsyntactic Constraints on Long-distance Reflexives
331
Moreover (and pace Chen, 1992), being a logophoric PIVOT is not a necessary condition for an NP to antecede ziji, as pointed out by Xue and Hukari (1993). In Sells, the notion of PIVOT is defined in a very physical sense, referring to the one with respect to whose (space-time) location the content of the proposition is evaluated. "If someone makes a report with Mary as the PIVOT, that person is understood as (literally) standing in Mary's shoes." (Sells, 1987:456). For example, since PIVOT is assumed as the locus to which deictic elements like 'come' and 'go' refer, the contrast between the Japanese examples in (27a) and (27b) is naturally explained if the PIVOT is the logophoric role that mediates coreference of unbound instances of zibun with their antecedents. (27) a. Takasii wa Yosiko ga zibuni o tazunete-kita node uresigatta. Takasi Top Yosiko Sub self obj visit-came because happy 'Takasij was happy because Yosiko came to visit sehy b. *Takasi{wa Yosiko ga zibun i o tazunete-itta node uresigatta. Takasi Top Yosiko Sub self obj visit-went because happy 'Takasii was happy because Yosiko went to visit selfi.' However, the corresponding sentences do not pattern analogously in Chinese, in which the following two sentences are equally possible. (28) a. Zhangsani hen gaoxing yinwei List yao lai kan ziji i. Zhangsan very happy because Lisi shall come see self 'Zhangsanj is very happy because Lisi shall come to see himi.' b. Zhangsani hen gaoxing yinwei Lisi yao qu kan ziji i Zhangsan very happy because Lisi shall go see self 'Zhangsanj is very happy because Lisi shall go to see hinij.' Although the deixis is the same as in the corresponding Japanese examples, there is no requirement that the antecedent of nonbound ziji be the PIVOT. Instead, the fact that, in both cases, Zhangsan is the SELF suffices to license the antecedency. In terms of deictic effects, the difference between Japanese and Chinese can be further shown by the following contrast (the Japanese examples are provided by Tsuneko Nakazawa (personal communication)): (29) a. Yamada-san-ga anata-ni kuruyou-ni it-te-imasuyo. Yamada-Mr-Nom you-Dat come-to say-ing 'Mr. Yamadaj is asking you to come to him}.' b. * Yamada-san-ga anata-ni ikuyou-ni it-te-imasuyo. Yamada-Mr.-Nom you-Dat go-to say-ing 'Mr. Yamadai is asking you to go to himi.'
332
Carl Pollard and Ping Xue
(30) a. Zhangsan qingni qu. Zhangsan ask you go 'Zhangsarij asked you to go to himi.' b. Zhangsan qingni lai. Zhangsan ask you come 'Zhangsani asked you to come to himi;.' It seems that in sentences like the above involving deictic elements, the external speaker will obligatorily take the viewpoint of the sentence internal referent (i.e., the subject) in (29) in Japanese, in order to get the interpretation indicated by the English translation. But in Chinese, the external speaker may or may not take the viewpoint of the sentence internal referent. This, together with facts discussed previously, indicates that while Japanese deictic effects may be closely associated with the notion of logophoricity, their Chinese counterparts are not. The crucial fact is that despite the difference in terms of deictic effects between Japanese and Chinese, both languages allow LD reflexive coreference into adverbial clauses. This suggests that the deictic effects cannot be the essential content of the logophoricity involved in adverbial cases crosslinguistically.11 In the examples discussed previously, the SELF is, in Sells' terminology, internal. However, the SELF can also be external, i.e., the point of view can be that of the utterance speaker. In this connection, it is interesting to note that there is a class of instances of reflexives which are interpreted as referring to the external speaker of the sentence. Particularly, there are certain uses of ziji which have sometimes been characterized as nonreflexives in the literature, such as ziji used as a "first-person pronoun,'" as discussed in Pan, and Gu Liu (1983), For example. (31) Lingdaode biaoyang dui ziji shi yi-ge bianche. leader DE compliment to me be one-CL impetus 'The leader's compliment was an encouragement to me.' (32) Zhe-ci de xianjin jingyan jiaoliu dahui dui dajia This-CL DE advanced experience exchange meeting to everyone dui ziji dou you shenke de jiaoyu yiyi. to me both have profound DE instructive meaning 'This advanced-experience exchange meeting has profound educational implications for everyone as well as for me.' In (31) and (32), the most natural interpretation is that ziji refers to the speaker, as indicated by the English translation. In other words, such a reflexive refers to the "I" or the narrator of the discourse, as discussed in Kuroda (1973), among many other researchers. Such cases have sometimes been dismissed (e.g., by Liu et al, 1983) as nonreflexive uses. However, even
Syntactic and Nonsyntactic Constraints on Long-distance Reflexives
333
though it is true that such cases lack linguistic antecedents, it is worth considering whether they should be assimilated to the logophoric case, inasmuch as "first-person" ziji does indeed refer to the individual whose point of view is being reflected.12 If this is correct, then it is altogether natural to expect that reflexive ziji should be able to function as a first-person pronoun. 3.2. Contrast, Intensification, and Discourse Prominence While logophoricity does play an important role in the interpretation of reflexives in Chinese, it seems clear, however, that logophoricity is not a necessary condition for nonsyntactic use of reflexives in this language.13 Consider the following example: (33) [Zhangsani neiyang zuo] zhihui did ziji i buli. Zhangsan that-waydo only-will toward self not-beneficial 'Zhangsanj acting that way won't do himi any good.' Here, the SELF, from whose perspective the subjective evaluation is presented, is not the referent of ziji but rather the external speaker. Thus, this use of ziji, while nonsyntactic, cannot be considered logophoric. On the other hand, it is clear that the discourse to which (33) belongs is one in which Zhangsan is the topic, in the sense that the discourse is principally concerned with Zhangsan. This notion of topicality is closely related to such notions as givenness and information saliency discussed in functional literature (e.g., Bates and MacWhinney, 1982; Givon, 1983; Chen, 1992).14 Topicality in this sense (what Baker calls "primary topic of concern") is just one factor that contributes to what has been widely referred to as discourse prominence. The precise characterization of this notion is a matter of widespread and ongoing debate, which we will not attempt to settle here.15 Thus, not all discourse-prominent entities are strictly topics, but it does seem to be true that in order to be sufficiently discourse-prominent to serve as referent of ziji, an entity must not only be present in the immediate discourse, but also should have been referred to before, since Chinese ziji, unlike ordinary stressed pronouns, cannot be used deictically: (34) a. Ni zhidao na ben shu hai le TAV na ge wuli xi you know that CL book hurt ASP he that CL physics dept. de xueshengi. DE student 'You know that book harmed HIMi, that physics studenti.' b. *Ni zhidao na ben shu hai le ZIJI i na ge wuli you know that CL book hurt ASP self that CL physics
334
Carl Pollard and Ping Xue
xi de xueshengi. dept. DE student 'You know that book harmed HIMSELF;, that physics student i ' We follow Baker in distinguishing contrastiveness from discourse prominence. What Baker refers to as intensive uses of reflexives in British English are those cases (e.g., (9)) where both factors are present; more precisely, the antecedent of the reflexive is a discourse-prominent entity which is involved in a comparison with some other entity. An example of this kind involving ziji is the following, due to Pan (1995): (35) [Jiaoshi bajiao], [shouhai de shouxian shi xuesheng]. teacher strike get-hurt DE first be student [Xueshengi bake], [daomei de shi (tamen) Zijii.] student strike bad-luck DE be them self 'If teachers are on strike, then the ones who get hurt are students. If students are on strike, then the ones who have bad luck are themselves.' In this example, xuesheng 'students' is a discourse-prominent entity. Note that this example reflects the point of view of the speaker, not the students, and therefore cannot be considered a logophoric use, as Pan points out. However, we reject (both for English and Chinese) Baker's view that all unbound reflexives are intensive. Relevant English examples are (16a), (16c), and (16d), which are neither bound nor contrastive (and therefore not intensive either). Relevant Chinese examples include such logophoric examples as in (23), (25), and (28). In this connection, it is interesting to note that contrastiveness (whether signified by pitch accent or by lexical/structural marking) can sometimes render acceptable (or more acceptable) a nonsyntactic use of ziji that would otherwise be unacceptable. Consider, for example, the following: (36) a. Zhangsani de xin anshi Lisij hai-le ziji*i/j. Zhangsan DE letter imply Lisi harm-ASP self 'Zhangsan'Sj letter implies that Lisi harmed him*i/himself.' b. Zhangsani de xin anshi Lisij hai-le ZIJIril]. Zhangsan DE letter imply Lisi harm-ASP self 'Zhangsan'Sj letter implies that Lisi harmed him?i/himself.' c. Zhangsani de xin anshi Lisij hai de neige ren shi Zhangsan DE letter imply Lisi harm DE that person FOC zijiwy self 'Zhangsan'Sj letter implies that it was him?i/himself that Lisi harmed.'
Syntactic and Nonsyntactic Constraints on Long-distance Reflexives
335
First, note the unacceptability of Zhangsan as the antecedent of ziji in (36a); evidently, it lacks sufficient discourse prominence. In fact, it is the logophoric SOURCE, but as we noted previously, in logophoric uses of ziji, it is the SELF, not the SOURCE (or PIVOT), that counts. Interestingly, however, acceptability is improved markedly by contrastiveness, whether marked by pitch accent (as in (36b)) or lexically/structurally (by use of the shi-construction, as in (36c)). Thus, it appears that contrastiveness, in addition to SELF-type logophoricity and discourse prominence, is among the pragmatic factors that figure in the interpretation of nonsyntactic ziji-16 It should be noted that the various nonsyntactic uses of ziji (e.g., logophoric, contrastive, prominence-signaling) need not be mutually exclusive. Consider the example in (23), repeated here: (23) Ziji{ de xiaohai mei de Jiang de xiaoxi shi Lisii hen nanguo. self DE child not get prize DE news make Lisi very sad 'The news of hiSi child not getting a prize made Lisii sad.' Although, as discussed previously, the use of ziji here is logophoric, the possibility is by no means ruled out that it is also contrastive, as the sentence can mean either 'Lisi was sad because his own child didn't get the prize while other children got the prize.' or 'Lisi was sad because his child didn't get the prize (regardless of whether other children got the prize or not).'17
4. CONCLUSION We have drawn a clear distinction between syntactic and nonsyntactic uses of reflexives in Chinese and English. The distinction does not necessarily correlate with one between locally bound reflexives and nonlocally bound reflexives, but rather with one between binding by a syntactic subject and reference to an entity in the discourse subject to pragmatic constraints (involving such factors as logophoricity, discourse prominence, and contrastiveness). Another recent account of ziji that bears comparison with the one we have presented here is that of Pan (1995), which also deals with other reflexives such as ta ziji, benren, benshen, zishen, and their compound forms. Like Baker's account of English reflexives, whose influence is acknowledged throughout, Pan's account recognizes a fundamental ambiguity between contrastive reflexives with a discourse-prominent antecedent (Baker's intensive pronouns) and noncontrastive reflexives. Unlike Baker though, Pan assumes that noncontrastive reflexives include not only anaphors, which are subject to both a tree-configurational closeness condition and a compatibil-
336
Carl Pollard and Ping Xue
ity condition (which encompasses animacy-compatability, among other things), but also what Pan calls self-ascription ziji, which he characterizes as a de se anaphor, and therefore requiring a self-conscious antecedent. More specifically, the antecedent of self-ascription ziji must be the holder of a de se belief, who is either a discourse participant (speaker or hearer) or the referent of the subject of an explicit attitude predicate whose complement expresses the belief in question. We agree with Pan that the holder of a de se belief can certainly antecede ziji, though we would simply assimilate these cases to the logophoric ones, inasmuch as the holder of a reported belief surely qualifies as the SELF or subject of consciousness. However, we see Pan's account as problematic in several respects. First, like Baker's, it is a lexical-ambiguity account, which we have explicitly argued against. Second, Pan's two-way classification of noncontrastive ziji seems both empirically flawed and unintelligible. The main empirical shortcoming is posed by examples such as (37): (37) Zhangsani zai mei you jian guo jiu le ziji i ming de na Zhangsan again not have see ASP save ASP self life DE that ge ren. CL person 'Zhangsanj didn't see again the person who saved his; life.' The problem is that there is no sense in which Zhangsan here can be considered a self-ascriber, and thus, according to Pan's account, ziji, which is noncontrastive, should be an anaphor and therefore bound by the closest compatible NP, i.e., the PRO subject of jiu 'save.' But here jiu heads a subject relative clause and so its subject must be coindexed with the noun modified by the relative clause; i.e. ziji must refer to the life-saver, not Zhangsan. By contrast, by our account, Zhangsan is a subject o-commander of ziji and therefore can syntactically bind it.18 Incomprehensibly (at least to us), Pan himself argues against logophoric accounts of ziji on the basis of examples like the following, inasmuch as the antecedent of ziji is not the subject of consciousness: (38) a. Johni mingling Billj [PRO gei ziji i/j guahuzi] order to self shave 'John ordered Bill to shave him/himself.' b. Johni bi Billj [PRO gei zijiyj guahuzi] force to self shave 'John forced Bill to shave him/himself.' c. Johni rang Billj [PRO gei ziji i/j guahuzi] let to self shave 'John let Bill to shave him/himself.'
Syntactic and Nonsyntactic Constraints on Long-distance Reflexives
337
Frankly, we fail to understand what Pan's account of such examples is. Given that ziji here is not contrastive, it seems to follow from Pan's theory that the antecedent must be either (i) a self-ascriber or (ii) closest to ziji. But in each case, it is clear that that the antecedent does not fit the definition of self-ascriber (as the matrix predicates are not attitude verbs whose complement clauses express de se beliefs), and the nearest NP is the PRO subject of the complement clause. By contrast, our account simply predicts that either the local PRO subject or the matrix subject can bind ziji. Equally unclear to us is Pan's discussion of example (39) in the case where John is the antecedent: (39) [npJohni de gou] yao-le zijir DE dog bite self 'John's dog bit him.' Pan notes, correctly, that John does not satisfy the closeness condition. It would seem to follow from his account, then, that in this case, either John is a self-ascriber or else the use of ziji is contrastive, but clearly neither of these is the case. Therefore, we fail to understand why Pan considers (39) merely a "potential problem" rather than a problem simpliciter. By contrast, our account simply treats John as discourse-prominent (he is the directly affected patient and likely the topic as well). We have demonstrated that, given the facts from Chinese as well as from English, any theory that assumes the notion of obligatory binding as formulated in Chomsky's binding theory will be empirically inadequate. We have also shown that an approach based on the hypothesis of lexical ambiguity results in unnecessary complications. There is no reason why an NP cannot be both syntactically prominent and pragmatically prominent in the discourse at the same time. Consider the example in (7), repeated here: (7)
Zongtongi qing wo zuo zai zijii de shenbian. president ask I sit at self DE side 'The president, asked me to sit beside hinij.'
As discussed previously any o-commanding animate subject can be the antecedent of a subordinate reflexive. In this case, however, the matrix subject zontong is both syntactically prominent and pragmatically prominent. On our account, this is the very reason why the matrix subject is preferred to be the antecedent. The local subject is ruled out as the antecedent not due to syntactic reasons but due to semantic anomaly. We have identified a number of discourse/pragmatic conditions on nonsyntactic ziji, among which are logophoricity, contrast, and various conditions that contribute to relative discourse prominence, consistent with
338
Carl Pollard and Ping Xue
what has been discussed in the literature. But our analysis indicates that while nonsyntactic uses of reflexives must satisfy one or more of these conditions, none of them alone seems to be a necessary condition, contrary to the conclusion reached by various attempts to seek a single explanation for nonsyntactic uses of reflexives (e.g., Baker, 1995; Zribi-Hertz, 1989). Thus, the relevant conditions for nonsyntactic uses of reflexives are more diverse than has been traditionally assumed. This fact, together with the independent status of syntactic binding as we have discussed, provides further argument against any unified (syntactic or nonsyntactic) analysis of reflexive pronouns.
ACKNOWLEDGEMENTS We wish to thank the anonymous referee for very perceptive and useful comments and suggestions.
NOTES ipor expository convenience, we employ here the standard characterization of nonagreement blocking. A more empirically adequate characterization is given in Pollard and Xue (1998:310). See also endnote 6 for related discussion. 2 Our use of the symbol "*" in front of an index is best construed not as an indication of ungrammaticality but as unavailability of the indicated interpretation. 3 See, e.g., Baker (1995), Pan (1995), Xue et al. (1994). 4 We simplify considerably here because of space considerations. See Pollard and Xue (1998) and Xue et al. (1994) for more detailed discussion. 5 Also see Chen (1992) and Y. Huang (1994) for similar examples and related discussion. 6 In fact, while not ruled out on syntactic grounds, sometimes ziji fails to corefer with an LD subject, or can corefer with it only with difficulty, because of the (not strictly syntactic) effect of nonagreement blocking, as discussed in the preceding sections. In this respect, the term "syntactic ziji" might appear somewhat misleading. We believe that this effect arises from a processing strategy to avoid conflicts in point of view. See Huang and Liu (this volume) for discussion. 7 Also see Xue et al. (1994) for detailed discussion of the relevance of argument structure to ziji binding. 8 As the anonymous referee points out, a set of independent diagnostic tests would greatly strengthen the case for the distinction between syntactic and nonsyntactic uses of reflexives and, concomitantly, the argument against a purely discourse-pragmatic account of ziji-
Syntactic and Nonsyntactic Constraints on Long-distance Reflexives
339
9 We exclude from consideration two kinds of uses of forms homophonous with reflexives which neither occur in syntactically bound positions nor refer to some entity picked out by nonsyntactic constraints:
(i)
ziji used as an adverb with such meanings as (all) alone, unaccompanied, (all) on one's/its own, unassisted, of one's/its own accord, (all) by one/itself, without help, in person, as in the following examples: Wo shi ziji laide I be self come+ASP 'I came alone/unaccompanied/on my own/unassisted.' Pingzi buhui ziji daoxialai, zhun shi youren peng-le ta. bottle not-can self fall-down-come, definitely be someone bump+ASP it "The bottle couldn't have tipped over on its own; someone must have bumped it.'
For further discussion, see Li and Thompson (1981), and Tang (1989), which employs the (in our opinion, misleading) term "emphatic" for such uses. (ii)
ziji used as an indefinite (or generic) pronoun, as in: Ziji zuo shi ziji dang. self do be self be-responsible 'If one does something, then one is responsible for it.'
For discussion, see Batistella and Xu (1990). 10 Conditional and concessive adverbial conjunctions such as zhiyao 'if,' napa 'even if,' and jishi 'even though' seem to have similar logophoric functions. This is particularly true when the adverbial clause occurs after the main clause. While an adverbial clause usually occurs before the main clause in Chinese, it may occur after the main clause. See Tang (1976) for discussion of possible positions of adverbial clauses. Precisely what adverbial conjunctions belong to this class in Chinese is a topic for further research. llr These examples also show that Sells' putative universal constraint that internal SELF always coincides with the deictic PIVOT, while perhaps valid for Japanese, is inapplicable to Chinese. See also Cole et at. (this volume) for discussion. 12 More generally, it could well be that the best approach to nonsyntactic reflexives is one that characterizes pragmatic constraints on their referents (as opposed to their antecedents) with those cases where there is a coreferential linguistic expression on the prior discourse being treated as a special case. 13 Baker (1995) concludes that in English, reflexive uses formerly analyzed as logophoric (e.g., by Zribi-Hertz, 1989) should all be assimilated to his intensive pronouns. We follow Golde (to appear) in assuming (both for English and Chinese) that, although Baker's analysis is probably correct for many of the examples he discusses, not all logophoric uses of reflexives are intensive. 14 Information saliency is related to such factors as the degree of semantic content, and perceptual and attentional vividness. See Chen (1992), Y. Huang (1994), and references cited there. Chen argues that givenness plays a particularly important role in characterizing
340
Carl Pollard and Ping Xue
what he calls "topicality," and that the possessive position of the subject is one typical position with high topicality (also see, e.g., Kuno, 1987). This seems to provide an explanation for why a subcommanding NP as discussed in Section 2.1 can serve as the antecedent of ziji in an appropriate discourse. 15 For expository convenience, we will adopt a usage rather close to that of Baker, which encompasses a number of other factors, including high external rank (e.g., the King, God) and thematic salience (e.g., directly responsible agent, directly affected patient). However, we depart from Baker's usage in not subsuming logophoricity (Baker's "subject of consciousness") under the notion of discourse prominence. This is an expository choice and is not intended to reflect a strong theoretical commitment. 16 Alternatively (and with all due respect to Baker), we could absorb contrastiveness into the notion of discourse prominence. This move would have the side effect of rendering Baker's notion of intensification (which crucially involves the interaction of contrastiveness with discourse prominence) unintelligible. 17 See Pan (1995) for related discussion and examples. However, we differ from Pan in that most of the examples he provides to argue that LD ziji need not be contrastive count as syntactic uses on our account. 18 In addition, the speaker in (37) might be adopting the point of view of Zhangsan, so that ziji is logophoric, i.e., identified with the (internal) SELF; but our account does not require this.
REFERENCES Baker, C. L. (1995). Contrast, discourse prominence, and intensification, with special reference to locally free reflexives in British English. Language 71, 66-101. Bates, E., and MacWhinney, B. (1982). Functional approaches to grammar. In E. Wanner and L. Gleitman, (eds.), Language Acquisition: The State of the Art, pp. 173-218. Cambridge University Press. Battistella, E. (1989). Chinese reflexivization: A movement to INFL approach. Linguistics 27,987-1012. Battistella, E., and Xu, Y.-H. (1990). Remarks on the reflexive in Chinese. Linguistics 28,205-240. Chen, P. (1992). The reflexive ziji in Chinese: Functional vs. formalist approaches. Research on Chinese Linguistics in Hong Kong. The Linguistic Society of Hong Kong. Chomsky, N. (1981). Lectures on Government and Binding. The Netherlands: Foris Publications. Chomsky, N. (1986). Knowledge of Language: Its Nature, Origin, and Use. New York: Praeger. Clements, N. G. (1975). The logophoric pronoun in Ewe: Its role in discourse. Journal of West African Languages 10,141-177.
Syntactic and Nonsyntactic Constraints on Long-distance Reflexives
341
Cole, P., Hermon, G., and Lee, C.L. (this volume). Grammatical and discourse conditions on long distance reflexives in two Chinese dialects. Cole, P., Hermon, G., and Sung, L.-M. (1990). Principles and parameters of long-distance reflexives. Linguistic Inquiry 21,1-22. Cole, P., and Sung, L.-M. (1994). Head movement and long-distance reflexives. Linguistic Inquiry 25,355-406. Everaert, M. (1991). Contextual determination of anaphor/pronominal distinction. In I Koster and E. Reuland (eds.), Long-Distance Anaphora, New York: Cambridge University Press. Givon, T. (1983). Topic Continuity in Discourse: Quantitative Cross-Language Studies. Amsterdam: John Benjamins. Golde, K. (to appear). A complete Principle A for English. In Proceedings of the 34th Colloquium of Linguistics. Berne: The Peter Lang-Verlag. Hellan, L. (1991). Containment and connectedness anaphors. In J. Koster and E. Reuland (eds.), Long-Distance Anaphora. New York: Cambridge University Press. Huang, C.-T. J. (1982). Logical Relations in Chinese and the Theory of Grammar, Ph.D. dissertation, MIT. Huang, C.-T. J., and Tang, C.-C. J. (1991). The local nature of the long-distance reflexive in Chinese. In J. Koster and E. Reuland (eds.), Long-distance Anaphora. New York: Cambridge University Press. Huang, C.-T. J., and Liu, C.-S. L. (this volume). Anaphoricity and logophoricity in Mandarin Chinese. Huang, Y. (1994). The Syntax and Pragmatics of Anaphora. New York: Cambridge University Press. Koster, I, and Reuland, E. (eds.) (1991). Long-Distance Anaphora, New York: Cambridge University Press. Kuno, S. (1987). Functional Syntax. Chicago: University of Chicago Press. Kuroda, S.-Y. (1973). Where epistemology, style, and grammar meet: A case study from Japanese. In S. R. Anderson and P. Kiparsky (eds.), A Festschrift for Morris Halle. New York: Holt, Rinehart and Winston. Li, C. N, and Thompson, S.-A. (1981). Mandarin Chinese: A Functional Reference Grammar. Berkeley: University of California Press. Liu, Y.-H., Pan, W.-Y, and Gu, H. (1983). Shiyong Xiandai Hanyu Yufa [Practical Grammar of Modern Chinese]. Beijing: Waiyu Jiaoxue Yu Yanjiu Chubanshe. Pan, H. H. (1995). Locality, Self-Ascription, Discourse Prominence, and Mandarin Reflexives. Ph.D. dissertation, University of Texas at Austin. Pollard, C. J., and Sag, I. A. (1992). Anaphors in English and the scope of the binding theory. Linguistic Inquiry 23,261-303. Pollard, C. J., and Sag, I. A. (1994). Head-Driven Phrase Structure Grammar. Chicago and Stanford: The University of Chicago Press and CSLI Publication. Pollard, C. J., and Xue, P. (1998). Chinese reflexive ziji: Syntactic reflexives vs. nonsyntactic reflexives. Journal of East Asian Linguistics 7,287-318. Reinhart, T., and Reuland, E. (1991). Anaphors and logophors: An argument structure perspective. In J. Koster and E. Reuland (eds.), Long-distance Anaphors. New York: Cambridge University Press.
342
Carl Pollard and Ping Xue
Reinhart, T., and Reuland, E. (1993). Reflexivity. Linguistic Inquiry 24,657-720. Reuland, E., and Koster, J. (1991). Long-distance anaphors: An overview. In J. Koster and E. Reuland (eds.), Long-distance Anaphors. New York: Cambridge University Press. Sells, P. (1987). Aspects of logophoricity. Linguistic Inquiry 18,445-479. Tang, C.-C. J. (1989). Chinese reflexives. Natural Language and Linguistic Theory 7, 93-121. Tang, T.-C. (1976). Studies in Transformational Grammar of Chinese, Vol. 1. Taipei: Student Book Co. Ltd. Yu, W. (1992). Challenging Chinese reflexive data. The Linguistic Review 9, 285-294. Xue, P., Pollard, C.J., and Sag, I.A. (1994). A new perspective on Chinese reflexive ziji. In Proceedings of the 13th West Coast Conference on Formal Linguistics. Standford University: CSLI Publications. Xue, P., and Hukari, T. (1993). Non-clause-bounded reflexives: Logophoricity and grammatical hierarchy. Manuscript, University of Victoria. Zribi-Hertz, A. (1989). Anaphora binding and narrative point of view: English reflexive pronouns in sentence and discourse. Language 65,695-727.
ANAPHORS, LOGOPHORS, AND BINDING ERIC REULAND Utrecht Institute of Linguistics OTS Utrecht, The Netherlands
1. INTRODUCTION The aim of this article is to contribute to our understanding of what has come to be known as the free, or "logophoric," use of anaphors. My starting point will be a comparison of the interpretation of sig in subjunctive and infinitival clauses, summarizing some of the considerations given in Reuland and Sigurjonsdottir (1997). Since the initial discussions of Icelandic long-distance anaphora by Thrainsson (1976) and Maling (1984), it has become widely acknowledged that the anaphor sig in subjunctives in Icelandic can take as its antecedent a non-c-commanding argument. Thus, in (1) the NP Jon can serve as the antecedent for sig, although it does not c-command and anaphor.1 (1) a. [NP Skodun Jonsi] er [ad sigi vanti hoefileika]2 Opinion John's is that SIG-Acc lacks-Subj talents 'John's opinion is that he lacks talents.' (Maling, 1984:222) b. [NP Alit Jonsi] virdist [tj vera [ad eg hati sigj]]] Belief John's seems be that I hate-Subj SIG 'John's belief seems to be that I hate him.' (him=John) c. Bjorn sagdi Petri fra [NP osk Jonsi ] ud [ad Ari syndi Bjorn told Peter about wish John's about that Ari showed-Subj seri virdingu]3 SIG respect 'Bjorn told Peter about John's wish that Ari showed him (=John) respect.' Syntax and Semantics, Volume 33 Long-Distance Reflexives
343
Copyright © 2001 by Academic Press All rights of reproduction in any form reserved. 0092-4563/00 $35.00
344
Eric Reuland
Furthermore, sig in subjunctives can refer to a matrix object, even though otherwise sig is strictly subject oriented. (2) ?J6ni er masokisti. Pad gledur Joni [ad eg muni lemja sigi i hausinn med spytu a morgun] 'John is a masochist. It pleases John that I will-Subj hit him on the head with a stick tomorrow.' (Sigurjonsdottir, 1992: 95)4 Comparing sig in subjunctive clauses with sig in infinitival clauses, we see that in clauses of the latter type, a c-command requirement on antecedents must be strictly observed (3a-c). The same holds true for subject orientation (3d). (3) a. *[NP Skodun J6nsi]j virdist [tj vera haettuleg fyrirsig i ] Opinion John's seems be-Inf dangerous for SIG 'John's opinion seems to be dangerous for him.' b. * [NP (Osk Jonsi]j er likleg til [t j adhafa slcemarafleidingar Wish John's is likely to to have-Inf bad consequences fyrir sigi] for SIG 'John's wish is likely to have bad consequences for him.' c. *[NP Alit Jonsi ]i er sagt [tj hcefa seri vel] Belief John's is said suit-Inf SIG well 'John's belief is said to suit him well.' d. *Egj hotadi Joni [ad PROj lemja sigi] I threatened John to hit-Inf SIG 'I threatened John to hit him.' That binding sig out of infinitival clauses is possible when c-command is obeyed is shown in (4). (examples cited in Thrainsson (1991)) (4) a. Peturi bad Jensj um [PRO j ad raka sigi,j] Peter asked Jens to shave self 'Peter asked Jens to shave him/himself.' b. Annai telur [pig hafa svikid sigi] Anna believes you-Acc have-Inf betrayed self 'Anna believes you to have betrayed her.' As has often been observed, the felicitousness of sentences as in (1) and (2) is governed by the discourse status of the antecedent. The antecedent of sig must be the person (distinct from the speaker-narrator) whose perspective or point of view is reflected in the sentence. This is illustrated by the following minimal pair reported in Sells (1987:451): (5) a. Barnidi let ekkii Ijos [ad pad hefdi verid hugsad The child put not in light that there had-Subj been thought
Anaphors, Logophors, and Binding
345
vel um sigi] well about SIG 'The child didn't reveal that she had been taken good care of.' b. *Barnidi bar pess ekki merki [ad pad hefdi verid The child bore of it not signs that there had-Subj been hugsad vel um sigi] thought well about SIG 'The child didn't look as if she had been taken good care of.' The difference in acceptability between (5a) and (5b) can be attributed to the fact that in the (a) sentence, the report is made from the child's point of view, i.e., it is the child, and not the speaker, who didn't reveal that he/she had been taken good care of, whereas in the (b) sentence, it is the speaker who reports that the child didn't look as if he/she had been taken good care of. To see this effect yet more clearly, consider the sentences in (6), comparing them to those in (1). (6) a. *[NP Skodun Jonsi fcer mig til adhalda [ad sigi Opinion John's leads me to to believe that SIG-Acc vanti hcefileika] lacks-Subj talents 'John's opinion leads me to believe that he lacks talent.' (Maling, 1984:222) b. *[NP Vinur Jonsi] telur [ad eghati sigi] Friend John's believes that I hate-Subj SIG 'John's friend believes that I hate him.' (him=John) c. *Petta vandamdli krafdist pess [ad vid hugsudum stodugt This problem demanded it that we thought-Subj constantly um sigi] about SIG 'This problem demanded that we constantly think about it.' (Sigurdsson, 1990:335) The sentences in (6) are infelicitous, yet the antecedent NP bears the same structural relation to sig as in the felicitous sentences in (1). In (1), however, Jon is the person (distinct from the speaker) whose opinion, belief, or wish is reported. Sentences (6a) and (6b), on the other hand, do not reflect Jon's point of view; rather, it is the first person pronoun mig 'me' in (6a) and the NP Vinur Jons 'John's' friend in (6b) which carry the perspective. This analysis is further supported by the example in (6c), which shows that an inanimate NP, which cannot possibly be the "source" of an utterance, cannot serve as the antecedent for sig in subjunctives. Since Hagege (1974) and
346
Eric Reuland
Clements (1975), this use of sig and similar elements in other languages is generally called logophoric. Consider next the sentences in (3), which we also found to be infelicitous. Crucially, these sentences have been constructed in such a way that the putative antecedents have the same discourse status as those in (1). In these sentences, the discourse factors that license the sentences in (1) are without effect. That is, in infinitival clauses, discourse factors are apparently unable to compensate for the lack of c-command between a putative antecedent and sig. The question is, then, how this difference between sig in infinitival clauses and sig in subjunctive clauses can be understood. For an answer, we must investigate the logophoric use of sig in more detail. The discussion of this use of sig is generally conducted against the background of two assumptions: (i) sig is anaphor and (ii) anaphors must be bound. The latter is taken to be a distinguishing characteristic of anaphors versus pronominals. That is, anaphors are inherently defective: they lack the content to be interpreted independently (Bouchard, 1984). Clearly, if these assumptions are correct, the logophoric use of sig is extremely puzzling. In view of the fact that binding requires c-command (Reinhart, 1983), it should be downright impossible for sig to receive an interpretation when it is not c-commanded by its antecedent. This point becomes even more pressing given that a linguistically expressed antecedent need not be available at all, as is illustrated in (7). Here, ser refers to the individual whose thoughts are being presented, although there is no other argument in (7) referring to this individual. That is, this individual is only manifested in the discourse representation associated with the sentence. (7) Maria var alltafsvo andstyggileg. Pegar Olafur j kosmi segdi him seri/*j areidanlega ad fara . . . 'Mary was always so nasty. When Olaf would come-Subj, she would certainly tell himself [the person whose thoughts are being presented—not Olaf] to leave.' (SigurSsson, 1990: 317) Such a use is characteristic for pronominals, not for anaphors. In order to understand this use of sig, two lines of enquiry must be pursued: (i) What types of interpretation are available to elements like sig? (ii) What is the role of the c-command condition? I will discuss these issues in that order.
2. INTERPRETING SIG Sig can be distinguished from other pronominal elements in that it lacks a grammatical singular/plural opposition and a grammatical gender op-
Anaphors, Logophors, and Binding
347
position. In agreement with anaphors in many other language, it lacks a form with nominative Case (see Everaert, 1986, 1990, and Anagnostopoulou and Everaert, 1998, for discussion). Suppose we adopt Bouchard's (1984) thesis that a DP must have a full specification of CD-features in order to be interpreted (henceforth, the deficiency thesis, for a precise formulation, see (15) following). It follows that sig cannot be interpreted independently, i.e., it must have an antecedent. A binding requirement on sig (and other anaphors) then follows immediately from the fact that in order to be interpreted, sig must first acquire CD-features. Directly or indirectly, it will be the role of the antecedent to provide them.5 This is essentially the line taken in much work on anaphors including that of Reinhart and Reuland (1991). This line, however, also raised the question of what enables sig to be interpreted in those contexts where an antecedent need not c-command sig, or can be absent entirely. If the deficiency thesis reflects a really deep property of language, we would expect that nothing can make up for the impossibility for sig to acquire CD-features in some given context. To say that some sort of discourse prominence can make up for the nonsatisfaction of a fundamental property of lexical or syntactic structure would effectively render the deficiency approach vacuous. Logically, there are, therefore, two options. One is that in those cases where sig appears to be interpretable without a c-commanding antecedent, there is a (structurally accessible) source of F-features, after all; the other is that the deficiency thesis is incorrect. One certainly cannot, in principle, exclude the possibility that a significant linguistic relation between an anaphor and a non-commanding antecedent can be established. For one thing, too little is known about the reasons why c-command holds at all, and furthermore, there are wellknown exceptions to the c-command requirement, notably in cases of variable binding. Where a linguistic antecedent is not realized, one faces the additional task of justifying some null-source of CD-features, but again, there are no a priori reasons to exclude the possibility that it can be done. Yet, even if both tasks can be carried out successfully, an additional task remains, namely, to explain the sensitivity to subjunctive and to discourse factors. In this respect, the option of dropping the deficiency thesis is no different. If this thesis is dropped, this amount to saying that under certain conditions anaphors like sig can be used as pronominals. But still, the sensitivity to subjunctive and discourse factors of this use remains to be explained. I will now show that it is the deficiency thesis which must be dropped, by eliminating the alternative. I will first discuss the issue of non-c-commanding antecedents.
348
Eric Reuland
3. NON-C-COMMANDING ANTECEDENTS To accommodate cases such as (1) where the visible does not c-command the anaphor, two options can logically be distinguished: (i) the element which is realized is the actual antecedent of the anaphor, or (ii) the structure of (1) is closer to that of (7) than appearances indicate, and the real antecedent is a null element (somehow linked to the visible antecedent) that does c-command the anaphor. Let us first address option (i). This option implies a certain degree of variability in the structural conditions on A-binding. However, it is found throughout that the relation between a and b in configurations such as (8) can be one of variable binding or coreference, but not one of A-binding. This is illustrated in (9). (8) [a'sN] V [b ... ] (9) a. * Mary's father loves herself b. Every girl's father hates her boyfriend c. Mary's father hates her boyfriend Both Reuland (1998) and Kayne (1994) present independent reasons for why A-binding requires the strict version of c-command. Here, I will not offer a discussion of the difference between these approaches, but I refer to Reuland (1998) for such a discussion.6 At this point, it suffices to note that under either approach, option (i) would call for additional assumptions, which would lack independent motivation. It would narrow down possibilities if we could independently determine whether the relation between sig and its antecedent in Icelandic subjunctives is modeled on anaphor binding, variable binding, or coreference. In line with Reinhart (1983), I will be assuming that strict versus sloppy identity in VP-deletion contexts provides a diagnostic for coreference versus variable binding. Consider (10): (10) Mary hates her boy friend and Cindy does too. As is well known, in (10), both a coreference and a bound variable reading are available. As observed by Thrainsson (1991: 60), where c-command is satisfied, the strict/sloppy identity ambiguity also shows up with sig in the long-distance subjunctive case. This is illustrated in (11): (11) Joni telur [ad professorinn muni fella sigi a profmu] John believes that the professor will-Subj fail SIG on the test og AriI telur pad lika andAri believes so too a. = Ari believes that the professor will fail Ari on the test b. = Ari believes that the professor will fail John on the test
Anaphors, Logophors, and Binding
349
However, when subjunctive sig is not c-commanded by its long-distance antecedent, the sloppy (i.e., bound) reading is, as Thramsson (1991) puts it, "more difficult, if not impossible to get." This is illustrated in (12) Thramsson (1991:60): (12) Skodun Jonsi er [ad sigi vanti hcefileika] og pad er Opinion John's is that SIG lacks-Subj talents and that is skodun Petursj lika opinion Peter's too 'John's opinion is that SIG lacks talents and that is Peter's opinion too.' a. Peter's opinion is that Peter lacks talents b. = Peter's opinion is that John lacks talents7 This is problematic for any approach that must claim that the interpretation of sig in (12) depends on the existence of a syntactic binding relation between Jon and sig, instead of coreference.8 The other alternative (mentioned under (ii)) is that sig is not directly licensed by Jon, but by some null element associated with the subjunctive clause, let's say, some null operator obtaining its range from the context, as in (13): (13) Skodun Jons er [Op i ad sigi vanti hcefileika] og pad er Opinion John's is that SIG lacks-Subj talents and it is skodun Peturs lika [Op i ad sigi vanti hcefileika] opinion Peter's too that SIG lacks-Subj talents However, it is easily seen that this does not solve the problem. Just as the element heading the original must get its value from Jon, the one in the copy would be able to gets its value from Petur. Thus, no explanation of the contrast between (11) and (12) follows. Moreover, it is unclear how such an operator could serve as a source of F-features for sig, except by stipulation. There is one further possibility to be considered: an operator associated with the full sentence, as in (14). (14) Opi[Skodun Jonsi er [ad sigi vanti hcefileika] og pad er Opinion John's is that SIG lacks-Subj talents and that is skodun Petursj lika] opinion Peter's too Assuming that the reconstruction of the VP in the conjunct uses material that is in the scope of the operator, one would indeed except that a sloppy reading does not occur. However, this structure does not resolve the question of how sig receives F-features any more than the structure of (13). Moreover, similar structures with overt operators are crossover configura-
350
Eric Reuland
tions. As illustrated by Who does John's mother love t? or The man who John's mother admires most t is him, this structure does not facilitate identity between elements in the position of Jon and sig.9 The conclusion is that this line does not offer any promise towards result either. Since this exhausts the alternatives discussed in Section 2, we conclude that the deficiency thesis must be dropped. Given a suitable environment, sig can be interpreted as a pronominal and can be related to its antecedent by coreference.
4. THE INTERPRETIVE PROCESS The result that the deficiency thesis must be dropped is quite welcome for independent reasons, as discussed in Reuland (1997a). Consider the specific formulation in (15): (15) If a lacks a specification for some F-feature F, then a cannot be interpreted independently. It is indeed true that anaphors (simplex anaphors, such as Icelandic sig, Dutch zich, and perhaps Latin se; clitic anaphors, such as Italian se/si, French se; and complex anaphors, such as Icelandic sjalfan sig, Dutch zichzelf) are underspecified in some respect (e.g., Burzio (1991)). Taking the set of F-features to be [aperson, bgender, gnumber], we observe that, e.g., zich is specified for person (namely, third-person), but is not specified for gender nor for number. Although zich requires a third-person antecedent, the latter can be masculine, feminine or neuter, and singular or plural. (Note the contrast with Slavic, where the anaphor is generally not specified for person either.) Although, at first glance, (15) may therefore seem natural, it is not hard to see that, in fact, it is just a stipulation. Proposition (15) cannot be reduced to known properties of features. The role of lexical and formal features is only to constrain the interpretation of the elements carrying them. On the basis of that role, one would rather expect (16) to hold, and not (15). (16) If a has fewer F-features than (3, there are fewer constraints on the interpretation of a than on the interpretation of (3 Proposition (16) should, then, be adopted as the null-hypothesis. Given (16), an anaphor such as Icelandic sig which carries only the feature third-person, should just have a broad range of application, namely, to all objects distinct from speaker and addressee. Thus, there is no intrinsic need for anaphors to have a syntactic binder. On the other hand, we know that free anaphors have a restricted distribution, just as they have a very specific type of interpretation. Thus, whereas
Anaphors, Logophors, and Binding
351
standard binding theory leads to the question of why, under certain conditions, logophoric interpretation can be allowed, from the present perspective, the questions to ask are (17a) and (17b): (17) a. Why is, under certain conditions, free interpretation of anaphors blocked? b. What are these conditions? In order to be able to address these questions fruitfully, I must first discuss the relation between pronominals and anaphors from a more general perspective, starting out with the interpretation of pronominal elements. 4.1. Interpreting Pronominal Elements Consider the tree structure a in (18), with its subtrees t and d. Let's focus on some argument a in 8:
If a has lexical features (Cindy, a cat, etc.) it will be directly interpreted by a suitable element in the domain of discourse (which reflects part of what one may call a knowledge base). Suppose a is an element without lexical features, which therefore functions as a semantic variable. Let's assume with Chierchia (1995) that a semantic variable may be either free or bound. If it ends up being free, it receives an E-type interpretation. That is, it is assigned as its value some element in the domain of discourse, just as in the case of a lexical. The alternative is interpretation by variable binding. Given some compositional interpretation process, a may, at any time, be construed as a bound variable (as when an expression [t [ § ... a ....]] is interpreted, with t a possible antecedent for a). Note, that as long as a has not been interpreted, it is held "in store." If the interpretation of a is "postponed," this will be visible on whatever expression reflects 8's interpretation, since 8 will be a semantically open expression. There may be various ways to represent this. Here, I will represent the semantically open character of 8 by a l-expression. So, in order for the interpretation of 8 to be completed, the l-expression must receive an argument enabling a to receive a value. Thus, whatever the precise nature of representations at the relevant level, the status of 8 will be equivalent to (19). (19) l x ( . . . x . . . )
352
Eric Reuland
As I said, I will be adopting the standard position that interpretation proceeds compositionally. Within the present framework, we can understand this as involving a semantic counterpart of the Merge operation in Chomsky (1995), let's call this procedure Interpretive Merge (IM) (see Reuland, 1998). This means that the fact that the subtree 8 is a semantically open expression will be visible at 8's root. The open position can become saturated at the next interpretive step, namely the one composing 8 with T. If 8 is represented as a l-expression, the process can be viewed as the application of t to 8. If, for some reason, application of t to d is impossible (for instance, if t has the wrong type), a will remain in store, which will be visible in the form of nonsaturation of a, etc. (As is shown in Reuland, 1998, a procedure along these lines effectively derives the structural conditions on variable binding, including the c-command condition where its holds, and violations of c-command in cases of inverse-linking and binding by a specifier). In Reuland (1997a) and Reuland and Sigurjonsdottir (1997), it is argued that the choice of a free versus a bound interpretation of pronominal elements is governed by an economy principle, not so much governing the choice of operations within one module of human communicative competence (such as the computational system CHL in Chomsky's (1995) sense), but rather the number of operations across modules that are involved. For instance, if some linguistic expression is assigned as its value an object from the knowledge base, given standard assumptions, this is an operation involving elements from different cognitive faculties. I am claiming that this is associated with a certain cost. It is generally assumed that an interpretation procedure that is able to capture the necessary semantic generalizations (for instance, those involving the contrast between collective and distributive readings) requires operations that are language independent and go beyond what can be stated in CHL (i.e., beyond move, check, merge). If this assumption is correct, which seems quite reasonable, this comes down to claiming that there is a separate semantic component among our cognitive faculties. If so, operations assigning to syntactic structures expressions in which variable binding or the distributive relations between arguments and predicates are represented are also cross-modular, and carry the concomitant cost as well. If expressions a and b are to receive the same value (to be covalued), the following operations may be involved: (i) a and b may independently be assigned the same object from the knowledge base. (ii) a and b may be identical variables bound by the same operator, (iii) a and b may be members of the same syntactic chain.
Anaphors, Logophors, and Binding
353
This leads to the following cases to be distinguished: a a (20) a. Knowledge base (objects) Semantic objects (variables) Syntactic objects (chains) Basic expressions b. Knowledge base (objects) Semantic objects (variables) Syntactic objects (chains) Basic expressions c. Knowledge base (objects)
x
l
X
2
C1 C2 a ... b a
X1
x1
C1
C2
a ... b
a
Semantic objects (variables) Syntactic objects (chains) C1 C1 Basic expressions a ... b As is easily seen, if a and b are to be assigned the same value via the knowledge base, as in (20a), they represent different chains and different variables; the number of cross-modular operations is 4. If a and b are covalued via variable binding, as in (20b), they are represented by different chains but an identical variable; the number of cross-modular operations is 3; hence, this operation is cheaper. The empirical claim is that this process is preferred where the option exists. (As regards the relation between semantics and pragmatics, this is essentially Rule I of Reinhart (1983) and Grodzinsky and Reinhart (1993)). The cheapest interpretation process is given in (20c): chain formation creates one syntactic object. Only two cross-modular operations are required. Thus, if it comes to the task of assigning a value to some element a, the cheapest way is to do it in the syntax (by linking a to some element b, creating a syntactic chain); the costliest way to do it is by interpreting a independently, accessing the knowledge base; intermediate in cost is to do it in the interpretive process, turning a into a variable which is kept in store until it can be logically bound by an antecedent. Note that the reason for chain formation is not that it is the least costly way of expressing a dependency. Chain formation is just the result of a blind process, driven by requirements of feature checking. Only where two derivations, one syntactic and the other semantic, lead to the same semantic object can there be a competition of the type discussed.
354
Eric Reuland
This leads us immediately back to the relation between pronominals and anaphors, and to a further discussion of syntactic chain formation. 4.2. Pronominal Binding and Anaphor Binding In Reuland (1996,1997a) it is shown that the binding of simplex anaphors (Dutch zich, Icelandic sig) follows from syntactic chain formation using no more than Merge, Move/attract, and feature checking and associated properties. Indices are not involved. Informally, this comes down to the linking of syntactic dependencies that exist independently. This is illustrated in (21) (see also Reuland and Sigurjonsdottir (1997)). (21) DP ... [ V / I ... jAgr jsig . . . ] . . . sig Example (21) shows two dependencies: the Case checking relation between verb and object (sig) and the dependency between verb and subject expressed by agreement (or by EPP; for expository purposes, we will limit ourselves to agreement). All are formal dependencies. The chains encoding binding relations are formed by linking two dependencies. More precisely, at the relevant level the verb has merged with inflection forming a V/I complex, and the Case checking relation between the verb and sig is encoded on that complex. The agreement features of the V/I complex are represented by jAgr, and the relation of the verb in the V/I complex with sig by the features jsig. Agreement between subject and verb involves the checking of F-features inside the V/I complex by the subject DP. For the mechanism effecting agreement, jAgr and jsig are equally visible. So, among the dependencies that can be formally represented is also the dependency , as a product of the same checking mechanism that establishes the dependency . <j sig , sig> is also among the formal dependencies. If this dependency relation is transitive, is among the formal dependencies as well. In fact, it has the status of a chain (see Reuland (1996,1997a) for more discussion). This chain is thus a syntactic object encoding the relation between the anaphor and its antecedent.10 Clearly, the possibility to form a chain must be restrict. For instance, if the object is a lexical NP or a pronoun, instead of sig, a chain should not be formed. In Reuland (1996, 1997a), it is argued that the relevant factor is grammatical number: sig is not marked for grammatical number, pronouns and lexical NPs are. For lack of space, I will not review the argumentation here, and simply assume the result of Reuland (1997a) that grammatical number on a in (22) will block checking, and make formation of a chain impossible: (22) DP ... [V/I jAgr ja] . .. a
Anaphors, Logophors, and Binding
355
The span of this type of dependency, and therefore the locality of the binding relation, is entirely determined by the factors governing the movement of the elements involved, such as Case checking and V-movements. (see Reinhart and Reuland (1991) for discussion), and the possibilities for the resulting objects to be linked. Where no chain can be formed, a dependency cannot be syntactically encoded. I am claiming that a syntactic chain constitutes one syntactic object which will be interpreted as one semantic object. This yields the effect of binding without invoking indices.11 To develop a systematic picture, the following cases are to be considered. Consider first (23):
(23)
...[DP1ll...[CP[DP2l2...Va]]]
Suppose a is the object of V, and DP2 is its subject. DP1 c-commands a from a higher clause. Say both are suitable antecedents (nondistinct in F-feature specification). Suppose that a lacks number but also that, in line with (16), it can, in principle, be interpreted independently (ignoring for the moment the special discourse restrictions). From the perspective of the interpretive procedure then, a can be bound by either DPl or DP2. Reinhart and Reuland (1991) argue that Scandinavian languages have a covert counterpart of the overt verb movement of Dutch or German. As they show, this makes possible a straightforward account of certain differences in binding domains (for lack of space, I will not repeat this discussion here). Suppose then that, in line with this argument, V can reach both I1 and I2 by (covert) movement (as when the complement is infinitival). If so, the dependencies (DPl ,a) and (DP2, a) can both be syntactically encoded, and there will be no grammatical preference. Suppose that V cannot reach I1 (as when the lower clause is an indicative CP). If so, the dependency (DP1,a) cannot be syntactically encoded, Hence, assigning the value of DP1 to a will require (at least) three cross-modular steps (two from CHL to the semantics, one from semantics to the knowledge base). Assigning a the value of DP2 requires only two cross-modular steps, since a chain (DP2, a) can be formed, mediated by V. (Cross-modular are the steps from DP2 to the semantics and from there to the knowledge base.) Hence, that course is cheaper and wins over the option to assign DP1 to a as its antecedent. What if DP2 is not a suitable antecedent for a, due to a feature mismatch (for instance, DP2 is first-person and a third-person)? In such cases DPl is nevertheless not a viable alternative antecedent, as we know. This fact follows from chain theory without further assumptions. As Chomsky (1995: 309) points out, chains with a feature mismatch are ill-formed objects. Derivations leading to such chains are cancelled. Cancelled derivations count for the computation of economy rankings, unlike crashed derivations. Hence, less economical alternatives remain blocked. This procedure will immedi-
356
Eric Reuland
ately carry over to the present case. The chain (DP2, a) will be formed, but being ill-formed, it will not be interpreted. Yet its presence will block the less economical alternative interpretation with a bound by DP1. 4.3. Syntactic Conditions on Free Anaphors Although we have given up the deficiency thesis in (15), it might seem that we are still in virtually the same position with regard to free anaphors, unless we can find a way to occasionally release them from the preference for derivations in which they are bound. The logic of our approach shows the way in which to look for a solution. Consider again (22). (22) D P . . . [ v / I . . . j A g r j a - - - ] - - . a Crucially, the V/I complex as a carrier of both F-features linking it to the subject and a Case feature linking it to the object acts as an intermediary for the chain between these two to be formed. If some property of V/I prevents it from acting as an intermediary, no chain between DP and a will result. If there is no chain, the cheapest way of arriving at an interpretation of a is not available. Hence, interpreting a both by a semantic strategy, or, if such a strategy is blocked, by directly accessing the knowledge base become available. Essentially, under such a condition, an element like sig comes to act like a pronominal. In Reuland (1997a) and Reuland and Sigurjonsdottir (1997), it is argued that subjunctive morphology on the verb has precisely this effect. In accordance with Kempchinsky (1986,1995), Manzini (1993), and many others one may assume that subjunctive morphology is licensed by an operator (or perhaps vice versa) (see also Quer (1998)). Licensers are elements such as negatives and modals. Assuming this to be a morphosyntactic property of subjunctive, this licensing relation must be syntactically encoded. Within the framework adopted, such encoding involves (covert) movement of the V/I complex to its licenser. The result is sketched in (24). (24) [0P [V/I • • • j.Agr ja ... ]i OP] DP ... t i . . . a So, this operation removes the V/I complex from the intermediate position between DP and a. Now, note that chain linking as it operates on (22) requires two chains, with an identical shared element (see the discussion in Chomsky (1995)). That is, if (a,b) is a chain, (t,d) is a chain, and b=t, then (a,8) is a linked chain. In (22), we can form the objects (DP, [V/I jAgrja]) and ([v/I jAgrja],a). For each object, its chain status is transparent, in the former by the relation between DP and Agr, in the latter by the relation between, ja and a. If the V/I complex with the information it caries is removed, only a trace is left. There are now two options as to the status of this trace. Suppose, a trace has no internal structure. In that case, the two chains are no longer
Anaphors, Logophors, and Binding
357
directly represented in the structure, and the information for chain linking to apply is no longer available. Alternatively, this type of movement allows reconstruction. So, in fact, ti is a copy of the moved element. However, that theory requires the inaccessibility of traces/copies to further operations (see Chomsky (1995) for discussion). Hence, the impossibility for ti or whatever is in its place to mediate chain linking also follows in this case. Thus, a chain between DP and an anaphor in the position of a cannot be formed in principle. It follows that there is no way to encode a dependency between a and a possible antecedent in the syntax. This opens the way for a pronominal interpretation of a.12 Of course, this procedure makes use of a very special property of the syntactic environment. Other languages also have forms that allow a dual use as free or bound anaphors, depending on their environment. It should be emphasized that what should be expected to be universal is not any particular factor blocking a syntactic encoding of a binding relation, such as subjunctive. Rather, whatever blocks syntactic encoding should follow from the nature of the encoding device. For instance, in the case of English, encoding the reflexivity of John saw himself presumably involves (abstract) movement of self onto the verb (yielding John self-saw him). If so, one would expect that free interpretation of himself is restricted to environments from which (abstract) syntactic movement of self is impossible. Such a restriction is precisely what appears to be involved (see Reinhart and Reuland (1991, 1993)). Consider also reflexives in Italian. In Italian, licensing reflexives involves a reflexive clitic (as discussed in Napoli (1979)). From our present perspective, it is not surprising then that a conditioning factor for a logophoric use of the anaphor (se) is that it occur in positions from which cliticization is impossible (Napoli, 1979; Reuland, 1990). What is common to all these cases is that a particular process which happens to be selected in a language for the syntactic encoding of an interpretive dependency is blocked in some set of environments for a general reason. It is in such environments that a logophoric interpretation is available for an anaphor. We are now ready to complete our investigation with a more detailed analysis of logophoricity.
5. WHAT MAKES A PRONOMINAL INTERPRETATION OF SIG LOGOPHORIC?13 The term "logophoric" is introduced in Hagege (1974). He uses it to characterize a class of pronouns that refer to the "auteur d'un discours" (the "source of a discourse" in the terms of Clements (1975)).
358
Eric Reuland
Both Hagege and Clements observe that the discourse function of logophoric pronouns is similar to that of what traditional grammarians called indirect reflexives, which occur in Latin and other languages, of which the use of Icelandic sig in (1) is an example. This "indirect" use of reflexives is not limited to Icelandic. It is, in fact, rather widespread over the languages of the world. Hagege observes that many languages have a formally distinct series of pronouns for this type of use (for instance, Mundang, Tuburi, and Ewe from the Niger-Congo family). These pronouns bear no formal resemblance to reflexives; hence, the term "indirect reflexive" would be inappropriate. It is for these that Hagege introduces the term "logophoric," setting them apart from the indirect reflexives. But the parallel between free anaphors and logophors is well illustrated by the following example from Ewe, discussed by Clements (1975): (25) Tsali gbl na-e be ye-e dyi ye gake ye-kpe dyi Tsalii say to-Pronj that Pronj beget LOGi but LOGi be victor 'Tsali told him (i.e., his father) that he begot him but he was the victor.' Here, LOG is the gloss for the logophoric pronoun ye.14 As Clements notes, only Tsali, the source of the reported discourse, can be the antecedent of ye. It is not necessary that the clause containing the logophoric pronoun be subjacent to the clause containing the antecedent. The logophoric pronoun may, in fact, occur at any depth of embedding. Whereas Hagege explicitly distinguishes free anaphors (indirect reflexives) from logophoric pronouns, Clements extends the notion of logophoricity so as to include free anaphors. For him, logophoricity is what characterizes both the phenomenon in (25) and that in (1), comprising uses of anaphors in languages such as Japanese, Icelandic, Italian, and English. Clements gives the following cross-linguistic characterization of logophoric pronouns (Clements, 1975,171-172): (i) logophoric pronouns are restricted to reportive contexts transmitting the words or thoughts of an individual or individuals other than the speaker narrator; (ii) the antecedent does not occur in the same reportive context as the logophoric pronoun; (iii) the antecedent designates the individual or individuals whose words or thoughts are transmitted in the reported context in which the logophoric pronoun occurs. Since logophoricity started receiving attention, research has been focusing on different aspects. There have been attempts (as in Sells, 1987) to pro-
Anaphors, Logophors, and Binding
359
vide a further understanding of the discourse properties of logophoricity. Another line of research has been addressing free anaphors. Specifically, there have been attempts to understand free anaphors from the perspective of the binding theory put forward in Chomsky (1981). It has also been debated whether Clements' proposal to group together logophoric pronouns of the Niger-Congo type with free anaphors can be maintained. For a recent contribution to this debate, see Frajzyngier (1997). From a syntactic and lexical perspective, logophoric pronouns and free anaphors pose very different theoretical problems. Logophoric pronouns have specific lexical properties, which impose a certain type of interpretation. For them, the question arises (i) how this type of interpretation can be characterized (and, perhaps, also how suprising it is to find elements with this type of interpretation); and (ii) why, in some languages, assigning this interpretation is still subject to syntactic licensing, even though it is lexically expressed. Free anaphors have no intrinsic lexical properties imposing a logophoric interpretation (since they do not have this interpretation in standard reflexive contexts). This raises the question of why they can be so interpreted at all. One question, however, applies to logophoric pronouns and free anaphors alike; How can this type of interpretation be characterized? For a proper perspective, let us first address the latter question. 5.1. The Nature of Logophoric Interpretation Any utterance is an event that happens at some particular time and place. This is virtually a triviality.15 However, many properties of utterances are independent of the particular time and place at which they were uttered. This is why for so many purposes an analysis of sentences abstracting away from these factors suffices. So, the interpretation of a sentence like John is an idiot can remain the same, regardless of whether it was uttered on January 10,11, or 12,1997. A sentence can be provided with expressions specifying values for time and place. These expressions may receive absolute interpretation, such as January 12,1997, or Room 0.19, Trans 10. However, the interpretation of other elements cannot be given independently of the context of utterance. Such relative interpretation is pervasive in the deictic system, be it local or temporal. So, the interpretation of John was an idiot yesterday is clearly affected when the utterance time is varied. Similarly, the interpretation of John saw an idiot over there is affected by varying utterance place. Of course, for the interpretation of many texts, actual time and location of production play a marginal role or none at all. Instead, values for time and place may be chosen that are distinct from the time and place of the utterance; they may be situated in the coordinate system of the actual world, but
360
Eric Reuland
also within some virtual coordinate system determined by the text itself. Let's call this the evaluation system, extending the standard notion of evaluation time (see for discussion of temporal structure the extensive literature on this subject, starting with Reichenbach (1947)).16 So, even if the interpretation domain is held constant, for certain items the interpretation will vary with the coordinates of a sentence in the evaluation system.17 Let's call these coordinates the position of a sentence. So, a certain type of context sensitivity reflects variation in position. The question is, then, whether all context sensitivity can be captured in terms of position. As is well known, the interpretation of personal pronouns shows that more is involved than position. Suppose that a saw b on January 12,1997, and a=John, b=Bill, but b did not see a on January 12 1997. On January 13, 1997, this verifies John saw Bill yesterday and falsifies Bill saw John yesterday. These arguments can also be expressed by personal pronouns such as / and you. Clearly, under the conditions sketched, it is possible for John to truthfully say to Bill / saw you yesterday. That is, I= John and you=Bill. However, it is equally possible that /= Bill and you= John. So, in this situation, Bill can truthfully say to John I didn't see you yesterday, but, of course, not I saw you yesterday. So, both a sentence and its negation can be truthfully uttered in one and the same position. Of course, from a common-sense perspective, all this is not surprising. The important point is that / and you are indexicals whose context sensitivity cannot be captured in terms of varying the utterance position, but which reflect a quite different aspect of the embedding of a sentence in a context. As trivial as the observation that any utterance is an event that happens at some particular time and place is the observation that any utterance is associated with a speaker who utters it. Similarly, if an utterance is perceived at all, it is associated with a person who perceives it. This implies that utterances have directionality. We can say that speaker and perceiver together define the orientation of an utterance in the world.18 Of course, the same caveats apply as with time and place. Just as actual time and place of an utterance must be generalized to the notion of an evaluation system, the notions of speaker and perceiver must be generalized so as to accommodate the possibility that they are positioned within some derived or virtual reality. Let's use source and target (allowing the possibility that source and target coincide) and continue to use the term orientation for what they contribute. So, for the system of temporal interpretation, utterance time, evaluation time, and event time must be distinguished, and for place interpretation corresponding distinctions must be made. Notice that there is no necessity for language systems to directly express these distinctions. Utterance time and evaluation time are generally not formally distinguished, nor are utterance place and evaluation place. However,
Anaphors, Logophors, and Binding
361
a distinction between evaluation time and event time is typically linguistically reflected in some way, just like a distinction between evaluation place and event place. Orientation may reflect the same type of systematic distinctions, namely, as utterance orientation, evaluation orientation, and event orientation. Just as in the case of position, utterance orientation and evaluation orientation mostly, or perhaps even generally, formally coincide. In many languages, there is no special system reflecting relativized event orientation either. My claim, however, is that logophoricity of the Niger-Congo type as described by Hagege and Clements is precisely that: a system linguistically expressing relativized event orientation. Hence, logophoricity is nothing mysterious. Making the interpretation of elements relative to a property which generalizes some aspect of the context of utterance is one of the common options among language systems. Logophoricity just employs this option with respect to the orientation of sentences. Logophoric pronouns are simply elements lexically marked as expressing event orientation. 5.2. Why Are Free Anaphors Interpreted Logophorically? What we have seen in Section 4 implies that free anaphors are essentially pronominal, i.e., potentially, they can refer independently. Their use, then, should be primarily explained in terms of discourse structure. It is well known that also in cross-sentential domains, where syntactic binding can play no role, the choice of an expression to refer to some object in the domain of discourse is not arbitrary. Ariel (1990) extensively discusses the conditions under which one and the same object is referred to by a full NP, an epithet, or a pronoun. Ariel establishes that there is an inverse relation between the descriptive content of a referring expression and what she calls the accessibility of the discourse referent. The more prominent an element is in the discourse, where prominence can, for instance, be measured in terms of textual distance, the more accessible it is in the theoretical sense Ariel develops. So, whenever the last mention of a discourse referent is sufficiently distant with respect to the mention to come, a full lexical NP or proper name will be used. If it is high on the scale of accessibility, a pronoun is preferred instead. Epithets are generally used for referents of intermediate accessibility. Since anaphors of the type discussed have even lower descriptive content than pronouns, Ariel's theory predicts that they require discourse referents of higher accessibility than pronouns do. The correctness of this prediction has been obscured by the fact that, in many languages, anaphors may not escape syntactic binding at all, and that the remaining cases were viewed as anomalies.19 The interpretation of free anaphors indicates that this predic-
362
Eric Reuland
tion is, in fact, correct. In the environments in which they occur, free anaphors take discourse referents that are, indeed, high on the scale of accessibility. That is, they refer to individuals within the domain of interpretation that can be characterized as being the center of consciousness, defining viewpoint, or as perspective holders. However, a bit more must be said. Although such individuals certainly figure prominently in the discourse, so far no independent explanation has been given as to why just these elements serve as discourse referents for free anaphors and why not just any forcefully foregrounded discourse entity. The following proposal sheds light on this issue. Note that, with respect to first- and second-person pronouns, it is uncontroversial that these, first and foremost express orientation. Suppose then, that the basic property of personal pronouns is to do precisely this. Whenever a pronoun has properties independent of reflecting orientation, this, following this line, should be due to other features it has. If such features are absent, as they are in the case of sig, se, etc., the only option for interpretation left should involve that property that is basic and, hence, unaffected, namely, orientation.20 Utterance and evaluation orientation, however, are typically expressed by first- and second-person pronouns. Hence, the only option left for a defective third-person pronoun is to reflect event orientation. This is what had to be explained. Such a view seems to correspond nicely with the intuition underlying the approach of Koopman and Sportiche (1989) to the pronominal system of Abe. As they note in their very detailed analysis, to which it is impossible to do full justice here, Abe has two pronominal classes, O-pronouns and n-pronouns. O-pronouns behave essentially like English pronominals. They respect condition B, and can be interpreted as bound variables, or referentially. N-pronouns also respect condition B, but they cannot be bound by either an R-expression or an O-pronoun, nor bind one themselves. At first sight, they allow a referential interpretation, but surprisingly, the number of indexings (taking these to reflect interpretations) available to them in a sentence is limited by the number of Comp nodes dominating them. For instance, two n-pronouns in a root clause are dominated by only one Comp node. Hence, they are obligatorily coindexed. Generalizing, in the case of n Comp nodes dominating n + i (occurrences of) n-pronouns, i n-pronouns will have to be coindexed with another n-pronoun. In clauses introduced by the complementizer kO (under reportive verbs), they are, however, coindexed with the NP representing the source of the report, regardless of whether this is an O-pronoun or R-expression. In this environment, they exhibit the standard features of logophoricity. The basis of the analysis in Koopman and Sportiche (1989) is their claim
Anaphors, Logophors, and Binding
363
that n-pronouns are what they call logical variables. In this, they say, they are similar to first- and second-person pronouns (n-pronouns are also characterized as + human). These authors discuss the status of logical variables in syntactic rather than semantic terms. A straightforward semantic characterization of what it means to be a logical variable can be based on precisely the idea arrived at earlier, namely, that they reflect event orientation. In reportive contexts, this gives us directly the consequence that they are identified with the source of the report. The behavior of n-pronouns in nonreportive contexts can be understood as follows. Koopman and Sportiche account for the relation between their interpretation and the number of dominating CPs by assuming that each C, being associated with an operator position, provides a possible A'-binder for the variable. One C, then, directly entails co-interpretation. The A'-binding part of the analysis is not essential, though. Let us assume that events are syntactically realized as CPs. Let us assume, furthermore, that the orientation of an utterance is not just defined by the pair source, target, but that, at least optionally, the person talked about can also be treated as a participant in the utterance, and thus be among the factors defining orientation, yielding a triple <source, target, 3rd>. Each event has, of necessity, precisely one orientation. Therefore, once the selection of this further participant has been made, enabling it to be referred to by the n-pronoun, this selection must be maintained for the whole CP Thus, the interpretation of the n-pronoun will be fixed to that one choice. In a complex sentence, with embedded expressions for events (CPs), each event will introduce a 3rd, yielding the pattern discussed. On the basis of the considerations discussed so far, the syntactic approach of Koopman and Sportiche and the semantic alternative sketched here may appear close to being notational variants. Although their analysis is more elaborate, a point in favor of the approach sketched here is that it gives independent content to the notion of logical variable. Further evidence bearing on the choice may be found by more carefully investigating the syntactic realizations of events, on the one hand, and the interaction between other properties of C and the licensing of n-pronouns on the other.
6. SUMMARY AND CONCLUSION The following picture has emerged. Simplex anaphors like Icelandic sig, but also its counterparts in other languages, can be interpreted like pronominals when they are free. There is no intrinsic necessity for them to be syntactically bound. Where anaphors must be bound, this is the result of an economy condition favoring operations applying within a module over
364
Eric Reuland
crossmodular operations. First- and second-person pronouns intrinsically reflect the orientation of an utterance. Third-person pronouns may do so as well in the absence of features that are able to fix their reference independently (such as the grammatical number feature). That is why pronouns that are impoverished in features are used logophorically just as first- and second-person pronouns. Logophoric use of third-person anaphors thus reflects their event orientation. Unlike the logophoric use of specialized pronouns of the type discussed by Hagege and Clements, the logophoric use of "indirect reflexives" can be considered a default interpretation.
ACKNOWLEDGEMENTS The research reported on in this article is part of an ongoing project on anaphoric relations of the Utrecht institute of Linguistics OTS. I am indebted to many people involved in this project. I am particularly grateful to Peter Coopmans, Denis Delfitto, Martin Everaert, Bill Philip, Tanya Reinhart, and Sigridur Sigurjonsdottir for sharing their insights with me, and to an anonymous reviewer for useful comments. Special thanks go to Martin Everaert for commenting on the prefinal version of this article. Needless to say that I alone am responsible for any errors. Some of the ideas reported in this article were developed during a stay at NIAS during the fall semester of 1996.I would like to express my gratitude to NIAS and in particular to the members of the SynCom theme group for providing such a stimulating research environment. Finally, I would like to thank Peter Cole, Gaby Hermon, and Jim Huang who invited me to join this volume and showed such a tremendous patience.
NOTES l
For expository reasons, I will follow common practice by using indices to represent interpretive dependencies of various kinds. However, the reader should attach no theoretical significance to this device. 2 Note that sig in Icelandic does not have a nominative form (see Everaert, 1990, for discussion). Hence, sig can occur in subject position only with those verbs that select a nonnominative subject, i.e., with the so-called "quirky" case verbs in Icelandic. The verb vanta "to lack, need" which appears in example (2a) is one of these verbs and takes an accusative subject. Quirky subjects in Icelandic have been discussed by a number of authors; see, for example, Andrews (1976), Thrainsson (1979), Bernodusson (1982), Zaenen, Maling, and Thrainsson (1985), and Sigurdsson (1989, 1992).
Anaphors, Logophors, and Binding
365
3
Note that there are two NPs in this sentence whose perspective or point of view are being reported, i.e., Jon and Bjorn. Hence,sig could also take Bjorn as its antecedent. 4 Note that objects generally do not qualify as holders of the perspective of a sentence, and hence, they rarely qualify as antecedents for sig in this type of context (see Sigurdsson, 1990:334). 5 Huang and Tang (1991) argue that, in Chinese, the source of F-features need not coincide with the actual antecedent. For present purposes, this complication can be ignored. 6 As discussed in Huang and Tang (1991), Chinese ziji can be bound by a subcommanding antecedent. The requirement expressed by subcommand is strikingly similar to that holding for pronominal binding in (8) and the examples of (9). Further research is required to explain the nature of this similarity. 7 It is interesting to note that Icelandic sig behaves quite similarly to Dutch pronominals in this respect. Whereas, in general, variable binding does not require strict c-command, in a context such as (12), it does. This is illustrated in the contrast between (i) and (ii). (i)
Johns mening is dat het hem aan talent ontbreekt en dat is ook John's opinion is that it lacks him in talent, and that is also Peters mening Peter's opinion a. Peter's opinion is that Peter lacks talents b. = Peter's opinion is that John lacks talents
(ii) John vindt dat het hem aan talent ontbreekt en Peter vindt dat ook John thinks that it lacks him in talent, and Peter thinks so too a. = Peter thinks that Peter lacks talents b. = Peter thinks that John lacks talents Although at this point I have nothing to say as to why variable binding requires strict c-command in cases like these, the pattern supports that thesis that logophoric sig behaves as a pronominal. 8 It should be noted that locally bound sig does not allow a strict reading. This is illustrated in (i): (i)
Joni rakadi sigi og Peturj geroibao lika John shaved SIG and Peter did so too ( Peter shaved John) Yet, in the long-distance infinitive case, both readings are possible.
(ii) Joni skipaoi professornum j [ao PRO. fella inf sigi a profinu] og Ari John ordered the professor to fail SIG on the test and Ari geroi pao lika did so too a. = Ari ordered the professor to fail Ari on the test b. = Ari ordered the professor to fail John on the test This may indicate that what forces the sloppy reading in (i) is not a property of the antecedent-anaphor relation, but a property of the predicate. In (i), the copied pred-
366
Eric Reuland
icate is intrinsically reflexive, whereas (ii) has no reflexive predicate (see Sigurjonsdottir (1992) for discussion of intrinsic reflexives in Icelandic). 9 I will not engage in a discussion of what precisely the theory would predict, since the status of the operator is unclear. In any case, the effect seems to have the strength of a WCO violation rather than a condition C-violation. 10 An anonymous reviewer raises the question of why one doesn't get a trace instead of sig, or the same Case on sig and its antecedent. As already argued in Reinhart and Reuland (1993), there is no general need to stipulate that the tail of a chain is empty or that a chain has only one Case. The idea that dependencies are, in principle, transitive (restricted by intrinsic properties of the terms involved) is essentially an elaboration of the notion of chain linking developed in Chomsky (1995). In the implementation of chain formation adopted here from Reuland (1997a), nothing in the operation of chain linking requires the tail to be empty or the chain to have only one Case. What is assumed is only that one may link a pair of dependencies sharing the same middle term. Another question by the reviewer concerns the status of quirky subjects, wondering how they are related to 'Agr' when they serve as antecedents. The relation of a subject to Agr, however, is not crucial. Quirky subjects, just like other subjects, are in a checking relation with the V/I complex, as they have to check the EPP feature. Suppose, we have a situation where the j-features of the verb have already been checked and erased by a VP-internal nominative. This would not change the fact the j-features from any other constituent moved onto the V/I complex would still be in a checking configuration with respect to the DP in the specifier position. Hence, a dependency can be established. 11 What about complex anaphors (English himself, Dutch zichzelf, Icelandic sjalfan sig)? Limiting ourselves to essentials (for details see Reuland, 1997a), the picture is as follows. Suppose we have a verb that is not only syntactically, but also semantically transitive, as in (i): (i)
* Cindy haat zich Cindy hates SE
If chain formation applies, a 2-place predicate has only one argument, which is sure to violate plausible interpretive requirements. Suppose, we add zelf, as in (ii), and allow chain formation to take place between zich and Cindy: (ii) CindyF haatF [t zichF [dzelf]] Cindy hates herself The arguments of haat are not Cindy and zich, but Cindy and t. These two do not form a chain; hence, the arity of the predicate is respected. What is the semantic effect of zelf? Well, in terms of value assignment in a model, the value of X zelf does, in general, not greatly differ from the value of X. So, for binding theoretic purposes, X zelf will do quite well as a stand-in for zich, as required. How, on the other hand, is binding enforced? Suppose, that zelf, as the head-noun of a DP in argument position, has insufficient lexical content to project an argument. One effect is that it will not be able to 0-mark its specifier zich, the other that it will not be able
Anaphors, Logophors, and Binding
367
to receive a -role from its governing verb. However, it may (covertly) incorporate into the verb, absorbing the relevant -role. If so, zich would seem to be left dangling, but given that it has formed a chain with the subject Cindy, its interpretation is also taken care of. Again, we have a binding relation that does not depend on any form of indexing. Note that ze//-incorporation involves a local relation with a predicate (by the Head Movement Constraint, or whatever principle of CHL derives it). Hence, the locality effects on SELF-anaphors follow. 12 More formally, in order to effectively enforce this conclusion we must explicitly assume that:(i) if x (here, the subjunctive) is part of g (i.e., a sublable of g), licensing £ (by attracting x involves obligatory pied-piping of all formal features of g (here, the V/I complex); (ii) subparts of traces are invisible to syntactic operations, in particular, the chain linking operation. Both have independent motivation. Point (i) is in line with the position argued for in Chomsky (1995) with respect to feature pied-piping; (ii) seems at least independently necessary, though at this point, it is not entirely clear to us whether it can be derived or must be stipulated. Note that this approach requires that"V2"-phenomena in Germanic (V/I to C) fall under the PF part of the grammatical system, in line with Chomsky (1995). Otherwise, the same effect should obtain in V2-clauses as in subjunctives. 13 Much of the following discussion is taken from Reuland (1997b). l4 Ye's nonhigh tone, represented by the grave accent, distinguishes it from the nonlogophoric strong form ye which bears high tone; the logophoric pronoun is also distinct from a reflexive; in Ewe, these are formed by affixing a genetive form of a pronoun to the noun d,okui 'self,' and, according to Clements, behave much like their English counterparts. 15 It would be completely trivial, were it not for the fact that where there is a significant delay between production and reception of a text occasionally, significance should be attached to time and place of reception, rather than to time and place of production. 16 And further references cited in Reuland (1997b). 17 This implies that we distinguish between the model and the space-time in which it is embedded. Clearly, formally one might collapse the distinction. However, there is independent evidence from behavior with respect to quantification that expressions of space-time are different from other argument expressions (see the contrast between John walked five kilometers versus ??/o/m walked every kilometer). They become part of the model only after anchoring. See De Jong (1987) and Reuland and Ter Meulen (1987) for discussion. 18 For present purposes, an informal notion suffices. 19 A reviewer suggests that there is no real contrast here, since anaphors have "high accessibility" referents anyway, namely, local binders. However, we cannot refer to a syntactic notion such as local binding to define the notion of a high accessibility antecedent. The notion of a high accessibility antecedent should be given in independent terms. Otherwise, the theory would become vacuous. 20 Note that the property of being able to bear an orientation feature should be distinguished from the property of actually bearing a feature such as source or target expressing a particular orientation. The claim is that the former property is basic and
368
Eric Reuland
remains active even in the absence of features actually expressing a particular orientation.
REFERENCES Anagnostopoulou, E. and Everaert, M. (1998). Towards a more complete typology of anaphoric expressions. Linguistic Inquiry 30 (1), 97-118 Andrews, A. (1976). The VP complement analysis in modern Icelandic. In Proceedings of NELS 6, GLSA, pp. 1-21. University of Massachusetts at Amherst. Ariel, M. (1990). Accessing Noun-Phrase Antecedents. London: Croom Helm. Bernodusson, H. (1982). Opersonulegar setningar. Master's thesis, University of Iceland, Reykjavik. Bouchard, D. (1984). On the Content of Empty Categories. Dordrecht: Foris. Burzio, L. (1991). The morphological basis of anaphora. Journal of Linguistics 27, 81-105 Chierchia, G. (1995). Dynamics of Meaning, Anaphora, Presupposition, and the Theory of Grammar. Chicago: Chicago University Press. Chomsky, N. (1981). Lectures on Government and Binding. Dordrecht: Foris. Chomsky, N. (1986). Knowledge of Language: Its Nature, Origin, and Use. New York: Praeger. Chomsky, N. (1995). The Minimalist Program. MIT Press, Cambridge, MA. Clements, G. N. (1975). The logophoric pronoun in Ewe: Its role in discourse. Journal of West African Languages 10,141-177. Everaert, M. (1986). The Syntax of Reflexivization. Dordrecht: Foris. Everaert, M. (1990). Nominative ananphors in Icelandic: Morphology or syntax? In Issues in Germanic Syntax (W. Abraham, W. Kosmeijer, and E.J. Reuland, eds.), pp. 277-307. Berlin: Mouton de Gruyter. Frajzyngier, Z. (1997). Pronouns and agreement: Systems interaction in the coding of reference. In Atomism and Binding (H. J. Bennis, P. Pica, and J. Rooryck, eds.), pp. 115-141. Dordrecht: Foris. Grodzinsky, Y., and Reinhart, T. (1993). The innateness of binding and coreference. Linguistic Inquiry 24,69-101 Hagege, C. (1974). Les pronoms logophoriques. Bulletin de la Societe de Linguistique de Paris 69,287-310 Hellan, L. (1988). Anaphora in Norwegian and the Theory of Grammar. Dordrecht: Foris. Hyams, N., and Sigurjonsdottir, S. (1990). The development of "long distance anaphora": Cross-linguistic comparison with special reference to Icelandic. Language Acquisition 1,57-93. Huang, C.-T. J., and Tang, C.-C. J. (1991). The local nature of the long-distance reflexive in Chinese. In Long Distance Anaphora, (J. Koster and E.J. Reuland, eds.), pp. 263-283. Cambridge: Cambridge University Press. Jong,F. M. G. de (1987). The compositional nature of (in)definiteness. In The Repre-
Anaphors, Logophors, and Binding
369
sentation of (In)definiteness, (E. J. Reuland and A. G. B. ter Meulen, eds.), pp. 270-286. Cambridge, MA: MIT Press. Kayne, R. (1994). The Antisymmetry of Syntax. Cambridge, MA: MIT Press. Kempchinsky. P. (1986). Romance subjunctive clauses and logical form. Unpublished doctoral dissertation. UCLA, Los Angeles. Kempchinsky, P. (1995). From the lexicon to the syntax: The problem of subjunctive clauses. In Evolution and Revolution in Linguistic Theory (H. Campos and P. Kempchinsky, eds.), pp. 228-250. Washington, DC: Georgetown University Press. Koopman, H., and Sportiche, D. (1989). Pronouns, logical variables and logophoricity in Abe. Linguistic Inquiry 20,555-589 Maling, J. (1984). Non-clause-bounded reflexives in modern Icelandic. Linguistics and Philosophy 7,211-241. Maling, J. (1986). Clause-bounded reflexives in modern Icelandic. In Topics in Scandinavian Syntax, (L. Hellan and K. K. Christensen, eds.), pp. 53-63. Dordrecht: Reidel. Manzini, M. R. (1993). The subjunctive. Paris 8 Working Papers, Vol. 1. Napoli, D. J. (1979). Reflexivization across clause boundaries in Italian. Journal of Linguistics 15,1-28 Quer, J. (1998). Mood at the Interface. [LOT Dissertation 1]. Dordrecht: Holland Academic Graphics. Reichenbach, H. (1947). Elements of Symbolic Logic. London and New York: Macmillan. Reinhart, T. (1983). Anaphora and Semantic Interpretation. London & Sydney: Croom Helm. Reinhart, T, and Reuland, E. J. (1991). Anaphors and logophors: An argument structure perspective. In Long Distance Anaphora, (J. Koster and E. J. Reuland, eds.), pp. 283-321. Cambridge: Cambridge University Press. Reinhart, T, and Reuland, E. J. (1993). Reflexivity. Linguistic Inquiry 24,657-720. Reuland, E. J. (1990). Reflexives and beyond: Non-local anaphora in Italian revisited. In Grammar in Progress (J. Mascaro and M. Nespor, eds.), pp. 351-361. Dordrecht: Foris. Reuland, E. J. (1996). Pronouns and features. In Proceedings of NELS 26,319-335. GLSA, University of Massachusetts at Amherst. Reuland, E. J. (1997a). Primitives of binding. Paper presented at GLOW 1995, Tromso, Norway. UiL OTS Working paper. Utrecht University, Utrecht. Reuland, E. J. (1997b). Logophoricity as orientation. In Yearbook 1997, (Jan Don and Ted Sanders, eds.), pp. 71-83. Utrecht Institute of Linguistics OTS, Utrecht University. Reuland, E. J. (1998). Structural conditions on chains and binding. In Proceedings of NELS 28, 341-356. GLSA, University of Massachusetts at Amherst, Amherst. Reuland, E. J. and Meulen, A. G. B. ter (1987) Introduction. In The Representation of (In)definiteness, (E. J. Reuland and A. G. B. ter Meulen, eds.), pp. 1-20, Cambridge, MA: MIT Press. Reuland, E. J. and Sigurjonsdottir S., (1997). Long-distance "binding" in Icelandic:
370
Eric Reuland
Syntax or discourse. In Atomism and Binding (H. J. Bennis, P. Pica, and J. Rooryck, eds.), pp. 323-341. Dordrecht: Foris. Rognvaldsson, E. (1986). Some comments on reflexivization in Icelandic. In Topics in Scandinavian Syntax, (L. Hellan and K. K. Christensen, eds.), pp. 89-102. Dordrecht: Reidel. Sells, P. (1987). Aspects of logophoricity, Linguistic Inquiry 18,445-479. Sigurosson, H. A. (1989). Verbal Syntax and Case in Icelandic. Doctoral dissertation, University of Lund, Sweden. Sigurdsson, H. A. (1990). Long distance reflexives and moods in Icelandic. In Modern Icelandic Syntax, (J. Maling and A. Zaenen, eds.), pp. 309-346. New York: Academic Press. Sigurdsson, H. A. (1992). The case of quirky subjects. In Working Papers in Scandinavian Syntax 49,1-26. Sigurjonsdottir, S. (1992). Binding in Icelandic: Evidence from Language Acquisition. Doctoral dissertation, University of California, Los Angeles. [Published in 1993 as UCLA Working Papers in Psycholinguistics 2.1. UCLA.] Thrainsson, H. (1976). Reflexives and subjunctives in Icelandic. In Proceedings of NELS 6,225-239. GLSA, University of Massachusetts at Amherst, Amherst. Thrainsson, H. (1979). On Complementation in Icelandic. New York: Garland. Thrainsson, H. (1990). A semantic reflexive in Icelandic. In Modern Icelandic Syntax (J. Maling and A. Zaenen, eds.), pp. 289-307. New York: Academic Press. Thrainsson, H. (1991). Long distance reflexives and the typology of NPs. In Long-Distance Anaphora (J. Koster and E. J. Reuland, eds.), pp. 49-75. Cambridge: Cambridge University Press. Zaenen, A., Maling, J., and Thrainsson, H. (1985). Case and grammatical functions: The Icelandic passive. Natural Language and Linguistic Theory 3,441-483.
INDEX
A
Agreement, 55-56 Anaphors anti-antilocality, 232-234 form of, Hindi/Urdu, long-distance anaphors, 53-55 conditions on, 53-55 logophors, binding, 343-370 free anaphors logophorical interpretation, 361-363 syntactic conditions on, 356-357 interpreting sig, 346-347 interpretive process, 350-357 logophoric interpretation, nature of, 359-361 non-c-commanding antecedents, 348-350 pronominal binding, 354-356 pronominal elements, interpreting, 351-354 pronominal interpretation of sig, logophoric, 357-363 long-distance, Hindi/Urdu, 47-82 Animacy, 268 Animate head, subcommand, 7 Antecedent as internal source, logophoric conditions, 5 logophoric conditions as internal self, 6 as internal source, Chinese, 5 neither self nor source, Chinese, 4 Anti-antilocality, 227-254 371
anaphors, semantic distinctions between, 232-234 binding theory, predicate-centered, 246-250 Chinese, 238-239 coargument-bound anaphors, as nearreflexives, 237-240 English, 238 hybrid pronoun, 242-246 Kannada, 227 anaphora in, 229-232 lexical reflexivity, variation in, 240-242 pure-reflexivity, vs. near-reflexivity, 234-236 Russian, 239-240 Antimacy condition, 69-70 Aspect, local binding domain, 65-67 Attitudes, 141-195 binding by internal source, self, 158-159 blocking effects, 142,161-165 consciousness, 159-160 and perspectivity, 179-182 direct discourse, attitudes de se, 154-157 extensions, 179-182 formal analyses, 144-147 functional analyses, 147-150 governing category, 168-170 local binding, 166-168 long distance ziji, 174-184 monomorphemicity, 142 nonuniform approaches, 150-154 perspectivity, 179-182 semantics, 174-176
372 sentence-free ziji, referring to speaker, 157 subcommanding antecedent, 142 subcommanding antecedents, 170-172 subject-orientation, 142 syntax, 176-179 ziji as logophor, 154-165 ziji as (syntactic) anaphor, 166-174 Attitudes de se, 154-157 logophoricity, 16-19 semantic, syntactic analysis, 19-23 B
Baker theory, 324 Bare reflexive, 216-217 Beliefs de re, de se, de dicto, 294-296 Binding domains, 61-65 Binding relations, 52-56 Binding theory, 246-250 Blocking effect, 279-316 attitudes, ziji at interface, logophoricity, 142 beliefs de re, de se, de dicto, 294-296 Chechen, Ingush, long-distance reflexivation in, 267-268 closeness, potential binders, 287-289 feature compatibility checking, 286-287 Hindi/Urdu, long-distance anaphors, 55-56 agreement/phi features, role of, 60-61 logophoricity attitudes, ziji at interface, 161-165 perspectivity, 289-293 long-distance bound ziji, not logophoric pronoun, 290-292 Mandarin reflexives, 293-294,294 not symmetrical, 282-286 self-ascription long-distance bound ziji, 293-305 ziji carrier of belief, 296-297 condition for, 297-305 C
C-command, 89-91 Causative, 62 Chained adjunct clauses, 263-264
Index
Chechen,255-276 animacy, 268 blocking effects, 267-268 chained adjunct clauses, 263-264 complement clauses, 264-265 conditions, 267-268 intervening different subject, 266 local reflexivization, 257-259,274 logophoric reflexivization, 259-261 main-clause antecedent, 273-274 modality, 267 multiple anaphora, conditions on, 268-274 multiple long distance reflexive, conditions on, 268-274 no upwards long distance reflexive, 266 non-main clause, long distance reflexive from, 265-266 ordinary long-distance reflexivization, 261-274 person hierarchy, 267 person interaction, 269-272 relative clauses, 265 sentence-bounded, 261-262 serialization, 262-263 Chinese dialects, 1-46 approaches to long distance reflexives, 1-4 binding theoretical status, long distance reflexives, 23-25 blocking effect, 28-35 c-command requirement, 25-26 informed subcommand, with animate head, 7 logophoric conditions antecedent as internal self, 6 as internal source, 5 neither self nor source, 4 on long distance reflexives, Mandarin, Teochew, 4-6 logophoricity, attitudes de se, 16-19 modifying phrases, 8-10 pivot restrictions, 11-16 subcommanding antecedents, of long distance reflexives, source/self requirements on antecedents, 6-8 syntactic, semantic analysis, of attitudes de se, 19-23 VP ellipsis, 26-28 Closeness, potential binders, 287-289 Coargument-bound anaphors, as near-reflex-
373
Index ives, 237-240 Compatibility checking, 286-287 Complement clauses, 264-265 Conditions on domains, 53-55 Conjunctive operator, 83-117 Consciousness, 159-160 and perspectivity, 179-182 D
De dicto, 294-296 De se, 19-23 Direct discourse, 154-157 Divergence of binding, 54-55 Domains conditions on, Hindi/Urdu, long-distance anaphors, 53-55 functional categories, binding possibilities, 62 E
Ellipsis, VP, 26-28 English anti-antilocality, 238 long-distance reflexives, syntactic, nonsyntactic constraints on, 326-328 reflexive pronouns, Baker theory, long-distance reflexives, syntactic, nonsyntactic constraints on, 324 reflexives, binding for, 326-328 Extensions, 179-182 F
Finite tense, 62 Free anaphors, syntactic conditions on, 356-357 Functional categories domains, 62 Hindi/Urdu, 62 Functional heads, 61-65 G
Governing category, 168-170
H
Head movement constraint, 71-73 Hindi/Urdu, long-distance anaphors, 47-82 agreement, role of, 55-56 anaphors, form of, 53-55 conditions on, 53-55 antimacy condition, 69-70 aspect, LF movement to, 57-59 binding domains, 61-65 binding relations, 52-56 blocking effect, 55-56 agreement/phi features, role of, 60-61 causative, 62 divergence of binding, agreement, 54-55 domains conditions on, 53-55 functional categories, binding possibilities, 62 finite tense, 62 head movement constraint, on LF movement, 71-73 inflectional categories, 50-51 inflectional morphology, lexical forms, 48-52 local binding, 53 local complementarity, 53 long-distance, 56-60 nonfinite aspect, 62 nonfinite binding domains, 54 nonfinite clause, with aspect, 63 nonfinite tense, 62 nonlocally bound anaphoric expressions, 47-48 noun phrase, no local binding, 64-65 object agreement, subject orientation, 77-78 pronoun-anaphor, 49 pronouns, and reflexives, 49-50 reciprocals, 49 reflexive antecedents, verbal inflection, interaction of, 51 reflexive movement, 73-75 reflexives, 49-50 functional projection host for, 60-68 small clause, 62 with no tense, aspect, 63-64 subject orientation, 53 syntactic/semantic factors, 68-75 tense, LF movement to, 57-59
374
Index antecedent, logophoric conditions, 5 binding by, attitudes, ziji at interface, logophoricity, self, 158-159
tense/aspect, 65-67 verb inflection, 50 verbal tense, agreement inflection, 50 XP reflexives, 70-71 subject orientation of, 71 Hybrid pronoun, 242-246
K
I
Kannada, 227 anaphora in, 229-232
Indonesian, Riau sendiri c-command, would-be antecedent, 89-91 conjunctive operator, 83-117, 98-106 long-distance reflexives, 83-85 nonreflexive interpretations, 92-98 as reflexive anaphor, 88-98 reflexive anaphor, 83-117 Inflectional categories, 50-51 Inflectional morphology, 48-52 lexical forms, Hindi/Urdu, long-distance anaphors, 48-52 Ingush, 255-276 animacy, 268 blocking effects, 267-268 chained adjunct clauses, 263-264 complement clauses, 264-265 conditions, 267-268 intervening different subject, 266 local reflexivization, 257-259,274 logophoric reflexivization, 259-261 main-clause antecedent, 273-274 modality, 267 multiple anaphora, conditions on, 268-274 multiple long distance reflexive, conditions on, 268-274 no upwards long distance reflexive, 266 non-main clause, long distance reflexive from, 265-266 ordinary long-distance reflexivization, 261-274 person hierarchy, 267 person interaction, 269-272 relative clauses, 265 sentence-bounded, 261-262 serialization, 262-263 Intensification, 333-335 Internal self, 6 Internal source
L Local binding Hindi/Urdu, long-distance anaphors, 53 logophoricity, attitudes, ziji at interface, 166-168 Local complementarity, 53 Local reflexivization, 257-259, 274 Logophoric reflexivization, 259-261 Logophoricity attitudes, ziji at interface, 141-195 binding by internal source, self, 158-159 blocking effects, 142, 161-165 consciousness, 159-160 and perspectivity, 179-182 direct discourse, attitudes de se and, 154-157 extensions, 179-182 formal analyses, 144-147 functional analyses, 147-150 governing category, 168-170 local binding, 166-168 long distance ziji, 174-184 monomorphemicity, 142 nonuniform approaches, 150-154 perspectivity, 179-182 semantics, 174-176 sentence-free ziji, referring to speaker, 157 subcommanding antecedent, 142 subcommanding antecedents, 170-172 subject-orientation, 142 syntax, 176-179 ziji as logophor, 154-165 ziji as (syntactic) anaphor, 166-174 long-distance reflexives, syntactic, nonsyntactic constraints on, 329-333 perspectivity, blocking effect, 289-293 Logophors, 343-370
Index
free anaphors logophorical interpretation, 361-363 syntactic conditions on, 356-357 interpreting sig, 346-347 interpretive process, 350-357 logophoric interpretation, nature of, 359-361 non-c-commanding antecedents, 348-350 pronominal binding, 354-356 pronominal elements, interpreting, 351-354 pronominal interpretation of sig, logophoric, 357-363 Long-distance anaphors, 47-82 agreement, role of, 55-56 anaphors, form of, 53-55 conditions on, 53-55 "antimacy" condition, 69-70 aspect, LF movement to, 57-59 binding domains, 61-65 binding relations, 52-56 blocking effect, 55-56 agreement/phi features, role of, 60-61 causative, 62 divergence of binding, agreement, 54-55 domains conditions on, 53-55 functional categories, binding possibilities, 62 finite tense, 62 head movement constraint, on LF movement, 71-73 inflectional categories, 50-51 inflectional morphology, lexical forms, 48-52 local binding, 53 local complementarity, 53 long-distance, 56-60 nonfinite aspect, 62 nonfinite binding domains, 54 nonfinite clause, with aspect, 63 nonfinite tense, 62 nonlocally bound anaphoric expressions, 47-8 noun phrase, no local binding, 64-65 object agreement, subject orientation, 77-78 pronoun-anaphor, 49 pronouns, and reflexives, 49-50 reciprocals, 49
375 reflexive antecedents, verbal inflection, interaction of, 51 reflexive movement, 73-75 reflexives, 49-50 functional projection host for, 60-68 small clause, 62 with no tense, aspect, 63-64 subject orientation, 53 syntactic/semantic factors, 68-75 tense, LF movement to, 57-59 tense/aspect, distinguishing from agreement, as host head, local binding domain, 65-67 verb inflection, 50 verbal tense, agreement inflection, 50 XP reflexives, 70-71 subject orientation of, 71 Long-distance bound ziji, 290-292 Long-distance interpretation, 104-106 Long-distance reflexives binding theoretical status, 23-25 logophoric conditions, Mandarin, Teochew, 4-6 Riau Indonesian, sendiri, 83-85 subcommanding antecedents, source/self requirements on antecedents, 6-8 syntactic, nonsyntactic constraints on, 317-342 contrast, intensification, discourse prominence, 333-335 English reflexive pronouns, Baker theory, 324 logophoricity, 329-333 nonsyntactic conditions on ziji, 328-335 nonsyntactic reflexives, 320-328 obliqueness hierarchy, 322 Turkish, 197-226 bare, 212-217 inflected reflexive contrasting, overt pronoun, 213-216 binding features, underdetermined, 205-206 complex reflexives, 217-218 inflected reflexive, 204-209 binding domain of, 202-203 narrative, standard styles, differences between, 212-213 nature of inflected reflexives, 201-205 phrasal analysis, 209-212
376 possessor raiser, bare reflexive, 216-217 properties of, 203 types of morphologically free reflexives, 197-199 Long distance ziji, 174-184
M
Main-clause antecedent, 273-274 Mandarin, 293-294 logophoric conditions, 4-6 Modality, 267 Monomorphemicity, 142
N Nonfinite aspect, 62 Nonfinite binding domains, 54 Nonfinite clause, 63 Nonfinite tense, 62 Nonlocally bound anaphoric expressions, 47-8 Nonreflexive interpretations, 92-98 Nonuniform approaches, 150-154 Norwegian child language, syntactic, vs logophoric binding, 119-140 Avrutin, Cunningham, 126-128 local, vs. nonlocal logophoric binding, 120-126 non-coargument theory, 120-123 possessive reflexives, reflexives in locative PPs, 129-134 reflexive-movement theory, 123-126 reflexives in picture NPS, 134-136 seg in locative PP conditions, 131 subjects, 129 within-subjects comparison, 136-138
Index Person hierarchy, 267 Person interaction, 269-272 Perspectivity, 179-182 Phi features, 60-61 Possessor raiser, 216-217 Projection host for reflexives, 60-68 Pronominal elements, 351-354 Pronouns, Hindi/Urdu, long-distance anaphors, and reflexives, 49-50
R Reciprocals, 49 Reflexive. See also Long-distance reflexive anaphor, Riau Indonesian, sendiri, 83-117,88-98 antecedents, verbal inflection, interaction of, Hindi/Urdu, long-distance anaphors, 51 Hindi/Urdu, long-distance anaphors, 49-50 functional projection host for, 60-68 movement, Hindi/Urdu, long-distance anaphors, 73-75 Relative clauses, 265 Riau Indonesian sendiri c-command, would-be antecedent, 89-91 conjunctive operator, 83-117,98-106 long-distance reflexives, 83-85 no long-distance interpretation, 104-106 nonreflexive interpretations, 92-98 as reflexive anaphor, 88-98 reflexive anaphor, 83-117 unified semantic analysis, 98-104 Russian, 239-240
S
O Object agreement, 77-78 Obliqueness hierarchy, 322
P
Pan, 182-184
Self internal, antecedent, logophoric conditions, 6 modifying phrases as tests for, 8-10 role of, additional evidence, 10-11 Self-ascription long-distance bound ziji, 293-305 ziji
377
Index carrier of belief, 296-297 condition for, 297-305 Self requirements on antecedents, 6-8 Semantics logophoricity, attitudes, ziji at interface, 174-176 syntactic analysis, of attitudes de se, 19-23 Sendiri, Riau Indonesian c-command, would-be antecedent, 89-91 conjunctive operator, 83-117 long-distance reflexives, 83-85 no long-distance interpretation, 104-106 nonreflexive interpretations, 92-98 as reflexive anaphor, 88-98 reflexive anaphor, 83-117 unified semantic analysis, 98-104 Serialization, 262-263 Small clause, 62 with no tense, aspect, 63-64 Source internal, antecedent, logophoric conditions, 5 requirements on antecedents, long distance reflexives, subcommanding antecedents, 6-8 Source/self, 8-10 Subcommanding antecedent, 142 Subcommanding antecedents, logophoricity, attitudes, ziji at interface, 170-172 informed, with animate head, 7 Subject, different, intervening, Chechen, Ingush, 266 Subject orientation Hindi/Urdu, long-distance anaphors, 53 logophoricity, attitudes, ziji at interface, 142 object agreement, Hindi/Urdu, long-distance anaphors, 77-78 Syntactic reflexives, 320-328
T
Tense distinguishing from agreement, as host head, local binding domain, Hindi/Urdu, long-distance anaphors, 65-67
LF movement to, Hindi/Urdu, long-distance anaphors, 57-59 Teochew, 4-6 Turkish, 197-226 bare, inflected reflexive contrasting, overt pronoun, 213-216 bare reflexive, 212-217 binding features, underdetermined, 205-206 complex reflexives, 217-218 inflected reflexive, 204-209 binding domain of, 202-203 narrative, standard styles, differences between, 212-213 nature of inflected reflexives, 201-205 phrasal analysis, 209-212 possessor raiser, bare reflexive, 216-217 properties of, 203 types of morphologically free reflexives, 197-199
U
Urdu Hindi, long-distance anaphors, 47-82 agreement, role of, 55-56 anaphors, form of, 53-55 conditions on, 53-55 "antimacy" condition, 69-70 aspect, LF movement to, 57-59 binding domains, 61-65 binding relations, 52-56 blocking effect, 55-56 agreement/phi features, role of, 60-61 causative, 62 divergence of binding, agreement, 54-55 domains conditions on, 53-55 functional categories, binding possibilities, 62 finite tense, 62 head movement constraint, on LF movement, 71-73 inflectional categories, 50-51 inflectional morphology, lexical forms, 48-52 local binding, 53 local complementarity, 53 long-distance, 56-60
378 nonfinite aspect, 62 nonfinite binding domains, 54 nonfinite clause, with aspect, 63 nonfinite tense, 62 nonlocally bound anaphoric expressions, 47-48 noun phrase, no local binding, 64-65 object agreement, subject orientation, 77-78 pronoun-anaphor, 49 pronouns, and reflexives, 49-50 reciprocals, 49 reflexive antecedents, verbal inflection, interaction of, 51 reflexive movement, 73-75 reflexives, 49-50 functional projection host for, 60-68 small clause, 62 with no tense, aspect, 63-64 subject orientation, 53 syntactic/semantic factors, 68-75 tense, LF movement to, 57-59 tense/aspect, distinguishing from agreement, as host head, local binding domain, 65-67 verb inflection, 50 verbal tense, agreement inflection, 50 XP reflexives, 70-71 subject orientation of, 71 long-distance anaphors, 47-82 V
Verb inflection Hindi/Urdu, long-distance anaphors, 50
Index reflexive antecedents, interaction of, Hindi/Urdu, long-distance anaphors, 51 Verbal tense, 50
Z
Ziji, at interface, 141-195 binding by internal source, self, 158-159 blocking effects, 142,161-165 consciousness, 159-160,179-182 direct discourse, attitudes de se and, 154-157 extensions, 179-182 formal analyses, 144-147 functional analyses, 147-150 governing category, 168-170 local binding, 166-168 logophor, 154-165 long distance ziji, 174-184 monomorphemicity, 142 nonuniform approaches, 150-154 perspectivity, 179-182 previous analyses, 143-154 semantics, 174-176 sentence-free ziji, referring to speaker, 157 subcommanding antecedent, 142 subcommanding antecedents, 170-172 subject-orientation, 142 (syntactic) anaphor, 166-174 syntax, 176-179
SYNTAX AND SEMANTICS
Volume 1 edited by John P. Kimball Volume 2 edited by John P. Kimball Volume 3: Speech Acts edited by Peter Cole and Jerry L. Morgan Volume 4 edited by John P. Kimball Volume 5: Japanese Generative Grammar edited by Masayoshi Shibatani Volume 6: The Grammar of Causative Constructions edited by Masayoshi Shibatani Volume 7: Notes from the Linguistic Underground edited by James D. McCawley Volume 8: Grammatical Relations edited by Peter Cole and Jerrold M. Sadock Volume 9: Pragmatics edited by Peter Cole Volume 10: Selections from the Third Groningen Round Table edited by Frank Heny and Helmut S. Schnelle Volume 11: Presupposition edited by Choon-Kyu Oh and David S. Dineen Volume 12: Discourse and Syntax edited by Talmy Givon Volume 13: Current Approaches to Syntax edited by Edith A. Moravcsik and Jessica R. Wirth Volume 14: Tense and Aspect edited by Philip J. Tedeschi and Annie Zaenen
Volume 15: Studies in Transitivity edited by Paul J. Hopper and Sandra A. Thompson Volume 16: The Syntax of Native American Languages edited by Eung-Do Cook and Donna B. Gerdts Volume 17: Composite Predicates in English Ray Cattell Volume 18: Diachronic Syntax: The Kartvelian Case Alice C. Harris Volume 19: The Syntax of Pronominal Clitics edited by Hagit Borer Volume 20: Discontinuous Constituency edited by Geoffrey J. Huck and Almerindo E. Ojeda Volume 21: Thematic Relations edited by Wendy Wilkins Volume 22: Structure and Case Marking in Japanese Shigeru Miyagawa Volume 23: The Syntax of the Modern Celtic Languages edited by Randall Hendrick Volume 24: Modern Icelandic Syntax edited by Joan Maling and Annie Zaenen Volume 25: Perspectives on Phrase Structure: Heads and Licensing edited by Susan D. Rothstein Volume 26: Syntax and the Lexicon edited by Tim Stowell and Eric Wehrli Volume 27: The Syntactic Structure of Hungarian edited by Ferenc Kiefer and Katalin E. Kiss Volume 28: Small Clauses edited by Anna Cardinaletti and Maria Teresa Guasti Volume 29: The Limits of Syntax edited by Peter Culicover and Louise McNally Volume 30: Complex Predicates in Nonderivational Syntax edited by Erhard Hinrichs, Adreas Kathol, and Tsuneko Nakazawa Volume 31: Sentence Processing: A Crosslinguistic Perspective edited by Dieter Hillert Volume 32: The Nature and Function of Syntactic Categories edited by Robert D. Borsley Volume 33: Long-Distance Reflexives edited by Peter Cole, Gabriella Hermon, and C.-T. James Huang